Primary cultures were prepared from P0/P1 rat pups (Sprague-Dawley) as described earlier (Goslin and Banker, 1989). The hippocampus was dissected out from the rodent brain in Hibernate Media (Thermofisher Scientific). The dissected hippocampi were trypsinized (0.25% Trypsin), washed 3×, and dissociated into single cells by syringing using needles of 2 pore sizes sequentially to remove all neurites (21 G and 25 G, BD Precision Glide Needle). The cells were plated on 18 mm diameter thickness corrected coverslips (Marienfeld Superior) coated with Poly-L-Lysine (1 mg/ml, Sigma Aldrich) after resuspending in Neurobasal media (Thermofisher Scientific) complemented with B27, Glutamate, and Normocin (Invivogen; referred to as Complete Neurobasal media) at a density of 50 cells/mm².

The cells were fixed at DIV1, DIV2, and DIV3 with 4% PFA+sucrose and labeled with antibodies against Map2 (Merck Millipore #AB5622, rabbit) and Tau (Merck Millipore #05–348, mouse). Alexa 488 and Alexa 647 secondary antibodies (Thermofisher Scientific) marked Map2 and Tau, respectively for dual color labeling.

For cholesterol sequestration, the cells were treated transiently with differing concentrations (2.5 mM, 5 mM, 7.5 mM, and 10 mM) of Methyl β cyclodextrin (MβCD, Sigma Aldrich C4555) for 10 min or 20 min at 37 deg at DIV1. The MβCD stock (100 mM) was reconstituted in water and the working concentrations were prepared in Complete Neurobasal media for treatment. After treatment, the cells were washed thrice in prewarmed Complete Neurobasal media and allowed to grow to DIV3 before fixing them. For cholesterol replenishment, the cells treated with 5 mM MβCD for 10 min were washed and further treated for 10 min with differing concentrations of exogenous cholesterol (Sigma Aldrich C4951, Cl:MβCD, 0.5 mM, 1.0 mM, 1.5 mM, and 2.0 mM) at 37 deg. The Cholesterol stock (100 mM) was reconstituted in water and the working concentrations were prepared in Complete Neurobasal media for treatment. The cells were again washed thrice in prewarmed Complete Neurobasal media and allowed to grow to DIV3 before fixing them. The treated cells were labeled with antibodies against Map2 and Tau marked by Alexa 488 and Alexa 647, respectively. The coverslips were mounted using Prolong antifade agent (Thermofisher Scientific).

For measuring unesterified cholesterol, the control and cells treated with varying concentrations of MβCD (2.5 mM, 5 mM, 7.5 mM, and 10 mM) for 10 min and 20 min were labeled at DIV3 using a membrane marker for lipid rafts tagged with Alexa 555 (Thermofisher Scientific V34404). Membrane labeling was done in Complete Neurobasal media for 10 min at 4 deg. The cells were then washed, fixed with 4% PFA+sucrose and labeled with Filipin III (Sigma Aldrich F4767) at a concentration of 50 μg/ml in PBS for 2 h at RT and overnight at 4 deg, after which they were washed and imaged in PBS.

The membrane damage to the cells induced on cholesterol sequestering for varying durations was assessed using a Live/Dead Cell imaging kit (Thermofisher Scientific R37601) which uses Calcein AM cell permeable dye and Bobo-3 Iodide as live and dead cell indicators, respectively. The live control and treated cells (10 mM MβCD, 10 min, 20 min, and 30 min) were labeled at DIV3 for 15 min at RT and imaged in an extracellular imaging buffer. For quantifying the viability of cells undergoing cholesterol sequestering, the control and cells treated for varying durations (10 mM MβCD, 10 min, 20 min, and 30 min) were fixed at DIV3 and labeled with an antibody (Cleaved Caspase 3, rabbit, #9661, Cell Signaling Technology) which recognize endogenous levels of activated Caspase 3 marked by Alexa 555. The coverslips were mounted using Prolong antifade agent containing DAPI (Thermofisher Scientific).

Images were acquired an inverted microscope Zeiss Axio Observer Z1 equipped with a 40× objective, halogen lamp and an ScMOS camera (pco.edge). Specific filters for Alexa 488 and Alexa 647 were used for acquisition. 3D stacks of total volume thickness of 3 μm with a plane thickness of 200 nm were acquired. The images were exported, and all processing and image analyses were performed using Metamorph 7.0. Maximum intensity projections of the images were generated. For analysis of neuronal development profile, length of the longest neurite, length of dendrites, number of neurites per cell and the neuronal polarity ratio were quantified. For specific sets of experiments, additional analysis included calculation of the percentage of cells with single neurite, percentage of neurons with multiple axons, and polarity ratio of neurons displaying multiple axons, as explained. Due to the minimal variability between neuronal cultures, Mean ± SEM is presented for all data. The statistical difference between the means was calculated by One-way Anova (Graph Pad Prism 9). ****p < 0.0001, ***p < 0.001, **p < 0.01, *p < 0.05 and ns non-significant.

Imaging of cells labeled with Filipin III was performed similarly using a 40× objective. Due to the high photobleaching of Filipin III, cells were focused using the lipid raft membrane marker tagged with Alexa 555 and 3D stacks of total volume thickness of 3 μm with a plane thickness of 200 nm were acquired for both channels. DsRed filter was used for Alexa 555 acquisition, whereas Filipin fluorescence was acquired using a filter combination 420–480 nm excitation and 500–550 nm emission. Quantification of Filipin fluorescence was done from maximum intensity projections of 3D stacks. The images were thresholded for automatic detection of light objects to form a mask. The regions mapping the cells were transferred to the original images to quantify the corresponding average intensity of the cells.

For quantifying the viability of cells undergoing cholesterol sequestering, images of control and treated cells were acquired using a 20× objective. For the live cell assay, images were acquired for green and orange channels using GFP and DsRed filters, respectively. The images were thresholded and assessed by the Count nuclei App in Metamorph software. The ratio of orange (dead) to green (live) cells in each field was quantified and normalized using the mean value of control cells. For the caspase 3 assay, images were acquired using DAPI and DsRed filters. The percentage of DAPI marked cells colabeled with Caspase 3 (positive, dead cells) vs. DAPI only cells (negative, live cells) was quantified for each field by a Cell Scoring App in Metamorph software. The ratio of the percentage of dead vs. live cells was assessed and normalized using the mean value of control cells.

The statistical difference between the means of each condition (fluorescence intensity/ratio) was calculated by One-way Anova (Graph Pad Prism 9). ****p < 0.0001, ***p < 0.001, **p < 0.01, *p < 0.05 and ns non-significant.

To understand the role of cholesterol on neuronal development, it was critical to assess a homogeneous neuronal population at different stages of development. Cell culture protocols were optimized to prepare homogenous primary hippocampal neuronal cultures from postnatal day 0 or 1 (P0/P1) rat pups. The morphology of pyramidal neurons in these cultures was assessed at DIV1, DIV2, and DIV3 (Figure 1). The homogeneity of cultures was verified using Tau (axonal marker) and Map2 (somato-dendritic marker) antibodies, which showed >90% of the plated neurons to be undifferentiated at DIV1 vs. >90% differentiated at DIV3. A gallery of low magnification images of cells at each developmental stage is presented in Supplementary Figure 1. 3D image stacks of total volume thickness of 3 μm of Tau and Map2 labeled neurons at different stages were acquired at higher magnification (Figure 1). All quantifications were performed on the maximum intensity projections of the stacks with an overlay of both channels. The experimental and analytical procedures are outlined in Supplementary Figure 2.

Neuronal development and growth profile were analyzed by quantifying the length of the longest neurite (axon), length of the dendrites (all neurites except the longest) and the number of neurites per cell. This classification of neurites into axon and dendrites was entirely based on their length and not their specific labeling. Neuronal polarity was assessed based on the ratio of the length of the longest neurite (axon) to the average length of the dendrites, which is referred to as the neuronal polarity ratio. The higher values of the neuronal polarity ratio indicated the asymmetric pattern of the neurons with a single long axon and short dendrites, while small values of polarity ratio showed a symmetric growth pattern, where the neurites were similar in length. The conventional semi-automated software for neuro-morphology assessment including Simple Neurite Tracer (SNT), Neural Circuit Tracer (NCT), AutoNeuriteJ, etc. are limited by various factors such as lack of sensitivity to detect fine neuronal processes for precise characterization, exclusion of overlapping cells, the requirement of additional labeling and lack of distinction of axo-dendritic processes (Chothani et al., 2011; Longair et al., 2011; Boulan et al., 2020). Therefore, manual tracing of neurites was performed from the periphery of the soma to the distal end of the neurites. All projections and bifurcations including the secondary and tertiary processes were considered and added to the length of the primary process. Thresholding of neurite length was performed where a cut-off length of 10 μm was introduced to exclude filopodia. The quantification of all parameters including the length of the longest neurite, length of dendrites, number of neurites per cell, and polarity ratio was done using the filtered values (Supplementary Figure 2B).

A gallery of rat hippocampal neurons of different stages namely, DIV1, DIV2, and DIV3 is presented (Figure 1A). A steep increase in the length of the longest neurite (referred to as the axon) was observed from DIV1 (51.73 ± 1.86 μm) to DIV3 (328.2 ± 14.53 μm), with DIV2 (212.7 ± 8.38 μm) being intermediate (Figure 1B, Table 1). The dendrites (rest of the neurites except the longest) also displayed a gradual increase in their length along development, though not as fast as the axon (Figure 1C, Table 1). The length of the dendrites increased from 21.27 ± 0.3 μm at DIV1 to 41.42 ± 1.35 μm at DIV3. The number of neurites per cell showed a slight increase with the maturity of neurons from 6 ± 0.2 at DIV1 to 8 ± 0.3 at DIV 2 and DIV3 (Figure 1D, Table 1). Neuronal development involves both growth of the neuronal processes as well as an axo-dendritic specification or neuronal polarity establishment. Different technical challenges including lack of reliable markers at the early developmental stages made it hard to determine whether the specification between the two neuronal processes has been established. Therefore, the neuronal polarity ratio or the ratio of the length of the longest neurite (axon) to the average length of the dendrites was assessed, which showed the polarity status of the neuron. The polarity ratio displayed a drastic increase from 2.47 ± 0.09 at DIV1 to 10.72 ± 0.75 at DIV3 (Figure 1E, Table 1), confirming this to be a reliable parameter to assess the switch of a neuron from a symmetric to asymmetric growth pattern. Along with the qualitative assessment of polarity establishment using Tau and Map2 as axonal and dendritic markers, quantifying the polarity ratio allowed us to make a reliable and additional assessment on the multiple morphological phenotypes observed by membrane lipid perturbations.

To account for the variability between cultures, all quantifications were done across the same cultures at different stages. The minimal error observed for the different parameters across the cultures showed the robustness of the data and purity of the cultures suggesting its homogenous nature, enabling us to quantify the reported parameters in a reliable manner. A significant increase in the length and number of neurites as well as in the percentage of polarized cells was observed during rat hippocampal neuronal development from DIV1 to DIV3.

Having a homogenous neuronal population, which could be developmentally assessed, allowed us to investigate the role of membrane cholesterol in two important aspects of neuronal development, in neurite outgrowth and polarity establishment. In contrast to the majority of studies which employed polarity established cells, the experiments described here were conducted at very early stages of development before the axodendritic specification was established (stages 1–3). Also, instead of chronic deprivation of cholesterol from the membrane surface, we adopted a minimally invasive approach where cholesterol was transiently sequestered for a few minutes. The treated cells were then washed and allowed to grow for 2 more days, after which they were fixed and immunolabeled with Tau and Map2 (Supplementary Figure 2Ai). This experimental paradigm allowed us to directly address the effects of a transient membrane lipid alteration on neuronal morphology and development.

Neurons at DIV1 were transiently treated with the cholesterol sequestering drug, Methyl β cyclodextrin (MβCD), at various concentrations (2.5 mM, 5 mM, 7.5 mM, and 10 mM) for 10 min and analyzed at DIV3 (Figure 2). Control cells were treated similarly with the highest amount of carrier alone (water). A gallery of cells including the control and those treated with varying concentrations of MβCD is presented (Figure 2A). We observed that transient cholesterol perturbation exerted effects on both neurite outgrowth and neuronal polarity establishment. This also revealed two distinct populations of neurons with different phenotypes on neuronal morphology.

The length of the longest neurite (referred to as the axon) was significantly attenuated with the increasing concentration of MβCD. The length of the axon decreased from 307.2 ± 13.12 μm in control neurons to 158.7 ± 5.49 μm when MβCD concentration was increased to 10 mM (Figure 2B, Table 2). Interestingly, cholesterol perturbation by MβCD increased the length of the dendrites (Figure 2C, Table 2). In contrast to the control neurons where the average length of dendrites was 37.04 ± 1.13 μm, the dendritic length increased to 62.38 ± 4.39 μm upon augmenting the MβCD concentration to 10 mM. A stark effect on neuronal arborization was observed on transient cholesterol sequestration by MβCD. The neurite number per cell reduced from 6 ± 0.3 to 2 ± 0.1 with increasing MβCD concentration (Figure 2D, Table 2). The strong effect of cholesterol sequestering on neuronal arborization resulted in a second phenotype, i.e., the population of cells with a single neurite (Figure 3A). The percentage of cells with a single neurite increased drastically with cholesterol perturbation (Figure 3B). In control cultures, the percentage of neurons with a single neurite was negligible (2 ± 2%). In striking contrast, the MβCD treated cultures exhibited a significant increase in neurons with a single neurite from 26 ± 7% in 2.5 mM MβCD to 55 ± 10% in 10 mM MβCD (Figure 3B, Table 2).

Quantification of all growth parameters was done from the same cultures to minimize variability between cultures. The minimal error in data obtained from these cultures confirmed the robustness of the data. These results showed distinct effects of transient cholesterol perturbation on the growth of axon vs. dendrites, significantly affecting both neurite outgrowth and neuronal arborization.

We next asked whether the membrane cholesterol perturbation affected neurite outgrowth alone or whether neuronal polarity establishment was also affected. The neuronal polarity ratio displayed a significant change from 11.79 ± 0.73 in control down to 5.97 ± 0.68 with cholesterol sequestering by MβCD (Figure 2E, Table 2), confirming that neuronal polarity was affected as well. As an independent assessment to dissect out the effect of cholesterol sequestration on neuronal polarity, we performed a qualitative comparison of cells immunolabeled for the axonal marker Tau and classified cells based on three criteria: (1) showing no polarity (undifferentiated), (2) polarized (with a single axon), and (3) polarized (with multiple or supernumerary axons).

Using Tau to label axons can be challenging at early developmental stages since Tau labels all neurites initially and then gets restricted to axons in mature neurons (Supplementary Figure 1). Since the developmental stage of interest in this study was very early when polarity is just getting established, we assessed the percentage of control neurons displaying multiple neurites with Tau labeling or with supernumerary axons (Table 2, Figures 3C,E). We observed that the percentage of cells displaying supernumerary axons significantly rose from 26 ± 5% to 60 ± 3% with increasing MβCD concentration (Table 2, Figure 3C). The multiple axons observed in cholesterol perturbed cells exhibited very strong and comparable levels of Tau labeling among them, whereas the majority of control cells displayed only a single neurite that was strongly labeled for Tau. For a better understanding of the effect of cholesterol sequestering on neuronal polarity, we compared the neuronal polarity ratio of the control cells and the MβCD treated cells displaying supernumerary axons. We observed that the polarity ratio of the MβCD treated cells with multiple axons significantly decreased (2.24 ± 0.29) compared to control cells displaying the same (4.51 ± 0.44, Table 2). Low values of the polarity ratio showed that the neurons exhibited a symmetric morphology following cholesterol sequestration compared to an asymmetric growth in control cells, which had a much longer axon compared to the dendrites. Therefore, in addition to a reduction in the length of the axon with increasing membrane cholesterol perturbation, there was also an increase in the percentage of neurons reversing to symmetric growth and displaying supernumerary axons. This unique population of cells with multiple axons would also explain the distinct effects observed previously, where there was an increase in the length of all the neurites except the axon upon cholesterol sequestration (Figure 2C). Therefore, the shift in polarity and an elevated population of neurons displaying supernumerary axons would largely account for the increase in length of dendrites observed with higher cholesterol sequestering.

Since distinct effects of cholesterol perturbation were observed on neuronal growth and polarity, we next addressed whether these effects were dependent on the duration of perturbation. For this, rat hippocampal neurons at DIV1 were treated with varying concentrations of MβCD, but the duration of cholesterol sequestering was doubled to 20 min (Figure 4). The cells were washed and allowed to grow to DIV3 when they were immunolabeled with Tau and Map2. All quantifications were done as described previously for the 10 min perturbation. A gallery of cells treated with differing concentrations of MβCD from 2.5 mM, 5 mM, 7.5 mM to 10 mM for 20 min is presented along with control neurons treated with the carrier alone (Figure 4A). Similar to that reported for the 10 min treatment, the growth of the axon was strongly attenuated for the 20 min treatment as well, with increasing MβCD concentrations (Figure 4B). The length of the axon significantly reduced in MβCD treated neurons (193.7 ± 8.63 μm) when compared to the control cells (344 ± 17.74 μm; Figure 4B, Table 3). Membrane cholesterol perturbation for 20 min caused an increase in the dendritic length, as observed in 10 mM MβCD treated cells (75.18 ± 8.36 μm), compared to that in the control condition (32.08 ± 1.21 μm; Figure 4C, Table 3). Neuronal arborization was also severely affected with a longer duration of cholesterol perturbation, wherein the number of neurites per cell reduced from 7 ± 0.4 to 2 ± 0.1 at higher MβCD concentrations (Figure 4D, Table 3). Similar to that observed with 10 min treatment, cholesterol perturbation for 20 min also resulted in an increase in the population of neurons with a single neurite (6 ± 6% in control to 64 ± 9% in treated cells; Figure 4F, Table 3). Neurons also displayed a significant decrease in their polarity ratio on longer cholesterol sequestering (14.32 ± 1.13 in control cells; 5.49 ± 1.12 in treated cells; Figure 4E, Table 3). Cholesterol sequestering for 20 min also resulted in an increase in the number of cells displaying supernumerary axons (26 ± 9% in control; 66 ± 17% in treated cells; Figure 4G, Table 3). Together these observations suggest that the neuronal developmental defects observed upon cholesterol perturbation were not affected by the duration of perturbation. The effects observed after 10 min and 20 min treatment were similar and followed the same pattern. However, in contrast to membrane cholesterol perturbation for 10 min duration, which resulted in a gradual transition of morphological changes between the different concentrations, the effects observed with 20 min perturbation were abrupt and stronger.

Since we observed acute neuro-morphological and polarity defects in response to varying levels of cholesterol sequestering, we quantified membrane cholesterol in parallel by an unesterified cholesterol assay by labeling with Filipin III after treatment with different MβCD concentrations (Figure 5). Filipin III is known to be highly photo-unstable; therefore, a lipid raft membrane marker which labels overlapping regions was used as a reference for focusing the cells. We observed that Filipin fluorescence decreased 2.5 to 3 times from the control condition in response to increased sequestering of membrane cholesterol (Figures 5A,B, Table 4). The results were consistent with the dose response curve of Figure 2. We also extended the study to Filipin labeling of cells which had undergone a longer duration of cholesterol sequestering such as 20 min. We observed a similar pattern for alteration in Filipin fluorescence, but a stronger response with a longer duration of perturbation, consistent with the dose response curve of Figure 3 (Figure 5C, Table 4). We also observed a higher membrane damage for the cells undergoing longer cholesterol perturbation (20 min) compared to the short duration (10 min), accounting for the higher variability in Filipin fluorescence in these cells (Table 4).

Therefore, we assessed whether the viability was compromised for cells undergoing cholesterol sequestering for longer durations. Cholesterol perturbation was performed at a high concentration (10 mM MβCD) for varying durations namely, 10 min, 20 min, and 30 min (Supplementary Figure 3). The viability assay was performed by two methods. In the first, the membrane damage to the cells induced upon transient cholesterol sequestering at DIV1 for varying durations was assessed at DIV3 in live cells. The control and cholesterol sequestered cells were labeled using a Live/Dead cell imaging kit which marked Calcein AM (green) and Bobo-3 Iodide (orange) as live and dead cell indicators, respectively (Supplementary Figure 3A). The normalized ratio of dead to live cells showed significantly higher levels for 20 min and 30 min compared to 10 min, whose levels were similar to control values (Supplementary Figure 3B, Supplementary Table 1). The assay verified a higher population of cells with membrane integrity compromised with a longer duration of cholesterol perturbation. In the second method, the control and treated cells at DIV3 were fixed and labeled with an antibody recognizing endogenous levels of activated Caspase 3 (orange), and counter labeled with DAPI (Blue, Supplementary Figure 3C). Similar to the live cell assay, the normalized ratio of the percentage of dead vs. live cells showed similar levels between control and 10 min perturbation, in contrast to a significantly higher population of cells entering apoptosis after 20 min and 30 min cholesterol sequestering (Supplementary Figure 3D, Supplementary Table 1). Since our results showed minimal membrane damage and viability issues with short cholesterol sequestering similar to control values, subsequent rescue experiments were performed with the short cholesterol perturbation of 10 min.

Cholesterol sequestering affected axonal and dendritic neurite outgrowth as well as neuronal polarity in a distinct manner. Therefore, we asked whether these developmental defects can be reversed by transient supplementation of membrane cholesterol. This would also address whether the effects we observed upon cholesterol sequestering were specific and direct rather than secondary, since MβCD is also known to affect the cytoskeleton. For this purpose, we treated the hippocampal neuronal cultures at DIV1 with 5 mM MβCD for 10 min, which showed a significant yet moderate effect on neuronal development (Figure 2). The treated cells were washed and immediately supplemented with exogenous cholesterol (Cl) at varying concentrations (0.5 mM, 1.0 mM, 1.5 mM, and 2.0 mM) for 10 min. The cells were washed and grown till DIV3, when they were fixed, immunolabeled with Tau and Map2 and imaged (Supplementary Figure 2Aii). We found that brief cholesterol supplementation of the MβCD treated cells significantly rescued many of the severe developmental defects generated by cholesterol sequestering (Figures 6 and 7).

A gallery of cells including control, MβCD treated (5 mM), and cholesterol supplemented cells is presented (Figures 6A and 7A). On transient cholesterol application, the length of the longest neurite increased from 149.9 ± 7.74 μm to 275.2 ± 14.78 μm, depending on the concentration of the exogenous cholesterol supplemented (Figure 6B, Table 5). Higher cholesterol concentration resulted in retarded axonal length, emphasizing the critical window of the membrane cholesterol content in supporting growth. Cholesterol replenishment, in general, reduced the average length of dendrites of cholesterol sequestered cells to control values (Figure 6C, Table 5). The number of neurites per cell also significantly increased from 2 ± 0.1 in MβCD treated cells to 7 ± 0.2 in cholesterol supplemented cells (Figure 6D, Table 5). The extent of rescue was critically dependent on the amount of replenished cholesterol. Consistently, we observed that the percentage of cells with a single neurite disappeared completely (0 ± 0) or to negligible values from a significant population of 27 ± 4% in MβCD treated cells upon brief cholesterol replenishment (Figure 7B, Table 5). On the contrary, the neuronal polarity ratio displayed a partial rescue upon replenishment (8.25 ± 0.61) when compared to cholesterol sequestered cells (6.22 ± 0.65; Figure 6E, Table 5). Similar to neurite outgrowth, the rescue of polarity was also critically dependent on the replenished cholesterol and behaved similar to cholesterol sequestered cells at high cholesterol concentrations (Figure 6E, Table 5). This strongly supports the critical nature of membrane cholesterol concentration on neuronal polarity. Consistently, we observed that the percentage of cells with supernumerary axons reduced drastically for a small window of replenished cholesterol concentration, emphasizing the critical role of fine membrane cholesterol balance in axo-dendritic specification and neuronal polarity establishment (Figure 7C, Table 5). The neuronal polarity ratio of the cholesterol replenished cells with multiple axons (4.51 ± 1.27) was higher than that of cholesterol sequestered cells (2.29 ± 0.29) and similar to that of control cells (4.52 ± 0.44), thereby verifying that the asymmetric nature of neurite growth had returned on retaining the membrane cholesterol balance (Figure 7D, Table 5). The results confirmed the critical nature of membrane cholesterol content in neurite outgrowth and neuronal polarity establishment.

Neuronal developmental defects induced by cholesterol sequestering were rescued by the transient application of exogenous cholesterol. Therefore, we wanted to address the effect of increasing the membrane cholesterol concentration on neuronal development. The rat hippocampal neurons at DIV1 were transiently treated with varying concentrations of exogenous cholesterol (0.5 mM, 1.0 mM, 1.5 mM, and 2.0 mM) for 10 min (Supplementary Figure 2Aiii). The cells were washed and grown till DIV3, when they were fixed, immunolabeled with axonal and dendritic markers, and imaged. A gallery of cells treated transiently with varying concentrations of cholesterol is presented in Figure 8A. We observed that the normal neuronal morphology was retained for a short concentration window of supplemented cholesterol and was lost at high cholesterol concentrations, emphasizing the relevance of the correct membrane lipid balance for the normal neuronal architecture. Cholesterol imbalance altered the shape of the soma and resulted in axons growing and curling back towards the soma instead of leading away from it, increasing the probability of generating ectopic synapses (Figure 8A). While supplementing extrinsic cholesterol rescued neurite outgrowth of cholesterol perturbed cells, cholesterol addition to the normal neurons resulted in attenuated growth. This attenuation was significant and augmented with increasing cholesterol concentration. The length of the longest neurite reduced from 361.9 ± 24.58 μm in control cells to 177.1 ± 9.06 μm in cells treated with higher amounts of cholesterol (Figure 8B, Table 6). In contrast to cholesterol sequestering which affected axon and dendrites in an opposing manner, cholesterol enhancement did not significantly alter the growth of dendrites (Figure 8C, Table 6). However, cholesterol enhancement negatively affected neurite outgrowth and neuronal arborization and reduced the average number of neurites per cell from 9 ± 0.5 in control cells to 6 ± 0.3 in treated cells (Figure 8D, Table 6). Contrary to cholesterol sequestering, cholesterol enhancement did not generate a significant population of neurons with a single neurite (Figure 8F, Table 6). The most significant effect of cholesterol enhancement was observed for the neuronal polarity ratio which sharply decreased from 12.94 ± 1.18 to 6.18 ± 0.5 upon increasing the membrane cholesterol concentration (Figure 8E, Table 6). The decrease in polarity ratio indicated the reversal of neurons to symmetric growth following cholesterol enhancement. Interestingly, contrary to cholesterol sequestering, cholesterol enhancement did not significantly shift polarity, though it altered neurite outgrowth and extension. Consistently, the percentage of cells displaying supernumerary axons was not significantly altered upon cholesterol enhancement, when compared to control neurons (Figure 8G, Table 6). These results showed that membrane cholesterol enhancement altered neurite outgrowth but not polarity. Together these observations support the critical nature of membrane cholesterol balance in regulating neurite outgrowth and polarity, and in maintaining the normal neuronal morphology and architecture.