|
Figure 1. Construction and development of a neurobot. (a) Neural precursor clumps were placed in the center of an animal cap “bowl”, excised from the animal pole of a Xenopus laevis embryo before it fully closed during healing. The composite forms gradually into first a sphere and then a more elongated shape, which is mobile by Day 3. (b) Top panels: examples of two neurobots, one more rounded than the other. Bottom panel: Roundness Index (RI) was calculated by fitting an ellipse on the image of the bot and calculating the ratio between the minor and major axes. Neurobots tended to be less rounded than biobots (Kruskal–Wallis test, p = 0.047). (c) Neurobots were significantly larger than biobots (Kruskal–Wallis test, p = 0.0007). The central line on the box plot shows the median, and the bottom and top edges of the box indicate the 25th and 75th percentiles, respectively. The whiskers show the extent of the extreme data points not considered outliers, and the outliers are shown using the ‘+’ symbol. |
|
Figure 2. Implanted neural precursors develop into neurons and extend their processes throughout the neurobot. (a,b) Z-projection of confocal image stack of a neurobot labeled with acetylated α-tubulin, which stains neurons and cilia of the multiciliated cells. (b) is the staining of the same neurobot as shown in (a) with fewer projected planes, rendering neural processes inside the bot visible. Color code corresponds to depth within the bot (confocal plane number). (c,d) Subregions of the same neurobot, showing neural processes projecting toward surface cells. Red shows the acetylated α-tubulin stain, and cyan depicts a nuclear (Hoechst) co-label (Nuc). Yellow arrows point to neural processes (Neu) or multiciliated cells whose cilia are stained (MCC). White arrows point to nuclear staining (Nuc). |
|
Figure 3. Immunostaining for acetylated α-tubulin, MAP2, and synapsin-1 reveals the presence of axons, dendrites, and synapses within neurobots. (a–c) Z-projection of 11 confocal sections (total depth: 22 µm). Acetylated α-tubulin (red) and MAP2 (green) are shown as: (a) merged image, (b) acetylated α-tubulin channel, and (c) MAP2 channel. Yellow arrowheads mark regions of overlap; white arrowheads indicate sites labeled only for acetylated α-tubulin; pink arrows denote regions labeled only for MAP2. Scale bar, 40 µm. (d–f) Distribution of putative synapses in a neurobot. Green puncta correspond to anti-synapsin-1 staining, red indicates acetylated α-tubulin, and blue denotes nuclei. (e) Higher magnification of the neurobot shown in (d). (f) Enlarged view of the boxed region in (e). Scale bars, 40 µm (e) and 10 µm (f). |
|
Figure 4. Calcium imaging in freely moving neurobots shows that the implanted cells are indeed active. (a) Average fluorescence of a freely moving neurobot containing neurons expressing GCaMP6s after motion correction (10 minutes of movement imaged at 5 frames per second). Colored circles correspond to regions of interest identified by the suite2p software, which could be single or multiple units. (b) Movement trajectory of the same neurobot. (c) The top curve shows the X-position of the neurobot over time. The 5 bottom curves show baseline-subtracted fluorescence activity of units labeled in panel (a). Arrowheads point to synchronized activity in some nearby (two shades of blue) and more distant (red and pink) ROIs. |
|
Figure 5. Neurobots show diverse and periodic patterns of spontaneous movement. (a) Exemplar trajectories of neurobots moving in an 8-well plate over a 30-minute trial. (b) Details of the trajectories of the same bots as shown in panel (b). The color gradient indicates time during the trial. |
|
Figure 6. The number of peaks in the power spectral density was used to quantify the complexity of movement trajectories. (a) Examples of simple (top panel) and complex (bottom) trajectories. (b) Time series of the movement amplitudes projected on the x-axis. (c) Power spectral densities corresponding to time series in panel (b) Red stars mark the location of significant peaks. |
|
Figure 7. Neurobots show differences in movement patterns compared to biobots. (a) Neurobots have more complex trajectories than biobots (Kruskal–Wallis test, p = 0.039), and (b) this complexity is not correlated with their roundness (Pearson correlation coefficient = −0.15, two tailed t-test t(45) = 45, p = 0.2), or (c) their area (Pearson correlation coefficient = −0.03, two tailed t-test t(df) = 45, p = 0.76). (d) Neurobots were more likely to be active (have non-zero minimum speed) than biobots (Kruskal–Wallis test, p = 0.037). |
|
Figure 8. PTZ differentially impacted the movement of neurobots and biobots. (a) Experimental protocol for testing the effect of PTZ on the movement of neurobots (n = 16) and biobots (n = 16). Movement of bots was measured in control media, across three transfers for 30 min. The bots were then transferred to a dish containing 15 mM PTZ solution, and their behavior was measured for another 30 min. (b) Relative complexity was defined as the complexity index measured while in PTZ, relative to the average complexity index measured for three consecutive controls. Although most biobots reduced their complexity index relative to control (cyan lines: decreased complexity, purple lines: increased complexity), the relative complexity index for neurobots was equally likely to increase or decrease (red and blue lines). The relative complexity of zero corresponds to the average CI for the three controls (black). Filled circles indicate neurobots cultured in zolmitriptan prior to testing. Neurobots showed significantly higher relative complexity compared to biobots (Kruskal–Wallis test, p = 0.01). (c,d) Show the results from control experiments where the PTZ treatment block was replaced with another control step. Trajectory complexity was equally likely to increase or decrease in the absence of PTZ. There was no significant difference between the relative complexity of biobots and neurobots (Kruskal–Wallis test, p = 0.01). |
|
Figure 9. Pair-wise correlations between size, shape, neural expression, and movement complexity in neurobots. (a,b) Examples of neurobots stained with acetylated α- tubulin which labels multiciliated cells and neurons. Overlaid white curves show the neural processes traced using Imaris software. Orange arrow heads point to exemplar multiciliated cells (a) Example of a neurobot with a high degree of innervation Nterminals = 327, LNeurite = 7635.9 mm. (b) Example of a neurobot with small degree of innervation Nterminals = 40, LNeurite = 753.8 mm. (c) Pairwise correlation between structural parameters of all stained neurobots and Complexity Index, Nterminals = total number of endings, LNeurite = total length of neurites, LNeuritenorm = total length of neurites normalized to area, NMCC = total number of multiciliated cells on the surface, NMCCnorm = NMCC normalized to area, RI = Roundness Index, Neu/Ect = ratio of the areas of neural implant to ectoderm shell. White and yellow arrows point to the position of manually traced neural processes and multiciliated cells. Pearson correlation coefficients are depicted for each pair. Values in red correspond to statistically significant correlations (two-tailed Student's t-test, P < 0.05). A robust linear regression was used for linear fits. The full model statistics are provided in Spreadsheet S1). |
|
Figure 10. Comparison of gene expression and its variability between neurobots, biobots, and sham neurobots. (a) Pearson correlation of gene expression within and across groups of neurobots (NB), sham neurobots (SH), and biobots (BB). The closer the value to 1, the more similar the expression patterns. (b) Principal component analysis of gene expression values. Each dot corresponds to one sample, which contained multiple bots of one kind. (c) Histograms of coefficients of variation (CV) in gene counts (FPKM) in neurobots (red), biobots (blue), and sham neurobots (yellow). Neurobots showed a significantly higher variability in their gene counts compared to both biobots and shams, and shams showed a higher variability compared to biobots (Kruskal–Wallis test with multiple comparisons using Tukey's honestly significant difference test; p < 0.00001). Genes with higher counts showed a higher degree of variability in their expression when comparing neurobots with biobots and shams (d,e). Dark blue bars mark the bins where the difference in CV was significantly different from what is expected if the ranking of genes were randomly shuffled. Similarly, genes with low levels of expression showed lower coefficient of variation than expected by chance. The p-value of each bin was defined as the proportion of the bin fractions from the distribution that were further in absolute value from the distribution mean than the true bin fraction. Bins with p-values of p < 0.05 were deemed statistically significant and were colored dark blue. All other bins were colored light blue. The mean bin value from the distributions was plotted as an orange line. (f) Same as (d,e) but comparing neurobots with shams. Bin values above 0.5 (light blue line) indicate that more than half of the CVs in that bin had higher values in that comparison. |
|
Figure 11. Distribution of differentially expressed genes between different bot groups. (a–c). The X-axis shows the fold change in gene expression between samples of different groups, and the Y-axis shows the statistical significance of the difference. Red dots represent genes that were significantly up (positive values on the X-axis) or downregulated (negative values on the X-axis); green dots represent genes with no significant change. Statistics were evaluated by DESeq2, which employs the two-sided Wald test. Multiple hypothesis testing corrections were used to obtain adjusted p-values. Red circles represent genes with adjusted p-values that were statistically significant (p < 0.05). |
|
Figure 12. Enrichment analysis performed using Gene Ontology annotations on differentially expressed genes with at least 4-log fold upregulation in expression. (a) neurobots versus biobots (b) neurobots versus shams, BP: Biological processes, CC: Cellular Components, MF: Molecular Function. |
|
Figure 13. Cluster-based network connectivity patterns among genes that were upregulated in neurobots compared to biobots. (a) Cluster 15 (b) Cluster 23 (c) Cluster 1. Network connectivity was calculated using the STRING online tool. The edges indicate both functional and physical protein associations the line thickness indicates the strength of data support. Only nodes with interaction scores with confidence higher than 0.4 are shown. Nodes of special interest are highlighted in color. |
|
Figure 14. Phylostratigraphic analysis of upregulated or downregulated transcripts in neurobots compared to biobots and shams. (a) 54% of upregulated genes in neurobots fall into the two categories of most ancient genes (“All living organisms” and “Eukaryota”). (b) Very few ancient genes are downregulated. In total 279 are downregulated in these two strata for the NB versus BB conditions and 233 for the NB versus SH condition, whereas 941 and 1109, genes are upregulated respectively. |
|
Figure S1. Comparison between physical and kinematic parameters between biobots and sham neurobots. There was no significant difference between roundness index (a), area (b), complexity index (c) and Minimum Speed (d) between biobots and sham neurobots (Kruskal- Wallis test, p=0.89, 0.943,0.11 and 0.35, respectively). |
|
Figure S2. Anti-synapsin-1 antibody labeling of putative synapses in neurobots. a. Confocal image from the center of a neurobot stained with anti-synapsin-1 primary antibody and Alexa Fluor 488–conjugated secondary antibody, showing punctate labeling consistent with putative synapses. b. Equivalent optical plane from a different neurobot stained with secondary antibody only, showing no detectable signal (secondary-only control). Scale bars, 40 µm. |
|
Figure S3. Examples of Z-projected confocal fluorescent images of neurobots. The first two columns show staining of acetylated alpha tubulin, which labels multiciliated cells and neurons (color code represents depth in the confocal stack). The last column shows the nuclear stain in the same neurobot shown on that row, with the same set of plane shown in the middle panel. The first column shows the Z-projection of the full stack, where as the next two columns show partial stacks to reveal the interior of the neurobot. All neurobots contain processes within the bot and those that extend towards the surface. They also contain a central region with seemingly no nuclei present, which we hypothesize might be filled with extracellular matrix materials. |
|
Figure S4. Structural composition of the neurobot cavity revealed by multiphoton imaging and immunostaining. Representative image of a collapsed z-stack (30 µm total thickness) of a neurobot showing: (a) nuclei stained with Hoechst 33342, (b) second harmonic generation (SHG) signal representing extracellular matrix collagen, (c) multi-ciliated cells and neurons labeled with acetylated α-tubulin, and (d) Alexa Fluor 647–conjugated phalloidin staining of filamentous actin in all cells. (e) Enlarged view showing all four channels overlaid. Collagen fibers and puncta-like structures (green arrows), as well as neuronal processes labeled by phalloidin alone or by both phalloidin and acetylated α-tubulin (blue and purple arrows, respectively), are visible within the cavity; however, most of the cavity appears free of nuclei, cell bodies, fibers, and collagen. Scale bars, 100 µm. |
|
Figure S5. Comparison of various kinematic parameters between biobots and neurobots. There was no significant difference in total distance traveled (a), average speed (b), average acceleration (c) and the percentage of the well area visited (d) between biobots (n=22) and neurobots (n=47). (Kruskal- Wallis test, p=0.5, 0.5,0.28 and 0.43, respectively) |
|
Figure S6. Neurobots had a significantly smaller density of multiciliated cells. Kruskal- Wallis test was used to obtain the p-value (p=0.027). |
|
Figure S7. Impact of treatment with zolmitriptan on neural expression patterns in neurobots. a-c Treatment with zolmitriptan increased number of terminals, total length of neurites and neurite density. There was no significant change in complexity index (Kruskal- Wallis test, p=0.05, 0.04,0.1 and 0.03, respectively) (d). e. Pairwise correlation between structural parameters of zolmitriptan-treated neurobots and Complexity Index, Nterminals = total number of endings, LNeurite=total length of neurites, LNeuritenorm=total length of neurites normalized to area, NMCC=total number of multiciliated cells on the top surface, NMCCnorm= NMCC normalized to area, RI= Roundness Index, Neu/Ect= ratio of the areas of neural implant to ectoderm shell. Pearson correlation coefficients are depicted for each pair. Values in red correspond to statistically significant correlations (two-tailed Student’s t-test, P<0.05). |
|
Figure S8. Enrichment analysis performed using Gene Ontology annotations a. sham neurobots vs biobots (4-fold upregulated pathways) and b. neurobots with biobots (4- fold downregulated pathways) |
