Unraveling Axonal Transcriptional Landscapes: Insights from iPSC-Derived Cortical Neurons and Implications for Motor Neuron Degeneration

This article has been Reviewed by the following groups

Read the full article See related articles

Discuss this preprint

Start a discussion What are Sciety discussions?

Listed in

Log in to save this article

Abstract

Neuronal function and pathology are deeply influenced by the distinct molecular profiles of the axon and soma. Traditional studies have often overlooked these differences due to the technical challenges of compartment specific analysis. In this study, we employ a robust RNA-sequencing (RNA-seq) approach, using microfluidic devices, to generate high-quality axonal transcriptomes from iPSC-derived cortical neurons (CNs). We achieve high specificity of axonal fractions, ensuring sample purity without contamination. Comparative analysis revealed a unique and specific transcriptional landscape in axonal compartments, characterized by diverse transcript types, including protein-coding mRNAs, RNAs encoding ribosomal proteins (RPs), mitochondrial-encoded RNAs, and long non-coding RNAs (lncRNAs). Previous works have reported the existence of transcription factors (TFs) in the axon. Here, we detect a set of TFs specific to the axon and indicative of their active participation in transcriptional regulation. To investigate transcripts and pathways essential for central motor neuron (MN) degeneration and maintenance we analyzed KIF1C-knockout (KO) CNs, modeling hereditary spastic paraplegia (HSP), a disorder associated with prominent length-dependent degeneration of central MN axons. We found that several key factors crucial for survival and health were absent in KIF1C-KO axons, highlighting a possible role of these also in other neurodegenerative diseases. Taken together, this study underscores the utility of microfluidic devices in studying compartment-specific transcriptomics in human neuronal models and reveals complex molecular dynamics of axonal biology. The impact of KIF1C on the axonal transcriptome not only deepens our understanding of MN diseases but also presents a promising avenue for exploration of compartment specific disease mechanisms.

Article activity feed

  1. Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

    Learn more at Review Commons


    Reply to the reviewers

    Reviewer #1

    Evidence, reproducibility and clarity

    Summary:

    This manuscript by Xu, Hörner, Schüle and colleagues is an RNA-seq study focusing characterization of axonal transcriptomes from human iPSC-derived cortical neurons. The authors have differentiated iPSC into neurons, cultured them in microfluidic devices and isolated axonal RNA, comparing this to corresponding cell soma transcriptomes. Second, axonal transcriptomes are compared between wild type and Kif1c knockout axons to determine Kif1c-dependently localized transcripts. Characterization of the latter allows the authors to suggest differentially expressed transcripts in Kif1c-KO axons can be mRNAs relevant for motor neuron degeneration owing to Kif1c mutations in hereditary spastic paraplegias.

    Major comments: Overall, his manuscript reads like work in (early) progress. This manuscript provides an interesting dataset, but needs substantial additional experimental and/or bioinformatic work to merit publication. The technical complexity of steps that have led to obtaining axonal transcriptomes can be appreciated, the soundness of generating these data is beyond doubt. However, the study stops at the point of generating axonal transcriptomes from wild type and Kif1c axons. No follow-up experiments are performed to study genes of interest found in RNA-seq. This could be compensated by in-depth bioinformatic analysis (e.g. comparisons with the many different datasets in known in the field), but this is clearly lacking as well. The results section only contains minimal bioinformatic analysis and nothing else. Introduction and discussion are well, clearly written and are in good dialogue with the existing body of work. To improve the manuscript, at minimum these two aspects should be addressed:

    1. Characterization of the iPSC-derived neurons is missing (immunostaining with neuronal markers, e.g. Tau, MAP2, exclusion of glial markers, and lack of stem cell markers)
    2. Validation of candidates of interest (e.g. FISH analysis in axons vs somata, Kif1c vs wt). Very specific requests from the review are useless at this point, as the authors should have the liberty to focus.

    Thank you for the review of our manuscript. We appreciate your recognition of the technical complexity involved in generating axonal transcriptomes and the clarity of our introduction and discussion sections.

    __Characterization of iPSC-derived neurons: __We acknowledge the importance of immunostaining with neuronal markers to ensure the purity of our neuronal population. We included this characterization in our revised manuscript and added it into the results and methods section of the paper (Supplementary Figure S1). Additionally, we included RT-qPCR analysis that confirmed the presence of cortical markers and added these to the results and method section of the paper (Supplementary Figure S2).

    Additional bioinformatic work: We agree that additional bioinformatic work will greatly benefit this paper. Therefore, we compared our datasets to all additional datasets that we were able to retrieve. This was added to the main text (results and discussion) and supplementary material (Supplementary Figure S5 and S6). We believe this strengthens the merit of our paper, and adds a lot of new unpublished information to the manuscript

    __Validation of candidates of interest: __We understand the necessity of validating our RNA-seq findings through experimental approaches such as FISH analysis and comparisons between KIF1C knockout and wild-type neurons. While we appreciate the comment and agree on the importance of high-resolution RNA FISH, we believe it is beyond the scope of this manuscript due to the considerable complexity of these experiments in human iPSC-derived cortical neurons. We will focus on incorporating this aspect into future studies and added a corresponding statement outlining the limitations of our study in the discussion stressing the importance of this.

    Minor comments:

    1. Details of RNA seq technicalities are redundant in the results section, e.g. „Our RNA-seq pipeline encompassed read quality control (QC), RNA-seq mapping, and gene quantification" (p. 7) is a trivial description - this and similar details should be skipped or described in methods.

    We will ensure that technical details are appropriately placed in the methods section and avoid redundancy in the results. Technical details included in the results section have been moved to the methods.

    1. Fig1A: Y axis should start from 0

    We adjusted Figure 1A to start the Y-axis from 0.

    1. Too much interpretational voice in figure legends (e.g. see Fig. 1, „PC1 clearly distinguishes the soma (blue)"

    We revised the interpretational voice in the figure legends to maintain objectivity.

    1. PCA analysis seems redundant in Fig. 2C

    We removed the PCA analysis in Fig. 2A (2C corresponds to Gene ontology term enrichment analysis).

    1. Subheading „Human motor axons show a unique transcription factor profile" is misleading - you are not dealing with motor iPS-derived motoneurons (Isl-1 positive), but cortical neurons (again, no marker information provided to assess this!)

    *The subheading **„Human motor axons show a unique transcription factor profile" *was adjusted. Furthermore, validation of neuronal identity has been added to the supplementary figures (Supplementary Figure S1 and S2), as well as main text and methods section.

    1. Fig. 3: Just by comparing top expressed factors in axonal samples is not informative - overall high expression of a certain transcript likely makes it easier for it to be picked up in the axonal compartment. Axon/soma ratios would perhaps be more appropriate.

    After careful consideration, we decided that we will not change the data presentation in Figure 3. Our aim in this figure was not to compare axon and soma but to see highly expressed transcripts in the axon, regardless of whether they are highly expressed in the soma as well. We think that looking at transcripts present in the axon can give information about axonal function, that we might lose when we only consider transcripts that are upregulated compared to the soma. The fact that 25 out of 50 transcription factor RNAs detected in the axon are actually specific to the axons supports this point of view. The comparison between transcripts expressed in axon and soma are presented in Figure 2.

    1. Figure 4 (KIF1C modulates the axonal transcriptome): you should show also data for the same genes in the soma, axonal data only is misleading (is overall expression changed?)

    We appreciate your suggestion. This data was already included in Supplementary Figure S6 (now Supplementary Figure S9). To make this easier to find, we've added a section to the results part to more clearly state how transcript expression changes in the soma.

    Significance

    Axonal transcriptomes have been studied since early 2010s by a number of groups and several datasets exist from different model systems. The authors know these studies well, address their findings and cite them appropriately. Is the dataset in this manuscript novel? Does it contribute to the field? Several axonal transcriptomes have been characterized in thorough studies, and even in the specific niche (human IPS-derived motoneurons) a point of reference exists - as the authors themselves point out, it is the Nijssen 2018 study. With appropriate presentation and follow-up experiments this material could have merit as a replication study.

    Audience: specialized

    We appreciate the reviewer's suggestion to clarify the differences between our findings and previously published data. In response, we have added a dedicated section to the discussion, where we provide a more detailed comparison of our results with existing research. This includes an in-depth examination of the methodologies, experimental conditions, and biological contexts that may explain the observed discrepancies (e.g., variations in methods, neuronal types, and disease contexts). As prior studies primarily focused on mouse-derived neurons, we have included a new section in both the results (Supplementary Figure S6) and the discussion to highlight the limited overlap in gene expression between the axons of mouse- and human-derived neurons. Furthermore, previous studies on human-derived cells either investigated i3 neurons -induced by transcription factors but not fully representative of human-derived CNS-resident neurons - or neurons of the peripheral nervous system (lower motor neurons). In contrast, our study focuses on human-derived CNS-resident cortical neurons (Supplementary Figure S1, S2; comparison shown in Supplementary Figure S5), emphasizing the greater translatability of our findings.

    Moreover, we have expanded our bioinformatic analyses and compared our dataset with additional datasets to further substantiate our conclusions (Supplementary Figure S5, S6)

    We believe that these revisions significantly enhance the clarity, quality, and impact of our manuscript. We sincerely thank the reviewer for their constructive feedback.

    Reviewer #2

    Evidence, reproducibility and clarity

    This study seeks to identify axonal transcriptome by RNA-sequencing of the iPSC-derived cortical neuron axons. This is achieved by comparing the RNA expressions between the axonal and soma compartments using microfluid system. The specific expression of axon specific RNAs in the axonal compartment validate the specificity of the approach. Some unique RNAs including TF specific RNAs are identified. Furthermore, this study compared the KIF1C-knockout neurons (which models hereditary spastic paraplegia characterized by axonal degeneration) with wildtype (WT) control neurons, which led to the identification of specific down-regulated RNAs involved in axonal development and guidance, neurotransmission, and synaptic formation.

    The data of this study are interesting and clearly presented. The major concerns are the lack of characterization of the neuron identities and the examination of functional deficits in the KIF1C-knockout neurons. For example: 1) are these neurons express layer V/VI markers at protein levels, and the proportion of positive neurons (efficiency of cortical neuron differentiation); 2) What are the phenotypic changes in the KIF1C-knockout neurons; are there change sin axonal growth or transport? 3) Day 58 was selected for collecting RNA for sequencing study: how this time point is selected? And are there phenotypic differences between the WT and knockout neurons at this time point?

    We appreciate the favorable review of our manuscript and the insightful comments:

    Characterization of neuron identities: We agree on the importance of validating neuron identities and included protein-level characterization of layer V/VI markers and efficiency of cortical neuron differentiation in our revised manuscript: We conducted immunohistochemical staining for layer V/VI and other neuronal markers, as well as qRT-PCR to validate the identity of the neurons, ensuring a comprehensive characterization of our neuronal population.

    Functional deficits in KIF1C-knockout neurons: We have conducted phenotypic examinations of the neurons but did not observe gross differences in differentiation, axon growth or axon length. We added a corresponding statement to the results section. Neurons were harvested at DAI 58 because at this time we achieved a nearly confluent chamber that yielded enough material for in-depth RNA-sequencing. We did not observe phenotypic differences between wt and KIF1C-KO neurons at this time point. We added a statement to the method section outlining this.

    Some minor comments:1. The protein levels of some critical factors needs to be validated.

    We validated neuronal identities on qRT-PCR level (Supplementary Figure S2). While we understand the necessity of validating our RNA-seq findings on protein level, we believe it is beyond the scope of this manuscript. However, we will focus on incorporating this aspect into future studies and added a corresponding statement outlining the limitations of our study in the discussion stressing the importance of this.

    1. Figure 4C, for the list genes, statistical analyses between WT and knockout groups are required.

    In Figure 4C we only included differentially expressed genes with a p-value We added a corresponding statement in the main text and figure legend.

    1. Page 15, the 5th to last sentence: "nucleus nucleus" (repeat)

    The repeat word on page 15 was deleted.

    1. The sequencing data requires public links to the deposited library

    We will provide public links to the deposited library for the sequencing data once the data is submitted to a journal (depending on journal guidelines).

    Significance

    The strength of this study is the combinations of iPSC differentiation, gene editing (KIF1C knockout iPSC) and microfluidic system. This allows the identification of specific axonal transcriptomes. Moreover, the comparisons of control and KIF1C knockout neurons at both axon and soma compartments enables the identification of RNAs and pathways caused by the loss of KIF1C.

    The limitation is the lack of functional assessment of the iPSC-derived neurons, especially phenotypic changes in the KIF1C-knockout neurons. Only one time point is selected for comparing the WT and KIF1C knockout neurons, and the relationship between this time point and disease phenotypes is unclear.

    This study will be of interest to researchers from both basic and translational fields, and in the fields of stem cells, neuroscience, neurology and genetics.

    My expertise includes stem cells, iPSC modeling, motor neuron diseases, and nerve degeneration.

    We appreciate the favorable significance statement and believe addressing these points will strengthen the scientific rigor and impact of our study. Thank you for your valuable feedback.

    Reviewer #3 (Evidence, reproducibility and clarity (Required)):  Using microfluidics chambers and RNA sequencing (RNA-seq) of axons from iPSC-derived human cortical neurons, authors use RNA profiling to investigate the RNAs present in the soma and axons and the impact of KIF1C molecular motor downregulation (KIF1CKO) on the axonal transcriptome. The rationale is that mutations in KIF1C are associated with an autosomal recessive form of hereditary spastic paraplegia, and KIF1C is implicated in the long-range directional transport of APC-dependent mRNAs and RNA-dependent transport of the exon junction complex into neurites.  Employing a well-defined RNA-seq pipeline for analysis, they obtained RNA sequences particular to axonal samples, outperforming previous studies. They detected over 16,000 genes in the soma (which includes axons) and RNA for more than 5,000 genes in axons. A comparison of the list of axonal genes revealed a strong correlation with previous publications, but they detected more genes overall. They identified transcripts enriched in axons compared to somas, notably those for ribosomal and mitochondrial proteins. Indeed, they observed enrichment for ribosomal subunits, respiratory chain complexes, ion transport, and mRNA splicing.  The study also found that human axons exhibit a unique RNA transcription profile of transcription factors (TFs), with TFs such as GTF3A and ATF4 predominant in axons. At the same time, CREB3 was highly expressed in the soma.  Upon analyzing the soma and axon transcriptomes from KIF1CKO cultures, they identified 189 differentially regulated transcripts: 89 downregulated and 100 upregulated in the KIF1CKO condition. Some of these transcripts are critical for synaptic growth and neurotransmission. Notably, only two targets of APC-target RNAs were downregulated, contrary to their expectation. Their data indicates that KIF1C downregulation significantly alters the axonal transcriptome landscape.  Reviewer #3 (Significance (Required)):  The study is well-performed and informative, particularly for researchers interested in the local translation of axonal proteins and the axonal transcriptome. However, the authors did not validate their findings for any transcripts and did not perform any functional assays, so the manuscript lacks mechanistic insight. Interestingly, GTF3A is a transcription factor that stimulates polymerase III transcription of ribosomal proteins, and mRNAs for ribosomal proteins are enriched in human axons. Maybe there is an interesting story there.

    We appreciate the favorable significance statement and the valuable feedback. We have conducted phenotypic examinations of the neurons but did not observe gross differences in differentiation, axon growth or axon length. We added a corresponding statement to the results section. While we understand the necessity of validating our RNA-seq findings on protein level, we believe it is beyond the scope of this manuscript. However, we will focus on incorporating this aspect into future studies and added a corresponding statement outlining the limitations of our study in the discussion stressing the importance of this.

  2. Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons


    Referee #3

    Evidence, reproducibility and clarity

    Using microfluidics chambers and RNA sequencing (RNA-seq) of axons from iPSC-derived human cortical neurons, authors use RNA profiling to investigate the RNAs present in the soma and axons and the impact of KIF1C molecular motor downregulation (KIF1CKO) on the axonal transcriptome. The rationale is that mutations in KIF1C are associated with an autosomal recessive form of hereditary spastic paraplegia, and KIF1C is implicated in the long-range directional transport of APC-dependent mRNAs and RNA-dependent transport of the exon junction complex into neurites.

    Employing a well-defined RNA-seq pipeline for analysis, they obtained RNA sequences particular to axonal samples, outperforming previous studies. They detected over 16,000 genes in the soma (which includes axons) and RNA for more than 5,000 genes in axons. A comparison of the list of axonal genes revealed a strong correlation with previous publications, but they detected more genes overall. They identified transcripts enriched in axons compared to somas, notably those for ribosomal and mitochondrial proteins. Indeed, they observed enrichment for ribosomal subunits, respiratory chain complexes, ion transport, and mRNA splicing. The study also found that human axons exhibit a unique RNA transcription profile of transcription factors (TFs), with TFs such as GTF3A and ATF4 predominant in axons. At the same time, CREB3 was highly expressed in the soma. Upon analyzing the soma and axon transcriptomes from KIF1CKO cultures, they identified 189 differentially regulated transcripts: 89 downregulated and 100 upregulated in the KIF1CKO condition. Some of these transcripts are critical for synaptic growth and neurotransmission. Notably, only two targets of APC-target RNAs were downregulated, contrary to their expectation. Their data indicates that KIF1C downregulation significantly alters the axonal transcriptome landscape.

    Significance

    The study is well-performed and informative, particularly for researchers interested in the local translation of axonal proteins and the axonal transcriptome. However, the authors did not validate their findings for any transcripts and did not perform any functional assays, so the manuscript lacks mechanistic insight. Interestingly, GTF3A is a transcription factor that stimulates polymerase III transcription of ribosomal proteins, and mRNAs for ribosomal proteins are enriched in human axons. Maybe there is an interesting story there.

  3. Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons


    Referee #2

    Evidence, reproducibility and clarity

    This study seeks to identify axonal transcriptome by RNA-sequencing of the iPSC-derived cortical neuron axons. This is achieved by comparing the RNA expressions between the axonal and soma compartments using microfluid system. The specific expression of axon specific RNAs in the axonal compartment validate the specificity of the approach. Some unique RNAs including TF specific RNAs are identified. Furthermore, this study compared the KIF1C-knockout neurons (which models hereditary spastic paraplegia characterized by axonal degeneration) with wildtype (WT) control neurons, which led to the identification of specific down-regulated RNAs involved in axonal development and guidance, neurotransmission, and synaptic formation.

    The data of this study are interesting and clearly presented. The major concerns are the lack of characterization of the neuron identities and the examination of functional deficits in the KIF1C-knockout neurons. For example: 1) are these neurons express layer V/VI markers at protein levels, and the proportion of positive neurons (efficiency of cortical neuron differentiation); 2) What are the phenotypic changes in the KIF1C-knockout neurons; are there change sin axonal growth or transport? 3) Day 58 was selected for collecting RNA for sequencing study: how this time point is selected? And are there phenotypic differences between the WT and knockout neurons at this time point?

    Some minor comments:

    1. The protein levels of some critical factors needs to be validated.
    2. Figure 4C, for the list genes, statistical analyses between WT and knockout groups are required.
    3. Page 15, the 5th to last sentence: "nucleus nucleus" (repeat)
    4. The sequencing data requires public links to the deposited library

    Significance

    The strength of this study is the combinations of iPSC differentiation, gene editing (KIF1C knockout iPSC) and microfluidic system. This allows the identification of specific axonal transcriptomes. Moreover, the comparisons of control and KIF1C knockout neurons at both axon and soma compartments enables the identification of RNAs and pathways caused by the loss of KIF1C.

    The limitation is the lack of functional assessment of the iPSC-derived neurons, especially phenotypic changes in the KIF1C-knockout neurons. Only one time point is selected for comparing the WT and KIF1C knockout neurons, and the relationship between this time point and disease phenotypes is unclear.

    This study will be of interest to researchers from both basic and translational fields, and in the fields of stem cells, neuroscience, neurology and genetics.

    My expertise includes stem cells, iPSC modeling, motor neuron diseases, and nerve degeneration.

  4. Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons


    Referee #1

    Evidence, reproducibility and clarity

    Summary:

    This manuscript by Xu, Hörner, Schüle and colleagues is an RNA-seq study focusing characterization of axonal transcriptomes from human iPSC-derived cortical neurons. The authors have differentiated iPSC into neurons, cultured them in microfluidic devices and isolated axonal RNA, comparing this to corresponding cell soma transcriptomes. Second, axonal transcriptomes are compared between wild type and Kif1c knockout axons to determine Kif1c-dependently localized transcripts. Characterization of the latter allows the authors to suggest differentially expressed transcripts in Kif1c-KO axons can be mRNAs relevant for motor neuron degeneration owing to Kif1c mutations in hereditary spastic paraplegias.

    Major comments:

    Overall, his manuscript reads like work in (early) progress. This manuscript provides an interesting dataset, but needs substantial additional experimental and/or bioinformatic work to merit publication. The technical complexity of steps that have led to obtaining axonal transcriptomes can be appreciated, the soundness of generating these data is beyond doubt. However, the study stops at the point of generating axonal transcriptomes from wild type and Kif1c axons. No follow-up experiments are performed to study genes of interest found in RNA-seq. This could be compensated by in-depth bioinformatic analysis (e.g. comparisons with the many different datasets in known in the field), but this is clearly lacking as well.

    The results section only contains minimal bioinformatic analysis and nothing else. Indroduction and discussion are well, clearly written and are in good dialogue with the existing body of work. To improve the manuscript, at minimum these two aspects should be addressed:

    1. Characterization of the iPSC-derived neurons is missing (immunostaining with neuronal markers, e.g. Tau, MAP2, exclusion of glial markers, and lack of stem cell markers)
    2. Validation of candidates of interest (e.g. FISH analysis in axons vs somata, Kif1c vs wt). Very specific requests from the review are useless at this point, as the authors should have the liberty to focus.

    Minor comments:

    1. Details of RNA seq technicalities are redundant in the results section, e.g. „Our RNA-seq pipeline encompassed read quality control (QC), RNA-seq mapping, and gene quantification" (p. 7) is a trivial description - this and similar details should be skipped or described in methods.
    2. Fig1A: Y axis should start from 0
    3. Too much interpretational voice in figure legends (e.g. see Fig. 1, „PC1 clearly distinguishes the soma (blue)"
    4. PCA analysis seems redundant in Fig. 2C
    5. Subheading „Human motor axons show a unique transcription factor profile" is misleading - you are not dealing with motor iPS-derived motoneurons (Isl-1 positive), but cortical neurons (again, no marker information provided to assess this!)
    6. Fig. 3: Just by comparing top expressed factors in axonal samples is not informative - overall high expression of a certain transcript likely makes it easier for it to be picked up in the axonal compartment. Axon/soma ratios would perhaps be more appropriate.
    7. Figure 4 (KIF1C modulates the axonal transcriptome): you should show also data for the same genes in the soma, axonal data only is misleading (is overall expression changed?)

    Significance

    Axonal transcriptomes have been studied since early 2010s by a number of groups and several datasets exist from different model systems. The authors know these studies well, address their findings and cite them appropriately. Is the dataset in this manuscript novel? Does it contribute to the field? Several axonal transcriptomes have been characterized in thorough studies, and even in the specific niche (human IPS-derived motoneurons) a point of reference exists - as the authors themselves point out, it is the Nijssen 2018 study. With appropriate presentation and follow-up experiments this material could have merit as a replication study.

    Audience: specialized