Plant Trans-Golgi Network/Early Endosome pH regulation requires Cation Chloride Cotransporter (CCC1)

Article activity feed

1. 

   eLife

   Jul 18, 2022

   Author Response

   Reviewer #1 (Public Review):

   In their manuscript “Plant Trans-Golgi Network/Early Endosome pH regulation requires Cation Chloride Cotransporter (CCC1)” the authors sought out to understand the importance of the cation chloride co-transporter CCC1 on plant function and intracellular ion homeostasis. The authors provide new data showing that CCC1 functions at the TGN/EE where it regulates ion homeostasis. Plants lacking CCC1 show a disruption to normal endomembrane trafficking, leading to defects in root hair cell elongation and patterning. Interestingly the authors show that the cell elongation defects can be rescued by supplementing the plants with an external osmolyte such as mannitol. Through the characterisation of CCC1 in A. thaliana, this paper shows that cation/anion transporters are essential in maintaining …

   Author Response

   Reviewer #1 (Public Review):

   In their manuscript “Plant Trans-Golgi Network/Early Endosome pH regulation requires Cation Chloride Cotransporter (CCC1)” the authors sought out to understand the importance of the cation chloride co-transporter CCC1 on plant function and intracellular ion homeostasis. The authors provide new data showing that CCC1 functions at the TGN/EE where it regulates ion homeostasis. Plants lacking CCC1 show a disruption to normal endomembrane trafficking, leading to defects in root hair cell elongation and patterning. Interestingly the authors show that the cell elongation defects can be rescued by supplementing the plants with an external osmolyte such as mannitol. Through the characterisation of CCC1 in A. thaliana, this paper shows that cation/anion transporters are essential in maintaining fine control over endosomal pH, in addition to previously characterised endosomal proton/cation transporters such as NHX5, NHX6, and CLCd.

   The paper is well written, and the experimental design is generally well thought out. The data mostly supports the authors conclusions, however there are some areas where changes are necessary to improve the clarity and completeness of the experimental work.

   1. The co-localisation experiment of CCC1 with VHA-a1 (TGN/EE marker) shows that they highly overlap, however, there are clear regions where the CCC1 and VHA-a1 marker do not co-localise, suggesting CCC1 has a broader localisation pattern which is also alluded to in the text.

   It is important to clearly determine which endomembrane compartments CCC1 localises to as this has large implications in interpretation of data regarding where the endomembrane trafficking defects originate from (eg: TGN/EE dysfunction, or other organelles, such as the Golgi and MVB/LE), and for comparisons with other intracellular transporters such as NHX5 and NHX6 (which have broader localisation at the Golgi, TGN/EE, and MVB/LE). A more detailed localisation approach by also assessing the co-localisation of CCC1 with Golgi and MVB/LE markers is necessary.

   Our data shows that the co-localisation of VHA-a1-RFP and GFP-CCC1 is extraordinarily high at 0.86 for CCC1/VHAa1, compared, for instance, with SYP43/VHAa1 from Shimizu et al. 2021 Nature Plants, which has a correlation coefficient of ~0.7.

   Proteomic studies (e.g. Groen et al 2014 J Proteomic Research), have shown that CCC1 is a high-confidence TGN/EE resident protein, co-localised with VHA-a1 and SYP61. We have included this information in the introduction and results (L86-90 and L182-186). We agree that further co-localisation studies will be useful in the future; at this time point, we focused on the role of CCC1 in the TGN/EE.

   1. The authors identify defects in cell elongation in ccc1-1 root epidermal cells, as well as defects in the formation of collet hairs. It is not clear whether the defects in collet hair formation is due to defects in cell elongation, or in root hair cell identity as root hair cell identity is disrupted in ccc mutants. Since under control conditions some ccc mutants do not form collet hair cells at all this would suggest that the hair cell identity is also disrupted, rather than just elongation. However, the root hair length quantification experiment does show very clear cell elongation defects in ccc1 mutants. The two phenotypes should be differentiated more clearly in the text.

   Thank you, we have amended the manuscript and now better differentiate between the collet hair phenotype and the root hair phenotype.

   We have now included evidence that collet hair elongation, and not cell identity, is disrupted in ccc1 collet hairs (Figure 4 – supplementary figure 1D). In ccc1 plants, collet hairs are initiated to some degree but do not elongate, while under increased external osmolarity, collet hairs elongate similar to what was observed in the wildtype.

   1. Figure 6 describes experiments designed to assess whether ccc1 mutants have defects in endo- and/or exocytosis. The authors assess endocytosis using an FM4-64 uptake experiment where they conclude that ccc1 mutants have defects in endocytosis. However, the data from the 10-minute time point (which is usually used to measure endocytosis) shows no difference between wild-type and mutant lines. There are clear differences in FM4-64 uptake to the BFA bodies after 60 minutes (Golgi+TGN) which instead suggests ccc1 mutants primarily have defects in post-Golgi trafficking, rather than endocytosis.

   We agree with the reviewer’s comment here and we think there was a misunderstanding due to the wording we used. Yes, a 10 minutes time point would measure immediate endocytosis and what we quantify at the 60 min time point is endocytic trafficking. We had used the terminology “endocytosis” throughout the manuscript, however, in the wake of these comments we realise that this terminology was not sufficiently precise. As the reviewer correctly points out, what we measured and what we are interested in is “endocytic trafficking”, a process previously shown to be disrupted in mutants with altered TGN/EE pH. We have improved the wording of the manuscript to better reflect this and more strictly adhered to the exact use of endocytic trafficking and endocytosis

   The authors should also assess whether secretion/recycling of PIP2;1 and PIN2-GFP is altered by quantifying the signal at the plasma membrane, and potentially by performing FRAP assays of PIP2;1 or PIN2-GFP at the plasma membrane.

   We appreciate the reviewers’ interest in this subject; however, the trafficking results are included to support the assertion that CCC1 has a role in TGN/EE pH and ion regulation. Detailed trafficking assays are therefore not a key or central theme in the manuscript and as such, we think that further focusing on the trafficking aspect would distract readers from the primary take home message of the work. Nevertheless, quantification of PIN2-GFP signal at the PM is now included in Figure 6 – supplementary figure 1A, as requested.

   The authors could also assess whether ccc1 mutants have general defects in secretion by visualisation of sec-RFP in ccc1 mutants. These experiments (in addition to the co-localisation experiments suggested above) would provide much stronger evidence to determine the exact source of trafficking defects.

   We agree that sec-RFP would be another means to assess general secretion defects on-top of what we have already provided. However, we believe that further characterisation of trafficking defects in ccc1 with sec-RFP will not aid in further determining the exact source of trafficking defects beyond what is already provided in the manuscript. We suggest that the trafficking defects are caused by changes to TGN/EE pH regulation and the mechanism by which pH impacts trafficking is not yet fully understood. To that end, we used FM4-64 and PIN2 to assay trafficking as these are markers used previously to assay trafficking of det3 and nhx5/nhx6 mutants. The assessment of CCC1’s role in TGN/EE pH and ion regulation is the central goal of this work.

   1. The calibration curves from Figure 7 are missing

   Calibration curves have now been included (Figure 7 – supplementary figure 2)

   1. The control image of PRP3::H2B in wild type seedlings is missing

   The control wildtype image has been added (Figure 2 – figure supplement 1C)

   Reviewer #2 (Public Review):

   In the submitted paper, the authors first show that activity of the CCC1 promoter is ubiquitous. They further analyze the phenotype of the mutant in the root and show a root cell elongation defect in epidermal cells as well as in root hairs. The ccc1 mutants also lack the collet root hairs and show trichoblast-atrichoblast cell fate identity defects in the primary root. The authors perform a set of elegant experiments where they show that, surprisingly, the ccc1 plants are resistant to hyperosmotic environment. The ccc1 cells show delayed plasmolysis, ccc1 seeds show better germination, and ccc1 root hair elongation is recovered in hyper-osmotic media. Interestingly, the absence of collet root hairs was also recovered in hyper-osmotic environment, even though it is not clear whether this was caused by 'reparation' of collet hair elongation or collet hair cell fate specification. The phenotypic analysis is carefully performed and the results are unexpected and intriguing.

   The authors further show that in root trichoblasts, GFP-CCC1 localizes to the TGN/EE compartment, and that in this tissue, the fusion protein recovers the root hair elongation of the mutant. Further, the authors focus on the subcellular phenotypes of the endomembrane system performance in the ccc1 mutant background. It is shown that PIP2 aquaporin internalized less in the ccc1 than in control, which hints to that endocytosis is reduced in the ccc1 cells. An alternative explanation however could be that the mutant is more osmotically tolerant also on the subcellular level. To test the endomembrane trafficking rate, PIN2 aggregation and recovery from BFA bodies is performed, as well as quantifications of FM4-64 uptake, and the authors conclude that the mutant has generic endomembrane trafficking defects.

   The authors hypothesize that the endomembrane defects might stem from a disturbance in TGN/EE luminal pH caused by an ion imbalance in the ccc1 cells. Therefore, they measured the luminal pH of TGN/EE using a genetically encoded phluorin and demonstrate a more alkaline pH values in the ccc1 mutants. Finally, the authors show that during hyperosmotic stress, the TGN/EE pH rises in the control plants, suggesting that this pH rise is functionally connected to the stress response. The second part of the manuscript that focuses on subcellular phenotypes uses advanced live-cell imaging tool and successfully measures pH in minute volumes of TGN/EE compartments. In addition, the specificity of the phenotype is demonstrated by careful analysis of vacuolar and cytoplasmic pH. These well performed experiments indeed point to the function of CCC1 in ion control in TGN/EE.

   Many thanks for your positive comments on our work. We are glad it is appreciated.

   Weaknesses of the manuscript:

   The functionality of the GFP-CCC1 fusion is questionable as it was impossible to obtain transgenic lines that would express GFP-CCC1 under the control of the native promoter, not allowing full complementation of the ccc1 phenotype. This hints to a possible dominant-negative effect of this particular protein fusion. The authors therefore express GFP-CCC1 using a trichoblast-specific promoter and show that the root hair elongation phenotype is complemented, demonstrating some functionality of this construct. Moreover, the root hair length data in the ccc1-1 mutants shown in figure 2D and 3C differ, which to some extent weakens the important conclusion that the GFP-CCC1 is functional at least in this cell type. Functionality of this construct is a crucial aspect for the manuscript. The possible dominant-negative effect of the construct weakens the conclusion about the subcellular protein localization, which in turn weakens the main conclusion of the paper - that CCC1 by regulating ion fluxes in the TGN/EE allows proper endomembrane functionality.

   The reviewer notes the observed difference in wildtype root hair length between experimental data shown in Figures 2D and 3C. Yes, this is correct. The experiments were done several months apart (different to the biological replicate experiments pooled for one graph, which were always conducted around the same time). Root hairs are highly reactive to environmental conditions and as such, plants grown at different times of the year commonly have differences in root hair length. As such, comparisons can only be made to a control, grown at the same time, when looking for differences in treatments or genotypes. Therefore, the data from 2D and 3C cannot be compared to each other quantitatively but rather, qualitatively.

   In regards to the localisation and constructs used, neither N- nor C- terminally tagging produces transformants. Our experimental approaches suggest that all terminal tagging of CCC1 is dominant negative if the construct is expressed from the embryo stage (native promoter or ubiquitous promoters such as 35S). Expression of untagged CCC1 by either the native or 35S promoter rescues the phenotype of ccc1 KO plants. We have now provided a table (Supplementary File 1a) that summarises all attempts to localise CCC1 and the outcome, including the generation of an antibody.

   We have additionally added details on previous proteomics studies, which identified CCC1 as a high-confidence TGN/EE resident protein (L186 and discussion)

   The subcellular localization of CCC1 should be demonstrated without any doubts, as it was previously localized to the PM and endomembranes in pollen tubes (Domingos 2019). If CCC1 localized at the PM, alternative explanations of the phenotypes of the nature of the mutant phenotype that would include regulation ion fluxes across the PM would appear more probable than the TGN/EE hypothesis.

   The reviewer highlights that CCC1 has been proposed to localise to the PM in pollen tubes in Domingos et al. (2019). Localisation of CCC1 shown in Domingos et al. (2019), lacks colocalisation with a marker and importantly, no complementation of the knockout phenotype is shown with the tagged protein. We have added a paragraph in the discussion on this topic (L488-507).

   Quantification of endomembrane trafficking represents another important argument in the proposed hypothesis. The section that demonstrates the reduction in exo- and endocytosis is however not utterly convincing. It has been shown that a major contribution to the BFA body PIN2 pool originates from de-novo synthesis of the protein (Jasik et al, PMID:27506239). In the figure 6A, it is apparent that the BFA washout leads to disappearance of BFA bodies in the ccc1 mutant, but the level of PM fluorescence was decreased, leading to an apparent 'minimal recovery' of the cytoplasm:PM ratio. In case of endocytosis, the experiment combining FM4-64 uptake with BFA is hard to interpret as endocytosis visualization, because TGN/EE aggregation might be disturbed in the ccc1, as the authors suggest. A more detailed endomembrane trafficking then simple cytoplasm/PM ratios of signal could be performed to address what is happening with trafficking in this interesting mutant.

   In the FM4-64 experiment, there is a poor formation of BFA bodies and a lower ratio in ccc1, however, despite having the same poor formation of BFA bodies in the PIN2-GFP experiment, the ratio is higher, indicating that the visibility of BFA bodies is not crucial to the accumulation or measurement of fluorescence (discussed in L381). The reviewer does rightly state that de-novo protein synthesis is a major contributor to intracellular protein accumulation in these assays and that is why the assay focuses on the rate of recovery. That is, we measured PIN2-GFP recovery to the PM regardless of the origin of the PIN2-GFP protein in BFA bodies (de novo or from endocytosis).

   Further characterisation of the impacts TGN/EE luminal pH changes have on endomembrane trafficking is undoubtedly an interesting topic of study, as highlighted by the reviewer’s comments. The work detailed in this manuscript is focused on assessing the role of CCC1 in TGN/EE pH and ion regulation. As such, the results presented will enhance the potential to investigate the impact of TGN/EE pH on endomembrane trafficking by providing details of another tool, ccc1, which can be used to investigate this link and by further detailing the impact of environmental conditions on TGN/EE pH. It is not the aim of this study here to investigate the link between TGN/EE pH and endomembrane trafficking and as such, we believe that further results detailing endomembrane trafficking will detract from the central results of this work. However, this aspect highlights the general interest and importance of our work for other fields of plant science, and to other fields as CCC proteins are present in all organisms.

   Read the original source
2. 

   eLife

   May 28, 2022

   Evaluation Summary:

   Presented here is a study on the role of cation chloride cotransporter CCC1 as a key regulator of the plant Trans-Golgi/Early Endosome trafficking network. While the work is well controlled and presented overall, the reviewers judged the data supporting localization of CCC1 to TGN/EE as not being sufficiently clear, as was the role of CCC1 in endocytosis, which is one of the main conclusions. These points can be clarified with future careful experimentation.

   (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript.The reviewers remained anonymous to the authors.)

   Read the original source
3. 

   eLife

   May 28, 2022

   Reviewer #2 (Public Review):

   In the submitted paper, the authors first show that activity of the CCC1 promoter is ubiquitous. They further analyze the phenotype of the mutant in the root and show a root cell elongation defect in epidermal cells as well as in root hairs. The ccc1 mutants also lack the collet root hairs and show trichoblast-atrichoblast cell fate identity defects in the primary root. The authors perform a set of elegant experiments where they show that, surprisingly, the ccc1 plants are resistant to hyperosmotic environment. The ccc1 cells show delayed plasmolysis, ccc1 seeds show better germination, and ccc1 root hair elongation is recovered in hyper-osmotic media. Interestingly, the absence of collet root hairs was also recovered in hyper-osmotic environment, even though it is not clear whether this was caused by …

   Reviewer #2 (Public Review):

   In the submitted paper, the authors first show that activity of the CCC1 promoter is ubiquitous. They further analyze the phenotype of the mutant in the root and show a root cell elongation defect in epidermal cells as well as in root hairs. The ccc1 mutants also lack the collet root hairs and show trichoblast-atrichoblast cell fate identity defects in the primary root. The authors perform a set of elegant experiments where they show that, surprisingly, the ccc1 plants are resistant to hyperosmotic environment. The ccc1 cells show delayed plasmolysis, ccc1 seeds show better germination, and ccc1 root hair elongation is recovered in hyper-osmotic media. Interestingly, the absence of collet root hairs was also recovered in hyper-osmotic environment, even though it is not clear whether this was caused by 'reparation' of collet hair elongation or collet hair cell fate specification. The phenotypic analysis is carefully performed and the results are unexpected and intriguing.

   The authors further show that in root trichoblasts, GFP-CCC1 localizes to the TGN/EE compartment, and that in this tissue, the fusion protein recovers the root hair elongation of the mutant. Further, the authors focus on the subcellular phenotypes of the endomembrane system performance in the ccc1 mutant background. It is shown that PIP2 aquaporin internalized less in the ccc1 than in control, which hints to that endocytosis is reduced in the ccc1 cells. An alternative explanation however could be that the mutant is more osmotically tolerant also on the subcellular level. To test the endomembrane trafficking rate, PIN2 aggregation and recovery from BFA bodies is performed, as well as quantifications of FM4-64 uptake, and the authors conclude that the mutant has generic endomembrane trafficking defects.

   The authors hypothesize that the endomembrane defects might stem from a disturbance in TGN/EE luminal pH caused by an ion imbalance in the ccc1 cells. Therefore, they measured the luminal pH of TGN/EE using a genetically encoded phluorin and demonstrate a more alkaline pH values in the ccc1 mutants. Finally, the authors show that during hyperosmotic stress, the TGN/EE pH rises in the control plants, suggesting that this pH rise is functionally connected to the stress response. The second part of the manuscript that focuses on subcellular phenotypes uses advanced live-cell imaging tool and successfully measures pH in minute volumes of TGN/EE compartments. In addition, the specificity of the phenotype is demonstrated by careful analysis of vacuolar and cytoplasmic pH. These well performed experiments indeed point to the function of CCC1 in ion control in TGN/EE.

   Weaknesses of the manuscript:

   The functionality of the GFP-CCC1 fusion is questionable as it was impossible to obtain transgenic lines that would express GFP-CCC1 under the control of the native promoter, not allowing full complementation of the ccc1 phenotype. This hints to a possible dominant-negative effect of this particular protein fusion. The authors therefore express GFP-CCC1 using a trichoblast-specific promoter and show that the root hair elongation phenotype is complemented, demonstrating some functionality of this construct. Moreover, the root hair length data in the ccc1-1 mutants shown in figure 2D and 3C differ, which to some extent weakens the important conclusion that the GFP-CCC1 is functional at least in this cell type. Functionality of this construct is a crucial aspect for the manuscript. The possible dominant-negative effect of the construct weakens the conclusion about the subcellular protein localization, which in turn weakens the main conclusion of the paper - that CCC1 by regulating ion fluxes in the TGN/EE allows proper endomembrane functionality. The subcellular localization of CCC1 should be demonstrated without any doubts, as it was previously localized to the PM and endomembranes in pollen tubes (Domingos 2019). If CCC1 localized at the PM, alternative explanations of the phenotypes of the nature of the mutant phenotype that would include regulation ion fluxes across the PM would appear more probable than the TGN/EE hypothesis.

   Quantification of endomembrane trafficking represents another important argument in the proposed hypothesis. The section that demonstrates the reduction in exo- and endocytosis is however not utterly convincing. It has been shown that a major contribution to the BFA body PIN2 pool originates from de-novo synthesis of the protein (Jasik et al, PMID:27506239). In the figure 6A, it is apparent that the BFA washout leads to disappearance of BFA bodies in the ccc1 mutant, but the level of PM fluorescence was decreased, leading to an apparent 'minimal recovery' of the cytoplasm:PM ratio. In case of endocytosis, the experiment combining FM4-64 uptake with BFA is hard to interpret as endocytosis visualization, because TGN/EE aggregation might be disturbed in the ccc1, as the authors suggest. A more detailed endomembrane trafficking then simple cytoplasm/PM ratios of signal could be performed to address what is happening with trafficking in this interesting mutant.

   In summary:

   The manuscript is clearly written, the logic of the text is comprehensible, the data seems robust and well presented. The manuscript attempts to explain the organ- and cellular scale phenotype of the root growth and root hair elongation by the subcellular defect in the TGN/EE luminal pH via defective endomembrane trafficking. The functional connection between the organellar and cellular phenotype and the function of CCC1 is however still somewhat preliminary.

   Read the original source
4. 

   eLife

   May 28, 2022

   Reviewer #1 (Public Review):

   In their manuscript "Plant Trans-Golgi Network/Early Endosome pH regulation requires Cation Chloride Cotransporter (CCC1)" the authors sought out to understand the importance of the cation chloride co-transporter CCC1 on plant function and intracellular ion homeostasis. The authors provide new data showing that CCC1 functions at the TGN/EE where it regulates ion homeostasis. Plants lacking CCC1 show a disruption to normal endomembrane trafficking, leading to defects in root hair cell elongation and patterning. Interestingly the authors show that the cell elongation defects can be rescued by supplementing the plants with an external osmolyte such as mannitol. Through the characterisation of CCC1 in A. thaliana, this paper shows that cation/anion transporters are essential in maintaining fine control over …

   Reviewer #1 (Public Review):

   In their manuscript "Plant Trans-Golgi Network/Early Endosome pH regulation requires Cation Chloride Cotransporter (CCC1)" the authors sought out to understand the importance of the cation chloride co-transporter CCC1 on plant function and intracellular ion homeostasis. The authors provide new data showing that CCC1 functions at the TGN/EE where it regulates ion homeostasis. Plants lacking CCC1 show a disruption to normal endomembrane trafficking, leading to defects in root hair cell elongation and patterning. Interestingly the authors show that the cell elongation defects can be rescued by supplementing the plants with an external osmolyte such as mannitol. Through the characterisation of CCC1 in A. thaliana, this paper shows that cation/anion transporters are essential in maintaining fine control over endosomal pH, in addition to previously characterised endosomal proton/cation transporters such as NHX5, NHX6, and CLCd.

   The paper is well written, and the experimental design is generally well thought out. The data mostly supports the authors conclusions, however there are some areas where changes are necessary to improve the clarity and completeness of the experimental work.

   1. The co-localisation experiment of CCC1 with VHA-a1 (TGN/EE marker) shows that they highly overlap, however, there are clear regions where the CCC1 and VHA-a1 marker do not co-localise, suggesting CCC1 has a broader localisation pattern which is also alluded to in the text.

   It is important to clearly determine which endomembrane compartments CCC1 localises to as this has large implications in interpretation of data regarding where the endomembrane trafficking defects originate from (eg: TGN/EE dysfunction, or other organelles, such as the Golgi and MVB/LE), and for comparisons with other intracellular transporters such as NHX5 and NHX6 (which have broader localisation at the Golgi, TGN/EE, and MVB/LE). A more detailed localisation approach by also assessing the co-localisation of CCC1 with Golgi and MVB/LE markers is necessary.

   1. The authors identify defects in cell elongation in ccc1-1 root epidermal cells, as well as defects in the formation of collet hairs. It is not clear whether the defects in collet hair formation is due to defects in cell elongation, or in root hair cell identity as root hair cell identity is disrupted in ccc mutants. Since under control conditions some ccc mutants do not form collet hair cells at all this would suggest that the hair cell identity is also disrupted, rather than just elongation. However, the root hair length quantification experiment does show very clear cell elongation defects in ccc1 mutants. The two phenotypes should be differentiated more clearly in the text.

   2. Figure 6 describes experiments designed to assess whether ccc1 mutants have defects in endo- and/or exocytosis. The authors assess endocytosis using an FM4-64 uptake experiment where they conclude that ccc1 mutants have defects in endocytosis. However, the data from the 10-minute time point (which is usually used to measure endocytosis) shows no difference between wild-type and mutant lines. There are clear differences in FM4-64 uptake to the BFA bodies after 60 minutes (Golgi+TGN) which instead suggests ccc1 mutants primarily have defects in post-Golgi trafficking, rather than endocytosis.
      The authors should also assess whether secretion/recycling of PIP2;1 and PIN2-GFP is altered by quantifying the signal at the plasma membrane, and potentially by performing FRAP assays of PIP2;1 or PIN2-GFP at the plasma membrane. The authors could also assess whether ccc1 mutants have general defects in secretion by visualisation of sec-RFP in ccc1 mutants. These experiments (in addition to the co-localisation experiments suggested above) would provide much stronger evidence to determine the exact source of trafficking defects.

   3. The calibration curves from Fig. 7 are missing

   4. The control image of PRP3::H2B in wild type seedlings is missing

   Read the original source
5. 

   Version published to 10.7554/elife.70701 on eLife

   Jan 6, 2022
6. 

   Version published to 10.1101/2020.01.02.893073v4 on bioRxiv

   Jun 19, 2021
7. 

   Version published to 10.1101/2020.01.02.893073v3 on bioRxiv

   Nov 24, 2020
8. 

   Version published to 10.1101/2020.01.02.893073v1 on bioRxiv

   Jan 2, 2020
9. 

   Version published to 10.1101/2020.01.02.893073v2 on bioRxiv

   Jan 2, 2020