Phosphoproteomics identifies microglial Siglec‐F inflammatory response during neurodegeneration

Abstract Alzheimer’s disease (AD) is characterized by the appearance of amyloid‐β plaques, neurofibrillary tangles, and inflammation in brain regions involved in memory. Using mass spectrometry, we have quantified the phosphoproteome of the CK‐p25, 5XFAD, and Tau P301S mouse models of neurodegeneration. We identified a shared response involving Siglec‐F which was upregulated on a subset of reactive microglia. The human paralog Siglec‐8 was also upregulated on microglia in AD. Siglec‐F and Siglec‐8 were upregulated following microglial activation with interferon gamma (IFNγ) in BV‐2 cell line and human stem cell‐derived microglia models. Siglec‐F overexpression activates an endocytic and pyroptotic inflammatory response in BV‐2 cells, dependent on its sialic acid substrates and immunoreceptor tyrosine‐based inhibition motif (ITIM) phosphorylation sites. Related human Siglecs induced a similar response in BV‐2 cells. Collectively, our results point to an important role for mouse Siglec‐F and human Siglec‐8 in regulating microglial activation during neurodegeneration.

13th Aug 2020 1st Editorial Decision Thank you again for submit ting your work to Molecular Syst ems Biology. We have now heard back from two of the three referees who agreed to evaluat e your st udy. Unfort unat ely, aft er a series of reminders we have not heard back from reviewer #3. In the int erest of time and since the recommendat ions of the ot her two reviewers are rat her similar we have decided to proceed wit h making a decision based on these two available report s. As you will see below, bot h reviewers are overall support ive. However, they raise a series of concerns, which we would ask you to address in a revision.
I think that the recommendat ions of the reviewers are rat her clear and there is therefore no need to repeat the point s list ed below. Please let me know in case you would like to discuss in furt her det ail any of the issues raised. On a more edit orial level, we would ask you to address the following issues. The aut hors take three different mouse models of Alzheimer's disease and perform phosphoprot eomics on them, wit h special emphasis on tyrosine phosphorylat ion (pY). The experiment s are done wit h a special prot ocol for pY enrichment that the Whit e laborat ory has developed over the years. The MS analysis is relat ively convent ional, using Orbit rap mass spect romet ers. The analysis uses Prot eome Discoverer, Mascot but then the result s are analyzed once again wit h cust om script s and in some cases by manual int erpret at ion, for inst ance of the quant ificat ion in TMT. These st eps are described in the Met hods sect ion but st ill seem somewhat unconvent ional or subject ive (alt hough the aut hors make all script s available). Maybe the aut hors can comment more on this aspect in the text and show that it is not subject ive. Overall, however, the prot eomics analysis appears very solid to me. Given the scope of the journal, I would have expect ed some more syst ems biology level int erpret at ion of the result s. As it is, the aut hors go very quickly from the MS result s to the involvement of Siglec recept ors, which account s for the vast majorit y of the paper. This part cont ains a lot of analyses and given the current int erest in microglia, microglia subset s and their involvement in neurodegenerat ion, should be of broad int erest . The fact that Siglec phosphorylat ion goes up in all three models is a st rong indicat ion of a cent ral role. However, the fact that there is no direct homolog in humans of the mouse one, makes direct funct ional int erpret at ion a bit difficult , alt hough the aut hors cert ainly make a tremendous effort and present many int erest ing result s. All in all, I find this work to be very int erest ing and solid and a good mat ch to the journal. As ment ioned above, the st ruct ure could perhaps be improved somewhat to balance the st ory and make it flow bet ter. A Cell st yle final cart oon of the proposed mechanism may help to put it all toget her in the end.
Minor point s: -the aut hors could eliminat e some jargon and abbreviat ions that may not be so easy to underst and for the broad readership of the journal ( example, 'm.o. mice').
-From Figure 1, it is not immediately clear where the control mouse channel goes and how samples are paired into TMT sets (different brain regions, mouse models and replicate animals) -Throughout the manuscript it would be good to get something more than upregulated or downregulated, i.e. p-values and effect sizes.
-Some data availability refers to nature.com; presumably from another submission (p.39) Reviewer #2: Signaling pathways underlying the development of Alzheimer's disease (AD) may involve multiple toxic insults (e.g. Abeta, tau, inflammation, kinase deregulation, etc.) and diverse cell types/brain regions during different disease stages. Although genetic studies reveal some causative/ risk genes, the investigation of the related pathways has been significantly delayed. In this manuscript, two expert groups have collaborated to use the latest phosphoproteomic strategy to probe the activation of signaling pathways in several commonly used AD mouse models (CK-p25, 5XFAD, and Tau P301S), analyzing 10,828, 24,365, 11,853 unique phosphopeptides in these animals, respectively. The 5XFAD/Δp35KI mice were also analyzed to study the role of p25CDK5 in the 5xFAD background. Dr. White is a pioneer in the development of phosphoproteomics, and in this study, his group uniquely highlighted the analysis of Tyr phosphorylation events, which is often underrepresented in other global phosphoproteomic projects. Cross-model comparisons led to the identification of shared kinases and microglial activation. Then the authors focused on the functional studies of activated Siglec-F (a CD33-related Siglec gene family member) in microglia in a murine microglial cell line (BV-2) and human stem-cell derived microglia. While Siglec-F was upregulated upon interferon gamma-induced microglial activation, Siglec-F overexpression activates an endocytic and cell death-related inflammatory pathway, dependent on its sialic acid substrates and ITIM motif phosphosites. They also showed that the human homolog Siglec-8 was upregulated in human AD brain samples. Overall, the proteomics study was comprehensive and carried out with sufficient replicates, followed by extensive hypothesis-driven experiments to study the role of Siglec-F in microglial activation, although the experiments were largely limited to cellular models. The topic of this manuscript is important and the results might be highly relevant to AD pathogenesis. The manuscript is suitable for publication if some concerns are addressed.
1. The initial part of the manuscript is an excellent resource of phosphoproteome in AD mouse models. It is expected that all raw MS data should be released to a public database (e.g. https://www.ebi.ac.uk/pride/archive) 2. The sample information is not clearly presented in TABLE_EV1-supp. In the worksheet of "Mouse Tissue", 53 samples are listed. In the worksheet of "MS Runs", 83 lanes are shown with pY and pST analysis. In the worksheet of results (e.g. 5xFAD), the title of each column is not well explained, such as "7387 5XFAD Δp35KI 9mo Hip". It is not obvious how the samples were distributed in different TMT batches, how to match each channel (e.g. 126, 127N, etc.) in each TMT batch with the 53 samples, and how to normalize batch effects. A single summary table of samples, MS runs, and MS results, would be more informative. 3. Because of the undersampling issue, comparable phosphopeptides in different TMT batches were reduced. When merging results from different TMT batches, missing values often occur. It is not clear how the authors set up cutoffs for the missing values. How many phosphopeptides were constantly detected and compared in all samples? Although this limitation may be unavoidable due to restrictions of the current MS method, the problem is gradually overcome by further development in the proteomics field. This point needs to be thoroughly discussed and clarified. 4. It appears that the authors used the TMT-MS2 method for quantification, in which the accuracy of quantification may be affected by ratio compression. It should be discussed how the authors reduced the impact of ratio compression in the analysis. 5. The false discovery rate for phosphopeptide identification was based on Percolator FDR < 5e-2, which is acceptable and larger than regular 0.01. This important cutoff may be directly presented in the main text. 6. During the DE analysis, it is not clear how the authors set up cutoffs (e.g. fold change, p values or q value/FDR). 7. For this journal of Molecular Systems Biology, it is expected that some types of pathway/network analysis might be performed to interpret these large scale datasets. 8. The author might present more rationale on selecting the Siglec-F for downstream studies (e.g. in the CD33-related Siglec gene family). How to prioritize the identified DE phosphoproteins relevant to AD pathogenesis? The Siglec gene family has many members including CD33. Are other Siglec proteins also found to be changed in the analysis? 9. This reviewer appreciates the thorough investigation of Siglec-F function in the cellular models. One concern is that the cellular model experiments generated largely correlative data. What is the effect of Siglec-F knockout in the cells during interferon gamma stimulation? Do multiple Siglec family proteins play a redundant role in the pathway?

Summary of Responses:
We thank the reviewers for their careful and thoughtful comments. We have provided a second version of the manuscript with the following changes: Reviewer #1: The authors take three different mouse models of Alzheimer's disease and perform phosphoproteomics on them, with special emphasis on tyrosine phosphorylation (pY). The experiments are done with a special protocol for pY enrichment that the White laboratory has developed over the years. The MS analysis is relatively conventional, using Orbitrap mass spectrometers. The analysis uses Proteome Discoverer, Mascot but then the results are analyzed once again with custom scripts and in some cases by manual interpretation, for instance of the quantification in TMT. These steps are described in the Methods section but still seem somewhat unconventional or subjective (although the authors make all scripts available). Maybe the authors can comment more on this aspect in the text and show that it is not subjective. Overall, however, the proteomics analysis appears very solid to me. Given the scope of the journal, I would have expected some more systems biology level interpretation of the results. As it is, the authors go very quickly from the MS results to the involvement of Siglec receptors, which accounts for the vast majority of the paper. This part contains a lot of analyses and given the current interest in microglia, microglia subsets and their involvement in neurodegeneration, should be of broad interest. The fact that Siglec phosphorylation goes up in all three models is a strong indication of a central role. However, the fact that there is no direct homolog in humans of the mouse one, makes direct functional interpretation a bit difficult, although the authors certainly make a tremendous effort and present many interesting results. All in all, I find this work to be very interesting and solid and a good match to the journal. As mentioned above, the structure could perhaps be improved somewhat to balance the story and make it flow better. A Cell style final cartoon of the proposed mechanism may help to put it all together in the end. Response: We thank the reviewer for these thoughtful comments, and have tried to address their concerns. To clarify the steps involved in quantification of TMT: We use an open-source set of scripts that read peptide-spectrum matches (PSMs) and TMT quantification values directly from Proteome Discoverer .msf files into a pandas DataFrame for analysis in python. Due to our interest in Siglec-F, we manually validated and quantified 5 scans of SVyTEIK (Siglec-F pY561) and inserted these values into the DataFrame for its respective MS analysis. These scans all matched the MS2 pattern that is shown in Figure EV3E To quantify TMT abundances in these scans, we manually measured the TMT peak intensities using Xcalibur Qual Browser. These values varied from the exact values calculated by Proteome Discoverer's Reporter Ion "Most Confident Centroid" method. We compared TMT quantification values for scans that we matched by Proteome Discoverer and observed <5% difference between the two methods, indicating that manual TMT quantification is generally accurate. Our downstream processing workflow to filter, normalize, and integrate PSMs was designed specifically for this study. The filtering parameters for MASCOT Ion Score, Percolator FDR, and Proteome Discoverer Isolation Interference cutoffs were taken from previous projects in our lab. In an independent, unpublished analysis we found low correlation between TMT abundances for identical peptides that had a median TMT quantification value below 1500. We therefore included this cutoff to improve quantification accuracy.
To normalize our dataset, we used a method derived from the CONSTrained STANdardization (CONSTANd) algorithm, referenced in the text. This procedure iteratively normalizes the TMT values for each PSM to their mean TMT intensities. It then normalizes each TMT channel to its mode TMT intensity (estimated with a KDE function). This corrects for differences in peptide loading input amounts and generates a matrix of TMT quantification values that is centered around 1.
To integrate TMT quantification values, we combine TMT quantification information for PSMs that have identical peptide sequence and phosphorylation / oxidation modifications. To calculate the integrated abundances, we used a weighted mean function which biases quantification values towards peptides with high TMT intensities and low isolation interference. Finally, we normalize quantification values to the median of the control animal / cell line group. While most of this was contained in an abbreviated manner in the Methods, we added the following sentence to alleviate concerns regarding manual quantification: "We compared TMT quantification values for scans that we matched by Proteome Discoverer and observed <5% difference between the two methods, indicating that manual TMT quantification is generally accurate." With regard to systems analysis, this point was a concern of Reviewer #2 as well. To address this concern, we performed multiple additional analyses of our data sets, including using PHOTON for clustering and protein-protein interaction networks. Unfortunately, clustering did not find any interesting peptide trends that were not captured by a simple group comparison and worked poorly for cross model comparisons due to the sparseness of the matrix (due to data-dependent analysis). PHOTON gave a number of predicted protein-protein interactions, but did not provide consistent changes between its predictions and the phosphosites that we directly measured in each model: In the end we found that gene ontology enrichment was most reflective of the actual peptide trends that we identified. These new figures have been added to Figure EV3A-EV3C. We have also added text to results section under "Cross-model comparison identifies shared kinase and microglial activation" and methods section under "Phosphosite enrichment and pathway analysis" discussing pathway analysis using gene ontology enrichment. -Throughout the manuscript it would be good to get something more than upregulated or downregulated, i.e. p-values and effect sizes. Response: We have added p-values and effect sizes to the text of the manuscript wherever upregulation or downregulation are mentioned.
-Some data availability refers to nature.com; presumably from another submission (p.39) Response: Thank you for catching this mistake, we have removed it in the second version of the manuscript.
Reviewer #2: Signaling pathways underlying the development of Alzheimer's disease (AD) may involve multiple toxic insults (e.g. Abeta, tau, inflammation, kinase deregulation, etc.) and diverse cell types/brain regions during different disease stages. Although genetic studies reveal some causative/ risk genes, the investigation of the related pathways has been significantly delayed. In this manuscript, two expert groups have collaborated to use the latest phosphoproteomic strategy to probe the activation of signaling pathways in several commonly used AD mouse models (CK-p25, 5XFAD, and Tau P301S), analyzing 10,828, 24,365, 11,853 unique phosphopeptides in these animals, respectively. The 5XFAD/Δp35KI mice were also analyzed to study the role of p25CDK5 in the 5xFAD background. Dr. White is a pioneer in the development of phosphoproteomics, and in this study, his group uniquely highlighted the analysis of Tyr phosphorylation events, which is often underrepresented in other global phosphoproteomic projects. Cross-model comparisons led to the identification of shared kinases and microglial activation. Then the authors focused on the functional studies of activated Siglec-F (a CD33-related Siglec gene family member) in microglia in a murine microglial cell line (BV-2) and human stem-cell derived microglia. While Siglec-F was upregulated upon interferon gamma-induced microglial activation, Siglec-F overexpression activates an endocytic and cell death-related inflammatory pathway, dependent on its sialic acid substrates and ITIM motif phosphosites. They also showed that the human homolog Siglec-8 was upregulated in human AD brain samples. Overall, the proteomics study was comprehensive and carried out with sufficient replicates, followed by extensive hypothesis-driven experiments to study the role of Siglec-F in microglial activation, although the experiments were largely limited to cellular models. The topic of this manuscript is important and the results might be highly relevant to AD pathogenesis. The manuscript is suitable for publication if some concerns are addressed.
1. The initial part of the manuscript is an excellent resource of phosphoproteome in AD mouse models. It is expected that all raw MS data should be released to a public database (e.g. https://www.ebi.ac.uk/pride/archive) Response: All proteomics data generated in this study has been deposited on PRIDE under the accession: PXD018757. This deposition includes raw MS data and searched ProteomeDiscoverer 2.2 .msf files. We mistakenly omitted the reviewer account login info from the initial submission and have added it to the second version of the manuscript. 2. The sample information is not clearly presented in TABLE_EV1-supp. In the worksheet of "Mouse Tissue", 53 samples are listed. In the worksheet of "MS Runs", 83 lanes are shown with pY and pST analysis. In the worksheet of results (e.g. 5xFAD), the title of each column is not well explained, such as "7387 5XFAD Δp35KI 9mo Hip". It is not obvious how the samples were distributed in different TMT batches, how to match each channel (e. g. 126, 127N Although this limitation may be unavoidable due to restrictions of the current MS method, the problem is gradually overcome by further development in the proteomics field. This point needs to be thoroughly discussed and clarified. Response: This comment highlights an important point about our data. Because we used a data-dependent MS acquisition method, many peptides were not identified in every MS analysis. The overlap between unique peptides identified in technical replicates (separate TMT 6/10-plex labeled experiments for the same proteolytic digests) was 45-70% for hippocampus and cortex samples from CK-p25 and 5XFAD mice. The overlap between peptides identified between hippocampus and cortex tissue from the same set of mice was 68% for CK-p25 and 64% for 5XFAD. And the overlap between peptides identified from different animal models was 60% between CK-p25 and 5XFAD. Out of 30,370 unique peptides that were identified in any MS analysis, 12% (305 pY, 2900 pST, 3684 total peptides) were identified in at least one comparison group for each of the three animal models. Because data-dependent mass spectrometry has well-characterized reproducibility limitations, we ran technical replicates to increase the total number of identified peptides. We then combined TMT quantification data from all runs into one normalized peptide abundance matrix. To identify trends, we compare TMT quantification values for samples within the same set of TMT 6/10-plex analyses.
We have avoided any conclusions that would be derived from an absence of MS evidence (i.e. If Gfap pY321 is only identified in CK-p25 hippocampus, this does not imply that Gfap does not change in any other tissue or mouse model).
We considered adding a discussion of these points to the manuscript, but we thought that this addition would negatively affect the flow of the manuscript.
4. It appears that the authors used the TMT-MS2 method for quantification, in which the accuracy of quantification may be affected by ratio compression. It should be discussed how the authors reduced the impact of ratio compression in the analysis.
Response: TMT ratio compression is indeed a problem that can affect quantification accuracy and suppress large changes in proteomics analyses. We have reduced the effects of ratio compression in our data by (1) simplifying the sample by enriching for low-abundance phosphopeptides to remove background proteome signal, (2) using a narrow precursor mass isolation window of 0.4 Da (total width), and (3) weighting against peptide isolation interference when combining quantification data from peptidespectrum matches for identical peptides. We added the following sentence to the Methods section: "Isolation width was set to 0.4 Da (full width) to decrease isolation interference and TMT ratio compression." Using this approach, we were able to identify fold changes between diseased and control animals as large as 8.9x (CK-p25: Cdk1/2/3 pT14), 4.2x (5XFAD: Siglec5 pY561), and 10.4x (Tau P301S: MAPT pT152). 5. The false discovery rate for phosphopeptide identification was based on Percolator FDR < 5e-2, which is acceptable and larger than regular 0.01. This important cutoff may be directly presented in the main text. Response: We have re-run our analysis with a Percolator FDR cutoff of 0.01 and observed only minor changes in the generated figures. The second version of the manuscript uses an FDR cutoff of 0.01. 6. During the DE analysis, it is not clear how the authors set up cutoffs (e.g. fold change, p values or q value/FDR). Response: For the Venn diagrams and heatmaps, we used a fold change cutoff of 1.25x and 2-sample t-test p-value cutoff of 0.01 to identify peptides with differential abundance. We have clarified this point in the text. 7. For this journal of Molecular Systems Biology, it is expected that some types of pathway/network analysis might be performed to interpret these large scale datasets. Response: See response to Reviewer #1, above. Briefly, after several analysis attempts, we settled on gene-ontology enrichment and have added analysis figures to Figure EV3A-C and changed the text in "Cross-model comparison identifies shared kinase and microglial activation" to add an additional systems analysis. 8. The author might present more rationale on selecting the Siglec-F for downstream studies (e.g. in the CD33-related Siglec gene family). How to prioritize the identified DE phosphoproteins relevant to AD pathogenesis? The Siglec gene family has many members including CD33. Are other Siglec proteins also found to be changed in the analysis? Response: We identified Siglec-F pY561 as being one of two phosphosites that were consistently upregulated across all diseased models and tissues. We selected Siglec-F for follow-up validation as it is a cell surface receptor that was predicted to be expressed on microglia and little was known at the time about its role in Alzheimer's disease. In our follow-up analysis of available single-cell RNA-seq data of microglia from CK-p25 and 5XFAD mice, we observed that Siglec-1, Cd22, and Siglec-F appear to be upregulated at the transcript level in Lpl + / Spp1 + late-response inflammatory microglia ( Figure EV5A). In 5XFAD animals, Cd22, and Siglec-F are upregulated at the transcript level in the Lpl + / Spp1 + disease-associated microglia population (Figure EV5B). 9. This reviewer appreciates the thorough investigation of Siglec-F function in the cellular models. One concern is that the cellular model experiments generated largely correlative data. What is the effect of Siglec-F knockout in the cells during interferon gamma stimulation? Do multiple Siglec family proteins play a redundant role in the pathway? Response: While we do see reduced growth rates in BV-2 cells that are treated with interferon gamma, we do not know if this is mediated by Siglec-F. Due to time constraints and COVID-19 lab shutdowns, we did not test the effects of Siglec-F / Siglec-8 knockout in the context of interferon gamma simulation. With regards to Siglec receptor redundancy, in mice, it appears that only Cd22 and Siglec-F are upregulated in the late response inflammatory microglia seen in a scRNAseq analysis of CK-p25 mice. Cd22 and Siglec-F also appear to be upregulated in the disease-associated microglia population identified in 5XFAD mice. One additional point is that within the Siglec family, Siglec-8 and Siglec-F both uniquely bind 6'-sulfo Sialyl Lewis X. Siglec-F has a slightly more permissive binding pattern that also includes Sialyl Lewis X. This suggests that Siglec-F and Siglec-8 bind to similar substrates in the brain, although the more permissive binding of Siglec-F may indicate that it plays a more general role in mice compared to Siglec-8 in humans.
1st Oct 2020 2nd Editorial Decision Thank you for sending us your revised manuscript . We think that the performed revisions sat isfact orily address the issues raised by the reviewers. I am glad to inform you that we can soon formally accept your manuscript for publicat ion, pending some edit orial issues list ed below.

2nd Authors' Response to Reviewers 2nd Oct 2020
The Authors have made the requested editorial changes.

Accepted 6th Oct 2020
Thank you again for sending us your revised manuscript . We are now sat isfied wit h the modificat ions made and I am pleased to inform you that your paper has been accept ed for publicat ion. Selected TMT channels for mass spectrometry experiments were excluded from downstream analysis due to the need for excessive channel normalization. The runs with excluded channels are: p35-X2: TMT 128N, 129C; FAD-X2: TMT 128N, 128C.
Phenotype measurements were similarly performed with at least n=3 experimental replicates to determine significance. In each case, the number of samples was chosen to allow for significance values to be calculated while minimizing the reagent and instrument time costs.
Cell culture experiments were seeded from the same initial stock of cells, therefore randomization in treatment groups among cell culture dishes was not necessary.

Manuscript Number: MSB-20-9819
For every relevant figure the statistical test was indicated.
Animals were grouped by genotype based on animal availablity. No additional treatments were performed on animals prior to sacrifice that required randomization. TMT channel assignment for animal proteomics experiments was not randomized.
Investigators were not blinded to group allocations within analyses, though is not relevant for this study given the automated nature of data collection and analyses performed. For mass spectrometry data analysis, filtering criteria were established prior to analysis. Similarly, RNAsequencing analysis follows a standard protocol established prior to data collection. No blinding was necessary analyzing the phenotype data, as all data was collected using plate readers and automated imaging systems. For flow cytometry analysis, the same set of gates defined using an untreated control sample of the same cell line was applied to all samples within that cell line and treatment conditions.
No blinding was performed for animal proteomics and immunofluorescence experiments.

Data
the data were obtained and processed according to the field's best practice and are presented to reflect the results of the experiments in an accurate and unbiased manner. figure panels include only data points, measurements or observations that can be compared to each other in a scientifically meaningful way.
The data shown in figures should satisfy the following conditions: Source Data should be included to report the data underlying graphs. Please follow the guidelines set out in the author ship guidelines on Data Presentation.
Please fill out these boxes ê (Do not worry if you cannot see all your text once you press return) a specification of the experimental system investigated (eg cell line, species name).
Sample sizes for quantitative mass spectrometry experiments were chosen as n=3 and n=4-5 for each condition to assess statistical significance within the capacity of TMT 6-plex and 10-plex isobaric mass tag labeling reagents.
For RNA-sequencing analysis, n=5-6 was also chosen to assess statistical significance between conditions. graphs include clearly labeled error bars for independent experiments and sample sizes. Unless justified, error bars should not be shown for technical replicates. if n< 5, the individual data points from each experiment should be plotted and any statistical test employed should be justified the exact sample size (n) for each experimental group/condition, given as a number, not a range; Each figure caption should contain the following information, for each panel where they are relevant:

B-Statistics and general methods
the assay(s) and method(s) used to carry out the reported observations and measurements an explicit mention of the biological and chemical entity(ies) that are being measured. an explicit mention of the biological and chemical entity(ies) that are altered/varied/perturbed in a controlled manner.
a statement of how many times the experiment shown was independently replicated in the laboratory.
Any descriptions too long for the figure legend should be included in the methods section and/or with the source data.
In the pink boxes below, please ensure that the answers to the following questions are reported in the manuscript itself. Every question should be answered. If the question is not relevant to your research, please write NA (non applicable). We encourage you to include a specific subsection in the methods section for statistics, reagents, animal models and human subjects.
definitions of statistical methods and measures: a description of the sample collection allowing the reader to understand whether the samples represent technical or biological replicates (including how many animals, litters, cultures, etc.).