Author response:
The following is the authors’ response to the original reviews.
eLife Assessment
The nematode C. elegans is an ideal model in which to achieve the ambitious goal of a genome-wide atlas of protein expression and localization. In this paper, the authors explore the utility of a new and efficient method for labeling proteins with fluorescent tags, evaluating its potential to be the basis for a larger, genome-wide effort that is likely to be very useful for the community. While the evidence for the method itself is solid, carrying out this project at a large scale will require significant additional feasibility studies.
We appreciate the editor’s recognition that the evidence for our method is solid and that a genome-wide protein atlas in C. elegans would be highly valuable to the community. However, we respectfully disagree that “significant additional feasibility studies” are required. Take the yeast proteome-wide GFP tagging project (Huh et al., Nature 2003). It achieved ~75% coverage of ~6,000 proteins directly from an established protocol without any prior significant feasibility studies, at least to our knowledge. While the C. elegans genome is 3 times in size, we would argue that our tagging protocol may even be less labor intensive as it does not involve any cloning and the screening is visual, requiring no molecular biology skills. Reviewer 3 notes: ‘They also provide convincing evidence that labelling the whole proteome is an achievable goal with relatively limited resources and time.’
Our pilot study validates all key parameters for genome-wide scaling: editing efficiency at novel loci with untested reagents, viability of tagged worms, and detectability of multiple spectrally separated fluorophores across expression ranges. These address the core technical, biological, and practical challenges of large-scale endogenous tagging in a multicellular organism, leaving no fundamental barriers in our view.
The proposed cost and timeline align quite favorably with established large-scale consortium projects: e.g., ENCODE pilot analyzed 1% of the human genome at ~$55 million over 4 years; Mouse Knockout Consortium scaled to ~20,000 genes over 20 years (ongoing) with ~$100 million; Human Protein Atlas mapped ~87% of proteins with antibodies in fixed cells (through much more labor intensive methods) over 20+ years at >$100 million. With ~8% of C. elegans genes already tagged (WormTagDB) and labs already tagging entire gene classes (PMID: 40463100), scaling our protocol to the proteome is feasible, potentially covering the genome in 5-6 years by a single lab or faster with distributed effort at a reagent cost of merely $2.2 million. The main barriers now are funding commitment and assembling collaborators, not further feasibility testing.
Public Reviews:
Reviewer #1 (Public review):
Summary:
Eroglu and Hobert demonstrate that injecting CRISPR guides and repair constructs to target three genes at a time, tagging each with a different fluorescent protein, and selecting which gene to tag with which fluorophore based on genes' expression levels, can improve the efficiency of gene tagging.
Strengths:
This manuscript demonstrates that three genes can be targeted efficiently with three different fluorophores. It also presents some practical considerations, like using the fluorophore least complicated by agar/worm autofluorescence for genes with low expression levels, and cost calculations if the same methods were used on all genes.
Weaknesses:
Eroglu has demonstrated in a previous publication that single-stranded DNA injection can increase the efficiency of CRISPR in C. elegans while inserting two fluorescent proteins and a co-CRISPR marker into three loci. The current work is, therefore, an incremental advance. In general, I applaud the authors' willingness to think ahead to how whole proteome tagging might be accomplished, but I predict that the advance here will be one of many small advances that will get the field to that goal.
Our manuscript indeed builds on prior multiplex editing (including our own co-CRISPR work), but the manuscript's primary contribution is not a novel technical breakthrough per se. Instead, our main goal was to pilot and strategize a feasible path to whole-proteome tagging in C. elegans and, most critically, test the following key parameters: (1) success rate of triple pools with prior untested reagents at novel targets; (2) utility of fluorophores across expression levels; (3) major effects on tagged protein function. In prior multiplexing, we used two targets which we already knew could be edited quite efficiently, with the 3rd target a point mutation with nearly 100% efficiency. Thus, it was not at all clear that picking 3 random genes and replacing the 3rd highly efficient locus with another less efficient large insertion would work or be sufficiently scalable for thousands of novel genes with unvalidated reagents at first pass.
The title vastly oversells the advance in my view, and the first sentence of the Discussion seems a more apt summary of the key advance here.
Some injections target genes on the same chromosome together, which will create unnecessary issues when doing necessary backcrossing, especially if the mutation rate is increased by CRISPR.
We disagree with the reviewer’s assessment of the need for backcrossing, for two reasons: (1) Prior studies have shown that off-target mutations are not a serious concern in C. elegans (reviewed in PMID: 26336798). For instance, WGS of strains after CRISPR/Cas9 found negligible off-target effects (PMID: 25249454, PMID: 30420468 – using similar RNP/ssDNA method and multiple guides; PMID: 23979577, PMID: 27650892 using other methods). Targeted sequencing studies have reported similar findings, using various CRISPR/Cas9 methods, with essentially no mutations at sites other than the intended target (PMID: 23995389; PMID: 23817069). (2) If the goal is to tag the entire genome, the introduction of backcrossing should not reasonably be a routine part of the initial tagging.
Lastly, if one really does want to backcross, the existence of tags on the same chromosome is actually an advantage because it permits selection for recombinants with wild-type chromosomes.
Also, the need for backcrossing and perhaps sequencing made me wonder if injecting 3 together really is helpful vs targeting each gene separately, since only 5 worms need to be injected.
Apart from our disagreement regarding backcrossing, we are puzzled by the reviewer’s comment. Why would one do single tagging at a time, rather than triple tagging if the whole point is to scale up tagging? It is important to keep in mind that the rate limiting step for tagging the whole genome is the number of injections that can be done per day. Since there is no cloning to generate the repair templates/guides and all other reagents are commercially available and not sample specific, these can be prepared quite rapidly. Being able to isolate multiple lines (together or independently) from the same injection increases throughput 3-fold and in our view does not provide any disadvantages as individual tags can be isolated independently if desired.
Beyond the numerous technical advantages pooling provides (also lower cost and throughput for making injection mixes as well as imaging), our results show that it yields epistemic benefits as well: we would never have noted the subcellular pattern in Fig. 6B, C with different sets of mitochondria being marked by different mitochondrial proteins had we imaged them separately or even aligned to a pan-mitochondrial landmark. As we mentioned in the discussion, grouping proteins predicted to localize to the same compartment together can simultaneously test how uniform or differentiated such compartments are during the screen.
The limited utility of current blue fluorescent proteins makes me wonder if it's worth using at all at this stage, before there are better blue (or far red) fluorescent proteins.
We do not think that the utility of current BFPs is that limiting. At least the theoretical brightness of mTagBFP2 is comparable to that of EGFP (PMID: 30886412), which was useful for the bulk of currently tagged proteins. Due to modestly higher autofluorescence in the blue spectrum, the practical brightness is somewhat less ideal, but we have shown that many proteins are expressed high enough to be detected quite well with mTagBFP2 by eye at low magnification. We also note that many tags that are not visible by eye under a dissection scope become visible with long exposure cameras of widefield microscopes or modern confocal (GaAsP) detectors, so the list of genes detectable with mTagBFP2 is likely to be much higher. We routinely use mTagBFP2 to super-resolve subnuclear structures with endogenous tags (e.g., in the nucleolus), with some tags having lower annotated FPKMs than the genes tested here.
Some literature reviews, particularly in the Introduction and Abstract, rely too much on recent examples from the authors' laboratory instead of presenting the state of the field. I'd like to have known what exactly has been done with simultaneous injection targeting multiple loci more thoroughly, comparing what has been accomplished to date by various laboratories' advances to date.
We are not sure what the reviewer is referring to. In the Abstract, we do not refer to any literature. In the Introduction, we cite 28 papers, 6 of those from our lab (4 of which providing examples of protein tags). We do not believe that this can be fairly called an unbalanced presentation of the state of the field.
This being said, we have gladly expanded our Introduction to provide more background on co-CRISPRing. Labs have routinely used co-conversion (“coCRISPR”) markers for picking out their intended edits (e.g., point mutations or insertions), as it has been shown by multiple groups that a CRISPR/Cas9 edit at one locus correlates with efficiency at other simultaneous targets (PMID: 25161212). Generally, making point mutations with the Cas9/RNP protocol is highly efficient, especially at specific loci such as dpy-10. However, multiple FP-sized insertions have not been routinely attempted. We and only one other group have successfully attempted it using previously working targets and reagents (e.g., 28% in PMID: 26187122). Importantly, the efficiency of such multiple insertions has never been assessed at scale and using entirely untested reagents at novel sites – critical parameters to determine for a whole genome approach. So, we test here (1) the efficiency of triple insertions and (2) the chance of getting them with new and untested guides and reagents.
In our view, since we have to use some injection/coCRISPR marker anyway for those genes which are not expressed at dissecting-scope visible levels (likely most genes), using highly expressed intended targets as improvised markers in a pooled approach makes our approach much more efficient. It allows us to find the worms with the highest chance of yielding CRISPR insertions, which we can screen with higher power methods for the dimmer targets, while enabling us to co-isolate other intended targets. Insertions, being often heterozygous in F1, can be segregated independently if desired, or homozygosed together to facilitate maintenance then outcrossed individually by those interested in studying specific genes in more detail.
In the revised version of this manuscript, we now discuss some of these points in the introduction section:
“Currently, around 1554 proteins representing 8% of the proteome are estimated to have been endogenously tagged (Leyhr et al., 2025). However, at current rates, tagging the proteome is projected to take around 100 years and likely involve numerous duplicate attempts on a small number of commonly studied proteins (Leyhr et al., 2025). It will thus be crucial for the field to coordinate tagging efforts and scale up tagging protocols to enable coverage of the entire genome at a reasonable timescale and cost. Given the number of injections is a major time-limiting factor, pooling multiple injections into one would at minimum cut tagging time by a factor of 3. In C. elegans, screening for novel CRISPR/Cas9-induced genomic edits is already facilitated either by use of co-injection markers (i.e., plasmids that form extrachromosomal arrays) that yield phenotypes or fluorescence in progeny of successfully injected worms, or co-editing well characterized loci using established and highly efficient reagents which likewise yield visible phenotypes. In the latter approach, termed “co-CRISPR”, worms edited at the marker locus are most likely to also carry the intended edit (Arribere et al., 2014). Recent methods for CRISPR/Cas9 mediated genomic insertions have pushed efficiencies to sufficient levels to simultaneously insert multiple fluorophores (e.g., mNeonGreen and mScarlet) as well as a co-CRISPR marker (dpy-10) at three independent loci in a single injection (Eroglu et al., 2023; Paix et al., 2015). These attempts pooled reagents previously established to work efficiently and targeted genes that were known to yield functional fusion proteins when tagged. Thus, while in principle current methods could allow tagging of at least 3 independent loci in one injection if a co-CRISPR marker is omitted, it is not known to what extent such an approach could be generalized across the genome with previously unvalidated reagents (i.e., guides and repair template homology arms) at novel loci to yield functional tags”
Reviewer #2 (Public review):
The manuscript by Eroglu and Hobert presents a set of strains each harboring up to three fluorescently tagged endogenous proteins. While there is technically nothing wrong with the method and the images are beautiful, we struggled to appreciate the advance of this work - who is this paper for?
We consider this paper to have two purposes: (1) motivate the community to come together to consider such genome-wide tagging approach; (2) provide a reference point for funding agencies that such an aim is not unreasonable and will provide novel interesting insights.
As a technical method, the advance is minimal since the first author had already demonstrated that three mutations (fluorophore insertion and co-CRISPR marker) could be introduced simultaneously.
We agree that the basic principle is similar. However, it was not clear that triple pooling three novel large edits would work, given the numbers in our original paper or that it would be scalable.
The dpy-10 coCRISPR marker previously used is a highly efficient single site, with close to 100% hit rate. We also knew in the earlier study that the two pooled insertions already worked quite efficiently and did not disrupt the function of targeted proteins. Exchanging these plus dpy-10 for three novel tags was not guaranteed to succeed for many potential reasons, including both biological and technical. For instance, such a “marker free” approach necessitates that a significant number of targets in the genome should be expressed highly enough to be visible by fluorescence stereomicroscopy when tagged with current best fluorophores. The chance of disrupting gene function by tagging was also not explored in detail in C. elegans, nor whether one untested guide is generally sufficient. We think that establishing these parameters was meaningful and necessary for the goal of whole genome tagging. We have clarified some of these points in the text.
As a pilot for creating genome-scale resources, it is not clear whether three different fluorophores in one animal, while elegantly designed and implemented, will be desired by the broader community.
The usage of three different fluorophores is largely driven by the ability to co-inject and therefore cut injection effort by a factor of three. Moreover, having all three fluorophores together facilitates imaging and maintenance. Lastly, co-labeling has the potential to reveal unexpected patterns of co-localization or lack thereof (example: two mitochondrial proteins that we found to not have overlapping distribution). We clarified this point in the revised text in both the results and discussion.
Finally, the interpretation of the patterns observed in the created lines is somewhat lacking. A Table with all the observations must be included. This can replace the descriptions of the observations with the different lines, which could be somewhat laborious for the reader, and are often wrong. There are numerous mistaken expectations of protein expression here, but two examples include:
We are not convinced that our expectations are mistaken. Below we respond to the reviewer’s specific examples, and we are open to hear from the reviewer about additional cases.
(1) The expectation that ACDH-10 is enriched in the intestine and epidermal tissues (hypodermis).
There are multiple paralogs of this protein (see WormPaths or WormFlux) that may share functions in different tissues. There is also no reason to assume that fatty acid metabolism does not occur in other tissues (including the germline). Finally, there are no published studies about this enzyme, so we really don't know for sure what it's doing.
The expression of acdh-10 is annotated in multiple scRNA datasets as intestine and epidermal enriched (CeNGEN/Taylor et al. 2021, highest in epidermis; Ghaddar et al 2023 highest in intestine). We did not mean to imply that fatty acid metabolism does not occur in the gonad, nor that a paralog of acdh-10 could not be performing the same function in tissues where acdh-10 is not expressed.
However, this raises an important question: why have different paralogs doing the same thing? Duplicate genes with the same function are generally not evolutionarily stable (PMID: 11073452, PMID: 24659815). That there are such striking tissue specific expression patterns of an essential or widely expressed protein class suggests that paralogs of the gene likely differ in some meaningful parameter that might align with tissue-specific functional needs or regulation. The reviewer’s statement that ‘there are no published studies about this enzyme, so we really don't know for sure what it's doing’ is in fact an excellent demonstration of our point; finding out where the duplicates are expressed can provide a starting point to uncover potential differences between the paralogs. At the very least it can delineate to what degree paralogs diverge in their expression across the proteome and identify which such cases merit further study. In a more ideal scenario, prior information of protein function could indicate that the involved pathway requires tissue specific regulation.
(2) The expectation that HXK-1 is ubiquitously expressed.
Three paralogous enzymes are all associated with the same reaction, and we have shown that these three function redundantly in vivo, perhaps in different tissues (PMID: 40011787).
The cited paper (PMID: 40011787) does not show where they are expressed. We discussed redundancy/paralogs above in point 1, and in our view the same applies here. They may perform the same reaction but are likely to differ in some meaningful way, be it regulation or rate of activity, for them to be stably maintained as functional genes over evolution.
Moreover, single-cell RNA-seq data (PMID: 38816550) also show enrichment of hxk-1 in gonadal sheath cells.
The Ghaddar et al. and CeNGEN/Taylor et al. datasets do not show this. The scRNA paper cited (PMID: 38816550) also shows enrichment in neurons, pharynx, coelomocyte and germ cells which we did not note. In our view, these in fact further support our goals: often, transcript datasets alone (frequently used to infer tissue function) do not sufficiently predict protein expression. One can post hoc find an scRNA-seq dataset that aligns somewhat with our protein observations, but how does one know which to trust a priori? Disagreements between transcript datasets will ultimately require resolution at the protein level, in our view.
To clarify these points, we added the following to the discussion section:
“We also noted unexpected cell type dependent distributions of proteins involved in broadly important metabolic processes such as ACDH-10, which was depleted from the germline compared to other tissues, and HXK-1, which was highly enriched in the gonadal sheath. Notably, for these as well as other cases, scRNA-seq datasets were not sufficient to deduce a priori the observed cell type specific differences at the protein level. Importantly, many genes encoding metabolic enzymes including acdh-10 and hxk-1 have paralogs that likely perform similar catalytic functions. Yet, duplicate genes with identical functions are generally not evolutionarily stable (Adler et al., 2014; Lynch and Conery, 2000); thus such genes are likely to differ in some meaningful parameter (e.g., regulation or activity) that might align with tissue-specific functional needs. Fully annotating the expression patterns of paralogs at the protein level could indicate which tissues require unique metabolic needs and indicate which paralogous genes have undergone sub- versus neo-functionalization. For those proteins that are less functionally understood, unexpected distributions might indicate which merit further study.”
The table should have at least the following information: gene/protein name - Wormbase ID - TPM levels of single cell data assigned to tissues for L2, L4, and adult (all published) - tissues in which expression is observed in the lines presented by the authors.
We added some of this information such as annotated expression levels in young adults from various scRNA datasets (but not larval datasets as we did not image these). We note that each of these studies use different pipelines and report different metrics (scaled TPM/Z-score versus Seurat average expression versus TPM), so comparisons between them are not informative unless they are integrated and analyzed together.
Reviewer #3 (Public review):
Summary:
The authors argue that establishing the expression pattern and subcellular localisation of an animal's proteome will highlight many hypotheses for further study. To make this point and show feasibility, they developed a pipeline to knock in DNA encoding fluorescent tags into C. elegans genes.
Strengths:
The authors effectively make the points above. For example, they provide evidence of two populations of mitochondria in the C. elegans germline that differ qualitatively in the proteins they express. They also provide convincing evidence that labelling the whole proteome is an achievable goal with relatively limited resources and time.
We appreciate the referee’s recognition that whole proteome tagging is feasible.
Weaknesses:
Cell biology in C. elegans is challenging because of the small size of many of its cells, notably neurons. This can make establishing the sub-cellular localisation of a fluorescently tagged protein, or co-localizing it with another protein, tricky. The authors point out in their introduction that advances in light microscopy, such as diSPIM, STED, and ISM (a close relative of SIM), have increased the resolution of light microscopy. They also point out that recent advances in expansion microscopy can similarly help overcome the resolution limit.
(1) Have the authors investigated if the three fluorescent tags they use are appropriate for super-resolution microscopy of C. elegans, e.g., STED or SIM? Would Elektra be better than mTAGBFP2? How does mScarlet3-S2 compare to mScarlet 3?
All three tags work for ISM (i.e., Airyscan). We previously tried Electra (not for the genes tested here) but could not isolate positive tags. Given Electra is not that much brighter on paper than mTagBFP2 we did not pursue it further, though we recognize that these may simply have been unlucky injections. mScarlet3-S2 is quite a bit dimmer than mScarlet3 on paper – the advantage is that it has higher photostability. In our view, the limiting factor will be having FPs that are bright enough to screen, image and scale to the whole genome, so brightness will likely provide an advantage over photostability at this stage.
(2) Have the authors investigated what tags could be used in expansion microscopy - that is, which retain antigenicity or even fluorescence after the protocol is applied? It may be useful to add different epitope tags to the knock-in cassettes for this purpose.
mSG and mSc3 retain fluorescence after fixing with formaldehyde. We have not tested mTagBFP2 fluorescence in fixed worms. We agree that adding different epitope tags would be useful.
The paper is fine as it stands. The experiments above could add value to it and future-proof it, but are not essential. If the experiments are not attempted, the authors could refer to the points above in the discussion.
Recommendations for the authors:
Reviewer #1 (Recommendations for the authors):
(1) Merged figures appear saturated, and use colors that won't work for red-green colorblind viewers.
For all figures, we also show individual channels separately, which is common practice for making fluorescence images accessible to colorblind readers (PMID: 33788834). Figures highlighting non-overlap like 6B and C are already in accessible colors when merged (blue/green) and include a numerical quantification. 3-color RGB images preserve the greatest information for the highest number of individuals.
(2) Targeting ubiquitously expressed genes as a proof of concept gives me some concern that this might underestimate the challenges that may be experienced with less widely expressed genes.
While the genes were predicted to be ubiquitously expressed, many were not in practice, like HXK-1 and F54C8.1, which were also among the lower expressed genes on our list and highly cell type restricted. As discussed, the more tissue restricted a gene, the likelier that bulk RNA levels underestimate expression. Such genes are therefore more likely to be detected in a specific tissue. We routinely isolate tissue restricted endogenous tags, including those expressed in only a few neurons, with bulk FPKMs lower than the ranges tested in this manuscript.
(3) Some results are not shown or referenced (autofluorescence, for example, is shown using a schematic in Figure 1C).
We now provide representative images alongside what would be expected to be observed by eye during screening.
(4) It would be useful to describe how to recover worms from what is shown in Figure 1A.
In the revised version, we added the following in the caption for Fig. 1A:
“Selected worms expressing the brighter tag can be screened for dimmer tags by higher magnification and long exposure imaging. Worms can be recovered directly from slides if immobilized by levamisole as described (Ghanta et al., 2021). Alternatively, single hermaphrodite worms can be isolated, allowed to lay eggs, then screened.”
(5) A blue bar of data must be missing from Figure 3B injection pool 5.
As stated in the text, “All but one tag (cox-6B::mTagBFP2) was visible in the F1 generation of injected P0 animals, and these were subsequently isolated among F2 worms positive for the other tags in the pool.”
To clarify that data points are not unintentionally omitted, we added the following text to the caption of Fig. 3B:
“For group 5 including cox-6B::mTagBFP2, worms with detectable levels of mTagBFP2 fluorescence were not recovered in the F1 generation but were isolated among progeny of F1s positive for mStayGold and mScarlet3; we were thus unable to quantify efficiency for this locus at F1.”
(6) Some expression or localization patterns were unexpected, but complications like germline silencing and protein mislocalization, with a small fraction localizing normally and rescuing function, were not presented as possibilities. Viability is used to confirm function, but without presenting whether this means 100% viability, less, or just the ability to maintain a strain.
We already do discuss mislocalization and functionality issues in the Discussion, as well as tradeoffs of alternate methods. Any existing method to observe biological molecules, be it protein, RNA or DNA, has multiple drawbacks and sources of artifacts, which are unlikely to be fully eliminated in the foreseeable future.
In regard to germline silencing of endogenously tagged genes in C. elegans, there is actually very little evidence for this. Collectively, various labs have now generated over 200 reporter alleles of germline-expressed genes (WormTagDB), with robust expression throughout the germline and retention of function. Likewise, numerous of our tags across fluorophores showed robust germline expressions including EEF-1A.1::mTagBFP2, Y22D7AL.10::mStayGold, and HAT-1::mScarlet3. In fact, overall transcript levels generally tended to underestimate germline enrichment at the protein level. We note that single-copy transgenes driven by eef-1A.1/eft-3 promoter by itself are frequently not expressed in the germline (PMID: 31064766); that we could detect EEF-1A.1 robustly in the germline when tagged endogenously is evidence that silencing is unlikely to be a widespread concern, and at the least less of a concern than single copy transgenes. We appreciate that for a transgene, presence/absence of specific sequence elements and genomic loci play a role in expression, but an endogenous tag captures all such information at a given locus.
Indeed, we found only two reports of endogenous tags being silenced in the germline, the first being a novel tag (not fluorophore) which initially prevented expression at the tagged locus (PMID: 30109984), but after making changes to the sequence to avoid silencing signals the authors could rescue expression and thereafter saw robust expression in various novel contexts with this tag. The second example (PMID: 34547227) leaves open the possibility that germline repression of that particular gene might be a part of its endogenous regulation.
Nevertheless, given it is probably rare if occurs at all, it will likely take a large scale tagging effort to uncover such cases at sufficient numbers to study. In our view, this further justifies tagging at large, ideally genomic, scales. If we do discover that there are numerous annotated germline proteins which we don’t observe by tagging, that would be interesting to study on its own.
(7) Halotag is presented in the Discussion as a small tag, but it is bigger than GFP.
Thank you for catching this. We have removed the discussion of Halotag. Given the comparable size to FPs, it would be unlikely to alleviate issues of tag functionality.
(8) It would be useful to include FPKMs and viability percentages in Table 1.
FPKM is included in column 6, but the title for this column is cut off. In the revised table FPKM values are now shown more clearly across stages.
We did not quantify viability percentage. In our view it does not yield an informative metric when there is little information about the protein’s required dosage for function, which was the case for most proteins here. A haplosufficient gene might yield a full brood size even if 50% of protein function is lost; conversely, a highly dose sensitive protein could yield penetrant and severe inviability with mild perturbation of function. It also is not actionable information at this stage if there is no alternate tagging strategy as a baseline of comparison. The worms we picked to image all have viable embryos as adults, so in those individuals the genes were likely to be sufficiently expressed and functional.
(9) Because establishing that a guide works well is a limiting step for many CRISPR experiments (once a guide works well, it's easy to inject 5 worms and get lines), I wondered if testing that for many genes is what is really needed in the field at this stage.
Guide quality is rarely an issue in C. elegans, as for all the genes here we tried only one guide, all of which were previously untested. We now clarified this in the discussion section:
“Notably, we find that previously untested guide RNAs and homology arms perform exceptionally well at novel loci, as we only tested one set of reagents for each locus which yielded satisfactory tagging rates.”
(10) For a manuscript where the injection is so central to what was done, I was surprised to read in the Acknowledgments that all of the injections were done by someone who is not included as an author.
We are likewise surprised by such a comment but gladly clarify: Chi Chen has been with us as an expert microinjection specialist for more than 25 years and her very important technical contributions have been acknowledged in many dozen papers. Multiple authorship guidelines, including COPE’s and ICMJE’s, state that technical contributions alone do not qualify for authorship.
Reviewer #2 (Recommendations for the authors):
(1) We would encourage the authors to provide systematic validation of the reported insertions. The manuscript reports that 24 of 30 tags were isolated and visible, but does not clearly state whether each isolated line was confirmed by sequence‑level validation to be correctly in‑frame and free of unintended mutations at the target locus.
We appreciate the reviewer’s concerns on fidelity. These parameters have been assessed in prior published work (e.g., PMID: 30504364, PMID: 34748534) and in our hands are in the range of 80% whenever we sequence non-fluorescent tags of similar sizes. The efficiencies we observed are high enough that one can expect to recover numerous worms with the exact intended sequence for each target, though we would argue mutations within the FP reporter are less likely to matter if it retains high fluorescence.
(2) The manuscript presents aggregated success counts (e.g., 8/10 mTagBFP2 tags, 9/10 mStayGold, 7/10 mScarlet3) and useful narrative descriptions of injection outcomes. We also suggest including per‑locus success rates.
Figure 3B shows per locus success rate and source data is provided for this figure. Each dot is an individual injection and the Y axis is per locus rate. We now worded this more clearly in the figure’s caption.
“Total insertion efficiencies per locus for the indicated targets across injection pools.”
(3) For pools that required re‑injection after initial failures, we would like to see a description of the specific changes that were made to the injection mixes or procedures (e.g., new repair template prep, different Cas9 reagent lot, guide redesign). This will be useful troubleshooting information for others.
We re-made the exact same injection mix but with nanodrop to ensure the purity of the repair templates as assessed by absorbance ratios (A260/230 and A260/280) were sufficient after each purification step. No other changes were made. This is now specified in the methods section in the following way:
“For re-runs of pools 4, 6 and 10 which failed initially, we regenerated the repair templates and ensured that after each column purification, the A260/230 ratio of the purified DNA was ≥2.2 and A260/280 was 1.8 ± 0.05 when measured with a Nanodrop spectrophotometer.”
(4) The authors state that the fluorophore sequences are codon-optimized for C. elegans. We suggest they provide the exact donor/tag sequences, specifically state whether the fluorophore sequences contain any synthetic/artificial introns, or whether other sequence modifications (e.g., silent PAM‑disrupting mutations) were included in the donor templates.
This information is provided in Supplementary Table 1.
(5) Page 3: Include a reference for "The C. elegans genome encodes around 20,000 genes"
We added a reference to the most recent release of the genome (WS237, May 2013). Spieth et al., 2014.
