A novel high-throughput single-cell DNA sequencing method reveals hidden genomic heterogeneity in the unicellular eukaryote Leishmania

  1. Gabriel H Negreira
  2. Pieter Monsieurs
  3. Jean-Claude Dujardin
  4. Malgorzata A Domagalska

  • Curated by eLife

    eLife logo

    eLife Assessment

    This important study tackles the challenge of analyzing genome integrity and instability in unicellular pathogens by introducing a novel single-cell genomics approach. It presents compelling evidence that this new tool outperforms standard whole-genome amplification techniques and describes a method that shows promise for probing questions regarding the biology of Leishmania and other microbial eukaryotes.

Reviewed by

Read the full article

Listed in

Log in to save this article

Abstract

Genome instability is considered a major driver of adaptation in eukaryotic microorganisms, but its study is hampered by the limited availability of genomic technologies with single-cell resolution. Here we present a novel high throughput method to reconstruct both structural and nucleotide information at single-cell level in Leishmania, a protozoan parasite with a remarkable genome plasticity characterized by frequent gene copy number variations (CNVs) and high aneuploidy mosaicism. By combining the use of semi-permeable capsules with primary template-directed amplification, we could determine the karyotypes of hundreds of Leishmania parasites, detect distinct CNVs between different cell populations, and identify sub population of cells harboring distinct nucleotide variants, including in genes associated with drug resistance. This approach provides a powerful new framework to uncover hidden evolutionary potential in complex microbial populations, with application in studying adaptation, drug resistance, and genome evolution in Leishmania and other pathogens.

Article activity feed

  1. Version published to 10.7554/elife.109350.2 on eLife
  2. Version published to 10.7554/elife.109350 on eLife
  3. eLife

    eLife Assessment

    This important study tackles the challenge of analyzing genome integrity and instability in unicellular pathogens by introducing a novel single-cell genomics approach. It presents compelling evidence that this new tool outperforms standard whole-genome amplification techniques and describes a method that shows promise for probing questions regarding the biology of Leishmania and other microbial eukaryotes.

  4. eLife

    Reviewer #1 (Public review):

    [Editors' note: this version has been assessed by the Reviewing Editor without further input from the original reviewers. The authors have addressed the comments raised in the previous round of review.]

    Summary:

    Negreira, G. et al clearly presented the challenges of conducting genomic studies in unicellular pathogens and of addressing questions related to the balance between genome integrity and instability, pivotal for survival under the stressful conditions these organisms face and for their evolutionary success. This underlies the need for powerful approaches to perform single-cell DNA analyses suited to the small and plastic Leishmania genome. Accordingly, their goal was to develop such a novel method and demonstrate its robustness.

    In this study, the authors combined semi-permeable capsules (SPCs) with primary template-directed amplification (PTA) and adapted the system to the Leishmania genome, which is about 100 times smaller than the human genome and exhibits remarkable plasticity and mosaic aneuploidy. Given the size and organization of the Leishmania genome, the challenges were substantial; nevertheless, the authors successfully demonstrated that PTA not only works for Leishmania but also represents a significantly improved whole-genome amplification (WGA) method compared with standard approaches. They showed that SPCs provide a superior alternative for cell encapsulation, increasing throughput. The methodology enabled high-resolution karyotyping and the detection of fine-scale copy number variations (CNVs) at the single-cell level. Furthermore, it allowed discrimination between genotypically distinct cells within mixed populations.

    Strengths:

    This is a high-impact study that will likely contribute to our understanding of DNA replication and the genetic plasticity of Leishmania, including its well-documented aneuploidy, somy variations, CNVs, and SNPs - all key elements for elucidating various aspects of the parasite's biology, such as genome evolution, genetic exchange, and mechanisms of drug resistance.

    Overall, the authors clearly achieved their objectives, providing a solid rationale for the study and demonstrating how this approach can advance the investigation of Leishmania's small, plastic genome and its frequent natural strain mixtures within hosts. This methodology may also prove valuable for genomic studies of other single-celled organisms.

  5. eLife

    Reviewer #2 (Public review):

    Summary:

    Negreira et al. present an application of a novel single-cell genomics approach to investigate the genetic heterogeneity of Leishmania parasites. Leishmania, while also representing a major global disease with hundreds of thousands of cases annually, serves as a model to test the rigor of the sequencing strategy. Its complex karyotypic nature necessitates a method that is capable of resolving natural variation to better understand genome dynamics. Importantly, an earlier single-cell genomics platform (10x Chromium) is no longer available, and new methods need to be evaluated to fill in this gap.

    The study was designed to evaluate whether a capsule-based cell capture method combined with primary template-directed amplification (PTA) could maintain levels of genomic heterogeneity represented in an equal mixture of two Leishmania strains. This was a high bar, given the relatively small protozoan genome and prior studies that showed limitations of single-cell genomics, especially for gene-level copy number changes. Overall, the study found that semi-permeable capsules (SPC) are an effective way to isolate high-quality single cells. Additionally, short reads from amplified genomes effectively maintained the relative levels of variation in the two strains on the chromosome, gene copy, and individual base level. Thus, this method will be useful to evaluate adaptive strategies of Leishmania. Many researchers will also refer to these studies to set up SPC collection and PTA methods for their organism of choice.

    Strengths:

    (1) The use of SPC and PTA in a non-bacterial organism is novel. The study displays the utility of these methods to isolate and amplify single genomes to a level that can be sequenced, despite being a motile organism with a GC-rich genome.

    (2) The authors clearly outlined their optimization strategy and provided numerous quality-control metrics that inspire confidence in the success of achieving even chromosomal coverage relative to ploidy.

    (3) The use of two distinct Leishmania strains with known clonal status provided strong evidence that PTA-based amplification could reflect genome differences and displayed the utility of the method for studies of rare genotypes.

    (4) Evaluating the SPCs pre- and post-amplification with microscopy is a practical and robust way of determining the success of SPC formation and PTA.

    (5) The authors show that the PTA-based approach easily resolved major genotypic ploidy in agreement with a prior 10x Chromium-based study. The new method had improved resolution of drug resistance genotypes in the form of both copy-number variations and single-nucleotide polymorphisms.

    (6) In general, the authors are very thorough in describing the methods, including those used to optimize PTA lysis and amplification steps (fresh vs frozen cells, naked DNA vs sorted cells, etc). This demonstrates a depth of knowledge about the procedure and leaves few unanswered questions.

    (7) The custom, multifaceted, computational assessment of coverage evenness is a major strength of the study and demonstrates that the authors acknowledge potential computational factors that could impact the analysis.

  6. eLife

    Reviewer #3 (Public review):

    In this manuscript, Negreira et al. propose a new scDNAseq method, using semi-permeable capsules (SPCs) and primary template-directed amplification (PTA). The authors optimize several metrics to improve their predictions, such as determining GC bias, Intra-Chromosomal fluctuation (ICF -metric to differentiate replicative and non-replicative cells) and Intra-chromosomal coefficient of variation (ICCV - chromosome read distribution). The coverage evenness was evaluated using the fini index and the median absolute pairwise difference between the counts of two consecutive bins. They validate the proposed method using two Leishmania donovani strains isolated from different countries, BPK081 (low genomic variability) and HU3 (high genomic variability). Then, they showed that the method outperforms WGA and has similar accuracy to the discontinued 10X-scDNA (10X Genomics), further improving on short CNV identification. The authors also show that the method can identify somy variations, insertions/deletions and SNP variations across cells. This is a timely and very relevant work that has a wide applicability in copy number variation assessment using single-cell data.

    I really appreciate this work. My congratulations to the authors.

  7. eLife

    Author response:

    The following is the authors’ response to the original reviews.

    Reviewer #1 (Public review):

    Summary:

    Negreira, G. et al clearly presented the challenges of conducting genomic studies in unicellular pathogens and of addressing questions related to the balance between genome integrity and instability, pivotal for survival under the stressful conditions these organisms face and for their evolutionary success. This underlies the need for powerful approaches to perform single-cell DNA analyses suited to the small and plastic Leishmania genome. Accordingly, their goal was to develop such a novel method and demonstrate its robustness.

    In this study, the authors combined semi-permeable capsules (SPCs) with primary template-directed amplification (PTA) and adapted the system to the Leishmania genome, which is about 100 times smaller than the human genome and exhibits remarkable plasticity and mosaic aneuploidy. Given the size and organization of the Leishmania genome, the challenges were substantial; nevertheless, the authors successfully demonstrated that PTA not only works for Leishmania but also represents a significantly improved whole-genome amplification (WGA) method compared with standard approaches. They showed that SPCs provide a superior alternative for cell encapsulation, increasing throughput. The methodology enabled high-resolution karyotyping and the detection of fine-scale copy number variations (CNVs) at the singlecell level. Furthermore, it allowed discrimination between genotypically distinct cells within mixed populations.

    Strengths:

    (1) This is a high-impact study that will likely contribute to our understanding of DNA replication and the genetic plasticity of Leishmania, including its welldocumented aneuploidy, somy variations, CNVs, and SNPs - all key elements for elucidating various aspects of the parasite's biology, such as genome evolution, genetic exchange, and mechanisms of drug resistance.

    Overall, the authors clearly achieved their objectives, providing a solid rationale for the study and demonstrating how this approach can advance the investigation of Leishmania's small, plastic genome and its frequent natural strain mixtures within hosts. This methodology may also prove valuable for genomic studies of other single-celled organisms.

    We thank the reviewer for the positive feedback and appreciation of the potential applications for the methodology we describe here.

    Weaknesses:

    (2) The discussion section could be enriched to help readers understand the significance of the work, for instance, by more clearly pointing out the obstacles to a better understanding of DNA replication in Leishmania. Or else, when they discuss the results obtained at the level of nucleotide information and the relevance of being able to compare, in their case, the two strains, they could refer to the implications of this level of precision to those studying clonal strains or field isolates, drug resistance or virulence in a more detailed way.

    We thank the reviewer for the suggestions. Indeed, single-cell DNA sequencing has successfully revealed cell-to-cell variability in replication timing and fork progression in mammalian cells [1,2] and we believe that the SPC-PTA workflow could be used in similar studies in Leishmania to complement bulk-based observations [3,4]. We have included a discussion about this in the revised version:

    “[…]scDNAseq has been used to study DNA replication programs, revealing fork progression speeds and replication timing along the cell cycle in mammalian cells [67,68]. Unlike most eukaryotes, Leishmania has a unique DNA replication program based on a single large replication origin per chromosome which is highly active during S-phase and which is complemented by thousands of minor low-efficiency origin sites that are active during G1 and G2/M phases [69,70]. However, it is still unclear if there is cell-to-cell variability in origin use and replication timing, and how this may link to genome instability, which could be addressed using the SPCPTA workflow in parasites at different stages of the cell cycle.”

    Regarding nucleotide information, it is indeed of high relevance to detect minor circulating variants with potential virulence impact and/or effect on drug resistance which could be missed by bulk sequencing. This includes the ability to detect co-occurring variants with potential epistatic effects. We have extensively expanded the discussion section in the “heterogeneity of nucleotide sequence” section to discuss these topics:

    “[…] Moreover, population-genomic studies of the L. donovani complex have identified isolates with allele-frequency patterns consistent with mixtures of genetically similar subclones, including mixtures within the same phylogenetic group [54]. Mixed-genotype infections are likely under-detected by routine diagnostics [55], and confounds bulk genomic analyses. This was evident when cultured parasites showed genomic signatures distinct from their clinical counterparts, likely due to the expansion of initially undetectable genotypes in culture [56]. Similar challenges have been described in other protozoan parasites such as Plasmodium falciparum, where polyclonal infections can contain mixed drug-sensitive and drug-resistant genotypes [57,58] and conventional bulk genotyping may reduce this complexity to binary calls [59], motivating the development of approaches that resolve subpopulation structure more explicitly. This complexity makes it challenging to link genotype to phenotype and to study evolutionary dynamics, especially for traits like drug resistance or virulence where minor variants can be clinically relevant […]”

    And:

    “In addition, another advantage of the single-cell approach is that it enables detecting co-existing mutations in the same cell, as illustrated in Figure 5D. This is particularly relevant in protozoan parasites, where resistance may depend on multi-locus haplotypes or combinations of mutations, as illustrated by resistance against sulfadoxine-pyrimethamine treatment in Plasmodium falciparum, which is characterized by mutations in dhfr and/or dhps genes, including triple and septuple mutant forms as well as mutations in both genes [62–64].”

    Reviewer #2 (Public review):

    Summary:

    Negreira et al. present an application of a novel single-cell genomics approach to investigate the genetic heterogeneity of Leishmania parasites. Leishmania, while also representing a major global disease with hundreds of thousands of cases annually, serves as a model to test the rigor of the sequencing strategy. Its complex karyotypic nature necessitates a method that is capable of resolving natural variation to better understand genome dynamics. Importantly, an earlier single-cell genomics platform (10x Chromium) is no longer available, and new methods need to be evaluated to fill in this gap.

    The study was designed to evaluate whether a capsule-based cell capture method combined with primary template-directed amplification (PTA) could maintain levels of genomic heterogeneity represented in an equal mixture of two Leishmania strains. This was a high bar, given the relatively small protozoan genome and prior studies that showed limitations of single-cell genomics, especially for gene-level copy number changes. Overall, the study found that semi-permeable capsules (SPC) are an effective way to isolate high-quality single cells. Additionally, short reads from amplified genomes effectively maintained the relative levels of variation in the two strains on the chromosome, gene copy, and individual base level. Thus, this method will be useful to evaluate adaptive strategies of Leishmania. Many researchers will also refer to these studies to set up SPC collection and PTA methods for their organism of choice.

    Strengths:

    (1) The use of SPC and PTA in a non-bacterial organism is novel. The study displays the utility of these methods to isolate and amplify single genomes to a level that can be sequenced, despite being a motile organism with a GC-rich genome.

    (2) The authors clearly outlined their optimization strategy and provided numerous quality-control metrics that inspire confidence in the success of achieving even chromosomal coverage relative to ploidy.

    (3) The use of two distinct Leishmania strains with known clonal status provided strong evidence that PTA-based amplification could reflect genome differences and displayed the utility of the method for studies of rare genotypes.

    (4) Evaluating the SPCs pre- and post-amplification with microscopy is a practical and robust way of determining the success of SPC formation and PTA.

    (5) The authors show that the PTA-based approach easily resolved major genotypic ploidy in agreement with a prior 10x Chromium-based study. The new method had improved resolution of drug resistance genotypes in the form of both copy-number variations and single-nucleotide polymorphisms.

    (6) In general, the authors are very thorough in describing the methods, including those used to optimize PTA lysis and amplification steps (fresh vs frozen cells, naked DNA vs sorted cells, etc). This demonstrates a depth of knowledge about the procedure and leaves few unanswered questions.

    (7) The custom, multifaceted, computational assessment of coverage evenness is a major strength of the study and demonstrates that the authors acknowledge potential computational factors that could impact the analysis.

    We deeply appreciate the positive and encouraging feedback on our manuscript.

    Weaknesses:

    (8) The rationale behind some experimental/analysis choices is not welldescribed. For example, the rationale behind methanol fixation and heat-lysis is unclear. Additionally, the choice of various methods to assess "evenness" is not justified (e.g. why are multiple methods needed? What is the strength of each method?). Also, there is no justification for using 100k reads for subsampling. Finally, what exactly constitutes a "confidently-called SNP"?

    The methanol fixation prior to lysis is part of the original protocol described in the Single-Microbe Genome Barcoding Kit manual and was meant to facilitate lysis and DNA denaturation in bacterial cells (for which the kit was originally developed). However, in our preliminary tests with bulk samples – described in the supplementary material – we noticed a strong negative effect on lysis efficiency/DNA recovery when parasites were fixed with methanol. Thus, we decided to test the effect of skipping this step in the single-cell DNA workflow. We kept the SPC_STD1 sample to have a safe control where the full workflow described in the kit manual was followed.

    As we were unsure if the standard lysis (25 ˚C for 15 minutes) would work efficiently for Leishmania, we included the heat-lysis (99˚C for 15 minutes) as well as the longer incubation lysis (25 ˚C for 1h). These modifications were listed as validated alternatives in the kit's manual. In the revised version, we have included a paragraph in the results section explaining the reason why we tested different lysis conditions:

    “We based our scDNAseq approach on the Single Microbe DNA Barcoding Kit from Atrandi BiosciencesTM, which uses semi-permeable capsules (SPCs) for high-throughput cell encapsulation30. The kit includes a pre-defined cell lysis protocol developed for bacteria, which starts with a methanol fixation step followed by incubation for 15 minutes at room temperature in the provided lysis buffer (0.8 M KOH, 20 mM EDTA, 200 mM DTT). Because it was unclear whether this protocol would be effective for Leishmania cells, we also tested modified lysis conditions by either extending the incubation time from 15 to 60 minutes or increasing the incubation temperature to 99 ˚C. In addition, we assessed if the protocol could be simplified by omitting the methanol fixation step. These modifications were first evaluated in pools of ~1,000,000 parasites per condition, by measuring with qPCR the quantity of free-floating DNA in the supernatant after parasite lysis and centrifugation. This analysis indicated that omitting methanol fixation in fact improved DNA release, whereas no clear differences were observed among the different lysis conditions (see Supplementary Text, Supplementary Figures 1 and 2, and Supplementary Tables 1 and 2).”

    Regarding read subsampling, in fact, we used two different thresholds, 20k reads for when 10X datasets were included in the comparisons, and 100k reads for when comparisons were done only among the SPC samples. These thresholds were always chosen to match the number of reads found in the true cell with the lowest read count across the datasets being compared: 20K reads in the 10X-scDNA data for BPK081 and 100K in the SPC-STD1 sample. This is now mentioned in the materials and methods:

    "Counts were then randomly subsampled to a total of 20,000 reads per cell – matching the number of reads found in the least covered true cell in the scDNA-10X datasets (see below) – to allow cross-sample comparisons. " and in the results section:

    "After equalizing each sample to 100,000 reads per cell (the number of reads observed in the least covered true cell in sample SPC-STD1), PTA still performed best"

    and:

    "Thus, we compared coverage uniformity in our dataset with that of two libraries generated using the Chromium Single-Cell CNV solution (10X-scDNA), considered the gold standard for this application. For this, we first subsampled each cell to 20,000 reads, matching the cell with the least read count in the 10X-scDNA dataset."

    Regarding variant calling, a variant was considered confidently called if it was covered, at single-cell level, by at least one deduplicated read with Phred quality above Q30 and mapping quality (MAPQ) also above 30. This is now mentioned in the materials and methods section:

    "A variant was considered confidently called when it was covered, at single-cell level, by at least one deduplicated read with Phred quality above Q30 and mapping quality (MAPQ) also above 30."

    (9) In the methods, the STD protocol lists a 15-minute amplification at 45C whereas the PTA protocol involves 10h at 37C. This is a dramatic difference in incubation time and should be addressed when comparing results from the two methods. It is not really a fair comparison when you look at coverage levels; of course, a 10-hour incubation is going to yield more reads than a 15-minute incubation.

    We agree with the reviewer that the longer incubation period of PTA might explain the higher read count seen in the PTA samples, although the differences in amplification kinetics (linear in PTA, exponential in STD) and potential differences in amplification saturation points make it difficult to compare them. For instance, an updated version of PTA (ResolveDNA V2) uses a lower amplification time (2.5 h) and achieves similar amplification levels compared to the 10h incubation time, suggesting PTA amplification saturates well before the 10h time. In any case, all quality check for coverage metrics were done with the cells subsampled to 20,000 reads to mitigate the effect of read count differences on the data quality.

    (10) There is a lack of quantitative evaluations of the SPCs. e.g. How many capsules were evaluated to assess doublets? How many capsules were detected as Syto5 positive in a successful vs an unsuccessful experiment?

    We agree with the reviewer but during experimental execution SPCs were only assessed qualitatively via microscopy following the Single-cell microbe DNA barcoding kit manual. No quantitative analysis was done and therefore we do not have this data. Regarding doublet, this was done in silico based on the detection of SPCs containing mixed genomes from the two strains used in the study as described in the Materials and Methods. As pointed by another reviewer, this only allow the detection of inter-strain doublets. We have now included in the revised version an estimate of total doublets in Table 1 and in the text:

    In the Materials and Methods:

    “Cells exhibiting a mixture of diagnostic SNPs from both strains above 5% were classified as inter-strain doublets. Total doublet frequency was estimated under the assumption of random cell pairing, by dividing the observed frequency of interstrain HU3/BPK081 doublets by the expected proportion of mixed doublets among all doublets, 2pq, where p and q are the relative frequencies of HU3 and BPK081 singlets respectively.”

    And in the results:

    “Successful WGA samples showed only 2–6 inter-strain doublets (4-12 estimated total doublets),

    (11) The authors do not address some of the amplification results obtained under various conditions. For example, why did temperature-based lysis of STD4 lead to amplification failure? Also, what is the reason for fewer "true" cells (higher background) in the PTA samples compared to the STD samples? Is this related to issues with barcoding or, alternatively, substandard amplification as indicated by lower read amounts in some capsules (knee plots in Figure 1C)?

    After exchange with the technical support team of the SPC generator kit, it was clarified that the heat lysis done in STD4 should have had a shorter incubation time (10 minutes instead of 15 minutes) and a lower KOH concentration (0.2 M instead of 0.8 M). We suspect that the longer incubation time, combined with the higher temperature and the harsh lysis condition with 0.8 M KOH might have damaged SPCs and therefore DNA might have leaked out of them before WGA. In the microscopy images, SPCs in STD4 show a swollen aspect not seen in the other samples. We added the following statement to the results section to explain this in the revised version:

    “SPC-STD4 also exhibited enlarged SPCs, which could indicate they were damaged during lysis.”

    and

    “According to follow-up guidance from the kit technical support team, it was suggested to use a shorter incubation time (10 min rather than 15 min) and a lower KOH concentration (0.2 M rather than 0.8 M) for heat-based lysis. We therefore infer that the combination of prolonged incubation, high temperature, and high KOH concentration in SPC-STD4 damaged the SPCs and promoted DNA leakage before WGA."

    (12) The paper presents limited biological relevance. Without this, the paper describes an improvement in genome amplification methods and some proof-ofconcept analyses. Using a 1:1 mixture of parasites with different genotypes, the authors display the utility of the method to resolve genetic diversity, but they don't seek to understand the limits of detecting this diversity. For some, the authors do not comment on the mixed karyotypes from the HU3 cells (Figure 3F) other than to state that this line was not clonal. For CNVs, the two loci evaluated were detected at relatively high copy number (according to Figure 4C, they are between 4 and 20 copies). Thus, the sensitivity of CNV detection from this data remains unclear; can this approach detect lower-level CNVs like duplications, or minor CNVs that do not show up in every cell?

    We have attempted an unbiased, genome-wide CNV calling approach across all samples to (i) confirm the two known loci and (ii) screen for novel short CNVs. CNV values were calculated as the ratio of median gene-level coverage to median chromosome-wide coverage. We have now added heatmaps summarizing this analysis for each sample in the supplementary material (Supplementary Figure 5). As expected, The SPC-STD1-3 samples sequencing depth was very sparse, with many genes showing low or even zero median coverage across the full chromosome. The SPC-PTA samples had substantially higher sequencing depth, making genome-wide CNV calculation feasible. However, the resulting heatmaps showed no clear, consistent signal across genes except for the two previously identified loci - the M-locus on chromosome 36 in BPK081 and the CNV on chromosome 29 in HU3. Thus, it is unclear if these are the only two differential CNVs present in these strains, or if our de novo detection method failed to retrieve other CNVs. We now describe this in the results section:

    “We also investigated if other CNVs could be present in our data using a gene-level unsupervised approach. For that, we divided the median read count of each annotated gene by the median count of the chromosome and looked for genes with median value 1.5 times higher than the chromosome median (supplementary figure 5). As expected, The SPC-PTA samples had substantially higher sequencing depth, making genome-wide CNV calculation feasible. Notably, the two previously identified loci - the M-locus on chromosome 36 in BPK081 and a known CNV on chromosome 29 in HU3 - stood out clearly against this background, consistent with the CNVs we had already characterized by bulk genome sequencing. However, this showed no clear, consistent signal across genes except for these two previously identified loci.”

    (13) The authors state that Leishmania can carry extrachromosomal copies of important genes. There is no discussion about how the presence of these molecules would affect the amplification steps and CNV detection. For example, the phi29 enzyme is very processive with circular molecules; does its presence lead to overamplification and overrepresentation in the data? Is this evident in the current study? This information would be useful for organisms that carry this type of genetic element.

    We believe our data, which uses short-read sequences, does not allow to differentiate between intra-chromosomal CNVs and linear or circular episomal CNVs derived from the nuclear genome. Of note, the Leishmania kinetoplast contains large circular DNA known as maxicircles (~10-40 kb), and we did not observe a clear dominance of maxicircle amplicons compared to nuclear genomic regions. The kinetoplast also contains thousands of minicircles (0.5- 2 kb) but those are not present in the reference genome and therefore it was not determined if they were detected and their prevalence in the data. As for the CNVs discussed in the paper, we have previously demonstrated that the M-locus CNV in chromosome 36 is intrachromosomal, not circular (episomal) [5].

    (14) The manuscript is missing a comparison with other similar studies in the field. For example, how does this coverage level compare to those achieved for other genomes? Can this method achieve amplification levels needed to assess larger genomes? Has there been any evaluation of base composition effects since Leishmania is a GC-rich genome?

    We believe the SPC-PTA workflow can be applied to organisms with larger genomes as PTA was developed specifically for mammalian cells [6], and also because, in our hands, it outperformed the 10X scDNA solution, which was developed for mammals.

    We believe direct comparison with other studies regarding coverage levels is elusive because other steps in the workflow apart from the WGA, such as the library preparation (PCR-based in our case), as well as genome features like GC content, size, and presence of repetitive regions, can also affect coverage levels and evenness. One strength of our approach was the use a single sample (the 50/50 mix between two L. donovani strain) for all conditions, thus removing potential parasite-specific biases. In addition, the application of a multiplexing system during barcoding allowed us to combine all samples prior to library preparation, thus removing potential differences introduced by this step.

    Regarding the effect of GC content, we did notice a negative bias in all samples in regions with higher GC content, which had to be corrected in silico. In the revised version, we have included a supplementary figure showing the GC bias and the correction result (supplementary figure 4).

    (15) Cost is mentioned as a benefit of the SPC platform, and savings are achieved when working in a plate format, but no details are included on how this was evaluated.

    We have improved the cost estimation in the discussion:

    “With SPCs we could apply PTA to a total of 399 cells (samples SPC-PTA1-2) while using the equivalent of 17 PTA reactions (17 cells if isolated in plates). Considering the costs of the PTA (~ €3100 for 96 WGA reactions) and library prep kits (~ €3100 for 96 libraries), a full plate-based system for 96 cells would have costed ~ €65 per cell excluding sequencing cost. In our hands, this cost was reduced to ~ €3,9 per cell including the required SPC generation kit (~ €230 for 1 encapsulation run) the PTA reactions (~ €550 eur for 17 reactions) the barcoding and library prep kits (~ €660 for 2 samples in the barcoding plate if using 6 sample multiplexing) and library prep (~ €130 for 4 final libraries). We also estimate that using the maximum capacity of 10.000 cells would further reduce price per cell to ~€0,39, although this comes with the caveats of a higher doublet rate and higher sequencing costs to achieve robust per-cell coverage. Overall, SPCs improves scalability and cost-efficiency, making high-throughput single-cell genome sequencing more accessible.”

    (16) The Zenodo link for custom scripts does not exist, and code cannot be evaluated.

    The full Zenodo link (https://doi.org/10.5281/zenodo.22876495) is now included in the revised version.

    Reviewer #2 (Recommendations for the authors):

    Suggestion for additional experiments:

    (17) In order to improve the CNV component of the work, the authors can employ a CNV caller as an unbiased method to identify all CNVs across the genome (including the 2 known loci). Assessing these patterns will be helpful to evaluate the sensitivity for CNV calling post-amplification.

    See response to Comment 12 of Reviewer 2.

    (18) Opportunities to include more biological relevance could include assessment of genetic changes under stress, or changes with parasite stage (current study is on promastigotes).

    We agree with the reviewer that these studies are highly relevant but we consider them out of the scope of the present work, which is more method-oriented. However, In the revised version we have expanded the discussion to include future potential applications for the SPC-PTA method:

    “Future perspectives

    The SPC-PTA scDNAseq workflow described here opens new possibilities for understanding the biology and adaptation mechanisms in Leishmania and other microbial eukaryotes. For instance, scDNAseq has been used to study DNA replication programs, revealing fork progression speeds and replication timing along the cell cycle in mammalian cells [67,68]. Unlike most eukaryotes, Leishmania has a unique DNA replication program based on a single large replication origin per chromosome which is highly active during S-phase and which is complemented by thousands of minor low-efficiency origin sites that are active during G1 and G2/M phases [69,70]. However, it is still unclear if there is cell-to-cell variability in origin use and replication timing, and how this may link to genome instability, which could be addressed using the SPC-PTA workflow in parasites at different stages of the cell cycle. Additionally, the SPC-PTA workflow could be applied to capture rare recombination events that occur in the insect vector – events that contribute to Leishmania genetic diversity but are hard to catch with bulk methods. It has been demonstrated that Leishmania undergo a cryptic sexual recombination following meiotic patterns of genome inheritance [71–73]. However, gamete-like cells were never identified. Our tool could be applied to identify cells harbouring whole-genome loss-of-heterozygosity signatures, which would be a strong indicative of meiosis.

    Another potential application is to combine the SPC-PTA workflow with loss-of-function assays to investigate key molecular factors involved in genome instability and DNA replication. For instance, small interfering RNA (siRNA) perturbations were combined with single-cell DNA sequencing to reveal a central role of Rap1 interacting factor 1 (RIF1) in coordinating replication timing in mouse embryos [74]. Although some Leishmania species lack a fully functional RNA interference pathway [75], loss-of-function assays can be achieved with CRISPR-Cas9-mediated gene knock-out [76]. More recently, genome-wide perturbation assays were made possible using CRISPR-Cas base editors opening the door to systematic, high-throughput functional screens in Leishmania [77,78]. Coupling these perturbation strategies with SPC-PTA could therefore enable direct testing of how some genes may shape replication dynamics, chromosome segregation, and genome instability at single-cell resolution.

    Lastly, the use of semipermeable capsules itself is also a key innovation and open the possibility of other applications. For instance, as we demonstrated, parasites remain alive after SPC encapsulation, and the semi-permeable nature of SPCs allow exchange of medium nutrients and cell metabolites, allowing cultivation of parasites inside SPCs [79]. This could be used to grow parasite clones trapped in SPCs, which would prevent cross-clone competition, allowing the recovery of slow-growing clones which would be otherwise outcompeted by faster-replicating clones. This is particularly important, e.g., for the study of quiescence, an emerging field of interest in Leishmania and other pathogens due to its adaptive role 80–83, as it would allow to cultivate and separate parasites with a low replicative phenotype.”

    Recommendations for improving the writing and presentation:

    (19) The paragraph in the Introduction beginning with "Although emulsion droplet..." could possibly be moved to the methods to provide a rationale for current procedures- as is, it dilutes the message at hand.

    We have removed this paragraph from the introduction and have replaced it with a short explanation about SPCs at the end of the introduction:

    “The main advantage of SPCs over emulsion droplet capturing lies in its semipermeable nature: large biomolecules (such as genomic DNA or whole organelles) are retained inside the capsule, while small molecules (e.g. enzymes, buffers, cofactors) can diffuse in and out [30]. Moreover, SPCs are resistant to harsh lysis chemicals, deep freezing and heat incubations. These properties permit multi-step reactions to be performed in the pools of encapsulated cells by simply washing or exchanging the surrounding solution, making custom protocol modifications easy.”

    (20) In the Introduction, the section describing the previous 10x study is relevant to understand the rationale, but again could be truncated, or moved to the methods or discussion.

    (21) The paragraph in the introduction beginning with "In this context, Leishmania..." could be combined with the paragraph introducing Leishmania ("To fully understand...." to improve flow and minimize repetitive information.

    We have restructured the introduction by rearraging the paragraphs and removing too technical information, so Leishmania is introduced earlier:

    “Introduction

    Genome instability plays a paradoxical and yet pivotal role in the evolutionary success of many unicellular pathogens. While genomic integrity is typically vital for cellular function, moderate levels of instability – either through intra-chromosomal copy number variations (CNV) or through dosage changes of whole chromosomes (aneuploidy) – can facilitate survival under environmental stress by quickly diversifying phenotypes in a cell population [1-3]. For instance, aneuploidy promotes survival to drug pressure and host immunity in Candida albicans [4,5], and CNVs were linked to survival under nitrogen-limiting conditions in Saccharomyces cerevisiae [6]. Aneuploidy and/or CNVs are also commonly found in cancer and are linked to tumor progression and drug resistance [7,8]. Moreover, these types of structural variation are also found in protist parasites such as Giardia intestinalis, Plasmodium spp., Trypanosoma cruzi, and Leishmania [9-13], representing potential adaptive mechanisms in clinically relevant organisms.

    Leishmania is a protist parasite that cause leishmaniasis, a neglected tropical disease which affects more than 99 countries with 700,000-1,000,000 new cases every year [14]. These parasites also represent an important model to study genome instability in eukaryotes, owing to its remarkable genomic plasticity and the central role that genome instability plays in its biology. First, while most eukaryotes maintain a euploid genome, i.e., with equal copy numbers of each chromosome, Leishmania breaks this rule, as aneuploidy is not an exception but the standard for these parasites. In its most basic form, aneuploidy is present as a characteristic polysomy of chromosome 31 contrasting with an otherwise fully diploid genome [15]. However, high levels of chromosome instability lead to frequent dosage changes involving virtually any of the 36 chromosomes, making any Leishmania population a ‘mosaic’ of distinct aneuploid karyotypes [16,17]. Aneuploidy patterns can shift in response to environmental stresses [18-21] and are directly reflected into RNA levels [18,22] and partially at protein levels [23]. In addition, Leishmania genome instability is also expressed in the form of frequent CNVs. Small segments of 20–70 kb can form extrachromosomal circular or linear amplicons carrying genes, including those encoding drug resistance factors [24]. Expansion of intra-chromosomal loci is also common. For instance, a common CNV found in chromosome 23 of some L. donovani strains was linked to resistance to antimonials in strains from Nepal [19,25]. In addition, the occurrence of mixed infections – where multiple Leishmania strains or even species coexist in the same host [26-28] – further highlight the need for genomic approaches that resolve parasite genotypes at single-cell resolution.

    In a previous study our group established the first single-cell DNA sequencing (scDNAseq) study with Leishmania [29]. Using single parasites isolated in plates with fluorescence-activated cell sorting (FACS), our group could successfully sequence 47 parasites, of which 28 had their karyotypes determined. This study also demonstrated that the whole-genome amplification (WGA) method must be chosen based on the downstream analysis, with multiple displacement amplification (MDA) being more suited for nucleotide sequence information, and PicoPlexTM performing better for structural variations such as chromosome copy number assessment and CNVs [29]. Later, we described the application of the Chromium single-cell CNV solution (10X GenomicsTM) as a high-throughput, emulsion droplet-based approach that could reconstruct karyotypes in thousands of Leishmania cells, revealing potential paths towards karyotype diversification, but with no access to nucleotide variant detection due to shallow genomic coverage [17]. Moreover, the Chromium single-cell CNV solution was discontinued in 2020, with no replacement currently available to our knowledge. This gap highlights the urgent need for novel, high-throughput scDNAseq approaches.

    Here we developed a novel high-throughput scDNAseq method tailored to detecting structural and nucleotide variants at single-cell level in eukaryotic parasites such as Leishmania. Our method is based on the commercially available Single-Microbe DNA Barcoding solution (Atrandi BiosciencesTM), which originally offers a workflow for cell encapsulation using semi-permeable capsules (SPCs), whole-genome amplification (WGA) and genome barcoding of bacterial cells. The main advantage of SPCs over emulsion droplet capturing lies in its semi-permeable nature: large biomolecules (such as genomic DNA or whole organelles) are retained inside the capsule, while small molecules (e.g. enzymes, buffers, cofactors) can diffuse in and out [30]. Moreover, SPCs are resistant to harsh lysis chemicals, deep freezing and heat incubations. These properties permit multi-step reactions to be performed in the pools of encapsulated cells by simply washing or exchanging the surrounding solution, making custom protocol modifications easy. We compared the performance of the method with libraries previously made with the Chromium single-cell CNV, showing that the original SPC-based protocol successfully generates scDNAseq libraries with Leishmania, but its uneven amplification prevents determination of structural variants. We solved this issue by using primary template-directed amplification (PTA) [31] as alternative WGA. This modification allowed us to fully reconstruct the karyotype of hundreds of L. donovani parasites and to detect distinct CNV patterns between cells. Moreover, the higher coverage seen in PTA unlocked the ability to simultaneously retrieve nucleotide sequence information from single-cells, which was not possible before. We demonstrate the ability of the method to distinguish parasite subpopulations of the same species, as well as small populations in the same strain, revealing minor genotypic variants among cells in clinical isolates, including variants in genes related to drug resistance in small subpopulations of parasites. Together, our results represent a significant advancement in single-cell genomics for Leishmania and other eukaryotes, providing a scalable, high-resolution tool to dissect both structural and sequence-level variation, and to deepen our understanding of parasite heterogeneity, evolution, and adaptation.”

    (22) The 4kb amplicon in HU3 is referred to as on chr 19 and 29 across the paper.

    We have fixed this mistake in the revised version.

    (23) Correct grammar: "A table summarizing the treatments that each sample was submitted to is provided in the results section (Table 1)."

    We replaced the sentence by:

    "A summary of lysis and WGA conditions is provided in the results section (Table 1)"

    (24) The Lorenz curve (Figure 2A) is not described in the methods. How was this analysis performed?

    We have now a description of the Lorenz curve in the materials and methods section

    “Lorenz curves were done for each cell by arranging genomic bins in ascending order for read counts and then calculating the cumulative read count sum. Cumulative sums were normalized by total count in each cell, and Lorenz curves were then summarized across cells of the same sample by calculating the mean cumulative fraction of counts at each genomic fraction.”

    (25) Move long lists of numbers across the Results into tables and cite them. It is hard to read as is.

    We have removed detailed statistical information from the results text. We also have included more information in Table 1.

    (26) For Figures 4A and B, please demark known amplicon boundaries to help interpret the plot. LD36 seems to have two nearby CNV regions, but the amplicon size is unclear.

    Figures 4A-B now include a dashed white line highlighting the CNV areas.

    Reviewer #3 (Public review):

    Summary:

    In this manuscript, Negreira et al. propose a new scDNAseq method, using semipermeable capsules (SPCs) and primary template-directed amplification (PTA). The authors optimize several metrics to improve their predictions, such as determining GC bias, Intra-Chromosomal fluctuation (ICF -metric to differentiate replicative and nonreplicative cells) and Intra-chromosomal coefficient of variation (ICCV - chromosome read distribution). The coverage evenness was evaluated using the fini index and the median absolute pairwise difference between the counts of two consecutive bins. They validate the proposed method using two Leishmania donovani strains isolated from different countries, BPK081 (low genomic variability) and HU3 (high genomic variability). Then, they showed that the method outperforms WGA and has similar accuracy to the discontinued 10X-scDNA (10X Genomics), further improving on short CNV identification. The authors also show that the method can identify somy variations, insertions/deletions and SNP variations across cells. This is a timely and very relevant work that has a wide applicability in copy number variation assessment using single-cell data.

    I really appreciate this work. My congratulations to the authors. All my comments below only aim to improve an already solid manuscript.

    We thank the reviewer for the enthusiasm and positive feedback.

    Remarks

    (1) Data availability: Although the authors provide a Zenodo link, the data is restricted. I also could not access the GitHub link in the Zenodo website: https://github.com/gabrielnegreira/2025_scDNA_paper. The authors should make these files available.

    Both the Zenodo (https://doi.org/10.5281/zenodo.22876495) and the GitHub (https://github.com/gabrielnegreira/2025_scDNA_paper) repositories are now publicly available.

    (2) SPC-PTA and SPC-STD cell count comparison: The authors have consistently proven that the SPC-PTA method was superior to SPC-STD. However, there are a few points that should be clarified regarding the SPC-PTA results. Is there an explanation for the lower proportion of SPC to true cells success in SPCSTD, which reflects the bimodal distribution for the reads per cell in SPC-PTA2 and a three-to-multimodal distribution in SPC-PTA1 in Figure 1B? Also, in Table 1, does the number of reads reflect the number of reads in all sequenced SPCs or only in the true cells? If it is in the SPCs, I suggest that the authors add a new column in the table with the "Number of reads in true cells" to account for this discrepancy.

    The reason for the higher presence of 'background' SPCs in the PTA samples is not clear, but we hypothesize that it could be due to PTA favoring amplification of small, free-floating DNA molecules that might have been trapped in cell-free SPCs, as PTA works with shorter amplicons. Also, the longer incubation time seen in PTA (10 h) might have allowed enhanced amplification of low quantities of free-floating DNA to detectable levels. Regarding Table 1, indeed it was only showing the total number of reads per sample. In the revised we have now included the suggested column to Table 1.

    (3) The authors should evaluate the results with a higher coverage for SCPPTA. I understand that the authors subsampled the total read to 100,000 to allow cross-sample comparisons, especially between SPC-STD and SPC-PTA. However, as they concluded that the SPC-PTA was far superior, and the samples SPC-PTA1 and SPC-PTA2 had an "elbow" of 650,493 and 448,041, respectively, it might be interesting to revisit some of the estimations using only SPC-PTA samples and a higher coverage cutoff, as 400,000.

    For the CNV analysis, we used all reads per cell, not the subsampled dataset, so SPC-PTA already benefit from their higher coverage. In addition, as mentioned in our response to Comment 8 of Reviewer 2, we have subsampled cells to 20,000 reads, not 100.000 reads per cell, to allow comparisons to the 10X datasets. We have corrected that information in the text now.

    (4) Doublet detection: I suggest that the authors be a little more careful with their definition of doublets. The doublet detection was based on diagnostic SNPs from the two strains, BPK081 and HU3, which identify doublets between two very different and well-characterised strains. However, this method will probably not identify strain-specific doublets. This is of minor importance for cloned and stable strains with few passages, as BPK081, but might be more relevant in more heterogeneous strains, as HU3. Strain-specific doublets might also be relevant in other scenarios, as multiclonal infections with different populations from the same strain in the same geographic area. One positive point is that the "between strain doublet count" was low, so probably the within-strain doublet count should be low too. The manuscript would benefit from a discussion on this regard.

    We fully agree with the reviewer. We have now included in the manuscript an estimation of total doublets based on the frequency of inter-strain doublets detected in each sample. See response to Comment 10 of Reviewer 2.

    (5) Nucleotide sequence variants and phylogeny: I believe that a more careful description of the phylogenetic analysis and some limitations of the sequence variant identification would benefit the manuscript.

    (a) As described in the methods, the authors intentionally selected two fairly different Leishmania donovani strains, HU3 and BPK081, and confirmed that the sequent variant methodology can separate cells from each strain. It is a solid proof of concept. However, most of the multiclonal infections in natural scenarios would be caused by parasite populations that diverge by fewer SNPs, and will be significantly harder to detect. Hence, I suggest that a short discussion about this is important.

    We agree that mixed infections might happen between closely related parasite populations and that the HU3/BPK081 mixture represents a relatively divergent strain pair. At the same time, our data suggest that the approach might not be restricted to highly divergent strains: in Fig. 5D, the two populations remain clearly separable using only a small subset of approximately 30 informative loci, and the within-BPK081 phylogeny in Fig. 5E further indicates that cell groups can be resolved based on comparatively few mutations. We therefore interpret the current dataset as evidence that the method retains discriminatory power with limited numbers of informative variants, while acknowledging that performance is expected to decrease as inter-clone divergence and per-cell SNP recovery decline. We have now added a small discussion about this:

    “Moreover, population-genomic studies of the L. donovani complex have identified isolates with allele-frequency patterns consistent with mixtures of genetically similar subclones, including mixtures within the same phylogenetic group [54].”

    And

    “Moreover, although the HU3/BPK081 mixture represents a relatively divergent two-strain model, the result shown in Figure 5D, which uses only a subset of 30 SNPs, indicates that separation can be achieved even with a limited set of informative loci, and the within-strain structure observed in the BPK081 tree in Figure 5E suggests sensitivity to relatively small mutational differences. This is of particular interest in cases where mixed infections happen between closely related parasite populations [54].”

    (b) The authors should expand on the description of the phylogenetic tree. In the HU3 on Figure 5F left panel, most of the variation is observed in ~8 cells, which goes from position 0 to position ~28.000. Most of the other cells are in very short branches, from ~29.000 to 30.4000 (5F right panel). Assuming that this representation is a phylogram, as the branches are short, these cells diverge by approximately 100-2000 SNPs. It is unexpected (but not impossible) that such ~8 divergent cells be maintained uniquely (or in very low counts) in the culture, unless this is a multiclonal infection. I would carefully investigate these cells. They might be doublets or have more missing data than other cells. I would also suggest that a quick discussion about this should be added to the manuscript.

    We have included in the Figure 5F a legend for the x-axis label as “cumulative inferred mutations”. We agree that the ~8 cells shown in figure 5F in HU3 with the largest cumulative inferred mutations needed closer inspection. We examined these cells in the sample depicted in figure 5F (SPC-PTA2), and observed a shift toward lower coverage and higher missingness (undefined loci), but these cells did not represent extreme QC outliers relative to the overall distributions, and other cells show comparable or even higher levels of missingness / reduced coverage but remain clustered with the main bulk rather than forming long branches in the phylogram. We have added graphs showing these values in the supplementary figure 6. In addition, we also investigated the corresponding pattern in the SPC-PTA1 sample, which represents a second independently sampled dataset from the same culture. In this second sample, we again observed a small subset of long-branch cells but this time showing a similar coverage and missingness values. Overall, this indicates that missingness and coverage may influence the inferred mutation load for the long-branch cells, but that elevated missingness alone is insufficient to fully explain why these specific cells extend to longer branches. Given that the long-branch shape is reproducible across independent samplings, we interpret the observation as most consistent with minor divergent subpopulation(s) persisting in the culture, while also acknowledging that locus-/depth-dependent calling sensitivity could contribute to the magnitude of inferred mutations in this subset. Lastly, because HU3 is a noncloned, long-term cultured population, polyclonality remains a plausible explanation for a small subset of more divergent cells. We have added this explanation in the results section of the revised version:

    “To determine whether the presence of 8 cells in long branches observed in HU3 in sample SPC-PTA2 could be driven by technical artefacts (e.g., reduced coverage, less variants called), we closely inspected the coverage values in these cells compared to the other cells in the same dataset. We did observe a consistent tendency toward lower coverage and higher variant missingness (undefined loci) in these 8 cells, but still within the range of variation seen across the other HU3 cells (supplementary figure 6A). Importantly, other cells show comparable or even higher levels of missingness/reduced coverage but remain clustered with the main bulk rather than forming long branches in the phylogram. Thus, it is unlikely that missingness/coverage is the sole cause of their phylogenetic separation. This is further supported by a similar analysis in sample SPC-PTA1, which represents and independent sampling of the same parasite population. Indeed, the phylogenetic tree of this sample showed the same overall pattern, where only a small subset of cells was present in long branches (supplementary figure 6B). However, in the case of SPC-PTA1 these cells did not show lower coverage or higher missingness compared to other cells (supplementary figure 6C). Thus, these results are most consistent with minor divergent subpopulation(s) persisting during long-term culture. As HU3 is a non-cloned isolate, polyclonality remains a plausible explanation for a small subset of more divergent cells. At the same time, we cannot exclude that locus-/depth-dependent calling sensitivity may contribute to the magnitude of inferred mutations in this subset.”

    Reviewer #3 (Recommendations for the authors):

    (6) The authors have shown the identification of two known copy number variants in these two strains. Has any effort been made to identify novel short CNVs?

    See response to Comment 12 of Reviewer 2.

    (7) In the results first section, in the passage "both strains differ in (bulk) aneuploidy profiles", please remove the parentheses from bulk.

    We have made the suggested modification.

    (8) Figure 3, D and F. Would it be possible to add legends for the colours, so that the data can be easily compared with the x-axis (karyotypes, as kar.1, kar.2, etc) in Figure 3, C and E?

    We have changed the figures 3D and F to highlight the top 5 karyotypes found in the 10X data. This helps comparing them to figures 3C and E.

    (9) Figure 5A. Would it be possible to include in the x and y axes the percentage of representation of PC1 and PC2? My guess is that most of the variability will be represented by PC1 (x-axis) as it is the one that separates the two populations and the hybrids.

    We have included the percentages of representation of the PCs in the figure in the revised version. Indeed, PC1 captured more than 90% of the variability.

    (10) In the discussion, section "SPCs as a viable tool for high throughput implementation of PTA" I would move this passage to the results (as it is a result): "By applying this strategy to Leishmania, we could sequence a total of 399 cells (from samples SPC-PTA1-2) while using the equivalent of 17 PTA reactions (17 cells) with cells isolated in plates, reducing the cost per cell by more than 20 fold."

    As mentioned in response to the Comment 15 of Reviewer 2 we have extended the discussion on costs and therefore kept this part in the discussion section.

    (11) Another advantage of the single-cell data that could be discussed by the authors is the possibility of checking if different variants occur in the same cell. For example, if a variation in position "X" in the resistance gene occurs in the same cell as the variation in position "Y". This is a great improvement compared to bulksequencing.

    We have now included a small discussion about this:

    “This is particularly relevant in protozoan parasites, where resistance may depend on multi-locus haplotypes or combinations of mutations, as illustrated by resistance against sulfadoxine-pyrimethamine treatment in Plasmodium falciparum, which is characterized by mutations in dhfr and/or dhps genes, including triple and septuple mutant forms as well as mutations in both genes6264.”

    REFERENCES

    (1) Dileep, V. & Gilbert, D. M. Single-cell replication profiling to measure stochastic variation in mammalian replication timing. Nat. Commun. 9, 427 (2018).

    (2) Miura, H. et al. Single-cell DNA replication profiling identifies spatiotemporal developmental dynamics of chromosome organization. Nat. Genet. 51, 1356–1368 (2019).

    (3) Marques, C. A., Dickens, N. J., Paape, D., Campbell, S. J. & McCulloch, R. Genomewide mapping reveals single-origin chromosome replication in Leishmania, a eukaryotic microbe. Genome Biol. 16, 230 (2015).

    (4) Damasceno, J. D. et al. Leishmania major chromosomes are replicated from a single high-efficiency locus supplemented by thousands of lower efficiency initiation events. Cell Rep. 44, (2025).

    (5) Imamura, H. et al. Evolutionary genomics of epidemic visceral leishmaniasis in the Indian subcontinent. eLife 5, e12613 (2016).

    (6) Gonzalez-Pena, V. et al. Accurate genomic variant detection in single cells with primary template-directed amplification. Proc. Natl. Acad. Sci. 118, e2024176118 (2021).

  8. Version published to 10.1101/2025.09.10.675331v3 on bioRxiv
  9. eLife

    eLife Assessment

    This important study, which tackles the challenge of analyzing genome integrity and instability in unicellular pathogens by introducing a novel single-cell genomics approach, presents compelling evidence that this new tool outperforms standard whole-genome amplification techniques. While thorough and rigorous, the work's impact would increase by providing scripts and data, as well as a description of the biological relevance that would make this method more appealing to the broad community studying genetic heterogeneity in diverse organisms.

  10. eLife

    Reviewer #1 (Public review):

    Summary:

    Negreira, G. et al clearly presented the challenges of conducting genomic studies in unicellular pathogens and of addressing questions related to the balance between genome integrity and instability, pivotal for survival under the stressful conditions these organisms face and for their evolutionary success. This underlies the need for powerful approaches to perform single-cell DNA analyses suited to the small and plastic Leishmania genome. Accordingly, their goal was to develop such a novel method and demonstrate its robustness.

    In this study, the authors combined semi-permeable capsules (SPCs) with primary template-directed amplification (PTA) and adapted the system to the Leishmania genome, which is about 100 times smaller than the human genome and exhibits remarkable plasticity and mosaic aneuploidy. Given the size and organization of the Leishmania genome, the challenges were substantial; nevertheless, the authors successfully demonstrated that PTA not only works for Leishmania but also represents a significantly improved whole-genome amplification (WGA) method compared with standard approaches. They showed that SPCs provide a superior alternative for cell encapsulation, increasing throughput. The methodology enabled high-resolution karyotyping and the detection of fine-scale copy number variations (CNVs) at the single-cell level. Furthermore, it allowed discrimination between genotypically distinct cells within mixed populations.

    Strengths:

    This is a high-impact study that will likely contribute to our understanding of DNA replication and the genetic plasticity of Leishmania, including its well-documented aneuploidy, somy variations, CNVs, and SNPs - all key elements for elucidating various aspects of the parasite's biology, such as genome evolution, genetic exchange, and mechanisms of drug resistance.

    Overall, the authors clearly achieved their objectives, providing a solid rationale for the study and demonstrating how this approach can advance the investigation of Leishmania's small, plastic genome and its frequent natural strain mixtures within hosts. This methodology may also prove valuable for genomic studies of other single-celled organisms.

    Weaknesses:

    The discussion section could be enriched to help readers understand the significance of the work, for instance, by more clearly pointing out the obstacles to a better understanding of DNA replication in Leishmania. Or else, when they discuss the results obtained at the level of nucleotide information and the relevance of being able to compare, in their case, the two strains, they could refer to the implications of this level of precision to those studying clonal strains or field isolates, drug resistance or virulence in a more detailed way.

  11. eLife

    Reviewer #2 (Public review):

    Summary:

    Negreira et al. present an application of a novel single-cell genomics approach to investigate the genetic heterogeneity of Leishmania parasites. Leishmania, while also representing a major global disease with hundreds of thousands of cases annually, serves as a model to test the rigor of the sequencing strategy. Its complex karyotypic nature necessitates a method that is capable of resolving natural variation to better understand genome dynamics. Importantly, an earlier single-cell genomics platform (10x Chromium) is no longer available, and new methods need to be evaluated to fill in this gap.

    The study was designed to evaluate whether a capsule-based cell capture method combined with primary template-directed amplification (PTA) could maintain levels of genomic heterogeneity represented in an equal mixture of two Leishmania strains. This was a high bar, given the relatively small protozoan genome and prior studies that showed limitations of single-cell genomics, especially for gene-level copy number changes. Overall, the study found that semi-permeable capsules (SPC) are an effective way to isolate high-quality single cells. Additionally, short reads from amplified genomes effectively maintained the relative levels of variation in the two strains on the chromosome, gene copy, and individual base level. Thus, this method will be useful to evaluate adaptive strategies of Leishmania. Many researchers will also refer to these studies to set up SPC collection and PTA methods for their organism of choice.

    Strengths:

    (1) The use of SPC and PTA in a non-bacterial organism is novel. The study displays the utility of these methods to isolate and amplify single genomes to a level that can be sequenced, despite being a motile organism with a GC-rich genome.

    (2) The authors clearly outlined their optimization strategy and provided numerous quality-control metrics that inspire confidence in the success of achieving even chromosomal coverage relative to ploidy.

    (3) The use of two distinct Leishmania strains with known clonal status provided strong evidence that PTA-based amplification could reflect genome differences and displayed the utility of the method for studies of rare genotypes.

    (4) Evaluating the SPCs pre- and post-amplification with microscopy is a practical and robust way of determining the success of SPC formation and PTA.

    (5) The authors show that the PTA-based approach easily resolved major genotypic ploidy in agreement with a prior 10x Chromium-based study. The new method had improved resolution of drug resistance genotypes in the form of both copy-number variations and single-nucleotide polymorphisms.

    (6) In general, the authors are very thorough in describing the methods, including those used to optimize PTA lysis and amplification steps (fresh vs frozen cells, naked DNA vs sorted cells, etc). This demonstrates a depth of knowledge about the procedure and leaves few unanswered questions.

    (7) The custom, multifaceted, computational assessment of coverage evenness is a major strength of the study and demonstrates that the authors acknowledge potential computational factors that could impact the analysis.

    Weaknesses:

    (1) The rationale behind some experimental/analysis choices is not well-described. For example, the rationale behind methanol fixation and heat-lysis is unclear. Additionally, the choice of various methods to assess "evenness" is not justified (e.g. why are multiple methods needed? What is the strength of each method?). Also, there is no justification for using 100k reads for subsampling. Finally, what exactly constitutes a "confidently-called SNP"?

    (2) In the methods, the STD protocol lists a 15-minute amplification at 45C whereas the PTA protocol involves 10h at 37C. This is a dramatic difference in incubation time and should be addressed when comparing results from the two methods. It is not really a fair comparison when you look at coverage levels; of course, a 10-hour incubation is going to yield more reads than a 15-minute incubation.

    (3) There is a lack of quantitative evaluations of the SPCs. e.g. How many capsules were evaluated to assess doublets? How many capsules were detected as Syto5 positive in a successful vs an unsuccessful experiment?

    (4) The authors do not address some of the amplification results obtained under various conditions. For example, why did temperature-based lysis of STD4 lead to amplification failure? Also, what is the reason for fewer "true" cells (higher background) in the PTA samples compared to the STD samples? Is this related to issues with barcoding or, alternatively, substandard amplification as indicated by lower read amounts in some capsules (knee plots in Figure 1C)?

    (5) The paper presents limited biological relevance. Without this, the paper describes an improvement in genome amplification methods and some proof-of-concept analyses. Using a 1:1 mixture of parasites with different genotypes, the authors display the utility of the method to resolve genetic diversity, but they don't seek to understand the limits of detecting this diversity. For some, the authors do not comment on the mixed karyotypes from the HU3 cells (Figure 3F) other than to state that this line was not clonal. For CNVs, the two loci evaluated were detected at relatively high copy number (according to Figure 4C, they are between 4 and 20 copies). Thus, the sensitivity of CNV detection from this data remains unclear; can this approach detect lower-level CNVs like duplications, or minor CNVs that do not show up in every cell?

    (6) The authors state that Leishmania can carry extrachromosomal copies of important genes. There is no discussion about how the presence of these molecules would affect the amplification steps and CNV detection. For example, the phi29 enzyme is very processive with circular molecules; does its presence lead to overamplification and overrepresentation in the data? Is this evident in the current study? This information would be useful for organisms that carry this type of genetic element.

    (7) The manuscript is missing a comparison with other similar studies in the field. For example, how does this coverage level compare to those achieved for other genomes? Can this method achieve amplification levels needed to assess larger genomes? Has there been any evaluation of base composition effects since Leishmania is a GC-rich genome?

    (8) Cost is mentioned as a benefit of the SPC platform, and savings are achieved when working in a plate format, but no details are included on how this was evaluated.

    (9) The Zenodo link for custom scripts does not exist, and code cannot be evaluated.

  12. eLife

    Reviewer #3 (Public review):

    In this manuscript, Negreira et al. propose a new scDNAseq method, using semi-permeable capsules (SPCs) and primary template-directed amplification (PTA). The authors optimize several metrics to improve their predictions, such as determining GC bias, Intra-Chromosomal fluctuation (ICF -metric to differentiate replicative and non-replicative cells) and Intra-chromosomal coefficient of variation (ICCV - chromosome read distribution). The coverage evenness was evaluated using the fini index and the median absolute pairwise difference between the counts of two consecutive bins. They validate the proposed method using two Leishmania donovani strains isolated from different countries, BPK081 (low genomic variability) and HU3 (high genomic variability). Then, they showed that the method outperforms WGA and has similar accuracy to the discontinued 10X-scDNA (10X Genomics), further improving on short CNV identification. The authors also show that the method can identify somy variations, insertions/deletions and SNP variations across cells. This is a timely and very relevant work that has a wide applicability in copy number variation assessment using single-cell data.

    I really appreciate this work. My congratulations to the authors. All my comments below only aim to improve an already solid manuscript.

    (1) Data availability: Although the authors provide a Zenodo link, the data is restricted. I also could not access the GitHub link in the Zenodo website: https://github.com/gabrielnegreira/2025_scDNA_paper. The authors should make these files available.

    (2) 2-SPC-PTA and SPC-STD cell count comparison: The authors have consistently proven that the SPC-PTA method was superior to SPC-STD. However, there are a few points that should be clarified regarding the SPC-PTA results. Is there an explanation for the lower proportion of SPC to true cells success in SPC-STD, which reflects the bimodal distribution for the reads per cell in SPC-PTA2 and a three-to-multimodal distribution in SPC-PTA1 in Figure 1B? Also, in Table 1, does the number of reads reflect the number of reads in all sequenced SPCs or only in the true cells? If it is in the SPCs, I suggest that the authors add a new column in the table with the "Number of reads in true cells" to account for this discrepancy.

    (3) The authors should evaluate the results with a higher coverage for SCP-PTA. I understand that the authors subsampled the total read to 100,000 to allow cross-sample comparisons, especially between SPC-STD and SPC-PTA. However, as they concluded that the SPC-PTA was far superior, and the samples SPC-PTA1 and SPC-PTA2 had an "elbow" of 650,493 and 448,041, respectively, it might be interesting to revisit some of the estimations using only SPC-PTA samples and a higher coverage cutoff, as 400,000.

    (4) Doublet detection: I suggest that the authors be a little more careful with their definition of doublets. The doublet detection was based on diagnostic SNPs from the two strains, BPK081 and HU3, which identify doublets between two very different and well-characterised strains. However, this method will probably not identify strain-specific doublets. This is of minor importance for cloned and stable strains with few passages, as BPK081, but might be more relevant in more heterogeneous strains, as HU3. Strain-specific doublets might also be relevant in other scenarios, as multiclonal infections with different populations from the same strain in the same geographic area. One positive point is that the "between strain doublet count" was low, so probably the within-strain doublet count should be low too. The manuscript would benefit from a discussion on this regard.

    (5) Nucleotide sequence variants and phylogeny: I believe that a more careful description of the phylogenetic analysis and some limitations of the sequence variant identification would benefit the manuscript.

    (5.1) As described in the methods, the authors intentionally selected two fairly different Leishmania donovani strains, HU3 and BPK081, and confirmed that the sequent variant methodology can separate cells from each strain. It is a solid proof of concept. However, most of the multiclonal infections in natural scenarios would be caused by parasite populations that diverge by fewer SNPs, and will be significantly harder to detect. Hence, I suggest that a short discussion about this is important.

    (5.2) The authors should expand on the description of the phylogenetic tree. In the HU3 on Figure 5F left panel, most of the variation is observed in ~8 cells, which goes from position 0 to position ~28.000.

    Most of the other cells are in very short branches, from ~29.000 to 30.4000 (5F right panel). Assuming that this representation is a phylogram, as the branches are short, these cells diverge by approximately 100-2000 SNPs. It is unexpected (but not impossible) that such ~8 divergent cells be maintained uniquely (or in very low counts) in the culture, unless this is a multiclonal infection. I would carefully investigate these cells. They might be doublets or have more missing data than other cells. I would also suggest that a quick discussion about this should be added to the manuscript.

  13. eLife

    Author response:

    Reviewer #1 (Public review):

    Summary:

    Negreira, G. et al clearly presented the challenges of conducting genomic studies in unicellular pathogens and of addressing questions related to the balance between genome integrity and instability, pivotal for survival under the stressful conditions these organisms face and for their evolutionary success. This underlies the need for powerful approaches to perform single-cell DNA analyses suited to the small and plastic Leishmania genome. Accordingly, their goal was to develop such a novel method and demonstrate its robustness.

    In this study, the authors combined semi-permeable capsules (SPCs) with primary template-directed amplification (PTA) and adapted the system to the Leishmania genome, which is about 100 times smaller than the human genome and exhibits remarkable plasticity and mosaic aneuploidy. Given the size and organization of the Leishmania genome, the challenges were substantial; nevertheless, the authors successfully demonstrated that PTA not only works for Leishmania but also represents a significantly improved whole-genome amplification (WGA) method compared with standard approaches. They showed that SPCs provide a superior alternative for cell encapsulation, increasing throughput. The methodology enabled high-resolution karyotyping and the detection of fine-scale copy number variations (CNVs) at the single-cell level. Furthermore, it allowed discrimination between genotypically distinct cells within mixed populations.

    Strengths:

    This is a high-impact study that will likely contribute to our understanding of DNA replication and the genetic plasticity of Leishmania, including its well-documented aneuploidy, somy variations, CNVs, and SNPs - all key elements for elucidating various aspects of the parasite's biology, such as genome evolution, genetic exchange, and mechanisms of drug resistance.

    Overall, the authors clearly achieved their objectives, providing a solid rationale for the study and demonstrating how this approach can advance the investigation of Leishmania's small, plastic genome and its frequent natural strain mixtures within hosts. This methodology may also prove valuable for genomic studies of other single-celled organisms.

    We thank the reviewer for the positive feedback and appreciation of the potential applications for the methodology we describe here.

    Weaknesses:

    The discussion section could be enriched to help readers understand the significance of the work, for instance, by more clearly pointing out the obstacles to a better understanding of DNA replication in Leishmania. Or else, when they discuss the results obtained at the level of nucleotide information and the relevance of being able to compare, in their case, the two strains, they could refer to the implications of this level of precision to those studying clonal strains or field isolates, drug resistance or virulence in a more detailed way.

    We thank the reviewer for the suggestions. Indeed, single-cell DNA sequencing has successfully revealed cell-to-cell variability in replication timing and fork progression in mammalian cells[1,2] and we believe that the SPC-PTA workflow could be used in similar studies in Leishmania to complement bulk-based observations[3,4]. Regarding nucleotide information, it is indeed of high relevance to detect minor circulating variants with potential virulence impact and/or effect on drug resistance which could be missed by bulk sequencing. This includes the ability to detect co-occurring variants with potential epistatic effects. These topics will be further developed in the revised version. Finally, we will explicitly discuss how this methodology can be applied beyond Leishmania, to investigate genome plasticity, adaptation, and evolutionary processes in other organisms.

    Reviewer #2 (Public review):

    Summary:

    Negreira et al. present an application of a novel single-cell genomics approach to investigate the genetic heterogeneity of Leishmania parasites. Leishmania, while also representing a major global disease with hundreds of thousands of cases annually, serves as a model to test the rigor of the sequencing strategy. Its complex karyotypic nature necessitates a method that is capable of resolving natural variation to better understand genome dynamics. Importantly, an earlier single-cell genomics platform (10x Chromium) is no longer available, and new methods need to be evaluated to fill in this gap.

    The study was designed to evaluate whether a capsule-based cell capture method combined with primary template-directed amplification (PTA) could maintain levels of genomic heterogeneity represented in an equal mixture of two Leishmania strains. This was a high bar, given the relatively small protozoan genome and prior studies that showed limitations of single-cell genomics, especially for gene-level copy number changes. Overall, the study found that semi-permeable capsules (SPC) are an effective way to isolate high-quality single cells. Additionally, short reads from amplified genomes effectively maintained the relative levels of variation in the two strains on the chromosome, gene copy, and individual base level. Thus, this method will be useful to evaluate adaptive strategies of Leishmania. Many researchers will also refer to these studies to set up SPC collection and PTA methods for their organism of choice.

    Strengths:

    (1) The use of SPC and PTA in a non-bacterial organism is novel. The study displays the utility of these methods to isolate and amplify single genomes to a level that can be sequenced, despite being a motile organism with a GC-rich genome.

    (2) The authors clearly outlined their optimization strategy and provided numerous quality-control metrics that inspire confidence in the success of achieving even chromosomal coverage relative to ploidy.

    (3) The use of two distinct Leishmania strains with known clonal status provided strong evidence that PTA-based amplification could reflect genome differences and displayed the utility of the method for studies of rare genotypes.

    (4) Evaluating the SPCs pre- and post-amplification with microscopy is a practical and robust way of determining the success of SPC formation and PTA.

    (5) The authors show that the PTA-based approach easily resolved major genotypic ploidy in agreement with a prior 10x Chromium-based study. The new method had improved resolution of drug resistance genotypes in the form of both copy-number variations and single-nucleotide polymorphisms.

    (6) In general, the authors are very thorough in describing the methods, including those used to optimize PTA lysis and amplification steps (fresh vs frozen cells, naked DNA vs sorted cells, etc). This demonstrates a depth of knowledge about the procedure and leaves few unanswered questions.

    (7) The custom, multifaceted, computational assessment of coverage evenness is a major strength of the study and demonstrates that the authors acknowledge potential computational factors that could impact the analysis.

    We deeply appreciate the positive and encouraging feedback on our manuscript.

    Weaknesses:

    (1) The rationale behind some experimental/analysis choices is not well-described. For example, the rationale behind methanol fixation and heat-lysis is unclear. Additionally, the choice of various methods to assess "evenness" is not justified (e.g. why are multiple methods needed? What is the strength of each method?). Also, there is no justification for using 100k reads for subsampling. Finally, what exactly constitutes a "confidently-called SNP"?

    The methanol fixation prior to lysis is part of the original protocol described in the Single-Microbe Genome Barcoding Kit manual and was meant to facilitate lysis and DNA denaturation in bacterial cells (for which the kit was originally developed). However, in our preliminary tests with bulk samples – described in the supplementary material – we noticed a strong negative effect on lysis efficiency/DNA recovery when parasites were fixed with methanol. Thus, we decided to test the effect of skipping this step in the single-cell DNA workflow. We kept the SPC_STD1 sample to have a safe control where the full workflow described in the kit manual was followed.

    As we were unsure if the standard lysis (25 ˚C for 15 minutes) would work efficiently for Leishmania, we included the heat-lysis (99˚C for 15 minutes) as well as the longer incubation lysis (25 ˚C for 1h). These modifications were listed as validated alternatives in the kit's manual.

    The 100k reads threshold was chosen based on the number of reads found in the 'true cell' with the lowest read count.

    Regarding variant calling, a variant was considered confidently called if it was covered, at single-cell level, by at least one deduplicated read with Phred quality above Q30 and mapping quality (MAPQ) also above 30.

    In the revised version, we will include these explanations and improve the explanation of the metrics used to estimate coverage quality.

    (2) In the methods, the STD protocol lists a 15-minute amplification at 45C whereas the PTA protocol involves 10h at 37C. This is a dramatic difference in incubation time and should be addressed when comparing results from the two methods. It is not really a fair comparison when you look at coverage levels; of course, a 10-hour incubation is going to yield more reads than a 15-minute incubation.

    We agree with the reviewer that the longer incubation period of PTA might explain the higher read count seen in the PTA samples, although the differences in amplification kinetics (linear in PTA, exponential in STD) and potential differences in amplification saturation points make it difficult to compare them. For instance, an updated version of PTA (ResolveDNA V2) uses a lower amplification time (2.5 h) and achieves similar amplification levels compared to the 10h incubation time, suggesting PTA amplification saturates well before the 10h time. In any case, all quality check metrics were done with the cells subsampled to 100 k reads to mitigate the effect of read count differences on the data quality.

    (3) There is a lack of quantitative evaluations of the SPCs. e.g. How many capsules were evaluated to assess doublets? How many capsules were detected as Syto5 positive in a successful vs an unsuccessful experiment?

    We agree with the reviewer but during experimental execution SPCs were only assessed qualitatively via microscopy following the Single-cell microbe DNA barcoding kit manual. No quantitative analysis was done and therefore we do not have this data. Regarding doublet, this was done in silico based on the detection of SPCs containing mixed genomes from the two strains used in the study as described in the Materials and Methods. As pointed by another reviewer, this only allow the detection of inter-strain doublets. In the revised version, we explain this and add an estimation of total doublets based on the inter-strain doublet rate.

    (4) The authors do not address some of the amplification results obtained under various conditions. For example, why did temperature-based lysis of STD4 lead to amplification failure? Also, what is the reason for fewer "true" cells (higher background) in the PTA samples compared to the STD samples? Is this related to issues with barcoding or, alternatively, substandard amplification as indicated by lower read amounts in some capsules (knee plots in Figure 1C)?

    After exchange with the technical support team of the SPC generator kit, it was clarified that the heat lysis done in STD4 should have had a shorter incubation time (10 minutes instead of 15 minutes). We suspect that the longer incubation time, combined with the higher temperature and the harsh lysis condition with 0.8M KOH might have damaged SPCs and therefore DNA might have leaked out of them before WGA. In the microscopy images, SPCs in STD4 show a swollen aspect not seen in the other samples. In the revised version we will explain this more clearly.

    (5) The paper presents limited biological relevance. Without this, the paper describes an improvement in genome amplification methods and some proof-of-concept analyses. Using a 1:1 mixture of parasites with different genotypes, the authors display the utility of the method to resolve genetic diversity, but they don't seek to understand the limits of detecting this diversity. For some, the authors do not comment on the mixed karyotypes from the HU3 cells (Figure 3F) other than to state that this line was not clonal. For CNVs, the two loci evaluated were detected at relatively high copy number (according to Figure 4C, they are between 4 and 20 copies). Thus, the sensitivity of CNV detection from this data remains unclear; can this approach detect lower-level CNVs like duplications, or minor CNVs that do not show up in every cell?

    As described above we will include more discussion on potential biological relevance of the method in the revised version of the manuscript. In the revised version we will attempt to use dedicated bioinformatic tools to discover de novo CNVs, as per the suggestion of other reviewers. This might also allow us to determine the detection limit of the methodology for CNVs.

    (6) The authors state that Leishmania can carry extrachromosomal copies of important genes. There is no discussion about how the presence of these molecules would affect the amplification steps and CNV detection. For example, the phi29 enzyme is very processive with circular molecules; does its presence lead to overamplification and overrepresentation in the data? Is this evident in the current study? This information would be useful for organisms that carry this type of genetic element.

    We believe our data, which uses short-read sequences, does not allow to differentiate between intra-chromosomal CNVs and linear or circular episomal CNVs, so we cannot define if circular CNVs are over-amplified. Of note, we have previously demonstrated that the M-locus CNV in chromosome 36 is intrachromosomal, not circular (episomal)[5].

    (7) The manuscript is missing a comparison with other similar studies in the field. For example, how does this coverage level compare to those achieved for other genomes? Can this method achieve amplification levels needed to assess larger genomes? Has there been any evaluation of base composition effects since Leishmania is a GC-rich genome?

    We believe the SPC-PTA workflow can be applied to organisms with larger genomes as PTA was developed specifically for mammalian cells[6], and also because, in our hands, it outperformed the 10X scDNA solution, which was developed for mammals.

    We believe direct comparison with other studies regarding coverage levels is elusive because other steps in the workflow apart from the WGA, such as the library preparation (PCR-based in our case), as well as genome features like GC content, size, and presence of repetitive regions, can also affect coverage levels and evenness. One strength of our approach was the use a single sample (the 50/50 mix between two L. donovani strain) for all conditions, thus removing potential parasite-specific biases. In addition, the application of a multiplexing system during barcoding allowed us to combine all samples prior to library preparation, thus removing potential differences introduced by this step.

    Regarding the effect of GC-content, we did notice a positive bias in all samples in regions with higher GC content, which had to be corrected in silico. This was the opposite to a negative bias observed in previous study[7] likely due to differences in WGA and/or library preparation. In the revised version, we will include a supplementary figure showing the GC bias.

    (8) Cost is mentioned as a benefit of the SPC platform, and savings are achieved when working in a plate format, but no details are included on how this was evaluated.

    In the revised version we will provide precise cost estimates and the rationale for the estimation.

    (9) The Zenodo link for custom scripts does not exist, and code cannot be evaluated.

    The full Zenodo link (https://doi.org/10.5281/zenodo.17094083) will be included in the revised version.

    Reviewer #3 (Public review):

    Summary

    In this manuscript, Negreira et al. propose a new scDNAseq method, using semi-permeable capsules (SPCs) and primary template-directed amplification (PTA). The authors optimize several metrics to improve their predictions, such as determining GC bias, Intra-Chromosomal fluctuation (ICF -metric to differentiate replicative and non-replicative cells) and Intra-chromosomal coefficient of variation (ICCV - chromosome read distribution). The coverage evenness was evaluated using the fini index and the median absolute pairwise difference between the counts of two consecutive bins. They validate the proposed method using two Leishmania donovani strains isolated from different countries, BPK081 (low genomic variability) and HU3 (high genomic variability). Then, they showed that the method outperforms WGA and has similar accuracy to the discontinued 10X-scDNA (10X Genomics), further improving on short CNV identification. The authors also show that the method can identify somy variations, insertions/deletions and SNP variations across cells. This is a timely and very relevant work that has a wide applicability in copy number variation assessment using single-cell data.

    Strengths

    I really appreciate this work. My congratulations to the authors. All my comments below only aim to improve an already solid manuscript.

    We thank the reviewer for the enthusiasm and positive feedback.

    Weaknesses

    (1) Data availability: Although the authors provide a Zenodo link, the data is restricted. I also could not access the GitHub link in the Zenodo website: https://github.com/gabrielnegreira/2025_scDNA_paper. The authors should make these files available.

    Both the Zenodo (https://doi.org/10.5281/zenodo.17094083) and the GitHub (https://github.com/gabrielnegreira/2025_scDNA_paper) repositories are now publicly available.

    (2) 2-SPC-PTA and SPC-STD cell count comparison: The authors have consistently proven that the SPC-PTA method was superior to SPC-STD. However, there are a few points that should be clarified regarding the SPC-PTA results. Is there an explanation for the lower proportion of SPC to true cells success in SPC-STD, which reflects the bimodal distribution for the reads per cell in SPC-PTA2 and a three-to-multimodal distribution in SPC-PTA1 in Figure 1B? Also, in Table 1, does the number of reads reflect the number of reads in all sequenced SPCs or only in the true cells? If it is in the SPCs, I suggest that the authors add a new column in the table with the "Number of reads in true cells" to account for this discrepancy.

    The reason for the higher presence of 'background' SPCs in the PTA samples is not clear, but we hypothesize that it could be due to PTA favoring amplification of small, free floating DNA molecules that might have been trapped in cell-free SPCs, as PTA works with shorter amplicons. Also, the longer incubation time seen in PTA (10 h) might have allowed enhanced amplification of low quantities of free-floating DNA to detectable levels. Regarding Table 1, indeed it only show the total number of reads per sample. In the revised version we will include the suggested column to Table 1.

    (3) The authors should evaluate the results with a higher coverage for SCP-PTA. I understand that the authors subsampled the total read to 100,000 to allow cross-sample comparisons, especially between SPC-STD and SPC-PTA. However, as they concluded that the SPC-PTA was far superior, and the samples SPC-PTA1 and SPC-PTA2 had an "elbow" of 650,493 and 448,041, respectively, it might be interesting to revisit some of the estimations using only SPC-PTA samples and a higher coverage cutoff, as 400,000.

    We believe the 100.000 cutoff is already high for aneuploidy analysis as we have successfully reconstructed parasite karyotype with 20.000 reads per cell8, so a higher cutoff will likely not improve it. For CNV analysis, in the revised version, we will try to identify de novo CNVs using dedicated bioinformatic tools as per other reviewer suggestions. There, we will also test if a higher CNV detection sensitivity is achieved using the suggested 400,000 reads cutoff for the PTA samples.

    (4) Doublet detection: I suggest that the authors be a little more careful with their definition of doublets. The doublet detection was based on diagnostic SNPs from the two strains, BPK081 and HU3, which identify doublets between two very different and well-characterised strains. However, this method will probably not identify strain-specific doublets. This is of minor importance for cloned and stable strains with few passages, as BPK081, but might be more relevant in more heterogeneous strains, as HU3. Strain-specific doublets might also be relevant in other scenarios, as multiclonal infections with different populations from the same strain in the same geographic area. One positive point is that the "between strain doublet count" was low, so probably the within-strain doublet count should be low too. The manuscript would benefit from a discussion on this regard.

    We fully agree with the reviewer. We will make it clear in the revised version that we quantify inter-strain doublets only, and we will also provide an estimation of total doublets based on the inter-strain doublet rate.

    (5) Nucleotide sequence variants and phylogeny: I believe that a more careful description of the phylogenetic analysis and some limitations of the sequence variant identification would benefit the manuscript.

    (5.1) As described in the methods, the authors intentionally selected two fairly different Leishmania donovani strains, HU3 and BPK081, and confirmed that the sequent variant methodology can separate cells from each strain. It is a solid proof of concept. However, most of the multiclonal infections in natural scenarios would be caused by parasite populations that diverge by fewer SNPs, and will be significantly harder to detect. Hence, I suggest that a short discussion about this is important.

    We will add a short discussion clarifying the limitations, while noting that our data demonstrate the ability of the approach to resolve very closely related cells, as illustrated by the fine-scale genetic differences observed within the clonal BPK081 population and by the detection of rare variants at targeted loci. We will also emphasize that the sensitivity to detect closely related genotypes depends on sequencing depth and the genomic regions considered.

    (5.2) The authors should expand on the description of the phylogenetic tree. In the HU3 on Figure 5F left panel, most of the variation is observed in ~8 cells, which goes from position 0 to position ~28.000. Most of the other cells are in very short branches, from ~29.000 to 30.4000 (5F right panel). Assuming that this representation is a phylogram, as the branches are short, these cells diverge by approximately 100-2000 SNPs. It is unexpected (but not impossible) that such ~8 divergent cells be maintained uniquely (or in very low counts) in the culture, unless this is a multiclonal infection. I would carefully investigate these cells. They might be doublets or have more missing data than other cells. I would also suggest that a quick discussion about this should be added to the manuscript.

    In the revised version we will improve the description of the phylogenetic analysis. We will also investigate deeper the 8 mentioned cells to define if they have confounding factors that might have led to their discrepancy. The possibility of multiclonal infection in HU3 is not excluded as this strain was not cloned after isolation.

    References:

    (1) Dileep, V., Gilbert, D. M., Dileep, V. & Gilbert, D. M. Single-cell replication profiling to measure stochastic variation in mammalian replication timing. Nat. Commun. 9, 427 (2018).

    (2) Miura, H. et al. Single-cell DNA replication profiling identifies spatiotemporal developmental dynamics of chromosome organization. Nat. Genet. 51, 1356–1368 (2019).

    (3) Marques, C. A. et al. Genome-wide mapping reveals single-origin chromosome replication in Leishmania, a eukaryotic microbe. Genome Biol. 16, 230 (2015).

    (4) Damasceno, J. D. et al. Leishmania major chromosomes are replicated from a single high-efficiency locus supplemented by thousands of lower efficiency initiation events. Cell Rep. 44, 116094 (2025).

    (5) Imamura, H. et al. Evolutionary genomics of epidemic visceral leishmaniasis in the Indian subcontinent. eLife 5, e12613 (2016).

    (6) Gonzalez-Pena, V. et al. Accurate genomic variant detection in single cells with primary template-directed amplification. Proc. Natl. Acad. Sci. 118, e2024176118 (2021).

    (7) Imamura, H. et al. Evaluation of whole genome amplification and bioinformatic methods for the characterization of Leishmania genomes at a single cell level. Sci. Rep. 10, 15043 (2020).

    (8) Negreira, G. H. et al. High throughput single-cell genome sequencing gives insights into the generation and evolution of mosaic aneuploidy in Leishmania donovani. Nucleic Acids Res. 50, 293–305 (2022).

  14. Version published to 10.7554/elife.109350.1 on eLife
  15. Version published to 10.1101/2025.09.10.675331v2 on bioRxiv
  16. Mentioned by @gnegreira.bsky.social on Bluesky

    This is just the beginning! We hope this method will help the community to explore genome evolution and parasite adaptation in Leishmania and beyond! For more info, check the preprint in @biorxivpreprint.bsky.social: www.biorxiv.org/content/10.1...

  17. Mentioned by @gnegreira.bsky.social on Bluesky

    Happy to share our latest preprint, where we present a novel high-throughput #single-cell DNA method that resolves genomic heterogeneity in #Leishmania, from #aneuploidy, #CNVs to #drug-resistance mutations. 🧵... Link here: www.biorxiv.org/content/10.1...

  18. Mentioned by @kinetoplastpapers.bsky.social on Bluesky

    A novel high-throughput single-cell DNA sequencing method reveals hidden genomic heterogeneity in the unicellular eukaryote Leishmania www.biorxiv.org/content/10.1101/2025.09.10.675331v1

  19. Mentioned by @prepub-singlecell.bsky.social on Bluesky

    A novel high-throughput single-cell DNA sequencing method reveals hidden genomic heterogeneity in the unicellular eukaryote Leishmania #SingleCell 🧪🧬🖥️ https://www.biorxiv.org/content/10.1101/2025.09.10.675331v1

  20. Mentioned by @biorxiv-molbio.bsky.social on Bluesky

    A novel high-throughput single-cell DNA sequencing method reveals hidden genomic heterogeneity in the unicellular eukaryote Leishmania https://www.biorxiv.org/content/10.1101/2025.09.10.675331v1

  21. Mentioned by @biorxivpreprint.bsky.social on Bluesky

    A novel high-throughput single-cell DNA sequencing method reveals hidden genomic heterogeneity in the unicellular eukaryote Leishmania https://www.biorxiv.org/content/10.1101/2025.09.10.675331v1

  22. Version published to 10.1101/2025.09.10.675331v1 on bioRxiv