Review Reports - Gradient-Free Isolation of Murine Pancreatic Islets for Single-Cell RNA Sequencing

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This manuscript presents a practical and clearly illustrated density-gradient-free workflow for pancreatic islet isolation and subsequent single-cell RNA sequencing. The protocol is well organized, and the workflow diagram and step-by-step images should be useful to researchers working with metabolically vulnerable islets. However, several methodological and reporting issues should be addressed to further improve the reproducibility and assessment of the practical utility of the protocol. Major revision is recommended.

The claimed time-saving advantage is not fully demonstrated, as the gradient-free workflow still requires 15-30 minutes of manual islet selection per dish. Please provide a clearer comparison of the total processing time and hands-on workload for the two workflows. A simple side-by-side summary would help readers evaluate the practical advantage of the proposed method.
Total cell number and viability alone may not fully reflect islet-isolation performance. If available, the authors are encouraged to provide the number of islets recovered per pancreas and an estimate of islet purity before dissociation. Capture efficiency should also be reported, given the difference between the number of cells loaded and those ultimately identified. In addition, the lack of direct assessment of islet function should be stated as a limitation.
The cell numbers reported across the scRNA-seq QC steps are not entirely clear, particularly because the final cell counts appear unchanged after doublet removal. Please clarify the order of the QC steps and report the number of cells retained or removed at each stage. A concise QC table summarizing the main sequencing and filtering metrics for each library would improve transparency and reproducibility.
The two scRNA-seq libraries were sequenced at substantially different depths, with approximately 87000 reads per cell in the Ficoll library and 47000 reads per cell in the gradient-free library. This difference may affect gene detection and stress-related gene-set scores. A depth-matched down-sampling or sequencing saturation analysis would help ensure a more appropriate comparison between the two libraries.
Several important protocol details should be reported more clearly, including the dose and route of Avertin, collagenase activity or lot information, preparation of the Ficoll solutions, shaker and rotor specifications, and the dimensions of the wide-bore tips. Reporting these details would further improve reproducibility.
The raw FASTQ files, raw and filtered expression matrices, cell annotations, mouse-level metadata, and analysis scripts should be deposited in a public repository to improve data transparency and reproducibility.

Author Response

Response to Reviewer 1 Comments

1. Summary

We sincerely thank the Reviewer for the careful evaluation of our manuscript and for the constructive comments and suggestions. We have carefully considered all points raised and revised the manuscript accordingly. Detailed responses to each comment are provided below, and the corresponding changes are highlighted in the revised manuscript. We believe that these revisions have improved the clarity, methodological transparency, and overall presentation of the work.

2. Questions for General Evaluation

Reviewer’s Evaluation

Response and Revisions

Does the introduction provide sufficient background and include all relevant references?

Yes/Can be improved/Must be improved/Not applicable

Are all the cited references relevant to the research?

Yes/Can be improved/Must be improved/Not applicable

Is the research design appropriate?

Yes/Can be improved/Must be improved/Not applicable

Are the methods adequately described?

Yes/Can be improved/Must be improved/Not applicable

Are the results clearly presented?

Yes/Can be improved/Must be improved/Not applicable

Are the conclusions supported by the results?

Yes/Can be improved/Must be improved/Not applicable

3. Point-by-point response to Comments and Suggestions for Authors

Comments 1: The claimed time-saving advantage is not fully demonstrated, as the gradient-free workflow still requires 15-30 minutes of manual islet selection per dish. Please provide a clearer comparison of the total processing time and hands-on workload for the two workflows. A simple side-by-side summary would help readers evaluate the practical advantage of the proposed method.

Response 1: We thank the reviewer for this comment. We have added a side-by-side comparison of the processing times for the gradient-free and Ficoll-based workflows (Table 4). Importantly, manual islet selection (15–30 min per dish) was performed in both workflows and therefore does not account for the difference in processing time. The gradient-free workflow required an estimated 75–110 min, whereas the Ficoll-based workflow required approximately 110–145 min. The difference results from the Ficoll purification module, which adds approximately 35 min, including 21.5 min of centrifugation and approximately 15 min of Ficoll-specific handling for gradient preparation, interphase collection, dilution, and transfers. We have also revised the text to describe this advantage specifically in terms of the
omitted Ficoll-related processing steps rather than implying a general reduction in all hands-on manipulation.

Comments 2: Total cell number and viability alone may not fully reflect islet-isolation performance. If available, the authors are encouraged to provide the number of islets recovered per pancreas and an estimate of islet purity before dissociation. Capture efficiency should also be reported, given the difference between the number of cells loaded and those ultimately identified. In addition, the lack of direct assessment of islet function should be stated as a limitation.

Response 2: We agree that islet yield and pre-dissociation purity would provide useful additional information, but these parameters were not prospectively recorded and cannot be reconstructed reliably. Because mouse islets vary markedly in size and cell content, islet number alone does not directly reflect the amount of material available for scRNA-seq. We therefore focused on viable cell recovery and viability after dissociation, which were the most relevant input measures for library preparation.

Capture efficiency was estimated from the recorded loading data. Approximately 24,000 viable cells were loaded for each library, yielding 5,568 initially identified cells in the gradient-free preparation and 4,358 in the Ficoll-based preparation, corresponding to 23.2% and 18.2%, respectively. After QC filtering, 2,659 and 2,715 cells were retained for analysis. These values have now been added to the manuscript.

Direct functional testing of the isolated islets, such as glucose-stimulated insulin secretion, was not performed. We therefore cannot draw conclusions about preservation of islet function, and this is now stated as a limitation.

Lines

It was originally in the text

Corrections

500-512

Before evaluating cell composition, the main QC parameters of the scRNA-seq libraries were examined. The distributions of the number of detected genes per cell largely overlapped between the Ficoll-based and gradient-free libraries (Figure 6A). A similar pattern was observed for the total number of transcript counts per cell (Figure 6B). The proportion of mitochondrial transcripts remained low in both libraries (Figure 6C), while the proportion of ribosomal transcripts was also within a comparable range (Figure 6D). The distributions of these QC metrics broadly overlapped between the two libraries.

Approximately 24,000 viable cells were loaded for each library. The gradient-free library yielded 5,568 initially identified cells, corresponding to an estimated capture efficiency of 23.2%, whereas the Ficoll-based library yielded 4,358 cells, corresponding to 18.2%.

Before evaluating cell composition, the main QC parameters of the scRNA-seq libraries were examined. After QC filtering, 2,659 cells from the gradient-free library and 2,715 cells from the Ficoll-based library were retained for downstream analysis. The distributions of the number of detected genes per cell largely overlapped between the Ficoll-based and gradient-free libraries (Figure 6A). A similar pattern was observed for the total number of transcript counts per cell (Figure 6B). The proportion of mitochondrial transcripts remained low in both libraries (Figure 6C), while the proportion of ribosomal transcripts was also within a comparable range (Figure 6D). The distributions of these QC metrics broadly overlapped between the two libraries.

679-694

The principal limitation is the absence of matched biological replicates for the two workflows. Each analyzed library was generated from pooled material, and variation among individual animals could therefore not be assessed. Consequently, differences in cell composition and transcriptional state should be regarded as descriptive rather than as evidence of a method-specific effect. A direct comparison using several independently prepared, matched libraries would be required to determine whether the gradient-free purification reproducibly reduces processing-associated stress while preserving cell recovery and cellular composition

The principal limitation is the absence of matched biological replicates for the two workflows. Each analyzed library was generated from pooled material, and variation among individual animals could therefore not be assessed. Inter-operator reproducibility was not assessed because all isolations were performed by a single operator. Consequently, differences in cell composition and transcriptional state should be regarded as descriptive rather than as evidence of a method-specific effect. In addition, islet yield per pancreas and pre-dissociation purity were not prospectively quantified. Although these parameters are useful for benchmarking isolation performance, islet number alone does not directly reflect the amount of material available for scRNA-seq because individual mouse islets vary in size and cell content; viable cell recovery after dissociation was therefore used as the primary input-related measure. Functional properties of the isolated islets were also not assessed, and the present data therefore do not establish preservation of glu-cose-responsive endocrine function. A direct comparison using several independently prepared, matched libraries would be required to determine whether the gradient-free workflow reproducibly affects cell recovery, cellular composition, or processing-associated transcriptional responses.

Comments 3: The cell numbers reported across the scRNA-seq QC steps are not entirely clear, particularly because the final cell counts appear unchanged after doublet removal. Please clarify the order of the QC steps and report the number of cells retained or removed at each stage. A concise QC table summarizing the main sequencing and filtering metrics for each library would improve transparency and reproducibility.

Response 3: We thank the reviewer for pointing out that the order of the scRNA-seq QC steps and the reported cell numbers were not sufficiently clear in the previous version. We have revised the Bioinformatic Processing and Cell-Type Annotation section to describe the QC procedure in its sequential order and to distinguish the initially identified cells, cells retained after primary QC, cells remaining after doublet removal, and the final endocrine/exocrine population used for downstream analysis. We have also removed the duplicated reporting of the final cell numbers.

In addition, we added Table 2, which summarizes the main cell-retention metrics at each QC stage for both libraries. The table also clarifies that the final dataset of 2,659 gradient-free cells and 2,715 Ficoll-based cells represents the pancreatic endocrine/exocrine subset used for downstream analysis rather than the total number of cells remaining immediately after doublet removal. These revisions were made to improve transparency and reproducibility of the scRNA-seq preprocessing workflow.

Lines

It was originally in the text

Corrections

386-406

The gradient-free library yielded 5,568 estimated cells with a mean sequencing depth of 46,913 reads per cell, whereas the Ficoll-based library contained 4,358 estimated cells with 87,251 reads per cell. The median number of detected genes per cell was 1,770 and 2,532, respectively.

Quality control filtering was performed based on transcript complexity and mitochondrial transcript content. Cells with fewer than 200 detected genes, fewer than 500 total transcripts, or mitochondrial transcript content exceeding 10% were excluded from downstream analysis. Following filtering, the final dataset consisted of 2,659 cells from the gradient-free library and 2,715 cells from the Ficoll-based library. The median number of detected genes per cell was 3,480 and 4,161, respectively, while the median mitochondrial transcript fraction remained low (1.89% and 1.94%). The distributions of transcript complexity and mitochondrial content broadly overlapped between the two libraries. Putative doublets were identified separately within each library using Scrublet [17] as implemented in Scanpy v1.12.1, with an expected doublet rate of 0.05, a simulated to observed doublet ratio of 2.0, 30 principal components, and an automatically determined doublet-score threshold. Predicted doublets were excluded from downstream analysis. Ambient RNA correction was performed separately for each library using the remove background module of CellBender [18] with the full noise model, 150 training epochs, a target false-positive rate of 0.01, and a learning rate of 1 × 10⁻⁴. The final analyzed dataset contained 2715 cells from the Ficoll-based library and 2659 cells from the gradient-free library.

The gradient-free library yielded 5,568 initially identified cells with a mean sequencing depth of 46,913 reads per cell, whereas the Ficoll-based library contained 4,358 initially identified cells with 87,251 reads per cell. The median number of detected genes per cell was 1,770 and 2,532, respectively.

Quality control filtering was performed separately for each library and followed a defined sequence. Cells with fewer than 200 detected genes, fewer than 500 total transcripts, or mitochondrial transcript content exceeding 10% were excluded from downstream analysis. Putative doublets were identified separately within each library using Scrublet [17] as implemented in Scanpy v1.12.1, with an expected doublet rate of 0.05, a simulated to observed doublet ratio of 2.0, 30 principal components, and an automatically determined doublet-score threshold. Ambient RNA correction was performed separately for each library using the remove background module of CellBender [18] with the full noise model, 150 training epochs, a target false-positive rate of 0.01, and a learning rate of 1 × 10⁻⁴. Following completion of the QC and preprocessing steps, 2,659 cells from the gradient-free library and 2,715 cells from the Ficoll-based library were retained for downstream analysis. The median number of detected genes per cell was 3,480 and 4,161, respectively, while the median mitochondrial transcript fraction remained low (1.89% and 1.94%). The distributions of transcript complexity and mitochondrial content broadly overlapped between the two libraries. Predicted doublets were excluded from downstream analysis. The final analyzed dataset contained 2715 cells from the Ficoll-based library and 2659 cells from the gradient-free library (Table 2).

Comments 4: The two scRNA-seq libraries were sequenced at substantially different depths, with approximately 87000 reads per cell in the Ficoll library and 47000 reads per cell in the gradient-free library. This difference may affect gene detection and stress-related gene-set scores. A depth-matched down-sampling or sequencing saturation analysis would help ensure a more appropriate comparison between the two libraries.

Response 4: We thank the reviewer for this important comment. We agree that the difference in sequencing depth between the two libraries may influence gene detection and stress-response scores. To address this point, we added a summary of sequencing and QC metrics for both libraries in Table 1, including mean reads per cell, sequencing saturation, valid barcode rates, Q30 metrics, genome mapping rates, and total genes detected. Although sequencing depth and saturation differed between the libraries, the other sequencing and mapping quality metrics were broadly similar. We have also revised the Discussion to explicitly acknowledge unequal sequencing depth as a limitation and to clarify that the observed gene-expression and stress-response differences should be interpreted descriptively. A depth-matched downsampling analysis was not performed in the present revision, and we have therefore avoided attributing these differences specifically to the isolation method.

Lines	It was originally in the text	Corrections
377-380	-	Sequencing and mapping quality metrics for both libraries are summarized in Table 1. Although sequencing depth and saturation differed between the libraries, the remaining sequencing and mapping quality metrics were broadly similar.
616-620	-	Despite differences in sequencing depth and saturation, other sequencing and mapping quality metrics were broadly similar between the libraries (Table 1). Unequal sequencing depth was nevertheless considered when interpreting gene-expression and stress-response differences.

Comments 5: Several important protocol details should be reported more clearly, including the dose and route of Avertin, collagenase activity or lot information, preparation of the Ficoll solutions, shaker and rotor specifications, and the dimensions of the wide-bore tips. Reporting these details would further improve reproducibility.

Response 5: We thank the reviewer for this suggestion. We have revised the Materials and Methods to provide the requested protocol details. Specifically, we added the dose, concentration, injection volume, and intraperitoneal route of Avertin; collagenase activity and lot information; clarification that the commercial Ficoll solutions were used as supplied and pre-cooled to 4 °C; shaker model, incubation conditions, centrifuge and rotor specifications; and the manufacturer and distal opening diameters of the 1000-µL and 200-µL wide-bore tips. These details have been added throughout the relevant protocol sections to improve reproducibility.

Lines	It was originally in the text	Corrections
177-183	-	Before pancreatic perfusion, mice were anesthetized with tribromoethanol (Avertin) administered intraperitoneally at a dose of 250 mg/kg. A 2.5% working solution was used, corresponding to an injection volume of 10 µL/g body weight. After confirmation of deep anesthesia, the subsequent procedure was performed in accordance with the approved institutional animal protocol. Pancreatic perfusion and tissue collection were performed as a terminal procedure, and animals were euthanized by cervical dislocation while under deep anesthesia.
162-164	-	A collagenase solution containing 3 mg/mL collagenase type I (act 270 U/mg, Lot. 44N24865, Worthington Biochemical, USA) and 3 mg/mL collagenase type II (act: 340 U/mg, dry. Lot. 41B20909, Worthington Biochemical, USA)
202-206	The pancreas was minced with scissors and transferred to a 50-mL tube containing 5 mL of collagenase solution (Figure 2E). The tissue was incubated at 37 °C for no longer than 15 min (Figure 3).	The pancreas was minced with scissors and transferred to a 50-mL tube containing 5 mL of collagenase solution (Figure 2E). The tissue was incubated at 37 °C for no longer than 15 min on MR-1 Mini-Rocker Shaker (BioSan, Latvia) at 30 rpm using heat-insulating beads, placed inside MCO-170AC-PE CO2 Incubator, maintained at 37 °C (PHCbi, Japan) (Figure 3).
215-216	The suspension was centrifuged at 200 × g for 2 min at 4 °C.	The suspension was centrifuged at 200 × g for 2 min at 4 °C, using an Eppendorf Cen-trifuge 5810 R (Eppendorf, Germany) equipped with an A-4-81 swing-bucket rotor.
228-230	The pellet was resuspended in 10 mL of cold HBSS containing 0.3% BSA and 100 U/mL DNase I using wide-bore pipette tips.	The pellet was resuspended in 10 mL of cold HBSS containing 0.3% BSA and 100 U/mL DNase I using 1000 µL wide-bore pipette tips (GenFollower, China, distal opening diameter, 2.05 mm).
233-235	The wash was combined with the main suspension, which was then gently pipetted 5–7 times. The suspension was centrifuged at 200 × g for 2 min at 4 °C.	The wash was combined with the main suspension, which was then gently pipetted 5–7 times. The suspension was centrifuged at 200 × g for 2 min at 4 °C, using an Eppendorf Centrifuge 5810 R equipped with an A-4-81 swing-bucket rotor.
240-242	-	Commercially prepared Ficoll solutions (PanEco, Russia) with densities of 1.100 and 1.077 g/mL were used as supplied by the manufacturer, without further dilution or density adjustment, and were pre-cooled to 4 °C before use.
247-252	The gradient was centrifuged at 900 × g for 18 min at 4 °C with the brake disabled. The islet-enriched interphase was collected into a new 50-mL tube, diluted to 50 mL with cold RPMI 1640, and washed twice at 200 × g and 4 °C for 2 min and 90 s, respectively.	The gradient was centrifuged at 900 × g for 18 min at 4 °C with the brake disabled, using an Eppendorf Centrifuge 5810 R equipped with an A-4-81 swing-bucket rotor. The islet-enriched interphase was collected into a new 50-mL tube, diluted to 50 mL with cold RPMI 1640, and washed twice at 200 × g and 4 °C for 2 min and 90 s, respectively, using an Eppendorf Centrifuge 5810 R equipped with an A-4-81 swing-bucket rotor.
291-299	For up to 300 islets, 1 mL of Accutase containing EDTA (STEMCELL Technologies, Canada) was used. The islets were washed, centrifuged at 100 × g for 3 min and the supernatant was removed. One millilitre of Accutase with EDTA pre-warmed to 37 °C was then added. The islets were incubated at 37 °C for 10–15 min. Every 3 min, the suspension was gently pipetted using a 200-µL wide-bore tip to promote gradual dissociation into single cells.	For up to 300 islets, 1 mL of Accutase containing EDTA (STEMCELL Technologies, Canada) was used. The islets were washed, centrifuged at 100 × g for 3 min, using an Eppendorf Centrifuge 5415 R equipped with an FA-45-24-11 rotor and the supernatant was removed. One millilitre of Accutase with EDTA pre-warmed to 37 °C was then added. The islets were incubated at 37 °C for 10–15 min on MR-1 Mini-Rocker Shaker at 30 rpm using heat-insulating beads, placed inside MCO-170AC-PE CO2 Incubator, maintained at 37 °C. Every 3 min, the suspension was gently pipetted using a 200-µL wide-bore tip (GenFollower, China, distal opening diameter, 1.50 mm) to promote gradual dissociation into single cells.
305-307	The suspension volume was adjusted to 15 mL with cold PBS without Ca²⁺ or Mg²⁺ supplemented with 0.1% BSA. Cells were centrifuged at 200 × g for 3 min at 4 °C.	The suspension volume was adjusted to 15 mL with cold PBS without Ca²⁺ or Mg²⁺ supplemented with 0.1% BSA. Cells were centrifuged at 200 × g for 3 min at 4 °C, using an Eppendorf Centrifuge 5810 R equipped with an A-4-81 swing-bucket rotor.
325-326	Cells were centrifuged at 200 × g for 3 min at 4 °C.	Cells were centrifuged at 200 × g for 3 min at 4 °C, using an Eppendorf Centrifuge 5415 R equipped with an FA-45-24-11 rotor.
342-344	The suspension was mixed gently by pipetting and centrifuged at 1000 × g for 5 min at 4 °C.	The suspension was mixed gently by pipetting and centrifuged at 1000 × g for 5 min at 4 °C, using an Eppendorf Centrifuge 5415 R equipped with an FA-45-24-11 rotor.

Comments 6: The raw FASTQ files, raw and filtered expression matrices, cell annotations, mouse-level metadata, and analysis scripts should be deposited in a public repository to improve data transparency and reproducibility.

Response 6: We thank the reviewer for this suggestion. The analysis scripts used for scRNA-seq preprocessing, cell-type annotation, and figure generation have been deposited in a public GitHub repository: https://github.com/RuchkoEvgeny/Gradient-Free-Isolation-of-Murine-Pancreatic-Islets-for-Single-Cell-RNA-Sequencing. The raw and processed sequencing data are currently being prepared for deposition in the NCBI Gene Expression Omnibus (GEO). We expect the GEO accession number to become available during September 2026 and will add it to the manuscript as soon as it is assigned. The dataset will be made publicly available prior to publication. The Data Availability Statement has been revised accordingly.

Lines

It was originally in the text

Corrections

766-770

Data Availability Statement: The data generated and analyzed during the current study are available from the corresponding author upon reasonable request.

Data Availability Statement: Raw and processed sequencing data generated in this study have been deposited at GEO NCBI (accession number: GSEXXXXX) and are publicly available as of the date of publication. All original code has been deposited at Github (https://github.com/RuchkoEvgeny/Gradient-Free-Isolation-of-Murine-Pancreatic-Islets-for-Single-Cell-RNA-Sequencing) and is publicly available as of the date of publication.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript presents a gradient-free workflow for isolation of murine pancreatic islets for subsequent single-cell RNA sequencing. The authors replace Ficoll-based density-gradient purification with stereomicroscope-guided manual islet selection and show that the resulting preparations have high cell viability, retain the major pancreatic endocrine populations, and generate scRNA-seq libraries of acceptable quality. The protocol is potentially useful; however, several claims regarding reduced processing, stress, and reproducibility require clarification or more cautious interpretation.

1. The manuscript should more clearly define whether the gradient-free procedure is intended as an alternative to Ficoll-based density-gradient purification or as a superior/improved method. If superiority is claimed, this would require stronger comparative evidence demonstrating clear advantages in parameters such as overall processing time, cell recovery, viability, reproducibility, and/or preservation of transcriptional state. Otherwise, the current data support presenting the gradient-free workflow as a feasible alternative that produces cells suitable for scRNA-seq.

2. The authors emphasize that the gradient-free protocol eliminates three centrifugation steps and at least 21.5 minutes of centrifuge run time. However, stereomicroscope-guided manual islet selection itself requires substantial hands-on time, reported as 15–30 minutes per dish, and therefore the manuscript should provide total elapsed and hands-on processing times for both workflows before concluding that the new procedure is faster or less time-consuming.

3. Manual islet selection is inherently operator-dependent and may introduce greater person-to-person variability than a standardized density-gradient procedure. The authors should clarify whether isolations were performed by a single operator or multiple operators and, if only one operator was involved, discuss the lack of inter-operator validation as an important limitation of the claimed reproducibility.

4. Relatedly, manual selection may introduce selection bias because operators may preferentially recover larger, more intact, or more easily recognizable islets while excluding less obvious or partially disrupted islets. This could influence islet yield, cell-type composition, and downstream transcriptomic profiles, and the authors should discuss how manual selection criteria were standardized and whether any objective criteria were used.

5. The interpretation that the gradient-free procedure may reduce processing-associated cellular stress should be substantially tempered. Manual microscopic selection, repeated aspiration and transfer, prolonged handling, and time outside optimal physiological conditions may themselves induce mechanical or transcriptional stress, so removal of Ficoll centrifugation cannot by itself be assumed to reduce total cellular stress.

6. The stress-response comparison is particularly limited because the three independently isolated mice were pooled to generate only one scRNA-seq library for each workflow. The authors appropriately acknowledge that these libraries represent single experimental units and that the observed differences cannot be attributed to the isolation method; therefore, statements in the Abstract, Results, and Discussion suggesting reduced stress with the gradient-free protocol should consistently be presented as descriptive observations rather than evidence of a method-specific effect.

7. The Introduction may benefit from briefly acknowledging the broader relevance of pancreatic molecular alterations and emerging molecular biomarkers. In this context, the authors may consider citing the recent review https://doi.org/10.3389/freae.2025.1723159, which provides a useful overview of molecular and epigenetic changes in pancreatic disease and their translational relevance.

8. Because density-gradient purification is a well-established and relatively standardized method, the practical value of replacing it with manual selection would be strengthened by a more comprehensive comparison of the two approaches. In addition to viability and scRNA-seq QC, useful parameters would include total processing time, hands-on time, islet yield, purity, operator-to-operator variability, and reproducibility across independent scRNA-seq libraries; without such comparisons, the gradient-free approach is best presented as a feasible alternative rather than an improved replacement for density-gradient purification.

Author Response

Response to Reviewer 2 Comments

1. Summary

We are grateful to the Reviewer for the thorough assessment of our manuscript and for the helpful comments and recommendations. Each point has been considered carefully, and the manuscript has been revised accordingly. Our detailed responses are provided below, with the corresponding modifications highlighted in the revised manuscript. We believe that the revised version more clearly defines the scope, limitations, and practical relevance of the study.

2. Questions for General Evaluation

Reviewer’s Evaluation

Response and Revisions

Does the introduction provide sufficient background and include all relevant references?

Yes/Can be improved/Must be improved/Not applicable

Are all the cited references relevant to the research?

Yes/Can be improved/Must be improved/Not applicable

Is the research design appropriate?

Yes/Can be improved/Must be improved/Not applicable

Are the methods adequately described?

Yes/Can be improved/Must be improved/Not applicable

Are the results clearly presented?

Yes/Can be improved/Must be improved/Not applicable

Are the conclusions supported by the results?

Yes/Can be improved/Must be improved/Not applicable

3. Point-by-point response to Comments and Suggestions for Authors

Comments 1: The manuscript should more clearly define whether the gradient-free procedure is intended as an alternative to Ficoll-based density-gradient purification or as a superior/improved method. If superiority is claimed, this would require stronger comparative evidence demonstrating clear advantages in parameters such as overall processing time, cell recovery, viability, reproducibility, and/or preservation of transcriptional state. Otherwise, the current data support presenting the gradient-free workflow as a feasible alternative that produces cells suitable for scRNA-seq.

Response 1: We thank the reviewer for this important comment. We have revised the manuscript throughout to clarify that the gradient-free workflow is presented as a feasible alternative to Ficoll-based purification, not as a superior method. Statements implying improved reproducibility, overall quality, or reduced cellular stress were removed or reformulated. The revised text now focuses on the demonstrated features of the workflow: omission of Ficoll-specific processing steps, recovery of viable cells suitable for scRNA-seq, fulfillment of the applied library QC criteria, and retention of the major endocrine populations.

We also explicitly state in the Introduction and Conclusions that the present study was not designed to establish superiority and that stronger comparative evidence with independently prepared matched libraries would be required for such a conclusion.

Lines	It was originally in the text	Corrections
14-16	This study aimed to adapt a reproducible protocol for preparing murine pancreatic islets for downstream single-cell transcriptomic analysis while reducing the duration and complexity of sample processing.	This study aimed to adapt a practical gradient-free alternative for preparing murine pancreatic islets for downstream single-cell transcriptomic analysis while reducing the number of sample-processing steps.
29-32	These findings support the use of gradient-free isolation for preparing metabolically vulnerable pancreatic islets for scRNA-seq analysis.	These findings support gradient-free isolation as a possible alternative to Ficoll-based purification for preparing murine pancreatic islets for scRNA-seq analysis.
119-123	The protocol was designed to eliminate Ficoll-based density-gradient purification, reduce the number and duration of processing steps, and limit additional stress imposed on the cells	The protocol was designed to omit Ficoll-based density-gradient purification, reduce the number of centrifugation, washing, and transfer steps before single-cell preparation. Its workability for scRNA-seq was assessed based on cell recovery, viability, library QC characteristics, cell-type representation, and descriptive analysis of stress-associated transcriptional features.
129-130	-	The aim was therefore to evaluate the gradient-free workflow as a practical alternative to Ficoll-based purification rather than to establish its superiority.
403-404	The distributions of transcript complexity and mitochondrial content indicated comparable library quality between the two workflows.	The distributions of transcript complexity and mitochondrial content broadly overlapped between the two libraries.
487-489	Cell viability exceeded 90% in both preparations (Figure 5B), indicating that omission of Ficoll-based purification did not compromise the quality of the resulting cell suspensions.	Cell viability exceeded 90% in both preparations (Figure 5B), showing that viable single-cell suspensions suitable for downstream processing could be obtained without Ficoll-based purification.
602-603	In the present protocol (Figure 1), comparable simplification was achieved through controlled digestion and stereomicroscope-guided manual selection.	In the present protocol (Figure 1), workflow simplification was achieved through controlled digestion and stereomicroscope-guided manual selection.
614-616	The gradient-free library showed QC characteristics comparable to those of the Ficoll-based library, supporting the technical suitability of the resulting material.	Both libraries met the applied QC criteria, supporting the technical suitability of material obtained using the gradient-free workflow for downstream scRNA-seq analysis (Figure 6).
701-705	The workflow yielded viable single-cell suspensions suitable for scRNA-seq, and the pooled gradient-free library met the applied QC criteria and contained the major endo-crine cell populations.	The workflow yielded viable single-cell suspensions suitable for scRNA-seq, and the pooled gradient-free library met the applied QC criteria and contained the major endo-crine cell populations. These findings support the gradient-free workflow as a feasible alternative to Ficoll-based purification, without establishing superiority of one isolation method over the other.

Comments 2: The authors emphasize that the gradient-free protocol eliminates three centrifugation steps and at least 21.5 minutes of centrifuge run time. However, stereomicroscope-guided manual islet selection itself requires substantial hands-on time, reported as 15–30 minutes per dish, and therefore the manuscript should provide total elapsed and hands-on processing times for both workflows before concluding that the new procedure is faster or less time-consuming.

Response 2: We thank the reviewer for this comment. We have added a side-by-side comparison of the processing times for both workflows (Table 4). The revised table shows that manual islet selection required 15–30 min per dish in both workflows and therefore did not contribute to the difference between them. The estimated processing time before the optional fixation steps was 75–110 min for the gradient-free workflow and 110–145 min for the Ficoll-based workflow. The additional time in the Ficoll-based protocol was associated with the purification module, which required 21.5 min of centrifugation and approximately 15 min of Ficoll-specific handling for gradient preparation, interphase collection, dilution, and transfers.

The manuscript has also been revised to describe the gradient-free workflow as reducing Ficoll-specific processing rather than simply being faster or less time-consuming overall.

Comments 3: Manual islet selection is inherently operator-dependent and may introduce greater person-to-person variability than a standardized density-gradient procedure. The authors should clarify whether isolations were performed by a single operator or multiple operators and, if only one operator was involved, discuss the lack of inter-operator validation as an important limitation of the claimed reproducibility.

Response 3: We agree that manual islet selection introduces an operator-dependent component into the workflow. All isolations and stereomicroscope-guided selection procedures in this study were performed by a single trained operator with experience in cell isolation. This minimized operator-related variability within the present study but did not allow us to assess inter-operator reproducibility. We have clarified this point in the manuscript and added a discussion of operator dependence, potential selection bias and the importance of standardized training and predefined selection criteria.

Lines	It was originally in the text	Corrections
644-650	-	However, the extent of residual exocrine contamination may partly depend on the efficiency and consistency of manual islet selection. Manual islet selection is inherently operator-dependent, as differences in experience and technique can influence the consistency of islet identification and purification. This source of pre-analytical variability should be taken into account when establishing isolation workflows, particularly for sensitive downstream applications such as scRNA-seq. Selection may also become biased if operators preferentially collect large, intact, or visually prominent islets while overlooking smaller or partially disrupted structures. To reduce this effect, manual selection should follow predefined criteria and include all visually identifiable islets rather than only those with an “ideal” morphology. Standardized training and clearly defined selection criteria are therefore important for improving consistency across operators.

Comments 4: Relatedly, manual selection may introduce selection bias because operators may preferentially recover larger, more intact, or more easily recognizable islets while excluding less obvious or partially disrupted islets. This could influence islet yield, cell-type composition, and downstream transcriptomic profiles, and the authors should discuss how manual selection criteria were standardized and whether any objective criteria were used.

Response 4: We agree that manual selection can introduce selection bias, particularly if operators preferentially collect large or morphologically intact islets. We have therefore clarified the selection procedure in the Materials and Methods. No predefined size cutoff was used, and all visually identifiable islets were considered for collection, including small, medium, large, elongated, and partially disrupted islets when present. Islets associated with acinar tissue were gently pipetted to remove surrounding exocrine tissue and were excluded only if they could not be sufficiently separated. The dish was also periodically redistributed and re-examined to reduce the likelihood of overlooking small or initially obscured islets. These criteria were applied consistently throughout all isolations. No quantitative morphometric or automated image-based criteria were used, and this is now stated explicitly.

Figure 4C and its legend were also revised to illustrate representative islets of different sizes and morphologies, including islets associated with residual acinar tissue. The potential for residual operator-dependent selection bias is additionally acknowledged in the Discussion.

Lines	It was originally in the text	Corrections
266-274	-	No predefined size cutoff was applied during selection. All visually identifiable islets were considered for collection, including small, medium, and large islets as well as round, elongated, and partially disrupted structures, when present (Figure 4C). Islets associated with adherent acinar tissue were first gently pipetted directly in the Petri dish to remove surrounding exocrine tissue. If the islet could not be sufficiently separated from the surrounding exocrine tissue by gentle pipetting, it was not transferred to the final preparation. During selection, the contents of the dish were gently redistributed every few minutes, after which the entire dish was re-examined under the stereomicroscope to reduce the likelihood of overlooking small or initially obscured islets.
650-659	-	Selection may also become biased if operators preferentially collect large, intact, or visually prominent islets while overlooking smaller or partially disrupted structures. To reduce this effect, manual selection should follow predefined criteria and include all visually identifiable islets rather than only those with an “ideal” morphology.
275-283	Figure 4. Stereomicroscope-guided manual selection of pancreatic islets. (A) Initial pancreatic tissue suspension containing islets among abundant exocrine tissue. (B) Intermediate washing step in which manually selected islets are transferred into a clean buffer drop to remove residual tissue fragments. (C) Final preparation of purified islets after repeated transfer and washing, showing a marked reduction in exocrine contamination.	Figure 4. Stereomicroscope-guided manual selection of pancreatic islets. (A) Initial pancreatic tissue suspension containing islets among abundant exocrine tissue. (B) Intermediate washing step in which manually selected islets are transferred into a clean buffer drop to remove residual tissue fragments. (C) Representative final field illustrating the range of islet sizes and morphologies considered during manual selection. The green arrow indicates a large, round islet free of visible acinar tissue; the yellow arrow indicates a medium-sized round islet; the blue arrow indicates a small islet; and the orange arrow indicates a medium-sized elongated islet. Red arrows indicate medium-sized islets associated with residual acinar tissue, which require additional gentle pipetting to facilitate separation before final transfer.

Comments 5: The interpretation that the gradient-free procedure may reduce processing-associated cellular stress should be substantially tempered. Manual microscopic selection, repeated aspiration and transfer, prolonged handling, and time outside optimal physiological conditions may themselves induce mechanical or transcriptional stress, so removal of Ficoll centrifugation cannot by itself be assumed to reduce total cellular stress.

Response 5: We agree with the reviewer that omission of Ficoll-based purification cannot by itself be interpreted as evidence of reduced overall processing-associated cellular stress. We have therefore substantially tempered the interpretation of the stress-response data throughout the manuscript. In the Introduction, we removed the suggestion that the workflow was designed to reduce cellular stress and now describe its objective in terms of omitting Ficoll-specific centrifugation, washing, and transfer steps. In the Abstract and Results, differences in stress-response scores and immediate early gene expression are now presented as descriptive observations between the two pooled libraries. We also explicitly state in the Results that only one pooled scRNA-seq library was generated for each workflow and that all stress-related comparisons are descriptive.

Lines	It was originally in the text	Corrections
29-31	-	These differences were considered descriptive because only one pooled scRNA-seq library was generated for each workflow.
117-123	Accordingly, the present study aimed to adapt a pancreatic islet isolation protocol for BKS.Cg-Dock7^m +/+ Lepr^db/J mice for subsequent scRNA-seq analysis. The protocol was designed to eliminate Ficoll-based density-gradient purification and reduce the number and duration of processing steps, and limit additional stress imposed on the cells.	Accordingly, the present study aimed to evaluate a gradient-free pancreatic islet isolation workflow for BKS.Cg-Dock7^m +/+ Lepr^db/J mice for subsequent scRNA-seq analysis. The protocol was designed to omit eliminate Ficoll-based density-gradient purification and reduce the number of Ficoll-specific centrifugation, washing, and transfer steps before single-cell preparation reduce the number and duration of processing steps, and limit additional stress imposed on the cells. Its feasibility for scRNA-seq was assessed based on cell recovery, viability, library QC characteristics, cell-type representation, and descriptive analysis of stress-associated transcriptional features.
558-561	To assess possible differences in cellular transcriptional state between the two scRNA-seq libraries, the expression of genes associated with cellular stress was analyzed.	To assess possible differences in cellular transcriptional state between the two scRNA-seq libraries, the expression of genes associated with cellular stress was analyzed. Because one pooled scRNA-seq library was generated for each workflow, all comparisons in this section are descriptive.
662-673	Transcriptional differences between the libraries were most evident in the immediate early response. The gradient-free library displayed a lower combined stress-response score and reduced activation of several immediate early genes. Howewer, heat shock-, endoplasmic reticulum stress-, and oxidative stress-associated programs did not change uniformly across cell populations. The results therefore do not support a simple global reduction in all forms of cellular stress, but rather suggest differences in specific components of the processing-associated response.	Transcriptional differences between the libraries were most evident in the immediate early response. The gradient-free library showed a lower combined stress-response score and lower expression of several immediate early genes, whereas heat shock-, ER-, and oxidative- stress associated genes showed no consistent direction of change. These findings therefore indicate differences in selected processing-associated transcriptional responses rather than a general reduction in cellular stress. Importantly, both workflows included stereomicroscope-guided manual selection with repeated aspiration, transfer, and ex vivo handling, which may themselves induce mechanical or transcriptional stress. Thus, omission of Ficoll reduces Ficoll-specific processing steps but does not demonstrate a reduction total cellular stress. The lower immediate early response in the pooled gradient-free library should therefore be regarded as a descriptive difference between the two preparations.
674	The lower immediate early response observed in the gradient-free library is com-patible with the hypothesis that fewer handling steps and closer control of digestion may reduce selected processing-associated transcriptional responses. This interpretation remains provisional, because the study design does not allow the observed differences to be attributed solely to the isolation method.	-
701-705	The workflow yielded viable single-cell suspensions suitable for scRNA-seq, and the pooled gradient-free library met the applied QC criteria and contained the major endo-crine cell populations.	The workflow yielded viable single-cell suspensions suitable for scRNA-seq, and the pooled gradient-free library met the applied QC criteria and contained the major endocrine cell populations. These findings support the gradient-free workflow as a feasible alternative to Ficoll-based purification, without establishing superiority of one isolation method over the other.

Comments 6: The stress-response comparison is particularly limited because the three independently isolated mice were pooled to generate only one scRNA-seq library for each workflow. The authors appropriately acknowledge that these libraries represent single experimental units and that the observed differences cannot be attributed to the isolation method; therefore, statements in the Abstract, Results, and Discussion suggesting reduced stress with the gradient-free protocol should consistently be presented as descriptive observations rather than evidence of a method-specific effect.

Response 6: We agree that the stress-response comparison should be interpreted cautiously because only one pooled scRNA-seq library was generated for each workflow. We revised the Abstract, Results, Discussion, and Conclusions to present all stress-related differences as descriptive observations rather than method-specific effects. We also clarified that manual selection and ex vivohandling may themselves contribute to stress and that independent matched libraries are required to determine whether the gradient-free workflow reproducibly affects cell recovery, cellular composition, or processing-associated transcriptional responses.

Lines	It was originally in the text	Corrections
27-31	The gradient-free sample also showed a lower overall stress-response score and reduced expression of several immediate early genes, although heat shock- and endoplasmic reticulum (ER) stress-associated genes displayed heterogeneous patterns.	The gradient-free sample also showed a lower overall stress-response score and lower expression of several immediate early genes, although heat shock- and endoplasmic reticulum (ER) stress-associated genes displayed heterogeneous patterns. These differences were considered descriptive because only one pooled scRNA-seq library was generated for each workflow.
558-561	To assess possible differences in cellular transcriptional state between the two scRNA-seq libraries, the expression of genes associated with cellular stress was analyzed.	To assess possible differences in cellular transcriptional state between the two scRNA-seq libraries, the expression of genes associated with cellular stress was analyzed. Because one pooled scRNA-seq library was generated for each workflow, all comparisons in this section are descriptive.
566-568	Expression of Fos, Jun, Atf3, and Dusp1 was generally higher in the Ficoll-based library (Figure 9A–D).	Expression of Fos, Jun, Atf3, and Dusp1 was generally higher in the pooled Ficoll-based library (Figure 9A–D).
569-572	The expression patterns of heat shock- and ER stress-associated genes were less uniform. Hspa1a and Ddit3 showed higher expression in the gradient-free library (Figure 9E, F), whereas Hsp90aa1, Xbp1, and Sod2 displayed broad, cell-type-specific expression without a consistent direction of change between the two libraries (Figure 9H).	The expression patterns of heat shock- and ER stress-associated genes were less uniform. Hspa1a and Ddit3 showed higher expression in the pooled gradient-free library (Figure 9E, F), whereas Hsp90aa1, Xbp1, and Sod2 displayed broad, cell-type-specific expression without a consistent direction of change between the two libraries (Figure 9H).
584-587	Thus, despite higher expression of selected heat shock- and ER stress-associated genes in the gradient-free sample. The distribution of the combined stress-associated gene score was shifted toward lower values in the gradient-free pooled library.	Thus, the combined score was lower int the pooled gradient-free library, despite higher expression of selected heat shock- and ER stress-associated genes. This comparison is descriptive and does not establish a method-specific effect.
588-591	Dot-plot analysis further demonstrated that the stress response was strongly cell-type specific. The most consistent differences between the libraries were observed for immediate early genes (IEGs), whereas genes associated with the HSR, ER stress, and oxidative stress showed more variable expression profiles across the different cell populations (Figure 9H).	Dot-plot analysis showed cell-type-specific variation in stress-associated gene expression. The most consistent differences between the two pooleslibraries were observed for immediate early genes (IEGs), whereas HSR-, ER stress-, and oxidative stress-associated genes showed more variable expression across cell populations (Figure 9H).
662-673	Transcriptional differences between the libraries were most evident in the immediate early response. The gradient-free library displayed a lower combined stress-response score and reduced activation of several immediate early genes. Howewer, heat shock-, endoplasmic reticulum stress-, and oxidative stress-associated programs did not change uniformly across cell populations. The results therefore do not support a simple global reduction in all forms of cellular stress, but rather suggest differences in specific components of the processing-associated response.	Transcriptional differences between the libraries were most evident in the immediate early response. The gradient-free library showed a lower combined stress-response score and lower activation of several immediate early genes, whereas heat shock-, ER-, and oxidative- stress associated genes showed no consistent direction of change. These findings therefore indicate differences in selected processing-associated transcriptional responses rather than a general reduction in cellular stress. Importantly, both workflows included stereomicroscope-guided manual selection with repeated aspiration, transfer, and ex vivo handling, which may themselves induce mechanical or transcriptional stress. Thus, omission of Ficoll reduces Ficoll-specific processing steps but cannot be assumed to reduce total cellular stress. The lower immediate early response in the pooled gradient-free library should therefore be regarded as a descriptive difference between the two preparations.
691-694	A direct comparison using several independently prepared, matched libraries would be required to determine whether the gradient-free purification reproducibly reduces processing-associated stress while preserving cell recovery and cellular composition	A direct comparison using several independently prepared, matched libraries would be required to determine whether the gradient-free workflow reproducibly affects cell recovery, cellular composition, or processing-associated transcriptional responses.
749-751	Independent, matched library preparations are needed to establish reproducibility and whether the gradient-free workflow consistently affects cell recovery, or processing-associated transcriptional responses.	Independent, matched library preparations are needed to establish reproducibility and determine whether the gradient-free workflow reproducibly affects cell recovery, cellular composition or processing-associated transcriptional responses.

Comments 7: The Introduction may benefit from briefly acknowledging the broader relevance of pancreatic molecular alterations and emerging molecular biomarkers. In this context, the authors may consider citing the recent review https://doi.org/10.3389/freae.2025.1723159, which provides a useful overview of molecular and epigenetic changes in pancreatic disease and their translational relevance.

Response 7: We thank the reviewer for this suggestion. We have expanded the Introduction to briefly acknowledge the broader relevance of molecular and epigenetic alterations in pancreatic disease and their potential value as diagnostic, prognostic, and predictive biomarkers. The recommended recent review has been cited in this context. We also linked this broader molecular perspective to the importance of standardized sample procurement and processing for reproducible downstream molecular analyses.

Lines	It was originally in the text	Corrections
97-107	-	Following pancreatic digestion, islets can be enriched by density-gradient purification, stereomicroscope-guided manual selection, or combinations of these approaches, each requiring different degrees of centrifugation, washing, transfer, and handling before dissociation [4]. These pre-analytical differences are particularly relevant for molecular studies, because the cellular transcriptome can be affected by tissue dissociation and ex vivo processing. More broadly, molecular and epigenetic alterations are increasingly being investigated as clinically relevant biomarkers in pancreatic disease, particularly pancreatic cancer. Aberrant DNA methylation, histone modifications, and dysregulated non-coding RNAs have been associated with diagnosis, prognosis, and therapeutic response, highlighting the importance of standardized sample procurement and processing for reproducible molecular measurements [12].

Comments 8: Because density-gradient purification is a well-established and relatively standardized method, the practical value of replacing it with manual selection would be strengthened by a more comprehensive comparison of the two approaches. In addition to viability and scRNA-seq QC, useful parameters would include total processing time, hands-on time, islet yield, purity, operator-to-operator variability, and reproducibility across independent scRNA-seq libraries; without such comparisons, the gradient-free approach is best presented as a feasible alternative rather than an improved replacement for density-gradient purification.

Response 8: We agree that a broader comparison would be required to establish the gradient-free workflow as an improved replacement for density-gradient purification. We have therefore revised the manuscript to present it consistently as a feasible alternative. A side-by-side comparison of processing time has been added (Table 2), and the manuscript now discusses viable cell recovery, viability, scRNA-seq QC, operator dependence, and the limitations of manual selection. We also explicitly acknowledge that islet yield per pancreas, pre-dissociation purity, inter-operator variability, and reproducibility across independently prepared scRNA-seq libraries were not assessed. These parameters are now identified as priorities for future comparative validation.

	Lines	It was originally in the text	Corrections
Manual selection	266-274	-	No predefined size cutoff was applied during selection. All visually identifiable islets were considered for collection, including small, medium, and large islets as well as round, elongated, and partially disrupted structures, when present (Figure 4C). Islets associated with adherent acinar tissue were first gently pipetted directly in the Petri dish to remove surrounding exocrine tissue. If the islet could not be sufficiently separated from the surrounding exocrine tissue by gentle pipetting, it was not transferred to the final preparation. During selection, the contents of the dish were gently redistributed every few minutes, after which the entire dish was re-examined under the stereomicroscope to reduce the likelihood of overlooking small or initially obscured islets.
Manual selection	650-659	-	Selection may also become biased if operators preferentially collect large, intact, or visually prominent islets while overlooking smaller or partially disrupted structures. To reduce this effect, manual selection should follow predefined criteria and include all visually identifiable islets rather than only those with an “ideal” morphology.
Islet yield per pancreas and Pre-dissociation purity	500-512	Before evaluating cell composition, the main QC parameters of the scRNA-seq libraries were examined. The distributions of the number of detected genes per cell largely overlapped between the Ficoll-based and gradient-free libraries (Figure 6A). A similar pattern was observed for the total number of transcript counts per cell (Figure 6B). The proportion of mitochondrial transcripts remained low in both libraries (Figure 6C), while the proportion of ribosomal transcripts was also within a comparable range (Figure 6D). The distributions of these QC metrics broadly overlapped between the two libraries.	Approximately 24,000 viable cells were loaded for each library. The gradient-free library yielded 5,568 initially identified cells, corresponding to an estimated capture efficiency of 23.2%, whereas the Ficoll-based library yielded 4,358 cells, corresponding to 18.2%. Before evaluating cell composition, the main QC parameters of the scRNA-seq libraries were examined. After QC filtering, 2,659 cells from the gradient-free library and 2,715 cells from the Ficoll-based library were retained for downstream analysis. The distributions of the number of detected genes per cell largely overlapped between the Ficoll-based and gradient-free libraries (Figure 6A). A similar pattern was observed for the total number of transcript counts per cell (Figure 6B). The proportion of mitochondrial transcripts remained low in both libraries (Figure 6C), while the proportion of ribosomal transcripts was also within a comparable range (Figure 6D). The distributions of these QC metrics broadly overlapped between the two libraries.
Islet yield per pancreas and Pre-dissociation purity	679-694	The principal limitation is the absence of matched biological replicates for the two workflows. Each analyzed library was generated from pooled material, and variation among individual animals could therefore not be assessed. Consequently, differences in cell composition and transcriptional state should be regarded as descriptive rather than as evidence of a method-specific effect. A direct comparison using several independently prepared, matched libraries would be required to determine whether the gradient-free purification reproducibly reduces processing-associated stress while preserving cell recovery and cellular composition	The principal limitation is the absence of matched biological replicates for the two workflows. Each analyzed library was generated from pooled material, and variation among individual animals could therefore not be assessed. Consequently, differences in cell composition and transcriptional state should be regarded as descriptive rather than as evidence of a method-specific effect. In addition, islet yield per pancreas and pre-dissociation purity were not prospectively quantified. Although these parameters are useful for benchmarking isolation performance, islet number alone does not directly reflect the amount of material available for scRNA-seq because individual mouse islets vary in size and cell content; viable cell recovery after dissociation was therefore used as the primary input-related measure. Functional properties of the isolated islets were also not assessed, and the present data therefore do not establish preservation of glu-cose-responsive endocrine function. A direct comparison using several independently prepared, matched libraries would be required to determine whether the gradient-free workflow reproducibly affects cell recovery, cellular composition, or processing-associated transcriptional responses.
Operator-to-operator variability	644-661	-	However, the extent of residual exocrine contamination may partly depend on the efficiency and consistency of manual islet selection. Manual islet selection is inherently operator-dependent, as differences in experience and technique can influence the consistency of islet identification and purification. This source of pre-analytical variability should be taken into account when establishing isolation workflows, particularly for sensitive downstream applications such as scRNA-seq. Selection may also become biased if operators preferentially collect large, intact, or visually prominent islets while overlooking smaller or partially disrupted structures. To reduce this effect, manual selection should follow predefined criteria and include all visually identifiable islets rather than only those with an “ideal” morphology. Standardized training and clearly defined selection criteria are therefore important for improving consistency across operators.
Reproducibility across independent scRNA-seq libraries	691-694	A direct comparison using several independently prepared, matched libraries would be required to determine whether the gradient-free purification reproducibly reduces processing-associated stress while preserving cell recovery and cellular composition	A direct comparison using several independently prepared, matched libraries would be required to determine whether the gradient-free workflow reproducibly affects cell recovery, cellular composition, or processing-associated transcriptional responses.
Reproducibility across independent scRNA-seq libraries	749-751	Independent, matched library preparations are needed to establish reproducibility and whether the gradient-free workflow consistently affects cell recovery, or processing-associated transcriptional responses.	Independent, matched library preparations are needed to establish reproducibility and determine whether the gradient-free workflow reproducibly affects cell recovery, cellular composition or processing-associated transcriptional responses.
Gradient-free approach is best presented as a feasible alternative	14-16	This study aimed to adapt a reproducible protocol for preparing murine pancreatic islets for downstream single-cell transcriptomic analysis while reducing the duration and complexity of sample processing.	This study aimed to adapt a practical gradient-free alternative for preparing murine pancreatic islets for downstream single-cell transcriptomic analysis while reducing the number of sample-processing steps.
	31-32	This study aimed to adapt a reproducible protocol for preparing murine pancreatic islets for downstream single-cell transcriptomic analysis while reducing the duration and complexity of sample processing.	These findings support gradient-free isolation as a possible alternative to Ficoll-based purification for preparing murine pancreatic islets for scRNA-seq analysis.
	129-130	-	The aim was therefore to evaluate the gradient-free workflow as a practical alternative to Ficoll-based purification rather than to establish its superiority.
	701-705	The workflow yielded viable single-cell suspensions suitable for scRNA-seq, and the pooled gradient-free library met the applied QC criteria and contained the major endo-crine cell populations.	The workflow yielded viable single-cell suspensions suitable for scRNA-seq, and the pooled gradient-free library met the applied QC criteria and contained the major endo-crine cell populations. These findings support the gradient-free workflow as a feasible alternative to Ficoll-based purification, without establishing superiority of one isolation method over the other.

We have added a side-by-side comparison of the processing times for the gradient-free and Ficoll-based workflows
(Table 4).

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Most of my previous comments have been addressed. I have two remaining points:

Please check the description of the time saved by omitting Ficoll purification. Table 4 shows that the Ficoll step takes about 35 min in total, including 21.5 min of centrifugation and about 15 min of Ficoll-specific handling. Therefore, statements referring to “35 min less centrifuge run time” are not accurate. Please revise these statements throughout the manuscript. It would be more accurate to state that the gradient-free workflow shortens the overall processing time by about 35 min, including 21.5 min of centrifugation.
The cell numbers in Table 2 appear to be assigned to the wrong workflows. The text reports 5,568 initially identified cells and 2,659 final cells for the gradient-free library, and 4,358 and 2,715 cells for the Ficoll-based library, whereas Table 2 shows the reverse. Please correct Table 2 accordingly.

Author Response

Response to Reviewer 1 Comments

1. Summary

We are very grateful for the thoughtful and constructive feedback, which helped us improve the manuscript. We have addressed all comments carefully, and the corresponding revisions are highlighted in the updated version.

We have also updated the Data Availability Statement to include the newly assigned GEO accession number, GSE347182 (Line 666). Our point-by-point responses are provided below.

2. Questions for General Evaluation

Reviewer’s Evaluation

Response and Revisions

Does the introduction provide sufficient background and include all relevant references?

Yes/Can be improved/Must be improved/Not applicable

Are all the cited references relevant to the research?

Yes/Can be improved/Must be improved/Not applicable

Is the research design appropriate?

Yes/Can be improved/Must be improved/Not applicable

Are the methods adequately described?

Yes/Can be improved/Must be improved/Not applicable

Are the results clearly presented?

Yes/Can be improved/Must be improved/Not applicable

Are the conclusions supported by the results?

Yes/Can be improved/Must be improved/Not applicable

3. Point-by-point response to Comments and Suggestions for Authors

Comments 1: Please check the description of the time saved by omitting Ficoll purification. Table 4 shows that the Ficoll step takes about 35 min in total, including 21.5 min of centrifugation and about 15 min of Ficoll-specific handling. Therefore, statements referring to “35 min less centrifuge run time” are not accurate. Please revise these statements throughout the manuscript. It would be more accurate to state that the gradient-free workflow shortens the overall processing time by about 35 min, including 21.5 min of centrifugation.

Response 1: We appreciate this careful observation. We agree that the previous wording incorrectly described the approximately 35-min difference as centrifuge run time. Table 4 shows that the Ficoll-specific purification module required approximately 35 min in total, comprising 21.5 min of centrifugation and approximately 15 min of additional handling.

We have therefore revised the corresponding statements throughout the Introduction, Results, Discussion, and Conclusions to state that omission of Ficoll purification shortened the estimated overall processing time by approximately 35 min, including 21.5 min of centrifugation.

Lines	It was originally in the text	Corrections
119-121	This removed three centrifugation steps, corresponding to at least 35 min of centrifuge run time, while providing sufficient viable cells for scRNA-seq library preparation.	This removed three centrifugation steps and shortened the estimated overall processing time by approximately 35 min, including 21.5 min of centrifugation, while providing sufficient viable cells for scRNA-seq library preparation.
433-435	The Ficoll-based module required one 18-min gradient centrifugation and two additional centrifugations of 2 min and 90 s. The gradient-free workflow therefore eliminated three centrifugation steps and at least 35 min of centrifuge run time.	The Ficoll-based module required one 18-min gradient centrifugation and two additional centrifugations of 2 min and 90 s. Together with approximately 15 min of Ficoll-specific handling, this purification module added approximately 35 min to the overall processing time, including 21.5 min of centrifugation.
543-544	The gradient-free workflow eliminated gradient assembly, interphase collection, large-volume dilution, and two subsequent washes. This corresponded to three fewer centrifugation steps and at least 35 min less centrifuge run time.	The gradient-free workflow eliminated gradient assembly, interphase collection, large-volume dilution, and two subsequent washes. This shortened the estimated overall processing time by approximately 35 min, including 21.5 min of centrifugation.
639-640	Omitting Ficoll-based purification eliminated gradient preparation, interphase recovery, and three centrifugation steps, reducing centrifuge run time by at least 35 min, including 21.5 min of centrifugation.	Omitting Ficoll-based purification eliminated gradient preparation, interphase recovery, and three centrifugation steps, shortening the estimated overall processing time by approximately 35 min, including 21.5 min of centrifugation.

Comments 2: The cell numbers in Table 2 appear to be assigned to the wrong workflows. The text reports 5,568 initially identified cells and 2,659 final cells for the gradient-free library, and 4,358 and 2,715 cells for the Ficoll-based library, whereas Table 2 shows the reverse. Please correct Table 2 accordingly.

Response 2: We appreciate this correction. The values in Table 2 were inadvertently assigned to the opposite workflow columns. We have corrected the table so that the gradient-free workflow now shows 5,568 initially identified cells and 2,659 final endocrine/exocrine cells, whereas the Ficoll-based workflow shows 4,358 initially identified cells and 2,715 final cells. The corresponding intermediate QC values were also reassigned to the correct workflows. The values in the revised Table 2 (Line 394) are now consistent with the text.

Author Response File: Author Response.pdf