Physically interacting beta-delta pairs in the regenerating pancreas revealed by single-cell sequencing

  1. Eran Yanowski
  2. Nancy-Sarah Yacovzada
  3. Eyal David
  4. Amir Giladi
  5. Diego Jaitin
  6. Lydia Farack
  7. Adi Egozi
  8. Danny Ben-Zvi
  9. Shalev Itzkovitz
  10. Ido Amit
  11. Eran Hornstein

  • Curated by eLife

    eLife logo

    Evaluation Summary:

    This paper provides an attempt to understand the crosstalk between to islet cell types during beta cell regeneration following partial pancreatectomy. It combines lineage tracing, single cell sequencing and light microscopy to describe islet cell heterogeneity and interactions in the regenerating mouse pancreas. The concept of protective signaling resulting from the direct interactions between beta and delta cells is compelling and would be of interest to scientists in the field of endocrine pancreas development and regeneration. However, the conclusions derived from the sequencing data require additional experimental support.

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

Reviewed by

Read the full article See related articles

Discuss this preprint

Start a discussion What are Sciety discussions?

Listed in

Log in to save this article

Abstract

The endocrine pancreas is able to regenerate in response to insult, including by driving beta-cells into the cell division cycle. Until recently, communication between neighboring cells in islets of Langerhans was overlooked by single-cell genomic technologies, which require rigorous tissue dissociation into single cells. Here, we utilize sorting of physically interacting cells (PICs) with single-cell RNA-sequencing to systematically map cellular interactions in the regenerating endocrine pancreas. The cellular landscape of the regenerated pancreas features regeneration-associated endocrine populations.

We explore the unexpected heterogeneity of beta-cells in regeneration, including an interaction-specific program between paired beta and delta-cells. Our analysis suggests that the particular cluster of beta-cells that pair with delta-cells benefits from stress protection, implying that the interaction between beta and delta-cells safeguards against regeneration-associated challenges.

Article activity feed

  1. eLife

    Evaluation Summary:

    This paper provides an attempt to understand the crosstalk between to islet cell types during beta cell regeneration following partial pancreatectomy. It combines lineage tracing, single cell sequencing and light microscopy to describe islet cell heterogeneity and interactions in the regenerating mouse pancreas. The concept of protective signaling resulting from the direct interactions between beta and delta cells is compelling and would be of interest to scientists in the field of endocrine pancreas development and regeneration. However, the conclusions derived from the sequencing data require additional experimental support.

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

  2. eLife

    Reviewer #1 (Public Review):

    The manuscript by Yanowski et al uses lineage tracing, single cell mRNA sequencing and in situ hybridization methods to study the developmental origin of delta cells and the transcriptional heterogeneity of beta cells in the regenerating pancreas. First, the authors use a CRE-lox approach driven by Sox9-CRE to trace the fate of Sox9-expressing cells in adult mice (12 weeks of age). They show that most Sox9-positive cells are exocrine cells - as expected. However, they also observe that most Sox9-positive cells inside the islet express Sst and therefore are delta cells. Next, they use a mouse model of beta cell regeneration triggered by pancreatic injury (partial pancreatectomy (ppx)) to characterize beta cell heterogeneity in the regenerating pancreas using single cell sequencing. In doing so, they identify beta-delta cell pairs that are characterized by lower expression of the ER stress marker Fkbp11 and higher Ins2 expression. These results are validated in situ using smFISH.

    These results lead to the conclusions that 1) delta cells originate from Sox-9 expressing cells and proliferate in ppx, 2) there are 3 distinct types/states of beta cells in the regenerating pancreas (stress, cell cycle and basal) and 3) beta-delta cell interactions appear to be specific to a beta cell subset called "int-beta".

    The study opens exciting possibilities and questions that indicate that a fraction of adult delta cells express Sox9, while implicating beta-delta cell signaling and/or physical interaction in the transcriptional heterogeneity "process" in beta cells in a regeneration setting. However further experiments are necessary to support the authors' conclusions and provide the physiological context of their observations

    Major strengths:

    - Study uses a genetic mouse model to express a fluorescent reporter in delta cells, which allows the authors to enrich isolated islet preparations with delta cells that can be studied with single cell technologies
    - The use of MARS-seq + PIC-seq is very exciting and it has the potential to identify important cell-cell interactions in the islet and that are mostly lost during tissue dissociation/islet isolation.
    - The smFISH technology is impressive. This approach allows the important validation step in situ of beta cell heterogeneity characterized by distinct gene expression profiles identified by single cell mRNA sequencing.
    - The combination of ppx with single cell technologies has the potential to identify the molecular signature of replicating islet cells which may play a role in pancreas regeneration, including beta cells and delta cells.
    - The finding that delta cells seem to be indeed capable of proliferation in adult mice is very exciting, specially since this cell types is mostly post-mitotic throughout the lifetime of adult mice

    Major weaknesses:

    - Although the study is data rich, it lacks the description of the molecular details of the different types and sub-types of cells identified and the manuscript as it is written is most times confusing in the presentation in the results and concepts. This makes it difficult for the reader to fully appreciate the degree of "heterogeneity" observed and the lack of a detailed description of these phenotypes precludes any meaningful comparison with other published results describing beta cell heterogeneity in the mouse. For example, the balance of PDX1/MAFA in the mouse islet has been shown to be important for islet function (see work by the Hodson lab), however these two transcription factors are not mentioned/analyzed in this manuscript.

    - The analysis of the imaging data, particularly of the number of delta cells in sham/ppx isolated islets is subpar. Here, the authors use whole islet fluorescence intensity (Fig2B-D) or smFISH (FigE-G) to quantify delta cell mass. These results are inconclusive since this enhanced fluorescence signal may also be explained by delta cell hypertrophy and/or an increase number of delta cell filopodia and not necessarily more delta cells.

    - The authors show that beta-cells with neighboring delta cells are likely the "int-beta" cell phenotype identified with MARS/PIC-seq. However, this important observation is only supported by the validation of two "int-beta" marker (Fkbp11, Ins2). Moreover, and given the peripheral distribution of delta cells, these results imply that beta cells in the periphery are molecularly different from other beta cells and may have overlapping features with virgin beta cells described by the Huising lab. However, no attempt is made to correlate these new findings with the published data and the physiological relevance of these findings remain mostly speculative.

  3. eLife

    Reviewer #2 (Public Review):

    This manuscript suggests crosstalk between beta and delta cells on the basis of the observation of doublets between beta and delta cells upon the incomplete dissociation of pancreatic islets. This confirms that beta and delta cells are neigboring cells, which histological analysis of pancreas morphology established a long time ago. It is further suggested that crosstalk plays a role in beta cell regeneration following partial pancreatectomy. However, the nature of this crosstalk - a dynamic exchange in information in time - is not addressed and is entirely based on approaches that provide a single snapshot. Then there is a suggestion that some delta cells expand during partial pancreatectomy and are descendants of the Sox9 expressing precursors, which has been shown a decade ago, but was not mentioned. The entire manuscript relies rather heavily on computational analysis of single cell transcriptomes with some validation experiments (such as smFISH for FKBP11 that unfortunately fail to advance the authors' argument). Controls to confirm that the cells the authors label as 'regenerating beta cells' are indeed regenerating beta cells are lacking as there is no direct demonstration of an increase in proliferating beta cells or increase in beta cell mass.

    The authors' starting point is that 'communication between neighboring cells in islets of Langerhans was overlooked by single cell genomic technologies. Perhaps this relates to the fact that communication between cells involves per definition a dynamic exchange of information. This is challenging to demonstrate directly using a technique that by design provides only a single snapshot in time, whether this is a single cell sequencing approach or a traditional histology readout. There is a rich literature on the crosstalk between different islet cell types, taking advantage of approaches better geared towards detecting crosstalk than single cell transcriptomics. The nature of the alleged crosstalk between beta cells and delta cells is not addressed, beyond the suggestion that paracrine Sst might be involved. Overall, the authors claims that beta cells adjacent to delta cells are protected from stress and contribute stronger to beta cell regeneration is not sufficiently supported by the observations in this manuscript.

    The detection of doublets between beta and delta cells in an incompletely digested islet preparation would confirm the fact that these cells are indeed adjacent in the islets. Heterotypic doublets between alpha and delta cells are similarly enriched, although this is not mentioned in the manuscript beyond the supplementary data.

    The fact that some endocrine cells in the islet derive from the Sox9 lineage and that these Sox9-positive islet cells are predominantly non-beta cells was published a decade ago in a paper that concluded that Sox9 lineage labeled cells did not contribute to beta cell regeneration. This paper by Kopp et al., curiously was not cited or discussed even though the authors did cite other key work from the same author group.

    On numerous occasions the authors refer to the phenotype of their beta cells as 'regeneration-driven', whereas they are more accurately described as 'pancreatectomy-driven'. Similarly, they call some subset of their beta cells 'regenerating beta cells'. I did not find any direct demonstration of regenerating beta cells in the study either by proliferation models or by quantification of beta cell mass. If the suggestion is that beta cell replication is indeed induced preferentially in delta cell adjacent beta cells, one could detect this in sections using traditional and proven approaches such as the detection of Ki67 or EdU in beta cells in proximity of delta cells.

    The data at several instances raise questions of rigor that remain unanswered. Some examples:
    -The authors stratification into Ins2-low, medium and high beta cells needs to be carefully vetted. The Ins2-low cells, which are labeled by the authors as enriched in stress-associated transcripts, also are enriched in the delta cell marker Rbp4.

    -In figure 5, the relative contribution of beta and delta cells is broken out across a series of beta/delta doublet where the beta cell contribution the mixed transcript pool averages around 75%. As there is no indication that beta cells express more genes or contain more RNA than delta cells, this would suggest aggregates or 3 beta cells per delta cell.

    -The suggestion is made that Fkbp11 mRNA is depleted in 'regenerating beta cells, that are suggested to be the ones adjacent to delta cells. This is followed by a smFISH confirmation, which to this reviewer does not demonstrate a clear difference in Fkbp11 levels among beta cells and further reveals similar Fkbp11 levels in delta cells (which are also stressed?)

    -Careful morphometry of the islet mass and composition is not done and would be required to start to support claims of delta cell contribution to a possibly regenerating beta cell mass. The quantification of delta cell numbers by a snapshot of the tdTomato expressing cells has the potential for confounds introduced by variability in the islet prep quality.

  4. eLife

    Reviewer #3 (Public Review):

    Yanowski et al present an interesting piece of work that identifies an important and under-investigated field of islet research; regeneration of the pancreas with a focus on non-beta-cells. They identify beta-delta cell pairs in their analysis. They confirm that these pairs are genuine pairs rather than an artefact. They then demonstrate that these pairs exhibit some beneficial properties (transcriptionally) in Ppx pancreas.

    I am not experienced in MARS-seq or PIC. Whilst these technologies look interesting and the data are exciting, I hope the other reviewers can comment on the validity of these data.

    Here are my main comments:

    1. Introduction: The introduction is lacking a clear introduction into Sox9. This is very important because in the results you suddenly 'jump' to looking at Sox9-CreER:TdTomato pancreas's (Figure 1) without any explanation as to why. You then (Figure 2) discuss Ppx in Sox9-CreER:TdTomato pancreas's, and again it is not clear why. What led you to consider this model in this context? You could consider reframing the results to give this narrative. Did you use Sox9 as a proxy for Sst? Why not just use Sst-ER:tDtomato mice?
    1. Line 155-157: This is a discussion point. Also, how can you, at this point at least, claim that there are beta-delta pairs in Figure 3. Could they, at this point, still be considered an artefact or cell dissociation or leaky RNA?
    1. How would your findings fit into the context of (or be explained through the lens of) what is currently known about interactions between beta and delta-cells? There are a few recent studies that have investigated the interaction between beta- and delta-cells, notably PMC5767697, PMC5026721. These should be cited as you have only given one reference which does not investigate beta-delta interactions, and a review.
  5. Version published to 10.1101/2021.02.22.432216 on bioRxiv