Cellular Endocrinology in Health and Disease

By Ya-Xiong Tao

Cellular Endocrinology in Health and Disease, Second Edition, describes the underlying basis of endocrine function, providing an important tool to understand the fundamentals of endocrine diseases. Delivering a comprehensive review of the basic science of endocrinology, from cell biology to human disease, this work explores and dissects the function of a number of cellular systems. The new edition provides an understanding of how endocrine glands function by integrating information resulting in biological effects on both local and systemic levels, also providing new information on the molecular physiopathogenesis of endocrine neoplasic cells.

The new edition expands the most used chapters from the first edition and proposes a series of substitutions and additions to the table of contents. New chapters cover signaling, brown adipose tissue, hypothalamic cell models, cellular basis of insulin resistance, genetics and epigenetics of neuroendocrine tumors, and a series of chapters on endocrine-related cancer.

Providing content that crosses disciplines, Cellular Endocrinology in Health and Disease, Second Edition, details how cellular endocrine function contributes to system physiology and mediates endocrine disorders. A methods section proves novel and useful approaches across research focus that will be attractive to medical students, residents, and specialists in the field of endocrinology, as well as to those interested in cellular regulation. Editors Alfredo Ulloa-Aguirre and Ya-Xiong Tao, experts in molecular and cellular aspects of endocrinology, deliver contributions carefully selected for relevance, impact, and clarity of expression from leading field experts

• Explores endocrine cells biology in normal and pathologic conditions
• Covers new aspects of endocrine cell function in distinct tissues
• Provides a view into the biological effect in local and systemic levels
• 15 new chapters covering the recent developments in the field

• Language: English
• Publisher: Academic Press
• Release date: Feb 2, 2021
• ISBN: 9780128198025

Book preview

Mexico

Preface to the Second Edition

Alfredo Ulloa-Aguirre; Ya-Xiong Tao

Endocrinology is the study of hormones and hormone actions, including cellular and molecular mechanisms of actions. First coined by Ernest Starling in 1905, hormone is secreted by endocrine glands and transported in the blood to target organs, where it exerts its action. Target cells have receptors, either on the plasma membrane or inside the cell. Traditional endocrine glands that produce hormones include the hypothalamus, pituitary gland, parathyroid and thyroid glands, pancreas, adrenal glands, and gonads. However, it has been demonstrated that many other organs such as adipose tissue, heart, gastrointestinal tract, and bones also produce hormones and therefore can be considered as endocrine glands. Furthermore, hormones that do not enter blood circulation, and work in autocrine, paracrine, or even intracrine manner are also studied by endocrinologists.

The first edition of Cellular Endocrinology in Health and Disease was published in 2014. The volume sought to summarize salient aspects of endocrinology that are relevant for physiology and pathophysiology at cellular level. Significant progress has been made in this field during the past 6 years thus a new edition is warranted. In this updated edition, which was suggested by André Wolff, from Elsevier S&T Books, who considering the success of the first edition asked us to assemble a second, updated edition, we included new chapters prepared by leading scientists on diverse aspects of cellular endocrinology. Therefore, in addition to significant updates on chapters included in the first edition, the current edition includes several new reviews representing rapidly progressing areas of endocrinology research.

We thank all the contributors for their excellent contributions. Their timely contributions, especially during this unprecedented and difficult to all novel coronavirus pandemic, are greatly appreciated.

We thank the colleagues at Elsevier, particularly André Gerhard Wolff and Tracy Tufaga for their help in stewarding this project from initiation to completion. Finally, we dedicate this volume to our mutual friend, mentor, and colleague, P. Michael Conn, who served as co-editor with Alfredo Ulloa-Aguirre for the first edition of Cellular Endocrinology in Health and Disease.

Chapter 1: Endocrinology of a Single Cell: Tools and Insights

Hanna Pincasa; Frederique Ruf-Zamojskia; Judith L. Turgeonb; Stuart C. Sealfona,c    a Department of Neurology, Center for Advanced Research on Diagnostic Assays, Icahn School of Medicine at Mount Sinai, New York, NY, United States

b Department of Internal Medicine, University of California, Davis, CA, United States

c Department of Neuroscience, Friedman Brain Institute, Icahn School of Medicine at Mount Sinai, New York, NY, United States

Abstract

Recent advances in single-cell (SC) technologies have enabled scientists to gain groundbreaking insights into cellular diversity, cell development, and cellular dynamics. In this chapter, we review recent SC technologies that address the key questions in the field of endocrine research, including characterization of cellular heterogeneity within a tissue, molecular classification of endocrine tumors, and analysis of signaling dynamics and stimulus response, such as hormone secretion. We provide examples of SC studies and highlight their main findings, focusing on recent discoveries on pancreatic β-cells and pituitary gonadotropes. Lastly, we discuss the pitfalls and limitations of SC approaches, as well as future avenues of research and technology development in this rapidly evolving field.

Keywords

Endocrine cellsSingle-cell analysis; Cell type; Cell state; Development; Transdifferentiation; Pancreatic β-cell; Pituitary gonadotrope

Why Study Single Cells?

A major challenge in biology is to elucidate the molecular mechanisms underlying cellular behavior and cell response to external cues. A tissue is composed of one or more cell types that can be defined by their physical, molecular, and functional characteristics. These cell type(s) may adopt distinct cellular states that vary in their transcriptional profiles and cellular characteristics (phenotypes) in response to various environmental stimuli.¹ Thus characterizing cellular heterogeneity has important implications for better understanding signaling dynamics and stimulus responses,² cell differentiation,³ cell development,⁴ and cancer pathophysiology.⁵ As intra-tumor heterogeneity affects drug sensitivity, from a therapeutic perspective, the development of an effective cancer treatment relies on an assessment of cellular heterogeneity (for a review, see Ref. 6).

While population-level studies are informative and often a first step in assessing basic molecular mechanisms, they only provide average measurements that may conceal the behavior of individual cells. As a consequence, critical biological phenomena may be hidden or only partially characterized. While some cell-to-cell variation has little functional significance, such as the variation in the expression of housekeeping genes,⁷, ⁸ many fundamental processes depend on regulatory changes within specific cells. Population-level studies obscure switch-like behaviors and important responses occurring in only a small proportion of cells or in rare cell subtypes. Some cell-to-cell differences may have functional consequences; for example, outlier cell behavior may account for cell differentiation. Thus, a general challenge is to characterize cellular heterogeneity and to determine the distribution of responses across cells and within specific cell types.

In recent years, single-cell (SC) technologies have become more widely utilized, and their importance for advancing biological knowledge has been recognized by research funding agencies. Between 2012 and 2017, the NIH funded the Single Cell Analysis Program (SCAP; https://commonfund.nih.gov/singlecell) that supports the development of various SC technologies to explore the heterogeneity of human tissues at the cellular level (Table 1.1).⁹ Researchers in the field of endocrinology have employed and benefited from SC technologies to: (i) analyze signaling dynamics and responses (e.g., hormone secretion), (ii) characterize the cellular composition of a tissue, and to (iii) recapitulate developmental processes within a tissue (Table 1.2). Hence, while population-based analysis of immediate early gene (IEG) induction responses to graded GnRH stimuli in LβT2 gonadotrope cells detected an apparent increase in IEG expression after the GnRH pulse,¹⁰ a SC qPCR approach unmasked an all-or-none pattern of gene activation, with an increased number of cells expressing IEG rather than an anticipated graded gene response.¹¹ In the field of diabetes, SC technologies are also an important approach. For example, the generation of insulin-producing cells in regenerative medicine necessitates the characterization of β-cell subpopulations, their putative precursors in the pancreas, and elucidating the molecular mechanisms involved in β-cell development (for a review, see Ref. 12). Similarly, the development of individualized therapy for endocrine cancer patients is being accelerated by improved molecular classification of tumor types and characterization of intra-tumoral heterogeneity using SC omics methods (for a review, see Ref. 13).

Table 1.1

Table 1.2

Studying Cellular State

Cells of a given cell type go through dynamic changes in gene expression and other cellular features in response to different stimuli, resulting in a complex landscape of cellular states.¹, ¹²¹ Recent advances in microfluidics,¹²² DNA barcoding and next-generation sequencing technologies gave rise to numerous omics methods, allowing the study of various aspects of cellular state at the SC level (see Table 1.1). These include characterization of the transcriptome, genome, whole-genome chromatin accessibility, chromatin state,⁴⁹ whole-genome DNA methylation, and proteome of individual cells within a tissue. The use of microfluidic devices to isolate and manipulate SCs greatly improves assay workflows and consistency, and enhances cell throughput. SC sequencing methods use barcoded sequences to label individual cells, thus enabling the multiplexing of thousands of cells (for a review, see Ref. 123).

Omics methods have enabled researchers to analyze intercellular variation within a tissue or cell line and generate an atlas of cell types, as well as gain insights into functional heterogeneity within an apparently homogeneous cell population. Additionally, techniques that can analyze multiple types of molecular profiles simultaneously, for instance, mRNA and DNA methylation, or mRNA and proteins, have been recently developed (for a review, see Ref. 124).

Cell-Type Classification and Characterization of Cell States (or Subtypes) Within a Tissue

How to distinguish a cell type (or subtype) from a transiently altered cell state is a difficult question.¹²⁵ Cell-type identification is typically based on well-established marker genes, and a cell type is defined as a group of cells sharing similar transcriptomic programs and carrying out the same function(s). Cell states may represent transitions, e.g., variations in the cell cycle, stress signatures, or temporary responses to the environment within a given cell type, thus implying an inherent plasticity (reviewed in Refs. 121, 126, 127). In practice, cell type vs. cell state identification by SC RNA-seq is often up to the laboratory conducting the study. With the increase in multi-omics studies, particularly the addition of SC ATAC-seq and single nucleus (SN) methylation to SC RNA-seq studies, it will become easier to discriminate cell types from cell states.

Endocrine organs are composed of several cell types, and studies using SC omics technologies have uncovered cellular heterogeneity within cell populations and detected variations in cell-type composition in disease. Studies of the pancreas provide a good example. The pancreas contains multiple cell types, including five hormone-secreting endocrine cell types (α-, β-, γ-, δ-, and ε-cells), three exocrine types (acinar, ductal, and centroacinar cells¹²⁸), as well as endothelial cells, macrophages, glia, and fibroblasts (reviewed in Ref. 129). In type 2 diabetes (T2D), β-cell function is impaired. Until SC RNA-seq studies were conducted in both mouse and human, the heterogeneity within the β-cell population had been underestimated.⁹¹–⁹⁵ Notably, Dorrell et al. report that differential expression of ST8S1A1 and CD9 distinguishes four β-cell subpopulations, while Muraro and colleagues detect subpopulations of β-cells expressing higher levels of FTH1 and acinar subpopulations expressing high levels of REG3A.

Importantly, analysis of SC transcriptomes from healthy and T2D human samples identifies each of the pancreatic endocrine and exocrine cell types, and shows the differential regulation of genes between T2D and healthy α-, β-, and δ-cells (Figure 1.1).⁹⁶ The results suggest that these differentially regulated genes are implicated in islet function, thus providing new avenues for the investigation of T2D pathogenesis.

Figure 1.1 Single-cell transcriptome analyses identify cell-type-specific expression changes in pancreatic islets from individuals with type 2 diabetes. (A) Type 2 diabetes (T2D) and non-diabetic (ND) SC transcriptomes cluster together by cell type following unsupervised hierarchical clustering. (B) The number of each ND and T2D cell type is classified by marker gene expression. Indicated in parentheses is the number of cells expected in each condition based on a χ ² test. (C–E, top) Scatter plots of log2 fold-change (FC) expression detected between T2D and ND samples from bulk RNA-seq from intact islets ( y -axis) and from Fluidigm C1 SC RNA-seq ( x -axis) from beta cells (left plot, red), alpha cells (middle plot, blue), and delta cells (right plot, green). (Bottom) Violin plots show examples of differentially expressed genes in one SC type. Dashed purple lines denote repressed genes in the respective T2D cell type, while dashed blue lines represent induced genes. (*) FDR < 0.05, (**) FDR < 0.01, (***) FDR < 0.001. (F) Venn diagram showing the intersections of differentially expressed genes identified between T2D and ND transcriptomes at SC type and islet SC ensemble resolution. The islet SC ensemble represents the pooled collection of beta, alpha, delta, and PP/gamma SCs. Reprinted from Ref. 96.

Pancreatic β-cells are not the only cell type showing intercellular heterogeneity in gene expression. A SC transcriptomic study in the human pancreas detected two subpopulations of α-cells: quiescent α-cells and a proliferating α-cell showing activation of the Sonic hedgehog signaling pathway.⁹⁷ In addition, SC RNA-seq studies of the pancreas have identified cells with a mixed α-/β-signature. Although further investigation is needed, these ambiguous cells may represent rare transitional cells and reflect a process of transdifferentiation of α-cells into β-cells (for a review, see Ref. 12).

It is noteworthy that SC RNA-seq analysis often results in the identification of multiple clusters, some corresponding to distinct cell types and others being different transient states of the same cell type. Cell stimulation can cause a cellular heterogeneous pattern of gene induction leading to the identification of distinct cell groups that nonetheless belong to the same cell type. SC epigenetic assays, such as SC ATAC or SN methylation, can identify stable chromatin differences that more accurately distinguish cell types than RNA expression assays (see Figure 1.2).¹³⁰ SC RNA-seq is also sensitive to stress response transcript changes occurring during cell dissociation. Working with nuclei enables the possibility of working on snap-frozen tissues, thereby eliminating the dissociation steps and potential artifacts not related to different cell types in the data.¹³¹ In the pancreas, Chiou and collaborators¹¹³ used SC epigenomics (SN ATAC-seq) to map the chromatin accessibility profiles of islet cells and define the regulatory programs of islet cell types and cell states and predicted the possible molecular mechanisms of T2D risk variants. They identified 13 cell clusters exhibiting different chromatin accessibility patterns at the promoter level, and could distinguish sub-clusters within the α-, β-, and δ-cell populations.

Figure 1.2 Schematic of the identification of cell clusters by SC transcriptomic and SC epigenomic analysis in the pituitary. (A) Graphic representation of the processing of pituitaries for SC omics assay and data analysis. Mouse pituitaries are collected and processed for SC assays, as indicated. Data analysis results in the identification of distinct cell clusters (B, C). (B) Example of an SC transcriptomic analysis resulting in the identification of 21 cell clusters in a pituitary sample. (C) SC epigenomic (DNA methylation or chromatin accessibility) analysis of the same pituitary sample as in B identifies only seven cell clusters. While the clusters identified by SC epigenomic analysis correspond to individual cell types, SC transcriptomic analysis distinguishes different cell states that are present within each cell type, thus the higher number of clusters.

SC studies have also been applied to other endocrine organs. The recent transcriptome mapping of human ovaries by SC RNA-seq identified five major cell types (granulosa cells, theca cells and stroma, smooth muscle cells, endothelial cells, and immune cells), different subpopulations of granulosa and theca cells, and several types of endothelial and smooth muscle cells, providing new insight into the dynamic cellular changes that occur during follicle maturation.⁹⁹ Remarkably, theca cells expressed several components of the complement system, suggesting it may have an impact on follicular remodeling and clinical implications with respect to female infertility.

SC omics studies have permitted the identification of cell types, unexpected cell subpopulations, and rare cellular states within a tissue. As noted above, Wang and colleagues discovered two subtypes of pancreatic α-cells, quiescent and activated α-cells, yet these might represent different states of the same cell type.⁹⁷ SC transcriptomics and SC epigenomics have the power to uncover rare cell types and cell states and to unravel some of the molecular mechanisms involved. The further integration of multi-omics SC assays, including proteome information, will lead to a more accurate identification of endocrine cell types and their regulatory processes.

SC Proteomic and Protein Secretion Profiling

SC assays are not restricted to transcriptome and epigenome profiling analyses. Because proteins directly correlate with a cell’s biological functions, proteomic studies provide critical information about cellular behaviors and cells’ phenotypic characteristics.¹²³ In particular, assessing the heterogeneity of SC secretomic signatures via secreted protein measurements facilitates the analysis of functional states, as released factors mediate both neighboring and long-distance cell-to-cell communication. Antibody-based methods such as mass cytometry by time-of-flight or CyTOF and SC barcode chips (SCBCs; for a review, see Ref. 90), as well as mass spectrometry (MS)-based methods are the main technologies used for measuring protein abundance in SCs (see Table 1.1).

CyTOF enables high-throughput detection and quantification of cell-surface, intracellular, and secreted proteins using rare metal isotope-conjugated antibodies.⁸⁹ This technique was used to quantify islet cell-type composition in human endocrine pancreas and led to the identification of multiple β-cell subtypes or states.¹¹⁴

MS is a label-free technology that allows for large-scale, untargeted detection, and relative quantitation of proteins with high sensitivity and specificity (for a review, see Ref. 100). Coupling it with microfluidics allows high-throughput SC MS analysis to be performed.¹²³ Matrix-assisted laser desorption/ionization (MALDI) MS is one of the most frequently used ionization methods in SC analysis (for a review, see Ref. 132). Characterization of cellular heterogeneity among rat islets of Langerhans was achieved by MALDI MS.¹¹⁵ Based on measurements of cell-type-specific canonical biomarkers like insulin and glucagon, the authors discovered that α-cells were more abundant in islets of the dorsal pancreas, and that γ-cells were predominant in ventral pancreatic islets. Moreover, they found endogenous peptides that were enriched in γ-cells from ventral islets compared to dorsal islets, and identified multiple β-cell states.

Reconstructing Developmental Trajectories

SC analysis allows reconstruction of developmental trajectories. Pseudotime is a computational method of ordering of cells based on expression profile similarities. It allows the identification of cell types at the beginning and end states of the trajectory, as well as in intermediate stages, thereby recapitulating entire or parts of developmental processes (for a review, see Ref. 133).

There are several examples of endocrine SC studies using pseudotime analysis. Notably, the maturation trajectory of mouse pancreatic β-cells was reconstructed by ordering β-cells isolated via fluorescence-activated cell sorting (FACS) either from fetal to adult stages¹³⁴ or from multiple postnatal time points.¹³⁵ Pseudotime analysis also demonstrated a subtle heterogeneity in neonatal and juvenile β-cells, likely reflecting distinct cell-cycling phases, origins, and maturation states, in contrast with a relative homogeneity of adult β-cells.¹³⁴ Similarly, it allowed Zeng and colleagues to capture precise signatures of immature, proliferative β-cells, and the overall contribution of reactive oxygen species and Srf/Jun/Fos transcription factors in postnatal β-cell proliferation.¹³⁵

Other studies using pseudotime analysis reported lineage dynamics during murine pancreatic morphogenesis.¹⁰⁶,¹⁰⁷ Scavuzzo et al. identified four endocrine progenitor (EP) subtypes during murine pancreatic morphogenesis.¹⁰⁷ These EP subtypes represent distinct stages of EP maturation of Ngn3-positive cells, with the first subtype expressing early EP markers, the second delamination and migration markers, and the third and fourth ones expressing markers of differentiated endocrine cells. Using a similar approach, another group characterized cell types during in vitro differentiation of human pluripotent stem cells into pancreatic β-cells and identified four major cell types: progenitors, endocrine cells (β-cells, α-like poly-hormonal cells), one type of non-endocrine cells, and cells resembling enterochromaffin cells (Figure 1.3).¹⁰⁸

Figure 1.3 High-resolution map of in vitro differentiation of human pancreatic islet cells. (A–C) SC RNA-seq data were used to reconstruct developmental trajectories. (A) Two-dimensional projection of gene expression (t-SNE plot) of 51,274 individual cells undergoing differentiation. The colors correspond to the development days of stage 5 (from day 0 in green to day 7 in blue). This stage was selected as stem-cell-derived β (SC-β-cell) and enterochromaffin cells (EC) are absent until the end of stage 4. The differentiation protocol was previously described in Ref. 136. (B) Expression of NEUROG3, a transiently expressed gene required for endocrine development, is shown in red in the same cell projection as in A. (C) The cell types from plot (A) are identified. The key lineage bifurcations in their trajectories from day 0 to day 7 are indicated with arrows. (D) Schematic of a proposed model for the differentiation of key cell types in the in vitro differentiation of human pancreatic islets. Reprinted and adapted from Ref. 108.

Altogether, these studies involving SC trajectory analysis may aid in understanding the molecular mechanisms that drive pancreatic islet lineage development and presage the identification of novel therapeutic targets to stimulate β-cell regeneration.

Multimodal Analysis of Cell States

While a SC sequencing assay allows the profiling of one type of biomolecule, concomitant assay of distinct molecular characteristics like RNA expression, chromatin conformation, epigenetic state, and protein expression gives researchers the opportunity to assemble a more comprehensive molecular picture of the cell. Multimodal SC measurement methods have been developed. As detailed in a recent review,¹²⁴ SC multimodal analysis can be classified into four main categories: (i) use of a non-destructive assay prior to the SC sequencing assay, (ii) use of parallel workflows on different cellular fractions from the same cell, (iii) use of a common workflow to measure distinct types of biomolecules within the same cell, and (iv) computation-based capture of additional information from SC RNA-seq data. A few examples of such approaches are described below.

In category (i), fluorescence-activated cell sorting (FACs) isolation of rare cell types or FACS-based protein measurements can be followed by SC RNA-seq. Different groups are adopting various approaches. Baron et al.¹⁰⁹ recently purified various cell types from the human pancreas using FACS index sorting and cell-type identification by SC RNA-seq. Another group performed whole-cell patch-clamp measurements (cell size, exocytosis, calcium, and sodium channel currents), followed by SC RNA-seq on pancreatic islet cells.¹¹⁰ This patch-seq approach for the pancreas enabled Camunas-Soler and collaborators to assess the electrophysiological activity of the major pancreatic islet cell types from healthy vs T2D donors, and identify a set of genes that predict β-cell electrophysiology.

A good illustration of a category (ii) approach is surveying the genome and transcriptome of the same cell in parallel, which allows the relationship between the genotype and the phenotype to be ascertained in an unambiguous manner. In G&T-seq, cytosolic mRNA and genomic DNA are separated using oligo-dT-coated beads that capture polyadenylated mRNA molecules, and then amplified and sequenced.⁵⁸ A SC triple omics sequencing technique, scTrio-seq, which was developed to simultaneously analyze the genome, methylome, and transcriptome of single cancer cells, allowed identification of two distinct hepatocellular carcinoma (HCC) subpopulations that differed significantly in DNA copy number, methylome, and transcriptome profiles.⁶⁰ Such techniques could be used to probe intra-tumor heterogeneity in other endocrine cancers.

Examples of category (iii) approaches include cellular indexing of transcriptomes and epitopes by sequencing (CITE-seq) as well as the RNA expression and protein sequencing assay (REAP-seq). Both measure polyadenylated mRNAs and protein levels.³⁷,³⁸ In these techniques, antibodies are conjugated to polyadenylated DNA barcodes that can be captured by the oligo-dT primers used in SC RNA-seq. A transcriptome-wide analysis of the effects of multiple gene perturbations in SCs can be achieved by combining CRISPR/Cas9-mediated multi-locus gene perturbations with SC RNA-seq.³⁹ In the CRISPR/Cas9 system, a small guide RNA (sgRNA) directs Cas9 nuclease to specifically edit or silence a gene of interest (for a review, see Ref. 137). In Perturb-seq, a polyadenylated expressed guide barcode (GBC) encodes the identity of each sgRNA, thus enabling its detection in the SC RNA-seq readout.

Finally, category (iv) consists of capturing additional information from SC RNA-seq data through a computational method, such as inferring somatic mutations or copy number variants. Enge and colleagues¹³⁸ analyzed pancreatic cells from individuals spanning a wide range of ages and characterized transcriptional signatures of aging and the pattern of somatic substitutions.

Spatial Transcriptomics

While SC RNA-seq represents a powerful high-throughput method for transcriptome-wide gene expression profiling, it requires tissue dissociation, which precludes determination of the spatial distribution of transcripts. For complex tissues, obtaining spatial information can help clarify cellular identification and understand cellular regulation. Spatial transcriptomic methods have been developed to capture gene expression profiles in cells within their native tissue context. These comprise fluorescence in situ hybridization (FISH)-based methods and in situ sequencing methods.

Large-scale FISH-based techniques include sequential FISH (seqFISH),¹³⁹, ¹⁴⁰ multiplexed error-robust FISH (MERFISH),¹⁴¹ and more recently seqFISH   +,¹⁴² which achieves transcriptome-level profiling in SCs. In contrast with FISH-based methods, in situ sequencing methods do not require a preselection of gene sets for measurement, yet they tend to either have a lower transcriptome coverage³³,¹⁴³ or lack cellular resolution.³⁶,¹⁴⁴ The recent development of Slide-seq technology enables spatially resolved gene expression profiling in various tissue contexts at nearly SC resolution.¹⁴⁵ Liu et al. studied the spatial distribution of transcripts measured by in situ sequencing in human fetal pancreas at three developmental ages.¹¹² They identified developing endocrine islets based on the expression of marker genes, determined the density profile of some transcripts in the vicinity of the forming islets or in relation to the closest nucleus and compared it at three developmental ages.

Integrative approaches that combine spatial transcriptomics with SC RNA-seq data from the same sample can overcome the limitations of each method and enable the construction of a single-cell resolution transcriptome-wide gene expression atlas for a tissue.¹¹¹,¹⁴⁶,¹⁴⁷ The Yanai group applied their integration of the spatial transcriptomics method³⁶ and SC RNA-seq data to distinguish and spatially map cell-type subpopulations across the tissue regions of pancreatic tumors (Figure 1.4).¹¹¹

Figure 1.4 Mapping of cell types in pancreatic tumors using spatial transcriptomics. (A) Annotated pancreatic ductal adenocarcinoma (PDAC)-A tumor cryosection on the spatial transcriptomics (ST) microarray slide. The annotations specify a region high in cancer cells and desmoplasia (red), duct epithelium (yellow), and normal pancreatic tissue (blue). Scale bar, 1 mm. (B) Standardized expression levels of three genes in the PDAC-A ST dataset. (C) Clustering of the PDAC-A ST spots. Colors denote the clustering assignments. Reprinted and adapted from Ref. 111.

While SC transcriptomics and spatially resolved transcriptomic profiling can generate an atlas of cell states and reveal a spatial pattern of gene expression within a tissue, they cannot capture the dynamics of gene expression and dynamic cell behaviors. In the next section, we describe methods that allow the study of cellular dynamics and stimulus responses.

Studying Cellular Dynamics and Stimulus Responses

Genetically identical cells may respond differently to the same stimulus. Notably, signaling proteins can exhibit cell-to-cell heterogeneity in their patterns of activity, e.g., pulses, bursts, oscillations, or switches (for review, see Ref. 148), as well as in their patterns of localization. Recent technological advances in live-cell fluorescence imaging-based approaches, optogenetics, and microfluidics have enabled measurements of the dynamics of signal transduction and stimulus responses in SCs (see Table 1.1) (for a review, see Ref. 149). Live-cell fluorescence imaging allows the visualization of spatiotemporal dynamics of signaling molecules using biosensors (for a review, see Ref. 77). Optogenetics allow the control of signaling dynamics. Integration with a microfluidic device provides the ability to control stimulus delivery, i.e., dose and timing of stimulus.¹⁵⁰ Another area of interest is the analysis of the temporal dynamics of protein secretion, which can provide valuable insights into the functional states of cells. We will give some examples in the following sub-sections.

Live-Cell Fluorescence Imaging

Chemical and genetically encoded fluorescent biosensors are both widely used in live-cell fluorescence imaging and provide high spatiotemporal resolution.⁷⁸,¹⁵¹ However, genetically encoded fluorescent probes (GEFPs) are much more specific than chemical fluorescent indicators.

Live-cell calcium imaging using a chemical calcium indicator was used to examine changes in intracellular calcium levels in response to multiple sequential glucose challenges in individual cells within intact pancreatic islets from wild-type vs diabetic mice.¹¹⁶ In the same study, analysis of endocrine marker expression in the imaged islets revealed that diabetic islets contained β and α cells with aberrant calcium responses. Another report describes the functional characterization of α and β cells from dispersed pancreatic islets by imaging of NADPH and calcium responses to glucose using chemical indicators, followed by immunohistochemistry (IHC)-based cell-type identification.⁸⁰ In order to monitor glucose-induced calcium influx in β cells within zebrafish primary islets, Janjuha and colleagues utilized a double transgenic zebrafish line expressing a genetically encoded calcium indicator and a nucleus marker, both driven by the β cell-specific insulin promoter.¹¹⁷

A large variety of single fluorophore-based as well as Förster resonance energy transfer (between two spectrally overlapping fluorescent proteins; abbreviated as FRET)-based kinase activity reporters have been used in live-cell imaging experiments.⁷² Such tools allow the dissection of kinase activation dynamics in live endocrine cells following hormone stimulation as well as in a pathological context. Furthermore, a new generation of genetically encoded FRET biosensors have been developed for the measurement of Gαi activation kinetics in SCs, which are potentially suited for the study of live endocrine cells.⁷³

Integration of Optogenetics with Fluorescence Imaging

The optogenetics technology uses genetically encoded photoactivatable proteins, referred to as optogenetic actuators (e.g., channelrhodopsin-2), to control intracellular signal transduction in living cells. Cellular responses to actuator-mediated perturbations are reported by genetically encoded optogenetic sensors (e.g., a calcium indicator) that modulate their fluorescence in response to a physiological variable (for reference, see Ref. 63). Actuators can be activated on a millisecond-time scale and targeted to a specific cell population using cell-type-specific promoters, making optogenetics achieve high temporal and spatial resolution (for review, see Ref. 152).

As pancreatic β cells are electrically excitable, optogenetic actuators can be used to modulate their electrical excitability and function.¹¹⁸,¹¹⁹ In β cells, insulin secretion depends on an elevation of intracellular calcium concentration. Integrated, non-overlapping optogenetics/fluorescence imaging of β cells was used to examine calcium dynamics and the effect of sulfonylurea (that is known to augment insulin release) on the calcium response.¹⁵³ Another study used a dual optogenetic/photopharmacological strategy including reversible silencing of calcium influx to study the roles of individual β cells within the pancreatic islet in regulating islet function. The authors found that a few hubs of immature β cells orchestrate islet calcium responses to glucose, thereby dictating emergent population behavior in response to glucose.¹⁵⁴ Thus, optogenetic approaches may help to study β cell function and facilitate the development of novel therapies.¹²⁰

Protein Secretion Dynamics

Several imaging-based and SC barcode microarray-based approaches can dissect the temporal dynamics of protein secretions, as well as integrate with other methods (for a review, see Ref. 85). Junkin et al. developed a two-layer microfluidic platform where SCs are in a culture chamber where they receive time-varying inputs, and nanoliter immunoassay-based measurements of secreted proteins occur at different times in a binding chamber. Additionally, time-lapse microscopy was used to measure SC transcription factor dynamics (i.e., NF-κB nuclear localization) simultaneously with protein secretion.⁸⁶ Other groups previously utilized real-time imaging techniques to measure SC protein secretions over time in conjunction with either fluorescence microscopy to examine cell morphology changes⁸⁷ or live-cell imaging to monitor plasma membrane integrity.⁸⁸ Another label-free technology has been reported that utilizes photonic crystal resonant imaging to analyze SC secretion in vitro in real-time with subcellular spatial resolution.¹⁵⁵ These techniques represent valuable tools for mapping the kinetics of protein secretion in individual cells within endocrine cell populations in both healthy and diseased states, as well as in response to drug treatments.

SC Studies: Application to the Anterior Pituitary and the Pituitary Gonadotropes

The anterior pituitary (AP) is a complex tissue, composed of at least five hormone-producing cell types (somatotropes, gonadotropes, lactotropes, thyrotropes, and corticotropes), as well as non-endocrine cell types that include folliculostellate cells (FSC) and adult progenitor cells. AP cell types, which are responsive to hypothalamic signals (or stimuli) and to peripheral regulators,¹⁵⁶ produce hormones that regulate important functions in target tissue or organs (e.g., tissue growth, milk production, gamete production, thyroid hormone release, cortisol release). Understanding the physiology of the AP requires a comprehensive, transcriptome-wide characterization of its cell populations. Moreover, SC transcriptomics enables researchers to study the dynamics of gene regulation in the AP.

Recent studies, including ours, used SC RNA-seq to map the cellular composition of either the mouse or rat AP,¹¹,¹⁰²–¹⁰⁵,¹⁵⁷ and employed qPCR, immunostaining, or in situ hybridization to validate their results. Bioinformatics analysis of the sequencing data allowed the identification of different cell clusters – between 9 and 13 depending on the study – that aligned closely with marker-based classification. Besides the detection of the classical secretory cell types, studies identified cell clusters corresponding to non-endocrine cells such as pericytes, macrophages, endothelial cells,¹⁰²–¹⁰⁵,¹⁵⁷ as well as poly-hormonal cells.¹¹,¹⁰⁴ As expected, nearly all studies identified a stem cell cluster and a cluster of proliferating Pou1f1-expressing cells.¹⁰²–¹⁰⁴,¹⁵⁷ While single clusters accounted for most cell types in the majority of studies, FSC formed two large clusters in the rat AP,¹⁰³ and two small clusters in one of the mouse studies.¹⁰⁴ By contrast, FSC were undetected in the other mouse studies.¹⁰²,¹⁰⁵ Although these discrepancies require further investigation, they suggest that FSC may be more prone to lysis following pituitary cell dissociation or cell preparation. Novel markers were identified in each cell type, for instance, Foxp2 in gonadotrope cells.¹⁰²,¹⁰³ Cheung et al. also identified subpopulations within the lactotropes and somatotropes, with one of the somatotrope subpopulations being enriched in sterol/cholesterol biosynthesis genes. The presence of sexual dimorphism in some cell types was another important finding highlighted in two studies: in rat, lactotropes, gonadotropes, and somatotropes formed distinct clusters in females as compared with males¹⁰³; in mouse, there was a marked sexual dimorphism in pituitary cell composition, with a higher proportion of Gh   +/Prl   − cells in male pituitaries, and more Gh   −/Prl   + cells in females.¹⁰⁴ Overall, these studies highlight the population heterogeneity of pituitary gland cells, notably among the hormone-producing cell types. Remarkably, the study of Ho et al. demonstrated the ability of SC transcriptomic analysis to detect dynamic shifts in specific cell populations and their transcriptomes in response to specific physiological demands (lactation and over-expression of Ghrh), thus uncovering cellular plasticity within the adult pituitary. When comparing lactating mice with female control, the authors found a higher percentage of lactotrope cells in lactating females, as well as increased Prl mRNA expression within the lactotrope cell population. Similarly, they observed an expansion of the somatotrope cluster and an increase in Gh mRNA expression in transgenic female mice over-expressing Gnrh as compared with non-transgenic females.

Besides cell-type characterization, SC studies have led to new insights on the dynamics of transcriptional responses to various stimuli. Several groups, including ours, have been involved in the study of transcriptional and signaling responses to GnRH in gonadotrope cells using a combination of bulk¹⁰,¹⁵⁸–¹⁷¹ and SC approaches.¹¹,¹⁷² Most of these studies utilized the LβT2 gonadotrope cell line, a well-established model system¹⁷³ for studying the regulation of gonadotropin subunit gene expression,¹⁵⁹,¹⁷⁴–¹⁷⁹ as the scarcity of gonadotrope cells in the AP renders primary culture studies difficult. Using bulk and SC assays, we recently demonstrated that LβT2 gonadotropes have a distinct pattern of immediate early gene (IEG) expression in response to GnRH pulse stimulation.¹⁰,¹¹ While bulk assays detected a low level of basal IEG (Egr1 or Fos) expression before each GnRH pulse and a peak of expression after the pulse, SC qPCR experiments showed an all-or-none switch-like mechanism of activation, with cells either expressing the IEG transcript or not. Thus, the peak of IEG expression reflected an increased number of cells expressing IEG rather than a higher level of transcript per cell. Furthermore, when cells were stimulated with increasing GnRH concentrations, more cells were activated, while IEG transcript level per cell remained unchanged (Figure 1.5). The identification of this all-or-none pattern of IEG induction highlights the improved resolution of SC studies over bulk studies for delineating gene regulatory mechanisms.

Figure 1.5 Characterization of SC heterogeneity in IEG responses to GnRH in LβT2 gonadotrope cells. (A) Cells were exposed to increasing GnRH concentrations and collected 35 min after the fourth GnRH pulse. Vertical scatter plots of Egr2 , Fosb , and Rps25 expression in SCs. The number of gene expressing (top) and non-gene expressing cells (bottom) is indicated in parentheses; each square symbolizes a cell; the green dotted line represents the average expression in all analyzed cells at a given GnRH concentration and corresponds to the population average. (B) Plots of the percentages of cells expressing an IEG, as indicated. Error bars are based on the binomial standard deviation on the number of gene expressing cells. (C) Bar graphs of average gene expression in gene expressing (i.e., activated) cells. Error bars signify standard deviation. ANOVA shows no significant differences. Overall, SC probability of IEG induction in LβT2 gonadotrope cells is concentration dependent and varies for each IEG. Cells are either gene-expressing (i.e., activated) or non-gene expressing, as shown in (A). The proportion of activated cells increases with increasing stimulus level (A, B), whereas the level of expression of each gene in activated cells is unaffected (C). Reprinted and adapted from Ref. 11.

Discussion

With the latest advances in SC technologies, it is possible to interrogate cellular heterogeneity in multiple ways, such as characterizing cellular states via the measurement of different modalities (e.g., transcriptomes, genomes, epigenomes, and proteomes), reconstructing developmental trajectories, or studying signaling dynamics and protein secretion dynamics. Despite considerable advances, some challenges remain. We discuss below the major pitfalls and limitations of current SC studies/methods and propose future directions.

While omics technologies like SC RNA-seq are characterized by high-throughput and multiplexing, live-cell imaging-based approaches that enable the measurement of signaling dynamics have either relatively low throughput (tens to hundreds of cells) or low multiplexing capabilities (a few measurements per cell), thus limiting the statistical power of a study (for a review, see Ref. 149). Among SC proteomic techniques, antibody-based CyTOF enables high-throughput analysis of cells, yet the specificity of detection tends to be low.⁸⁹ By contrast, MS-based methods achieve high specificity and higher proteome coverage than CyTOF, yet with a limited cell throughput. The incorporation of automation technologies for liquid handling, the implementation of parallel ion accumulation, and the use of larger barcode sets for multiplexing should enhance cell throughput in MS.¹⁸⁰

Another challenge that is inherent to SC analysis, particularly omics data, is bioinformatic or/and mechanistic interpretation, notably the ability to discriminate meaningful biological states from stochastic, functionally meaningless biochemical fluctuations such as gene expression noise.⁷,⁸,¹⁸¹ Moreover, the variety of experimental and computational analysis approaches in SC omics studies makes SC data analysis and data interpretation challenging. Due to the absence of standards for data pre-processing and noise removal, comparing the results from different studies is more complicated (for a review, see Ref. 182). Thus, it is important to evaluate the performance of algorithms/computational methods for data derived from SC RNA-seq experiments.¹⁸³ Furthermore, validation of SC data should be performed using independent SC approaches such as RNA FISH, immunofluorescence, or live-cell imaging.

Despite the substantial progress in SC multimodal omics technologies, these methods can be costly for large-scale analyses, and low coverage makes it difficult to discriminate signal from technical noise (for a review, see Ref. 184). Thus, new strategies may be needed to enhance these technologies. Spatially resolved transcriptomics approaches, on the other hand, circumvent the dissociation-induced cell perturbations that are associated with conventional SC sequencing methods and capture the spatial location of cells within frozen or fixed tissue.¹⁸⁵ Improvements in spatial resolution¹⁸⁶ and data analysis are underway that will broaden the use of spatial transcriptomics, allow researchers to build robust organism expression atlases,¹⁸⁷ and gain major insights into cell differentiation and development.¹⁸⁸ Finally, the future of SC omics studies also depends on the promotion of cross-disciplinary collaborations between biologists and computational scientists.¹⁸⁹,¹⁹⁰

The pace of development and improvement in the techniques for the analysis of individual cells has been breathtaking. The individual cell is the basic computational unit of endocrine systems. Over the next few years, as high-content multimodal data at SC resolution in the context of spatial organization accrue and the methods to interpret these massive datasets improve, the insight into the organization and physiology of endocrine systems in health and disease should dramatically increase.

References

1. Morris S.A. The evolving concept of cell identity in the single cell era. Development. 2019;146(12):1–5. doi:10.1242/dev.169748.

2. Jeknic S., Kudo T., Covert M.W. Techniques for studying decoding of single cell dynamics. Front Immunol. 2019;10:755.

3. Krieger T., Simons B.D. Dynamic stem cell heterogeneity. Development. 2015;142(8):1396–1406.

4. Turco M.Y., Furia L., Dietze A., Fernandez Diaz L., Ronzoni S., Sciullo A. Cellular heterogeneity during embryonic stem cell differentiation to epiblast stem cells is revealed by the ShcD/RaLP adaptor protein. Stem Cells. 2012;30(11):2423–2436.

5. Gay L., Baker A.M., Graham T.A. Tumour cell heterogeneity. F1000Res. 2016;5:1–14. doi:10.12688/f1000research.7210.1.

6. Sasaki N., Clevers H. Studying cellular heterogeneity and drug sensitivity in colorectal cancer using organoid technology. Curr Opin Genet Dev. 2018;52:117–122.

7. Altschuler S.J., Wu L.F. Cellular heterogeneity: do differences make a difference?. Cell. 2010;141(4):559–563.

8. Goldman S.L., MacKay M., Afshinnekoo E., Melnick A.M., Wu S., Mason C.E. The impact of heterogeneity on single-cell sequencing. Front Genet. 2019;10:1–8. doi:10.3389/fgene.2019.00008.

9. Roy A.L., Conroy R., Smith J., Yao Y., Beckel-Mitchener A.C., Anderson J.M. Accelerating a paradigm shift: The Common Fund Single Cell Analysis Program. Sci Adv. 2018;4(8):eaat8573.

10. Stern E., Ruf-Zamojski F., Zalepa-King L., Pincas H., Choi S.G., Peskin C.S. Modeling and high-throughput experimental data uncover the mechanisms underlying Fshb gene sensitivity to gonadotropin-releasing hormone pulse frequency. J Biol Chem. 2017;292(23):9815–9829.

11. Ruf-Zamojski F., Ge Y., Nair V., Zamojski M., Pincas H., Toufaily C. Single-cell stabilization method identifies gonadotrope transcriptional dynamics and pituitary cell type heterogeneity. Nucleic Acids Res. 2018;46(21):11370–11380.

12. Tritschler S., Theis F.J., Lickert H., Bottcher A. Systematic single-cell analysis provides new insights into heterogeneity and plasticity of the pancreas. Mol Metab. 2017;6(9):974–990.

13. Koh J., Allbritton N.L., Sosa J.A. Single-cell approaches for molecular classification of endocrine tumors. Curr Opin Oncol. 2016;28(1):43–49.

14. Gawad C., Koh W., Quake S.R. Single-cell genome sequencing: current state of the science. Nat Rev Genet. 2016;17(3):175–188.

15. Huang L., Ma F., Chapman A., Lu S., Xie X.S. Single-cell whole-genome amplification and sequencing: methodology and applications. Annu Rev Genomics Hum Genet. 2015;16:79–102.

16. Todorovic V. Amplification-free single-cell whole-genome sequencing gets a makeover. Nat Methods. 2020;17(1):27.

17. Knouse K.A., Wu J., Amon A. Assessment of megabase-scale somatic copy number variation using single-cell sequencing. Genome Res. 2016;26(3):376–384.

18. Macosko E.Z., Basu A., Satija R., Nemesh J., Shekhar K., Goldman M. Highly parallel genome-wide expression profiling of individual cells using nanoliter droplets. Cell. 2015;161(5):1202–1214.

19. Zheng G.X., Terry J.M., Belgrader P., Ryvkin P., Bent Z.W., Wilson R. Massively parallel digital transcriptional profiling of single cells. Nat Commun. 2017;8:14049.

20. Islam S., Kjallquist U., Moliner A., Zajac P., Fan J.B., Lonnerberg P. Highly multiplexed and strand-specific single-cell RNA 5' end sequencing. Nat Protoc. 2012;7(5):813–828.

21. Singh M., Al-Eryani G., Carswell S., Ferguson J.M., Blackburn J., Barton K. High-throughput targeted long-read single cell sequencing reveals the clonal and transcriptional landscape of lymphocytes. Nat Commun. 2019;10(1):3120.

22. Genshaft A.S., Li S., Gallant C.J., Darmanis S., Prakadan S.M., Ziegler C.G. Multiplexed, targeted profiling of single-cell proteomes and transcriptomes in a single reaction. Genome Biol. 2016;17(1):188.

23. Hashimshony T., Senderovich N., Avital G., Klochendler A., de Leeuw Y., Anavy L. CEL-Seq2: sensitive highly-multiplexed single-cell RNA-Seq. Genome Biol. 2016;17:77.

24. Sheng K., Cao W., Niu Y., Deng Q., Zong C. Effective detection of variation in single-cell transcriptomes using MATQ-seq. Nat Methods. 2017;14(3):267–270.

25. Gierahn T.M., Wadsworth 2nd M.H., Hughes T.K., Bryson B.D., Butler A., Satija R. Seq-Well: portable, low-cost RNA sequencing of single cells at high throughput. Nat Methods. 2017;14(4):395–398.

26. Pollen A.A., Nowakowski T.J., Shuga J., Wang X., Leyrat A.A., Lui J.H. Low-coverage single-cell mRNA sequencing reveals cellular heterogeneity and activated signaling pathways in developing cerebral cortex. Nat Biotechnol. 2014;32(10):1053–1058.

27. Natarajan K.N. Single-cell tagged reverse transcription (STRT-Seq). Methods Mol Biol. 2019;1979:133–153.

28. Khan S., Kaihara K.A. Single-cell RNA-sequencing of peripheral blood mononuclear cells with ddSEQ. Methods Mol Biol. 2019;1979:155–176.

29. Jaitin D.A., Kenigsberg E., Keren-Shaul H., Elefant N., Paul F., Zaretsky I. Massively parallel single-cell RNA-seq for marker-free decomposition of tissues into cell types. Science. 2014;343(6172):776–779.

30. Picelli S., Faridani O.R., Bjorklund A.K., Winberg G., Sagasser S., Sandberg R. Full-length RNA-seq from single cells using Smart-seq2. Nat Protoc. 2014;9(1):171–181.

31. See P., Lum J., Chen J., Ginhoux F. A single-cell sequencing guide for immunologists. Front Immunol. 2018;9:2425.

32. Bagnoli J.W., Ziegenhain C., Janjic A., Wange L.E., Vieth B., Parekh S. Sensitive and powerful single-cell RNA sequencing using mcSCRB-seq. Nat Commun. 2018;9(1):2937.

33. Wang X., Allen W.E., Wright M.A., Sylwestrak E.L., Samusik N., Vesuna S. Three-dimensional intact-tissue sequencing of single-cell transcriptional states. Science. 2018;361(6400):1–22. doi:10.1126/science.aat5691 eaat5691.

34. Burgess D.J. Spatial transcriptomics coming of age. Nat Rev Genet. 2019;20(6):317.

35. Lyubimova A., Itzkovitz S., Junker J.P., Fan Z.P., Wu X., van Oudenaarden A. Single-molecule mRNA detection and counting in mammalian tissue. Nat Protoc. 2013;8(9):1743–1758.

36. Stahl P.L., Salmen F., Vickovic S., Lundmark A., Navarro J.F., Magnusson J. Visualization and analysis of gene expression in tissue sections by spatial transcriptomics. Science. 2016;353(6294):78–82.

37. Stoeckius M., Hafemeister C., Stephenson W., Houck-Loomis B., Chattopadhyay P.K., Swerdlow H. Simultaneous epitope and transcriptome measurement in single cells. Nat Methods. 2017;14(9):865–868.

38. Peterson V.M., Zhang K.X., Kumar N., Wong J., Li L., Wilson D.C. Multiplexed quantification of proteins and transcripts in single cells. Nat Biotechnol.