Modeling bi-modality improves characterization of cell cycle on gene expression in single cells

SMURF: embedding single-cell RNA-seq data with matrix factorization preserving self-consistency

The advance in single-cell RNA-sequencing (scRNA-seq) sheds light on cell-specific transcriptomic studies of cell developments, complex diseases and cancers. Nevertheless, scRNA-seq techniques suffer from 'dropout' events, and imputation tools are proposed to address the sparsity. Here, rather than imputation, we propose a tool, SMURF, to extract the low-dimensional embeddings from cells and genes utilizing matrix factorization with a mixture of Poisson-Gamma divergent as objective while preserving self-consistency. SMURF exhibits feasible cell subpopulation discovery efficacy with obtained cell embeddings on replicated in silico and eight web lab scRNA datasets with ground truth cell types. Furthermore, SMURF can reduce the cell embedding to a 1D-oval space to recover the time course of cell cycle. SMURF can also serve as an imputation tool; the in silico data assessment shows that SMURF parades the most robust gene expression recovery power with low root mean square error and high Pearson correlation. Moreover, SMURF recovers the gene distribution for the WM989 Drop-seq data. SMURF is available at https://github.com/deepomicslab/SMURF.

Incorporating cell hierarchy to decipher the functional diversity of single cells

Cells possess functional diversity hierarchically. However, most single-cell analyses neglect the nested structures while detecting and visualizing the functional diversity. Here, we incorporate cell hierarchy to study functional diversity at subpopulation, club (i.e., sub-subpopulation), and cell layers. Accordingly, we implement a package, SEAT, to construct cell hierarchies utilizing structure entropy by minimizing the global uncertainty in cell-cell graphs. With cell hierarchies, SEAT deciphers functional diversity in 36 datasets covering scRNA, scDNA, scATAC, and scRNA-scATAC multiome. First, SEAT finds optimal cell subpopulations with high clustering accuracy. It identifies cell types or fates from omics profiles and boosts accuracy from 0.34 to 1. Second, SEAT detects insightful functional diversity among cell clubs. The hierarchy of breast cancer cells reveals that the specific tumor cell club drives AREG-EGFT signaling. We identify a dense co-accessibility network of cis-regulatory elements specified by one cell club in GM12878. Third, the cell order from the hierarchy infers periodic pseudo-time of cells, improving accuracy from 0.79 to 0.89. Moreover, we incorporate cell hierarchy layers as prior knowledge to refine nonlinear dimension reduction, enabling us to visualize hierarchical cell layouts in low-dimensional space.

G1/S restriction point coordinates phasic gene expression and cell differentiation

Pluripotent embryonic stem cells have a unique cell cycle structure with a suppressed G1/S restriction point and little differential expression across the cell cycle phases. Here, we evaluate the link between G1/S restriction point activation, phasic gene expression, and cellular differentiation. Expression analysis reveals a gain in phasic gene expression across lineages between embryonic days E7.5 and E9.5. Genetic manipulation of the G1/S restriction point regulators miR-302 and P27 respectively accelerates or delays the onset of phasic gene expression in mouse embryos. Loss of miR-302-mediated p21 or p27 suppression expedites embryonic stem cell differentiation, while a constitutive Cyclin E mutant blocks it. Together, these findings uncover a causal relationship between emergence of the G1/S restriction point with a gain in phasic gene expression and cellular differentiation.

Distinguishing activated T regulatory cell and T conventional cells by single cell technologies

Resting conventional T cells (Tconv) can be distinguished from T regulatory cells (Treg) by the canonical markers FOXP3, CD25 and CD127. However, the expression of these proteins alters after T‐cell activation leading to overlap between Tconv and Treg. The objective of this study was to distinguish resting and antigen‐responsive T effector (Tconv) and Treg using single‐cell technologies. CD4⁺ Treg and Tconv cells were stimulated with antigen and responsive and non‐responsive populations processed for targeted and non‐targeted single‐cell RNAseq. Machine learning was used to generate a limited set of genes that could distinguish responding and non‐responding Treg and Tconv cells and which was used for single‐cell multiplex qPCR and to design a flow cytometry panel. Targeted scRNAseq clearly distinguished the four‐cell populations. A minimal set of 27 genes was identified by machine learning algorithms to provide discrimination of the four populations at >95% accuracy. In all, 15 of the genes were validated to be differentially expressed by single‐cell multiplex qPCR. Discrimination of responding Treg from responding Tconv could be achieved by a flow cytometry strategy that included staining for CD25, CD127, FOXP3, IKZF2, ITGA4, and the novel marker TRIM which was strongly expressed in Tconv and weakly expressed in both responding and non‐responding Treg. A minimal set of genes was identified that discriminates responding and non‐responding CD4⁺ Treg and Tconv cells and, which have identified TRIM as a marker to distinguish Treg by flow cytometry.

Maternal Type 1 Diabetes Reduces Autoantigen-Responsive CD4+ T Cells in Offspring

Autoimmunity against pancreatic β-cell autoantigens is a characteristic of childhood type 1 diabetes (T1D). Autoimmunity usually appears in genetically susceptible children with the development of autoantibodies against (pro)insulin in early childhood. The offspring of mothers with T1D are protected from this process. The aim of this study was to determine whether the protection conferred by maternal T1D is associated with improved neonatal tolerance against (pro)insulin. Consistent with improved neonatal tolerance, the offspring of mothers with T1D had reduced cord blood CD4⁺ T-cell responses to proinsulin and insulin, a reduction in the inflammatory profile of their proinsulin-responsive CD4⁺ T cells, and improved regulation of CD4⁺ T cell responses to proinsulin at 9 months of age, as compared with offspring with a father or sibling with T1D. Maternal T1D was also associated with a modest reduction in CpG methylation of the INS gene in cord blood mononuclear cells from offspring with a susceptible INS genotype. Our findings support the concept that a maternal T1D environment improves neonatal immune tolerance against the autoantigen (pro)insulin.

Latent periodic process inference from single-cell RNA-seq data

The development of a phenotype in a multicellular organism often involves multiple, simultaneously occurring biological processes. Advances in single-cell RNA-sequencing make it possible to infer latent developmental processes from the transcriptomic profiles of cells at various developmental stages. Accurate characterization is challenging however, particularly for periodic processes such as cell cycle. To address this, we develop Cyclum, an autoencoder approach identifying circular trajectories in the gene expression space. Cyclum substantially improves the accuracy and robustness of cell-cycle characterization beyond existing approaches. Applying Cyclum to removing cell-cycle effects substantially improves delineations of cell subpopulations, which is useful for establishing various cell atlases and studying tumor heterogeneity.

Deciphering the transcriptomic landscape of tumor-infiltrating CD8 lymphocytes in B16 melanoma tumors with single-cell RNA-Seq

Recent studies have proposed that tumor-specific tumor-infiltrating CD8⁺ T lymphocytes (CD8 TIL) can be classified into two main groups: "exhausted" TILs, characterized by high expression of the inhibitory receptors PD-1 and TIM-3 and lack of transcription factor 1 (Tcf1); and "memory-like" TILs, with self-renewal capacity and co-expressing Tcf1 and PD-1. However, a comprehensive definition of the heterogeneity existing within CD8 TILs has yet to be clearly established. To investigate this heterogeneity at the transcriptomic level, we performed paired single-cell RNA and TCR sequencing of CD8 T cells infiltrating B16 murine melanoma tumors, including cells of known tumor specificity. Unsupervised clustering and gene-signature analysis revealed four distinct CD8 TIL states - exhausted, memory-like, naïve and effector memory-like (EM-like) - and predicted novel markers, including Ly6C for the EM-like cells, that were validated by flow cytometry. Tumor-specific PMEL T cells were predominantly found within the exhausted and memory-like states but also within the EM-like state. Further, T cell receptor sequencing revealed a large clonal expansion of exhausted, memory-like and EM-like cells with partial clonal relatedness between them. Finally, meta-analyses of public bulk and single-cell RNA-seq data suggested that anti-PD-1 treatment induces the expansion of EM-like cells. Our reference map of the transcriptomic landscape of murine CD8 TILs will help interpreting future bulk and single-cell transcriptomic studies and may guide the analysis of CD8IL subpopulations in response to therapeutic interventions.

Scientific Regulatory Policy Committee Points to Consider*: Nuisance Factors, Block Effects, and Batch Effects in Nonclinical Safety Assessment Studies

Detection of test article-related effects and the determination of the adversity of those changes are the primary goals of nonclinical safety assessment studies for drugs and chemicals in development. During these studies, variables that are not of primary interest to investigators may change and influence data interpretation. These variables, often referred to as "nuisance factors," may influence other groups of data and result in "block or batch effects" that complicate data interpretation. Definitions of the terms "nuisance factors," "block effects," and "batch effects," as they apply to nonclinical safety assessment studies, are reviewed. Multiple case examples of block and batch effects in safety assessment studies are provided, and the challenges these bring to pathology data interpretation are discussed. Methods to mitigate the occurrence of block and batch effects in safety assessment studies, including statistical blocking and utilization of study designs that minimize potential confounding variables, incorporation of adequate randomization, and use of an appropriate number of animals or repeated measurement of specific parameters for increased precision, are reviewed. [Box: see text].

Single-Cell Expression Variability Implies Cell Function

As single-cell RNA sequencing (scRNA-seq) data becomes widely available, cell-to-cell variability in gene expression, or single-cell expression variability (scEV), has been increasingly appreciated. However, it remains unclear whether this variability is functionally important and, if so, what are its implications for multi-cellular organisms. Here, we analyzed multiple scRNA-seq data sets from lymphoblastoid cell lines (LCLs), lung airway epithelial cells (LAECs), and dermal fibroblasts (DFs) and, for each cell type, selected a group of homogenous cells with highly similar expression profiles. We estimated the scEV levels for genes after correcting the mean-variance dependency in that data and identified 465, 466, and 364 highly variable genes (HVGs) in LCLs, LAECs, and DFs, respectively. Functions of these HVGs were found to be enriched with those biological processes precisely relevant to the corresponding cell type’s function, from which the scRNA-seq data used to identify HVGs were generated—e.g., cytokine signaling pathways were enriched in HVGs identified in LCLs, collagen formation in LAECs, and keratinization in DFs. We repeated the same analysis with scRNA-seq data from induced pluripotent stem cells (iPSCs) and identified only 79 HVGs with no statistically significant enriched functions; the overall scEV in iPSCs was of negligible magnitude. Our results support the “variation is function” hypothesis, arguing that scEV is required for cell type-specific, higher-level system function. Thus, quantifying and characterizing scEV are of importance for our understating of normal and pathological cellular processes.

Gene Expression-Based Identification of Antigen-Responsive CD8+ T Cells on a Single-Cell Level

CD8⁺ T cells are important effectors of adaptive immunity against pathogens, tumors, and self antigens. Here, we asked how human cognate antigen-responsive CD8⁺ T cells and their receptors could be identified in unselected single-cell gene expression data. Single-cell RNA sequencing and qPCR of dye-labeled antigen-specific cells identified large gene sets that were congruently up- or downregulated in virus-responsive CD8⁺ T cells under different antigen presentation conditions. Combined expression of TNFRSF9, XCL1, XCL2, and CRTAM was the most distinct marker of virus-responsive cells on a single-cell level. Using transcriptomic data, we developed a machine learning-based classifier that provides sensitive and specific detection of virus-responsive CD8⁺ T cells from unselected populations. Gene response profiles of CD8⁺ T cells specific for the autoantigen islet-specific glucose-6-phosphatase catalytic subunit-related protein differed markedly from virus-specific cells. These findings provide single-cell gene expression parameters for comprehensive identification of rare antigen-responsive cells and T cell receptors.

Gene Regulatory Network Inference from Single-Cell Data Using Multivariate Information Measures

While single-cell gene expression experiments present new challenges for data processing, the cell-to-cell variability observed also reveals statistical relationships that can be used by information theory. Here, we use multivariate information theory to explore the statistical dependencies between triplets of genes in single-cell gene expression datasets. We develop PIDC, a fast, efficient algorithm that uses partial information decomposition (PID) to identify regulatory relationships between genes. We thoroughly evaluate the performance of our algorithm and demonstrate that the higher-order information captured by PIDC allows it to outperform pairwise mutual information-based algorithms when recovering true relationships present in simulated data. We also infer gene regulatory networks from three experimental single-cell datasets and illustrate how network context, choices made during analysis, and sources of variability affect network inference. PIDC tutorials and open-source software for estimating PID are available. PIDC should facilitate the identification of putative functional relationships and mechanistic hypotheses from single-cell transcriptomic data.

•

PIDC infers gene regulatory networks from single-cell transcriptomic data

•

Multivariate information measures and context in PIDC improve network inference

•

Heterogeneity in single-cell data carries information about gene-gene interactions

•

Fast, efficient, open-source software is made freely available

GrandPrix: scaling up the Bayesian GPLVM for single-cell data

Motivation

The Gaussian Process Latent Variable Model (GPLVM) is a popular approach for dimensionality reduction of single-cell data and has been used for pseudotime estimation with capture time information. However, current implementations are computationally intensive and will not scale up to modern droplet-based single-cell datasets which routinely profile many tens of thousands of cells.

Results

We provide an efficient implementation which allows scaling up this approach to modern single-cell datasets. We also generalize the application of pseudotime inference to cases where there are other sources of variation such as branching dynamics. We apply our method on microarray, nCounter, RNA-seq, qPCR and droplet-based datasets from different organisms. The model converges an order of magnitude faster compared to existing methods whilst achieving similar levels of estimation accuracy. Further, we demonstrate the flexibility of our approach by extending the model to higher-dimensional latent spaces that can be used to simultaneously infer pseudotime and other structure such as branching. Thus, the model has the capability of producing meaningful biological insights about cell ordering as well as cell fate regulation.

Availability and implementation

Software available at github.com/ManchesterBioinference/GrandPrix.

Supplementary information

Supplementary data are available at Bioinformatics online.

Modulation of Myelopoiesis Progenitors Is an Integral Component of Trained Immunity

Trained innate immunity fosters a sustained favorable response of myeloid cells to a secondary challenge, despite their short lifespan in circulation. We thus hypothesized that trained immunity acts via modulation of hematopoietic stem and progenitor cells (HSPCs). Administration of β-glucan (prototypical trained-immunity-inducing agonist) to mice induced expansion of progenitors of the myeloid lineage, which was associated with elevated signaling by innate immune mediators, such as IL-1β and granulocyte-macrophage colony-stimulating factor (GM-CSF), and with adaptations in glucose metabolism and cholesterol biosynthesis. The trained-immunity-related increase in myelopoiesis resulted in a beneficial response to secondary LPS challenge and protection from chemotherapy-induced myelosuppression in mice. Therefore, modulation of myeloid progenitors in the bone marrow is an integral component of trained immunity, which to date, was considered to involve functional changes of mature myeloid cells in the periphery.

•

Trained immunity (TI) modulates hematopoietic progenitors in bone marrow

•

TI is associated with adaptations in cell metabolism in progenitors

•

TI increases expansion of hematopoietic progenitors and myelopoiesis

•

TI promotes beneficial responses to systemic inflammation and chemotherapy

Antigen discovery and specification of immunodominance hierarchies for MHCII-restricted epitopes

Identifying immunodominant T cell epitopes remains a significant challenge in the context of infectious disease, autoimmunity, and immuno-oncology. To address the challenge of antigen discovery, we developed a quantitative proteomic approach that enabled unbiased identification of major histocompatibility complex class II (MHCII)-associated peptide epitopes and biochemical features of antigenicity. On the basis of these data, we trained a deep neural network model for genome-scale predictions of immunodominant MHCII-restricted epitopes. We named this model bacteria originated T cell antigen (BOTA) predictor. In validation studies, BOTA accurately predicted novel CD4 T cell epitopes derived from the model pathogen Listeria monocytogenes and the commensal microorganism Muribaculum intestinale. To conclusively define immunodominant T cell epitopes predicted by BOTA, we developed a high-throughput approach to screen DNA-encoded peptide-MHCII libraries for functional recognition by T cell receptors identified from single-cell RNA sequencing. Collectively, these studies provide a framework for defining the immunodominance landscape across a broad range of immune pathologies.

Single cell molecular alterations reveal target cells and pathways of concussive brain injury

The complex neuropathology of traumatic brain injury (TBI) is difficult to dissect, given the convoluted cytoarchitecture of affected brain regions such as the hippocampus. Hippocampal dysfunction during TBI results in cognitive decline that may escalate to other neurological disorders, the molecular basis of which is hidden in the genomic programs of individual cells. Using the unbiased single cell sequencing method Drop-seq, we report that concussive TBI affects previously undefined cell populations, in addition to classical hippocampal cell types. TBI also impacts cell type-specific genes and pathways and alters gene co-expression across cell types, suggesting hidden pathogenic mechanisms and therapeutic target pathways. Modulating the thyroid hormone pathway as informed by the T4 transporter transthyretin Ttr mitigates TBI-associated genomic and behavioral abnormalities. Thus, single cell genomics provides unique information about how TBI impacts diverse hippocampal cell types, adding new insights into the pathogenic pathways amenable to therapeutics in TBI and related disorders.

Single-cell Quantitation of mRNA and Surface Protein Expression in Simian Immunodeficiency Virus-infected CD4+ T Cells Isolated from Rhesus macaques

Single-cell analysis is an important tool for dissecting heterogeneous populations of cells. The identification and isolation of rare cells can be difficult. To overcome this challenge, a methodology combining indexed flow cytometry and high-throughput multiplexed quantitative polymerase chain reaction (qPCR) was developed. The objective was to identify and characterize simian immunodeficiency virus (SIV)-infected cells present within rhesus macaques. Through quantitation of surface protein by fluorescence-activated cell sorting (FACS) and mRNA by qPCR, virus-infected cells are identified by viral gene expression, which is combined with host gene and protein measurements to create a multidimensional profile. We term the approach, targeted Single-Cell Proteo-transcriptional Evaluation, or tSCEPTRE. To perform the method, viable cells are stained with fluorescent antibodies specific for surface markers used for FACS isolation of a cell subset and/or downstream phenotypic analysis. Single cells are sorted followed by immediate lysis, multiplex reverse transcription (RT), PCR pre-amplification, and high throughput qPCR of up to 96 transcripts. FACS measurements are recorded at the time of sorting and subsequently linked to the gene expression data by well position to create a combined protein and transcriptional profile. To study SIV-infected cells directly ex vivo, cells were identified by qPCR detection of multiple viral RNA species. The combination of viral transcripts and the quantity of each provide a framework for classifying cells into distinct stages of the viral life cycle (e.g., productive versus non-productive). Moreover, tSCEPTRE of SIV⁺ cells were compared to uninfected cells isolated from the same specimen to assess differentially expressed host genes and proteins. The analysis revealed previously unappreciated viral RNA expression heterogeneity among infected cells as well as in vivo SIV-mediated post-transcriptional gene regulation with single-cell resolution. The tSCEPTRE method is relevant for the analysis of any cell population amenable to identification by expression of surface protein marker(s), host or pathogen gene(s), or combinations thereof.

Hematopoietic stem cells can differentiate into restricted myeloid progenitors before cell division in mice

Hematopoietic stem cells (HSCs) continuously replenish all blood cell types through a series of differentiation steps and repeated cell divisions that involve the generation of lineage-committed progenitors. However, whether cell division in HSCs precedes differentiation is unclear. To this end, we used an HSC cell-tracing approach and Ki67^RFP knock-in mice, in a non-conditioned transplantation model, to assess divisional history, cell cycle progression, and differentiation of adult HSCs. Our results reveal that HSCs are able to differentiate into restricted progenitors, especially common myeloid, megakaryocyte-erythroid and pre-megakaryocyte progenitors, without undergoing cell division and even before entering the S phase of the cell cycle. Additionally, the phenotype of the undivided but differentiated progenitors correlated with the expression of lineage-specific genes and loss of multipotency. Thus HSC fate decisions can be uncoupled from physical cell division. These results facilitate a better understanding of the mechanisms that control fate decisions in hematopoietic cells.

Dependence of hematopoietic stem cell (HSC) fate on the phase of the cell cycle has not been demonstrated in vivo. Here, the authors find that HSCs can differentiate into a downstream progenitor without physical division, even before progressing into the S phase of the cell cycle.

Automated cell cycle and cell size measurements for single-cell gene expression studies

OBJECTIVES

Recent rise of single-cell studies revealed the importance of understanding the role of cell-to-cell variability, especially at the transcriptomic level. One of the numerous sources of cell-to-cell variation in gene expression is the heterogeneity in cell proliferation state. In order to identify how cell cycle and cell size influences gene expression variability at the single-cell level, we provide an universal and automatic toxic-free label method, compatible with single-cell high-throughput RT-qPCR. The method consists of isolating cells after a double-stained, analyzing their morphological parameters and performing a transcriptomic analysis on the same identified cells.

RESULTS

This led to an unbiased gene expression analysis and could be also used for improving single-cell tracking and imaging when combined with cell isolation. As an application for this technique, we showed that cell-to-cell variability in chicken erythroid progenitors was negligibly influenced by cell size nor cell cycle.

GM-CSF producing autoreactive CD4+ T cells in type 1 diabetes

The phenotype of autoreactive T cells in type 1 diabetes is described as Th1, Th17 and/or Th21, but is largely uncharacterized. We combined multi-parameter cytokine profiling and proliferation, and identified GM-CSF producing cells as a component of the response to beta cell autoantigens proinsulin and GAD65. Overall cytokine profiles of CD4⁺ T cell were not altered in type 1 diabetes. In contrast, patients with recent onset type 1 diabetes had increased frequencies of proinsulin-responsive CD4⁺CD45RA^- T cells producing GM-CSF (p=0.002), IFNγ (p=0.004), IL-17A (p=0.008), IL-21 (p=0.011), and IL-22 (p=0.007), and GAD65-responsive CD4⁺CD45RA^- T cells producing IL-21 (p=0.039). CD4⁺ T cells with a GM-CSF⁺IFNγ^-IL-17A^-IL-21^-IL-22^- phenotype were increased in patients for responses to both proinsulin (p=0.006) and GAD65 (p=0.037). GM-CSF producing T cells are a novel phenotype in the repertoire of T helper cells in type 1 diabetes and consolidate a Th1/Th17 pro-inflammatory pathogenesis in the disease.

Combined single-cell quantitation of host and SIV genes and proteins ex vivo reveals host-pathogen interactions in individual cells

CD4 T cells harboring HIV-1/SIV represent a formidable hurdle to eradicating infection, and yet their detailed phenotype remains unknown. Here we integrate two single-cell technologies, flow cytometry and highly multiplexed quantitative RT-PCR, to characterize SIV-infected CD4 T cells directly ex vivo. Within individual cells, we correlate the cellular phenotype, in terms of host protein and RNA expression, with stages of the viral life cycle defined by combinatorial expression of viral RNAs. Spliced RNA⁺ infected cells display multiple memory and activation phenotypes, indicating virus production by diverse CD4 T cell subsets. In most (but not all) cells, progressive infection accompanies post-transcriptional downregulation of CD4 protein, while surface MHC class I is largely retained. Interferon-stimulated genes were also commonly upregulated. Thus, we demonstrate that combined quantitation of transcriptional and post-transcriptional regulation at the single-cell level informs in vivo mechanisms of viral replication and immune evasion.

HIV-1, and its simian counterpart, SIV, infect and kill CD4 T cells, resulting in their massive depletion that ultimately leads to AIDS in the absence of antiretroviral therapy. With effective therapy, these cells are largely preserved, but a subset harbors latent virus that can persist for decades and reemerge upon therapy interruption, preventing HIV-1 cure. To prevent or eliminate productive cellular infection, there is tremendous demand to identify host factors expressed by these cells in vivo, which may serve as unique biomarkers or drug targets. Here we provide the first detailed combined transcriptomic and protein expression profile of SIV-infected cells directly ex vivo using novel single-cell technologies. Our survey of activation markers, interferon-stimulated genes, and viral restriction factors identified multiple host genes differentially expressed by SIV-infected cells and will inform future therapeutic strategies.

Reconstructing cell cycle pseudo time-series via single-cell transcriptome data

Single-cell mRNA sequencing, which permits whole transcriptional profiling of individual cells, has been widely applied to study growth and development of tissues and tumors. Resolving cell cycle for such groups of cells is significant, but may not be adequately achieved by commonly used approaches. Here we develop a traveling salesman problem and hidden Markov model-based computational method named reCAT, to recover cell cycle along time for unsynchronized single-cell transcriptome data. We independently test reCAT for accuracy and reliability using several data sets. We find that cell cycle genes cluster into two major waves of expression, which correspond to the two well-known checkpoints, G1 and G2. Moreover, we leverage reCAT to exhibit methylation variation along the recovered cell cycle. Thus, reCAT shows the potential to elucidate diverse profiles of cell cycle, as well as other cyclic or circadian processes (e.g., in liver), on single-cell resolution.In single-cell RNA sequencing data of heterogeneous cell populations, cell cycle stage of individual cells would often be informative. Here, the authors introduce a computational model to reconstruct a pseudo-time series from single cell transcriptome data, identify the cell cycle stages, identify candidate cell cycle-regulated genes and recover the methylome changes during the cell cycle.

Reconstructing cell cycle pseudo time-series via single-cell transcriptome data

Single-cell mRNA sequencing, which permits whole transcriptional profiling of individual cells, has been widely applied to study growth and development of tissues and tumors. Resolving cell cycle for such groups of cells is significant, but may not be adequately achieved by commonly used approaches. Here we develop a traveling salesman problem and hidden Markov model-based computational method named reCAT, to recover cell cycle along time for unsynchronized single-cell transcriptome data. We independently test reCAT for accuracy and reliability using several data sets. We find that cell cycle genes cluster into two major waves of expression, which correspond to the two well-known checkpoints, G1 and G2. Moreover, we leverage reCAT to exhibit methylation variation along the recovered cell cycle. Thus, reCAT shows the potential to elucidate diverse profiles of cell cycle, as well as other cyclic or circadian processes (e.g., in liver), on single-cell resolution.

In single-cell RNA sequencing data of heterogeneous cell populations, cell cycle stage of individual cells would often be informative. Here, the authors introduce a computational model to reconstruct a pseudo-time series from single cell transcriptome data, identify the cell cycle stages, identify candidate cell cycle-regulated genes and recover the methylome changes during the cell cycle.

Revealing the vectors of cellular identity with single-cell genomics

Single-cell genomics has now made it possible to create a comprehensive atlas of human cells. At the same time, it has reopened definitions of a cell's identity and of the ways in which identity is regulated by the cell's molecular circuitry. Emerging computational analysis methods, especially in single-cell RNA sequencing (scRNA-seq), have already begun to reveal, in a data-driven way, the diverse simultaneous facets of a cell's identity, from discrete cell types to continuous dynamic transitions and spatial locations. These developments will eventually allow a cell to be represented as a superposition of 'basis vectors', each determining a different (but possibly dependent) aspect of cellular organization and function. However, computational methods must also overcome considerable challenges-from handling technical noise and data scale to forming new abstractions of biology. As the scale of single-cell experiments continues to increase, new computational approaches will be essential for constructing and characterizing a reference map of cell identities.

Single-Cell RNA Sequencing Reveals Expanded Clones of Islet Antigen-Reactive CD4+ T Cells in Peripheral Blood of Subjects with Type 1 Diabetes

The significance of islet Ag-reactive T cells found in peripheral blood of type 1 diabetes (T1D) subjects is unclear, partly because similar cells are also found in healthy control (HC) subjects. We hypothesized that key disease-associated cells would show evidence of prior Ag exposure, inferred from expanded TCR clonotypes, and essential phenotypic properties in their transcriptomes. To test this, we developed single-cell RNA sequencing procedures for identifying TCR clonotypes and transcript phenotypes in individual T cells. We applied these procedures to analysis of islet Ag-reactive CD4⁺ memory T cells from the blood of T1D and HC individuals after activation with pooled immunodominant islet peptides. We found extensive TCR clonotype sharing in Ag-activated cells, especially from individual T1D subjects, consistent with in vivo T cell expansion during disease progression. The expanded clonotype from one T1D subject was detected at repeat visits spanning >15 mo, demonstrating clonotype stability. Notably, we found no clonotype sharing between subjects, indicating a predominance of "private" TCR specificities. Expanded clones from two T1D subjects recognized distinct IGRP peptides, implicating this molecule as a trigger for CD4⁺ T cell expansion. Although overall transcript profiles of cells from HC and T1D subjects were similar, profiles from the most expanded clones were distinctive. Our findings demonstrate that islet Ag-reactive CD4⁺ memory T cells with unique Ag specificities and phenotypes are expanded during disease progression and can be detected by single-cell analysis of peripheral blood.

A novel method for quantitative measurements of gene expression in single living cells

Gene expression is at the heart of virtually any biological process, and its deregulation is at the source of numerous pathological conditions. While impressive progress has been made in genome-wide measurements of mRNA and protein expression levels, it is still challenging to obtain highly quantitative measurements in single living cells. Here we describe a novel approach based on internal tagging of endogenous proteins with a reporter allowing luminescence and fluorescence time-lapse microscopy. Using luminescence microscopy, fluctuations of protein expression levels can be monitored in single living cells with high sensitivity and temporal resolution over extended time periods. The integrated protein decay reporter allows measuring protein degradation rates in the absence of protein synthesis inhibitors, and in combination with absolute protein levels allows determining absolute amounts of proteins synthesized over the cell cycle. Finally, the internal tag can be excised by inducible expression of Cre recombinase, which enables to estimate endogenous mRNA half-lives. Our method thus opens new avenues in quantitative analysis of gene expression in single living cells.

CIDR: Ultrafast and accurate clustering through imputation for single-cell RNA-seq data

Most existing dimensionality reduction and clustering packages for single-cell RNA-seq (scRNA-seq) data deal with dropouts by heavy modeling and computational machinery. Here, we introduce CIDR (Clustering through Imputation and Dimensionality Reduction), an ultrafast algorithm that uses a novel yet very simple implicit imputation approach to alleviate the impact of dropouts in scRNA-seq data in a principled manner. Using a range of simulated and real data, we show that CIDR improves the standard principal component analysis and outperforms the state-of-the-art methods, namely t-SNE, ZIFA, and RaceID, in terms of clustering accuracy. CIDR typically completes within seconds when processing a data set of hundreds of cells and minutes for a data set of thousands of cells. CIDR can be downloaded at https://github.com/VCCRI/CIDR.

Integrated single cell data analysis reveals cell specific networks and novel coactivation markers

BACKGROUND

Large scale single cell transcriptome profiling has exploded in recent years and has enabled unprecedented insight into the behavior of individual cells. Identifying genes with high levels of expression using data from single cell RNA sequencing can be useful to characterize very active genes and cells in which this occurs. In particular single cell RNA-Seq allows for cell-specific characterization of high gene expression, as well as gene coexpression.

RESULTS

We offer a versatile modeling framework to identify transcriptional states as well as structures of coactivation for different neuronal cell types across multiple datasets. We employed a gamma-normal mixture model to identify active gene expression across cells, and used these to characterize markers for olfactory sensory neuron cell maturity, and to build cell-specific coactivation networks. We found that combined analysis of multiple datasets results in more known maturity markers being identified, as well as pointing towards some novel genes that may be involved in neuronal maturation. We also observed that the cell-specific coactivation networks of mature neurons tended to have a higher centralization network measure than immature neurons.

CONCLUSION

Integration of multiple datasets promises to bring about more statistical power to identify genes and patterns of interest. We found that transforming the data into active and inactive gene states allowed for more direct comparison of datasets, leading to identification of maturity marker genes and cell-specific network observations, taking into account the unique characteristics of single cell transcriptomics data.

TSPAN8 Expression Distinguishes Spermatogonial Stem Cells in the Prepubertal Mouse Testis

Precise separation of spermatogonial stem cells (SSCs) from progenitor spermatogonia that lack stem cell activity and are committed to differentiation remains a challenge. To distinguish between these spermatogonial subtypes, we identified genes that exhibited bimodal mRNA levels at the single-cell level among undifferentiated spermatogonia from Postnatal Day 6 mouse testes, including Tspan8, Epha2, and Pvr, each of which encode cell surface proteins useful for cell selection. Transplantation studies provided definitive evidence that a TSPAN8-high subpopulation is enriched for SSCs. RNA-seq analyses identified genes differentially expressed between TSPAN8-high and -low subpopulations that clustered into multiple biological pathways potentially involved in SSC renewal or differentiation, respectively. Methyl-seq analysis identified hypomethylated domains in the promoters of these genes in both subpopulations that colocalized with peaks of histone modifications defined by ChIP-seq analysis. Taken together, these results demonstrate functional heterogeneity among mouse undifferentiated spermatogonia and point to key biological characteristics that distinguish SSCs from progenitor spermatogonia.

Pseudotime estimation: deconfounding single cell time series

MOTIVATION

Repeated cross-sectional time series single cell data confound several sources of variation, with contributions from measurement noise, stochastic cell-to-cell variation and cell progression at different rates. Time series from single cell assays are particularly susceptible to confounding as the measurements are not averaged over populations of cells. When several genes are assayed in parallel these effects can be estimated and corrected for under certain smoothness assumptions on cell progression.

RESULTS

We present a principled probabilistic model with a Bayesian inference scheme to analyse such data. We demonstrate our method's utility on public microarray, nCounter and RNA-seq datasets from three organisms. Our method almost perfectly recovers withheld capture times in an Arabidopsis dataset, it accurately estimates cell cycle peak times in a human prostate cancer cell line and it correctly identifies two precocious cells in a study of paracrine signalling in mouse dendritic cells. Furthermore, our method compares favourably with Monocle, a state-of-the-art technique. We also show using held-out data that uncertainty in the temporal dimension is a common confounder and should be accounted for in analyses of repeated cross-sectional time series.

AVAILABILITY AND IMPLEMENTATION

Our method is available on CRAN in the DeLorean package.

CONTACT

john.reid@mrc-bsu.cam.ac.uk

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Gene Expression Analysis of Neurons and Astrocytes Isolated by Laser Capture Microdissection from Frozen Human Brain Tissues

Different cell types and multiple cellular connections characterize the human brain. Gene expression analysis using a specific population of cells is more accurate than conducting analysis of the whole tissue homogenate, particularly in the context of neurodegenerative diseases, where a specific subset of cells is affected by the different pathology. Due to the difficulty of obtaining homogenous cell populations, gene expression in specific cell-types (neurons, astrocytes, etc.) has been understudied. To leverage the use of archive resources of frozen human brains in studies of neurodegenerative diseases, we developed and calibrated a method to quantify cell-type specific—neuronal, astrocytes—expression profiles of genes implicated in neurodegenerative diseases, including Parkinson's and Alzheimer's diseases. Archive human frozen brain tissues were used to prepare slides for rapid immunostaining using cell-specific antibodies. The immunoreactive-cells were isolated by Laser Capture Microdissection (LCM). The enrichment for a particular cell-type of interest was validated in post-analysis stage by the expression of cell-specific markers. We optimized the technique to preserve the RNA integrity, so that the RNA was suitable for downstream expression analyses. Following RNA extraction, the expression levels were determined digitally using nCounter Single Cell Gene Expression assay (NanoString Technologies®). The results demonstrated that using our optimized technique we successfully isolated single neurons and astrocytes from human frozen brain tissues and obtained RNA of a good quality that was suitable for mRNA expression analysis. We present here new advancements compared to previous reported methods, which improve the method's feasibility and its applicability for a variety of downstream molecular analyses. Our new developed method can be implemented in genetic and functional genomic research of neurodegenerative diseases and has the potential to significantly advance the field.

Robust Inference of Cell-to-Cell Expression Variations from Single- and K-Cell Profiling

Quantifying heterogeneity in gene expression among single cells can reveal information inaccessible to cell-population averaged measurements. However, the expression level of many genes in single cells fall below the detection limit of even the most sensitive technologies currently available. One proposed approach to overcome this challenge is to measure random pools of k cells (e.g., 10) to increase sensitivity, followed by computational “deconvolution” of cellular heterogeneity parameters (CHPs), such as the biological variance of single-cell expression levels. Existing approaches infer CHPs using either single-cell or k-cell data alone, and typically within a single population of cells. However, integrating both single- and k-cell data may reap additional benefits, and quantifying differences in CHPs across cell populations or conditions could reveal novel biological information. Here we present a Bayesian approach that can utilize single-cell, k-cell, or both simultaneously to infer CHPs within a single condition or their differences across two conditions. Using simulated as well as experimentally generated single- and k-cell data, we found situations where each data type would offer advantages, but using both together can improve precision and better reconcile CHP information contained in single- and k-cell data. We illustrate the utility of our approach by applying it to jointly generated single- and k-cell data to reveal CHP differences in several key inflammatory genes between resting and inflammatory cytokine-activated human macrophages, delineating differences in the distribution of ‘ON’ versus ‘OFF’ cells and in continuous variation of expression level among cells. Our approach thus offers a practical and robust framework to assess and compare cellular heterogeneity within and across biological conditions using modern multiplexed technologies.

Different cells can make different amounts of biomolecules such as RNA transcripts of genes. New technologies are emerging to measure the transcript level of many genes in single cells. However, accurate quantification of the biological variation from cell to cell can be challenging due to the low transcript level of many genes and the presence of substantial measurement noise. Here we present a flexible, novel computational approach to quantify biological cell-to-cell variation that can use different types of data, namely measurements directly obtained from single cells, and/or those from random pools of k-cells (e.g., k = 10). Assessment of these different inputs using simulated and real data revealed that each data type can offer advantages under different scenarios, but combining both single- and k-cell measurements tend to offer the best of both. Application of our approach to single- and k-cell data obtained from resting and inflammatory macrophages, an important type of immune cells implicated in diverse diseases, revealed interesting changes in cell-to-cell variation in transcript levels upon inflammatory stimulation, thus suggesting that inflammation can shape not only the average expression level of a gene but also the gene’s degree of expression variation among single cells.

Cellular Deconstruction: Finding Meaning in Individual Cell Variation

The advent of single cell transcriptome analysis has permitted the discovery of cell-to-cell variation in transcriptome expression of even presumptively identical cells. We hypothesize that this variability reflects a many-to-one relation between transcriptome states and the phenotype of a cell. In this relation, the molecular ratios of the subsets of RNA are determined by the stoichiometric constraints of the cell systems, which underdetermine the transcriptome state. Furthermore, the variability is, in part, induced by the tissue context and is important for system-level function. This theory is analogous to theories of literary deconstruction, where multiple 'signifiers' work in opposition to one another to create meaning. By analogy, transcriptome phenotypes should be defined as subsets of RNAs comprising selected RNA systems where the system-associated RNAs are balanced with each other to produce the associated cellular function. This idea provides a framework for understanding cellular heterogeneity in phenotypic responses to variant conditions, such as disease challenge.

Distinct activation thresholds of human conventional and innate-like memory T cells

Conventional memory CD8⁺ T cells and mucosal-associated invariant T cells (MAIT cells) are found in blood, liver, and mucosal tissues and have similar effector potential following activation, specifically expression of IFN-γ and granzyme B. To better understand each subset's unique contributions to immunity and pathology, we interrogated inflammation- and TCR-driven activation requirements using human memory CD8⁺ T and MAIT cells isolated from blood and mucosal tissue biopsies in ex vivo functional assays and single cell gene expression experiments. We found that MAIT cells had a robust IFN-γ and granzyme B response to inflammatory signals but limited responsiveness when stimulated directly via their TCR. Importantly, this is not due to an overall hyporesponsiveness to TCR signals. When delivered together, TCR and inflammatory signals synergize to elicit potent effector function in MAIT cells. This unique control of effector function allows MAIT cells to respond to the same TCR signal in a dichotomous and situation-specific manner. We propose that this could serve to prevent responses to antigen in noninflamed healthy mucosal tissue, while maintaining responsiveness and great sensitivity to inflammation-eliciting infections. We discuss the implications of these findings in context of inflammation-inducing damage to tissues such as BM transplant conditioning or HIV infection.

Single-Cell Analysis in Cancer Genomics

Genetic changes and environmental differences result in cellular heterogeneity among cancer cells within the same tumor, thereby complicating treatment outcomes. Recent advances in single-cell technologies have opened new avenues to characterize the intra-tumor cellular heterogeneity, identify rare cell types, measure mutation rates, and, ultimately, guide diagnosis and treatment. In this paper we review the recent single-cell technological and computational advances at the genomic, transcriptomic, and proteomic levels, and discuss their applications in cancer research.

A Double-Hybridization Approach for the Transcription- and Amplification-Free Detection of Specific mRNA on a Microarray

A double-hybridization approach was developed for the enzyme-free detection of specific mRNA of a housekeeping gene. Targeted mRNA was immobilized by hybridization to complementary DNA capture probes spotted onto a microarray. A second hybridization step of Cy5-conjugated label DNA to another section of the mRNA enabled specific labeling of the target. Thus, enzymatic artifacts could be avoided by omitting transcription and amplification steps. This manuscript describes the development of capture probe molecules used in the transcription- and amplification-free analysis of RPLP0 mRNA in isolated total RNA. An increase in specific signal was found with increasing length of the target-specific section of capture probes. Unspecific signal comprising spot autofluorescence and unspecific label binding did not correlate with the capture length. An additional spacer between the specific part of the capture probe and the substrate attachment site increased the signal significantly only on a short capture probe of approximately 30 nt length.

Promises and Pitfalls of High-Throughput Biological Assays

This chapter discusses some of the pitfalls encountered when performing biomedical research involving high-throughput "omics" data and presents some strategies and guidelines that researchers should follow when undertaking such studies. We discuss common errors in experimental design and data analysis that lead to irreproducible and non-replicable research and provide some guidelines to avoid these common mistakes so that researchers may have confidence in study outcomes, even if the results are negative. We discuss the importance of ranking and prespecifying hypotheses, performing power analysis, careful experimental design, and preplanning of statistical analyses in order to avoid the "fishing expedition" data analysis strategy, which is doomed to fail. The impact of multiple testing on false-positive rates is discussed, particularly in the context of the analysis of high-throughput data, and methods to correct for it are presented, as well as approaches to detect and correct for experimental biases and batch effects, which often plague high-throughput assays. We highlight the importance of sharing data and analysis code to facilitate reproducibility and present tools and software that are appropriate for this purpose.

Mitochondria and the non-genetic origins of cell-to-cell variability: More is different

Gene expression activity is heterogeneous in a population of isogenic cells. Identifying the molecular basis of this variability will improve our understanding of phenomena like tumor resistance to drugs, virus infection, or cell fate choice. The complexity of the molecular steps and machines involved in transcription and translation could introduce sources of randomness at many levels, but a common constraint to most of these processes is its energy dependence. In eukaryotic cells, most of this energy is provided by mitochondria. A clonal population of cells may show a large variability in the number and functionality of mitochondria. Here, we discuss how differences in the mitochondrial content of each cell contribute to heterogeneity in gene products. Changes in the amount of mitochondria can also entail drastic alterations of a cell's gene expression program, which ultimately leads to phenotypic diversity. Also watch the Video Abstract.

Variation is function: Are single cell differences functionally important?: Testing the hypothesis that single cell variation is required for aggregate function

There is a growing appreciation of the extent of transcriptome variation across individual cells of the same cell type. While expression variation may be a byproduct of, for example, dynamic or homeostatic processes, here we consider whether single-cell molecular variation per se might be crucial for population-level function. Under this hypothesis, molecular variation indicates a diversity of hidden functional capacities within an ensemble of identical cells, and this functional diversity facilitates collective behavior that would be inaccessible to a homogenous population. In reviewing this topic, we explore possible functions that might be carried by a heterogeneous ensemble of cells; however, this question has proven difficult to test, both because methods to manipulate molecular variation are limited and because it is complicated to define, and measure, population-level function. We consider several possible methods to further pursue the hypothesis that variation is function through the use of comparative analysis and novel experimental techniques.

Full-length single-cell RNA-seq applied to a viral human cancer: applications to HPV expression and splicing analysis in HeLa S3 cells

Background

Viral infection causes multiple forms of human cancer, and HPV infection is the primary factor in cervical carcinomas. Recent single-cell RNA-seq studies highlight the tumor heterogeneity present in most cancers, but virally induced tumors have not been studied. HeLa is a well characterized HPV+ cervical cancer cell line.

Result

We developed a new high throughput platform to prepare single-cell RNA on a nanoliter scale based on a customized microwell chip. Using this method, we successfully amplified full-length transcripts of 669 single HeLa S3 cells and 40 of them were randomly selected to perform single-cell RNA sequencing. Based on these data, we obtained a comprehensive understanding of the heterogeneity of HeLa S3 cells in gene expression, alternative splicing and fusions. Furthermore, we identified a high diversity of HPV-18 expression and splicing at the single-cell level. By co-expression analysis we identified 283 E6, E7 co-regulated genes, including CDC25, PCNA, PLK4, BUB1B and IRF1 known to interact with HPV viral proteins.

Conclusion

Our results reveal the heterogeneity of a virus-infected cell line. It not only provides a transcriptome characterization of HeLa S3 cells at the single cell level, but is a demonstration of the power of single cell RNA-seq analysis of virally infected cells and cancers.

Electronic supplementary material

The online version of this article (doi:10.1186/s13742-015-0091-4) contains supplementary material, which is available to authorized users.

Geometry of the Gene Expression Space of Individual Cells

There is a revolution in the ability to analyze gene expression of single cells in a tissue. To understand this data we must comprehend how cells are distributed in a high-dimensional gene expression space. One open question is whether cell types form discrete clusters or whether gene expression forms a continuum of states. If such a continuum exists, what is its geometry? Recent theory on evolutionary trade-offs suggests that cells that need to perform multiple tasks are arranged in a polygon or polyhedron (line, triangle, tetrahedron and so on, generally called polytopes) in gene expression space, whose vertices are the expression profiles optimal for each task. Here, we analyze single-cell data from human and mouse tissues profiled using a variety of single-cell technologies. We fit the data to shapes with different numbers of vertices, compute their statistical significance, and infer their tasks. We find cases in which single cells fill out a continuum of expression states within a polyhedron. This occurs in intestinal progenitor cells, which fill out a tetrahedron in gene expression space. The four vertices of this tetrahedron are each enriched with genes for a specific task related to stemness and early differentiation. A polyhedral continuum of states is also found in spleen dendritic cells, known to perform multiple immune tasks: cells fill out a tetrahedron whose vertices correspond to key tasks related to maturation, pathogen sensing and communication with lymphocytes. A mixture of continuum-like distributions and discrete clusters is found in other cell types, including bone marrow and differentiated intestinal crypt cells. This approach can be used to understand the geometry and biological tasks of a wide range of single-cell datasets. The present results suggest that the concept of cell type may be expanded. In addition to discreet clusters in gene-expression space, we suggest a new possibility: a continuum of states within a polyhedron, in which the vertices represent specialists at key tasks.

In the past, biological experiments usually pooled together millions of cells, masking the differences between individual cells. Current technology takes a big step forward by measuring gene expression from individual cells. Interpreting this data is challenging because we need to understand how cells are arranged in a high dimensional gene expression space. Here we test recent theory that suggests that cells facing multiple tasks should be arranged in simple low dimensional polygons or polyhedra (generally called polytopes). The vertices of the polytopes are gene expression profiles optimal for each of the tasks. We find evidence for such simplicity in a variety of tissues—spleen, bone marrow, intestine—analyzed by different single-cell technologies. We find that cells are distributed inside polytopes, such as tetrahedrons or four-dimensional simplexes, with cells closest to each vertex responsible for a different key task. For example, intestinal progenitor cells that give rise to the other cell types show a continuous distribution in a tetrahedron whose vertices correspond to several key sub-tasks. Immune dendritic cells likewise are continuously distributed between key immune tasks. This approach of testing whether data falls in polytopes may be useful for interpreting a variety of single-cell datasets in terms of biological tasks.

Transcriptional and translational heterogeneity among neonatal mouse spermatogonia

Spermatogonial stem cells (SSCs) are a subset of undifferentiated spermatogonia responsible for ongoing spermatogenesis in mammalian testes. Spermatogonial stem cells arise from morphologically homogeneous prospermatogonia, but growing evidence suggests that only a subset of prospermatogonia develops into the foundational SSC pool. This predicts that subtypes of undifferentiated spermatogonia with discrete mRNA and protein signatures should be distinguishable in neonatal testes. We used single-cell quantitative RT-PCR to examine mRNA levels of 172 genes in individual spermatogonia from 6-day postnatal (P6) mouse testes. Cells enriched from P6 testes using the StaPut or THY1(+) magnetic cell sorting methods exhibited considerable heterogeneity in the abundance of specific germ cell and stem cell mRNAs, segregating into one somatic and three distinct spermatogonial clusters. However, P6 Id4-eGFP(+) transgenic spermatogonia, which are known to be enriched for SSCs, were more homogeneous in their mRNA levels, exhibiting uniform levels for the majority of genes examined (122 of 172). Interestingly, these cells displayed nonuniform (50 of 172) expression of a smaller cohort of these genes, suggesting there is substantial heterogeneity even within the Id4-eGFP(+) population. Further, although immunofluorescence staining largely demonstrated conformity between mRNA and protein levels, some proteins were observed in patterns that were disparate from those detected for the corresponding mRNAs in Id4-eGFP(+) spermatogonia (e.g., Kit, Sohlh2, Stra8), suggesting additional heterogeneity is introduced at the posttranscriptional level. Taken together, these data demonstrate the existence of multiple spermatogonial subtypes in P6 mouse testes and raise the intriguing possibility that these subpopulations may correlate with the development of functionally distinct spermatogenic cell types.