Single-Cell RNA Sequencing and Its Applications in Pituitary Research

The intricacy of the human body is epitomized by its composition of approximately 30–40 trillion cells, encompassing over 200 distinct cell types [1, 2]. This cellular diversity underpins critical biological processes, including development, regeneration, and aging, by affecting intercellular signaling, metabolism, and cell division [3]. In the context of the pituitary gland – an endocrine beacon orchestrating a plethora of physiological processes – cellular heterogeneity assumes a paramount role. The pituitary gland’s multifaceted cell types contribute to its complex function, development, and pathophysiology [4]. Traditional research methodologies, while foundational, often fall short in dissecting this cellular complexity due to their inherent limitations in resolving the nuances of cell-to-cell interactions and gene expression changes.

The last decade has witnessed remarkable technological advancements, notably single-cell RNA sequencing (scRNA-seq), which has revolutionized our understanding of cellular diversity [5]. By enabling the detailed analysis of individual cells, scRNA-seq has emerged as a powerful tool for unveiling previously unrecognized cell subtypes, delineating intricate cellular interactions, and understanding the molecular mechanisms underpinning cellular functions [6]. However, the application of these technologies to endocrine organs, especially the pituitary gland, remains nascent.

This review aims to illuminate the pivotal role of scRNA-seq in advancing pituitary gland research. By unraveling the gland’s cellular landscape and molecular dynamics, scRNA-seq offers novel insights into the normal functioning of the pituitary, as well as the genesis and progression of pituitary disorders. It facilitates a granular understanding of cellular heterogeneity, which is crucial for the identification of therapeutic targets and the development of personalized medicine approaches. Moreover, in delving into the applications of scRNA-seq in pituitary research, this review not only highlights the current state of knowledge but also sets the stage for future investigations that could redefine endocrine research. By offering a deeper understanding of the pituitary gland’s cellular diversity and its implications for health and disease, scRNA-seq paves the way for innovative therapeutic strategies and enhances our ability to tackle complex pituitary disorders.

Before scRNA-seq, identifying cell surface markers without prior genetic or protein knowledge was challenging, often relying on methods like FACS or cytometry [7]. scRNA-seq offers an alternative which is ideal for analyzing cellular transcriptomes and cell-type compositions, even with limited samples [8]. Standard scRNA-seq workflow includes sample preparation, cell isolation, reverse transcription and amplification, sequencing, and bioinformatic analysis.

Single-cell isolation in scRNA-seq requires precise techniques [9]. Options vary based on experimental needs and include serial dilution, micropipette aspiration, FACS, LCM, and microfluidic systems [10, 11]. FACS sorts cells efficiently but can cause damage, while LCM preserves spatial data but is technically demanding [12, 13]. Microfluidic methods like Fluidigm C1 offer high throughput and automation [14]. Each method’s selection depends on specific experimental requirements, balancing precision, efficiency, and cost.

CDNA synthesis integrates unique molecular identifiers (UMIs), which are random sequences that act as barcodes to track transcripts [15]. Sequencing adapters are added for platform compatibility. PCR amplification, crucial for library construction, uses techniques like template switching and in vitro transcription to enhance the sequencing process [16].

Researchers have developed over 15 distinct scRNA-seq methodologies (shown in Fig. 1; Table 1), categorized into full-length and tag-based approaches. Full-length methods cover the entire gene body to facilitate gene expression analysis and detection of SNPs, splice variants, and mutations [17, 18]. Tag-based methods focus on sequencing the 5′ or 3′ ends, trading full-length coverage for improved transcript abundance estimation and enhanced cDNA synthesis multiplexing [19, 20]. Tang et al. [21, 22] initiated scRNA-seq innovation by isolating single cells for cDNA synthesis using oligo-dT primers with anchor sequences for PCR amplification, followed by enzymatic polyA tailing. Smart-seq by Ramsköld et al. [18] in 2012 and its iterations, Smart-seq2 and Smart-seq3, have advanced tumor cell research by identifying variable splicing and allele-specific expression, with Smart-seq3 integrating full-length coverage with a UMI strategy [23, 24]. Meanwhile, Quartz-seq simplifies the protocol in a single PCR tube without purification, enhancing sensitivity and reproducibility [25]. CEL-seq uses linear in vitro transcription with a barcode and T7 promoter for RNA amplification, with CEL-seq2 introducing UMIs to reduce costs and operational time [26]. The progression from CEL-seq to MARS-seq emphasizes automation with cell barcodes and UMIs for multiplexing, enhancing throughput and cost efficiency [27]. STRT-seq has adapted for use with microfluidic platforms, with STRT/C1 incorporating UMI tags for precision in library enrichment [28, 29]. Microfluidic methods like CytoSeq and BD Rhapsody, and droplet-based techniques like Drop-seq and inDrop, enable high-throughput cell sequencing with minimal reagent use and reduced costs [20, 30, 31]. Each scRNA-seq method offers unique benefits, with Smart-seq2 noted for detecting the highest number of genes per cell and microfluidic methods reducing amplification noise with UMIs [19, 20].

The analysis of scRNA-seq data is broadly segmented into two principal phases: preprocessing and downstream analysis [32].

Quality control invariably marks the commencement of scRNA-seq data analysis [33, 34]. This critical step involves the removal of data from inferior-quality cells to minimize their adverse effects on the results of downstream analysis. A common practice involves evaluating the proportion of mitochondrial reads; a high prevalence of mitochondrial transcripts indicates cellular distress, leading to the exclusion of cells exceeding a specific mitochondrial transcript threshold [34, 35]. Similarly, cells with excessive ribosomal reads are also discarded despite ribosomal RNA depletion efforts, aligning with scRNA-seq’s focus on functional mRNA [36, 37]. “scRNA Batch QC” method aids in cross-dataset quality assessments to identify biases and outliers. Batch effects are addressed using tools like Seurat3, Liger, and Harmony, with Harmony noted for computational efficiency [38].

Normalization is crucial for comparing gene expression within or among samples [39]. While traditional RNA-seq normalization methods are common, SCONE and regularized negative binomial regression are specifically designed for scRNA-seq, addressing technical variances and preserving biological heterogeneity [40, 41]. A comparative analysis of seven distinct normalization methods for scRNA-seq data revealed that each method exhibits specific strengths for normalizing different data types [39].

Following preprocessing, scRNA-seq data undergo further scrutiny, which outlines prevalent downstream analysis techniques [42] (shown in Fig. 2). Here, we delve into several pivotal methods essential for in-depth understanding.

Identifying cell types within samples is paramount, aimed at unraveling tissue diversity and complexity [43]. The development of algorithms like SIMLR, SAFE, CIDR, CountClust, and others reflects ongoing efforts in this domain [44‒46]. Comprehensive evaluations of these clustering algorithms have been conducted [47, 48], alongside innovative methods such as DivBiclust [49], which utilizes biclustering, SAME for amalgamating solutions from various methods, and PARC [50], tailored for large dataset analysis. To circumvent manual cell-type labeling’s constraints, Shao et al. [51] introduced “scCATCH,” an automated, highly repeatable clustering-based labeling tool.

Reconstructing cell trajectories and pseudotime calculations is vital for simulating dynamic cellular processes, especially in understanding tumor cell state transitions [52]. Algorithms such as Monocle 2, Monocle 3, TSCAN, Slingshot, SLICE, p-Creode, and Waddington-OT have been pivotal [53‒55]. The surge in scRNA-seq data has led to “LISA,” an unsupervised method for large datasets [56]. p-Creode offers another unsupervised trajectory prediction [57], while Waddington-OT applies optimal transport techniques for ancestral and descendant cell fate determination [58]. Algorithms such as TSCAN, Monocle 2/3, and Slingshot are proficient in cell trajectory reconstruction and time estimation [59]. However, the use of pseudotime and RNA velocity analyses must be approached with an awareness of their inherent limitations. Notably, RNA velocity analysis heavily relies on the construction of a k-nearest neighbors graph, which can lead to substantial errors in estimating both the direction and speed of RNA velocities, especially when the k-nearest neighbor graph fails to accurately represent the true structure of the data. Additionally, the application of RNA velocity in validating the correctness of low-dimensional embeddings may result in circular reasoning, as the method can only confirm the embeddings to the extent that they support the velocity estimates. Overinterpretation of the expression dynamics inferred from RNA velocity could lead to misleading conclusions about the biological processes under study. Therefore, caution is advised when interpreting RNA velocity results, particularly regarding the estimation of gene expression changes' speed, which is often inaccurately assessed by RNA velocity methods [60].

Differentially expressed gene (DEG) analysis is a common scRNA-seq application [60], with zero expression levels indicating either “true” zero expression or “dropout” due to technical reasons [61]. “DEsingle,” developed by Miao et al. [62], accurately differentiates between these “zero expressions” [63]. DECENT, utilizing UMI, analyzes RNA molecule distribution pre-dropout. Addressing dropout and multimodal data distribution challenges, ZIAQ was the first to consider dropout rates and scRNA-seq data distribution for DEG identification [64, 65]. Data imputation is a method to reduce dropout noise in scRNA-seq, categorized into model-based, smoothing-based, deep learning-based, and low-rank matrix-based approaches. Xu et al. [66] found that imputation generally improves clustering and data visualization, but its effectiveness varies with dataset size. While imputation enhances the accuracy of analyses by reducing false positives and negatives, it can introduce biases if overused. Therefore, careful selection and application of imputation methods are essential to avoid misleading biological interpretations [67]. Gene set enrichment analysis methods such as PAGE, DAVID, and CAMERA, though primarily for bulk sequencing, are crucial for interpreting DEGs in biological processes [68]. Ma et al. [69]’s IDEA integrates differential expression analysis with gene set enrichment analysis, significantly enhancing outcomes.

Transcription factor (TF) analysis is predominantly characterized by their potential gene regulatory networks. The expression of target genes and other specific genes is driven by TFs, thus establishing profiles of gene expression [70]. Single-cell regulatory network inference and clustering (SCENIC) is a methodology to reconstruct gene regulatory networks and identify cell states based on co-expression and motif analysis in scRNA-seq data [71]. This method includes evaluating the expression of TFs and their target genes in individual cells and identifying significant cell clusters. Analyzing disparities in TF activity using scRNA-seq enhances our comprehension of cell heterogeneity, providing a robust means for exploring the important regulatory mechanisms of specific TFs [72].

CellPhoneDB contains a database of ligands, receptors, and their interactions [73]. This database enables comprehensive analysis of molecules participating in intercellular communication, elucidating communication networks among various cell types. The database also considers the subunit structures of ligands and receptors, representing heterodimers accurately, which are crucial for precise analysis of cell communication mediated by multimeric complexes.

CellChat is able to identify cell-to-cell communication signals accurately, uncover novel communication networks between cells, and map within diverse tissues through systematic analysis [74]. CellChat computes the probability of cell-to-cell communication by combining known receptors, ligands, and their associated factors with expression profiles. Due to the capacity of this method to calculate the strength or quantity of ligand-receptor relationship and summarize the integrated cell-to-cell communication network, its predictive capabilities extend beyond classifying cell populations and establishing their lineages.

In a study, 16 cell-cell communication inference resources and 7 methods were compared, finding that scRNA-seq predominantly captures local cell signals, neglecting processes such as protein translation and only reflecting protein levels. Furthermore, these methods are constrained to single species and do not encompass interspecies communication [75].

scRNA-seq stands out as a powerful tool for delving into the complexities of individual cells, offering unparalleled high-throughput capabilities with minimal sample requirements [76]. This technique is adept at facilitating multi-omic investigations, covering the spectrum from genomics and transcriptomics to epigenomics, proteomics, and metabolomics [77]. It shines in its ability to uncover new cell types, delineate cell lineage trajectories, and chart pathways of cell development. When compared to conventional sequencing methods, scRNA-seq is superior in discovering new genes, pinpointing rare mutations, and accurately quantifying transcript levels. A distinctive benefit of this method is its capability to circumvent biases typically introduced during PCR amplification, by quantifying nucleic acids prior to amplification [78].

However, the application of scRNA-seq is not without its hurdles. Foremost among these is the critical need for high cell viability and quality, as the success of the technique hinges on this factor. Achieving a high-quality single-cell suspension can prove to be a significant challenge [79]. Additionally, a notable limitation of scRNA-seq is the induction of “artificial transcriptional stress responses” due to stress from tissue dissociation [80], necessitating careful optimization of the dissociation process to ensure accurate results [81]. It is also worth noting that scRNA-seq cannot provide information about cell morphology or the spatial distribution of cells within tissues. This technology only analyzes the gene expression patterns of cells at the molecular level, without addressing their actual structural morphology or their specific locations within the tissue. Due to this limitation, relying solely on gene expression signatures to identify cell types without considering histomorphological features can lead to incorrect assumptions about the cells in the sample. For example, certain gene expressions may be present in different cell types, and without considering their spatial relationships and morphological characteristics within the tissue, they might be erroneously classified as the same cell type. Therefore, to more accurately understand and interpret scRNA-seq data, it is recommended to use additional techniques, such as spatial transcriptomics or flow cytometry, to complement the limitations of scRNA-seq. This multimodal approach can provide more comprehensive information about cells and tissues, aiding in the correct identification and interpretation of cell types and their functional states [82]. Furthermore, scRNA-seq is subject to technical biases including gene dropout, library preparation biases, capture efficiency variations, batch effects, amplification biases, and issues related to sequencing depth and coverage. These biases can distort gene expression data, but can be mitigated through strategies such as the use of UMIs, advanced normalization techniques, batch effect correction methods, and continuous refinement of protocols. Acknowledging and addressing these biases are essential for improving the accuracy and reliability of scRNA-seq data, thereby enabling more robust biological insights [83]. Lastly, while scRNA-seq offers powerful insights into cellular heterogeneity, cell-type identification can be challenging, especially when distinguishing between different cell states within the same cell type. It is important to note that cell-type identification is not always robust due to inherent technical biases and variability in gene expression. Initial quality control and filtering steps, while necessary, can introduce biases that affect the accuracy of cell type identification. Therefore, to ensure the robustness of scRNA-seq findings, it is essential to validate these results using complementary techniques such as flow cytometry, immunohistochemistry, or spatial transcriptomics. These orthogonal methods can provide additional layers of verification, enhancing the reliability of cell type identification and offering a more comprehensive understanding of the biological context [84]. Despite these challenges, the growing standardization of single-cell analyses and the formation of dedicated analysis cores within research institutions are progressively alleviating these obstacles.

The pituitary gland, a central endocrine organ, plays a crucial role in maintaining physiological balance through a complex cellular composition. It coordinates various bodily functions such as growth, metabolism, and reproduction by different types of endocrine cells, including thyrotrophs, lactotrophs, and corticotrophs [4]. Therefore, accurately analyzing the types, developmental trajectories, and interactions of pituitary cells is essential for understanding both normal functions and pathological conditions of the gland.

Traditional gene expression analysis techniques, such as microarrays and bulk RNA sequencing, have been standard methods for studying cellular responses and functions over the past decades [85]. However, these methods provide only average signals from mixed cell populations, which presents several significant limitations [86]: First, traditional sequencing methods struggle to resolve specific gene expression differences among various cell types within the pituitary. For instance, distinguishing between diseased and normal cells in PitNETs is challenging, limiting comprehensive understanding of disease mechanisms. Second, pituitary cells exhibit significant dynamic changes in different physiological or pathological states. Traditional methods lack the resolution to detect these subtle changes, hindering the understanding of state transitions during disease progression. Lastly, pituitary research is often constrained by limited sample sizes, especially in clinical samples. Traditional techniques require larger sample volumes, posing substantial obstacles in studying rare cell types or small sample quantities.

The introduction of scRNA-seq has revolutionized pituitary research by offering a powerful tool for precise gene expression measurement at the single-cell level [5]. This technology reveals cellular heterogeneity, identifies new cell subtypes, traces cell fate decisions, and analyzes cell-to-cell interactions, providing valuable insights for developing new diagnostic and therapeutic strategies. Specific advantages include the following: Firstl, scRNA-seq enables high-resolution cell classification and phenotyping. Tools like Seurat and Scanpy facilitate clustering analysis, allowing researchers to identify different cell types and subtypes within the pituitary, unveiling their specific roles in gland function. For example, scRNA-seq can discover new cell types or functional subgroups that play crucial roles under specific physiological or pathological conditions. Second, scRNA-seq can construct developmental trajectories using tools like Monocle, mapping continuous state transitions, and differentiation pathways of cells [54]. Understanding how progenitor cells develop into mature hormone-secreting cells and how these pathways are disrupted in pathological conditions can provide insights into disease mechanisms. Additionally, scRNA-seq data allow exploration of gene regulatory networks within pituitary cells. Tools like SCENIC can identify TF networks regulating gene expression, revealing which genes and pathways are activated or repressed under specific physiological or pathological states [71]. This is crucial for understanding molecular mechanisms of pituitary diseases and identifying new therapeutic targets. While scRNA-seq itself does not provide spatial information, it can be combined with other techniques to reveal cell-to-cell interactions. Understanding how cells interact within the pituitary tissue and their precise locations helps elucidate the cooperative mechanisms in maintaining tissue function and responding to environmental changes.

Through these advanced analytical methods, scRNA-seq not only provides detailed molecular-level maps of pituitary cells but also reveals how cells respond to diseases or other physiological states. This offers new strategies and targets for diagnosing and treating pituitary-related diseases. The application of these technologies significantly advances the depth and breadth of pituitary research, demonstrating the powerful potential of single-cell techniques in biomedical research.

In summary, the importance of scRNA-seq in pituitary research lies not only in its precision and sensitivity but also in its ability to provide deep molecular insights into pituitary biology that traditional sequencing techniques cannot match. This technology holds promise for advancing early diagnosis, treatment, and prognosis evaluation of pituitary diseases, ultimately offering more precise medical solutions for patients.

While scRNA-seq has significantly advanced our understanding of the cellular heterogeneity within the pituitary gland, it is not without limitations. Complementary single-cell multi-omics technologies offer additional layers of information that can address these shortcomings and further propel pituitary research.

Single-cell genomics-plus-transcriptomics methods, such as G&T-seq and SIDR-seq, allow for the simultaneous analysis of both the genome and transcriptome of individual cells [87, 88]. This dual analysis enables researchers to investigate the impact of genetic variations, such as mutations or copy number alterations, on gene expression within the same cell. Understanding how these genetic changes influence pituitary cell function can provide deeper insights into the pathogenesis of pituitary disorders.

Advances in single-cell epigenomics-plus-transcriptomics methods, such as scM&T-seq and scNMT-seq, enable the parallel profiling of the epigenome (e.g., DNA methylation, chromatin accessibility) and the transcriptome [89, 90]. Additionally, single-cell ATAC-seq provides insights into chromatin accessibility at the single-cell level, revealing how DNA is packaged and regulated in the nucleus [91]. This technique is particularly valuable for identifying regulatory elements such as enhancers and promoters that are active in specific cell types. When combined with scRNA-seq, single-cell ATAC-seq allows researchers to correlate chromatin accessibility with gene expression, providing a comprehensive view of the regulatory landscape in pituitary cells [92]. These methods are crucial for understanding how epigenetic modifications regulate gene expression in pituitary cells. For instance, they can reveal how changes in DNA methylation or histone modifications influence cell differentiation and function, thereby providing insights into the regulatory mechanisms underlying pituitary development and disease.

Multi-omics approaches, such as spatial transcriptomics combined with protein profiling, provide spatially resolved data that highlight the physical and functional interactions between pituitary cells [93]. These methods enable researchers to map the precise locations and interactions of different cell types within the tissue, uncovering how these interactions contribute to pituitary function and pathology. Understanding cell-cell communication and the tissue microenvironment is particularly important for elucidating the mechanisms of disease progression and identifying potential therapeutic targets.

In summary, by integrating multiple layers of omics data, single-cell multi-omics approaches can mitigate some of the limitations inherent in scRNA-seq. For example, while scRNA-seq excels at capturing gene expression profiles, it does not provide direct information about the genomic context or the epigenetic landscape that influences these profiles. Multi-omics approaches bridge this gap, offering a more comprehensive view of the molecular mechanisms at play. Additionally, these methods can improve the accuracy of cell-type identification and functional annotation by incorporating genomic and epigenomic data, thereby reducing the biases introduced by single-modality analyses.

The application of multi-omics technologies in pituitary research holds great promise for advancing our understanding of this complex endocrine organ. These advanced methodologies enable researchers to provide high-resolution molecular profiles, which can identify previously unrecognized cell types and functional states within the pituitary [94]. This contributes to a more detailed cellular atlas of the gland, significantly enhancing our knowledge of its intricate cellular diversity [4].

Integrating transcriptomic and epigenomic data through these multi-omics approaches allows for the reconstruction of gene regulatory networks [92]. This integration highlights key TFs and regulatory elements that drive pituitary cell function and differentiation. Such detailed regulatory maps are crucial for understanding the molecular underpinnings of pituitary biology and its various functional states. Moreover, multi-omics data can elucidate the molecular alterations associated with pituitary diseases, such as PitNETs. By linking genetic and epigenetic changes to aberrant gene expression patterns and cellular behaviors, researchers can gain a comprehensive understanding of disease mechanisms [95]. This knowledge is vital for identifying potential biomarkers and therapeutic targets. Insights gained from multi-omics studies also inform the development of targeted therapies by identifying specific molecular pathways and cell types involved in disease processes. This precision in identifying therapeutic targets can lead to more effective and personalized treatment strategies for pituitary disorders.

In summary, the integration of single-cell multi-omics technologies provides a comprehensive toolkit for dissecting the complex biology of the pituitary gland. These approaches complement scRNA-seq by offering additional layers of molecular information, thereby addressing its limitations and opening new avenues for research into pituitary function and disease. This holistic view of pituitary biology and pathology is essential for advancing both basic and translational research in endocrinology.

In the past decades, research has predominantly harnessed tissue- or organ-level aggregate data to elucidate cell population functionality and traits. The advent of scRNA-seq technology, however, has ushered in the capability to probe into previously indistinguishable cell populations, which were masked by analogous histological characteristics, genetic markers, and proximal tissue locations. This method affords a granular view, uncovering cellular heterogeneity and novel cellular identities of pituitary (shown in Fig. 3; Table 2).

In a pioneering study, Cheung et al. [96] implemented scRNA-seq on the pituitary glands of 7-week-old male mice, unveiling not only endocrine and stem cells in a digital format but also auxiliary cells including endothelial, connective tissue, and hematopoietic cells. Their research highlights scRNA-seq’s capacity to explore the pituitary gland’s cellular dynamics and functional evolution. Nonetheless, the exclusive focus on male specimens constituted a notable study limitation. Addressing this, Ho et al. [97] extended the application of scRNA-seq to delineate cell-type diversity and functional state heterogeneity within both male and female mouse pituitaries. Their analysis discerned sex-specific cellular architectures in physiological pituitary regulation and pinpointed cell clusters enriched with various hormones alongside nonhormonal interstitial and support cells. Remarkably, they discovered a Pou1f1-expressing cell group characterized by a distinctive hormonal gene expression profile. This discovery reveals the pituitary’s cellular complexity and functional plasticity, highlighting how multifunctional cell groups can challenge traditional views on pituitary cell lineages and suggest dynamic adaptations in physiological and pathological states. These insights underscore the adult pituitary’s complex cellular composition and adaptive capacity.

In addition to the anterior lobe, the posterior and intermediate lobes of the pituitary have garnered significant interest among researchers. A study by Chen et al. [98] employed scRNA-seq on the pituitaries of adult male mice, uncovering the cellular diversity encompassing seven unique cell types within the posterior and intermediate lobes. Notably, this investigation identified new markers including Col25a1, Scn7a, and Srebf1 for pituicytes that surpass the specificity of those previously known, thus offering indispensable resources for the exploration of distinct cellular functions and physiological processes.

Apart from profiling major cell lineages within the pituitary, we can also utilize scRNA-seq for an in-depth investigation of specific cell populations. Cheung et al. [99] applied scRNA-seq combined with lineage tracing to delineate and corroborate a previously unidentified endocrine subset in the juvenile mouse pituitary, elucidating a novel developmental pathway leading to thyrotropes. This discovery furnishes a foundation for further exploration into the biology of thyrotropes. The interest of the researchers has also extended to lactotropes. Through scRNA-seq of the rat pituitary, Kučka et al. [110] underscored the pivotal involvement of PI4KA in the coupling of calcium secretion in pituitary lactotropes, subsequent to voltage-gated and PI(4,5)P2-mediated calcium signaling. Furthermore, an insightful study by Chen et al. [111] challenged established views regarding the embryonic origins of neurohypophyseal and adenohypophyseal cells by demonstrating that neural plate progenitors contribute to both cell types, as evidenced through scRNA-seq analysis of the zebrafish pituitary.

Collectively, the aforementioned studies underscore the profound impact of single-cell sequencing technologies on the identification and characterization of cellular subsets, thereby enriching our understanding of cellular composition of the pituitary. While the current depiction of the pituitary gland’s cellular composition is comprehensive, forthcoming investigations should aim at uncovering finer details through more in-depth mechanistic studies. Potential directions include the discovery of novel cellular subsets influencing hormone release or the recognition of previously unknown roles of established cell types.

The evolution of scRNA-seq technology has revolutionized our capacity to attribute gene expression profiles to distinct cell populations, especially within intricately structured tissues such as the pituitary gland. Professor Ruf-Zamojski and her colleagues have embarked on a series of studies. In their 2018 investigation, they leveraged scRNA-seq on the LβT2 gonadotroph cell line to uncover a cell-specific, all-or-none transcript induction pattern in reaction to GnRH gradient [101]. This finding introduced the concept that immediate-early gene activation at the single-cell level might operate via binary switch-like gene bits. Subsequent research by the team refined this understanding, illustrating that LβT2 cell responses to GnRH are not linearly amplifying but rather follow a bimodal pattern [100]. In 2021, they unveiled transcriptional and chromatin accessibility profiles unique to each primary pituitary cell type, including sex-specific regulatory elements, by analyzing over 70,000 nuclei from adult mice pituitaries, underlining the pivotal influence of chromatin accessibility on cell identity transcriptional blueprints [102]. Advancing their research to human subjects, they identified critical regulatory domains and TFs correlating with gene expression in pituitary stem cells (PSCs) through a similar methodology [108]. Their latest innovation, CREMA, addresses the shortcomings of traditional approaches by enabling the identification of functional regulatory elements beyond conventional chromatin “peaks” [103]. Complementing their work, Mayran et al. [104] elucidated the necessity of both Pax7 and Tpit for the activation and chromatin accessibility of melanotrope-specific enhancers, underscoring the synergistic action of pioneer and non-pioneer factors in lineage-specific chromatin remodeling.

In summary, scRNA-seq presents an unparalleled lens through which to examine gene expression regulation within specific cellular subsets, offering profound insights into the biological intricacies of the pituitary gland. Future studies are likely to harness this technology for more in-depth research to uncover the dynamic changes and mechanisms of the pituitary gland under various physiological and pathological conditions. For example, such studies could track changes in gene expression patterns of the pituitary gland during different developmental stages, disease progression, or treatment responses, thereby providing new insights and targets for clinical intervention strategies.

Several studies have characterized PSC using scRNA-seq and these studies illustrate the potential role of PSC in the regenerative medicine. For instance, Laporte et al. [105] utilized scRNA-seq on the neonatal pituitary gland of mouse, revealing proliferative stem cell populations that displayed a hybrid epithelial/mesenchymal phenotype, characteristic of development-involved tissue stem cells. They also recapitulated the stem cells’ phenotype in organoid culturing. Interestingly, they found that the neonatal gland efficiently regenerated after local damage without the presence of upregulated IL-6, while Vennekens et al. had the opponent finding on the adult pituitary gland. They applied scRNA-seq on the adult pituitary, finding the acute PSC activation process with the presence of upregulated IL-6 upon targeted endocrine cell-ablation damage. The difference may be caused by the different status of pituitary gland, as the neonatal pituitary gland is in the already high stem cell activation status and the adult pituitary gland is in the dormant status. Notably, they found that administering IL-6 to young mice promptly triggered PSC proliferation upon local damage in vivo and in vitro using stem cell-derived organoids. Their study identified IL-6 as a PSC activator upon local damage. Another exciting finding is that they found the aging pituitary’s stem cells retain intrinsic activatability, resurfacing once released from their impeding tissue milieu. Taken together, above findings indicate that PSC plays a crucial role in regenerative medicine. However, it is well known that obtaining mature cell types from PSCs in vitro remains challenging, and further research is needed to overcome these difficulties and harness their full potential for therapeutic purposes.

One of the paramount benefits of scRNA-seq is its capability for pseudotime analysis. In a landmark study, Zhang et al. [109] conducted scRNA-seq on 4,113 cells from human fetal pituitaries, delineating varied developmental pathways and distinct transitional states across five hormone-secreting cell lineages. A significant aspect of their research was the delineation of PSC heterogeneity, identifying a unique hybrid epithelial/mesenchymal phase alongside an early-to-late state transition, which is in line with Laporte’s finding. Historically, delineating the roles of stem cell differentiation and the maintenance of their progeny in cell turnover has been challenging. Utilizing scRNA-seq coupled with lineage tracing, Rizzoti et al. [107] mapped the diversity and mobilization of PSCs, uncovering that PSC differentiation is more prevalent than previously believed. Under adaptive conditions, this differentiation is not only heightened but also more varied than lineage tracing alone has shown. Their thorough analysis of PSC progeny underlines the critical support role of the microenvironment in sustaining selected emerging cells, thereby influencing PSC output. A notable limitation of scRNA-seq is the requirement for fresh samples, which considerably limits its applicability. In response, advancements in single-nucleus sequencing technologies have emerged. Zhang et al. [108] utilized single-nucleus RNA-seq and ATAC-seq to map cell-type-specific gene expression and chromatin accessibility across all major pituitary cell lineages in postmortem samples from pediatric to aged individuals. Their work highlighted the presence of uncommitted PSCs, transitioning progenitor cells, and gender-based differences. Through linear modeling of multiome data, they pinpointed regulatory elements and TFs linked with PSC gene expression, also uncovering deterministic factors that account for variability in PSC marker expression. These insights provide a deeper understanding of the mechanisms regulating essential PSC genes and cell identity.

Taken together, scRNA-seq offers unprecedented depth in exploring the complex cellular ecosystems within the pituitary gland, yielding vital breakthroughs in understanding PSC biology. Such research paves the way for novel stem cell-based therapeutic strategies targeting the replenishment of lost or impaired endocrine cells in the pituitary gland, thereby offering potential treatments for conditions such as pituitary hormone deficiencies and tumors.

Pituitary neuroendocrine tumors (PitNETs), which constitute common intracranial neoplasms accounting for an incidence rate of about 10%, are classified into various distinct categories, such as those aligning with PIT1, TPIT, and SF1 lineages, in addition to plurihormonal and null cell tumors [124]. The management of PitNETs typically involves a combination of surgical resection, pharmacological interventions, and radiotherapy. Despite these strategies, the patient faces significant challenges including resistance to pharmacological treatments, a scarcity of targeted therapeutic options, risks of tumor recurrence postsurgery, and a lack of comprehensive insights into the molecular drivers of these tumors [125]. Numerous researches have been committed to unraveling the complex molecular basis of PitNETs, yet the multifaceted gene interactions and the distinctive attributes of anterior pituitary cells within the tumor’s microenvironment have hindered a full elucidation of the disease’s pathogenesis [126‒128]. With the advent of recent technological advancements, the application of single-cell sequencing stands out as a transformative approach for delving into the molecular intricacies of PitNETs (shown in Fig. 4) [129, 130].

In their groundbreaking study, Cui et al. [116] utilized scRNA-seq on 23 samples of PitNETs, revealing for the first time the transcriptomic profiles of both tumor and adjacent normal cells. This analysis laid the groundwork for the delineation of PitNET characteristics and heterogeneity, crucial for the discovery of novel biomarkers and therapeutic avenues. Similarly, Yan et al. [117] undertook scRNA-seq on 4 healthy and 24 PitNET specimens, elucidating the fundamental cell types within the healthy pituitary gland, the intrinsic heterogeneity of PitNETs, and characteristics of the tumor immune microenvironment. Intriguingly, they also identified an aggressive epithelial cell subset present in all cases examined. Furthermore, Batchu et al. [118]’s analysis of scRNA-seq data from 21 PitNET samples, sourced from an accessible public database, mapped out the metabolic profiles of different pituitary cell lineages. Their insights are pivotal for the formulation of targeted therapies specific to various PitNET lineages or subgroups.

In addition to showing the landscape of PitNETs, scRNA-seq facilitates an in-depth analysis of their pathogenesis. Asuzu et al. [119] utilized scRNA-seq on 27,594 cells from PitNETs and adjacent normal glands, revealing a previously unrecognized population of proliferating cells may drive tumorigenesis. A critical discovery was the identification of a pathway for apoptosis evasion through proteasome-mediated degradation of noxa, highlighting the therapeutic promise of proteasome inhibitors. In a focused study, Zhang et al. [120] analyzed scRNA-seq data from three silent ACTH-secreting tumors and five functioning ACTH-secreting tumors, revealing that silent ACTH-secreting tumors exhibit signs of epithelial-to-mesenchymal transition, characterized by enhanced mesenchymal gene expression and diminished transcripts involved in hormone synthesis and secretion. This study posited that a unified transcriptional reprogramming might simultaneously hinder POMC processing and foster tumor invasiveness. Additionally, another study by Zhang et al. [121] using scRNA-seq on 3 healthy and 21 PitNET samples, distinguished specific genes indicative of major subtypes within both well-differentiated and poorly differentiated PitNETs across each lineage, also confirming the prognostic utility of differentiation markers in an independent dataset.

The treatment of PitNETs is becoming increasingly complex, with more reports of PitNETs lacking effective treatments, especially refractory PitNETs. This underscores the urgency for innovative therapeutic approaches. Moncho-Amor et al. [115] utilized scRNA-seq on P27-null mouse, identifying the pivotal role of an SOX2-driven MAPK pathway in stem cells for tumorigenesis. This research suggests that targeting PSCs could be a viable antitumor strategy. In a related vein, Cheng et al. [122] explored the basis of dopamine agonist (DA) resistance by analyzing scRNA-seq data from both DA-resistant and DA-sensitive PitNETs, pinpointing the association of DA resistance with the focal adhesion signaling pathway and proposing Genistein as a potential therapeutic agent due to its ability to inhibit focal adhesion pathway expression. Furthermore, Lyu et al. [123] leveraged scRNA-seq to examine 4 PIT1-positive PitNETs, shedding light on the cellular and functional diversity of tumor-associated fibroblasts and immune cells. Their findings underscore the antitumor potential of IFN-γ through its role in modulating the tumor-promoting activities of tumor-associated fibroblasts.

In summary, scRNA-seq has emerged as a cornerstone technology in translational medical research, offering unparalleled insights into the complex landscape of cellular heterogeneity and dynamics within PitNETs. This cutting-edge approach allows for the detailed profiling of individual cells, uncovering rare cell types, defining various cell states, and mapping intricate cellular interactions with precision previously beyond reach. For instance, several studies have challenged the concept of prevailing mono-lineage PitNETs and reported frequent co-expression of TFs, particularly of PIT1 and SF1 [131‒133]. scRNA-seq provides a powerful tool to investigate these findings in greater detail. By analyzing cell-specific gene expression, scRNA-seq can identify cells that co-express PIT1 and SF1 within the same cell population. This enables researchers to confirm the presence of these co-expressing cells and study their prevalence and characteristics in pituitary tumors. Additionally, scRNA-seq data can construct developmental trajectories to track gene expression changes from precursor cells to mature tumor cells, providing insights into the origins of PIT1/SF1-co-expressing cells and their roles in tumor progression. The rich molecular insights gained from scRNA-seq provide critical resources for deciphering the pathogenesis of PitNETs, propelling forward the field of precision medicine. By facilitating the development of therapies specifically designed to target the distinct molecular characteristics of individual tumors, scRNA-seq heralds a new era in the diagnosis, treatment, and management of PitNETs, promising significant advancements in patient care and therapeutic outcomes.

While the focus on PitNETs remains significant, the broader spectrum of pituitary gland disorders, such as obesity and systemic inflammation, has attracted increased scrutiny in recent research efforts. Miles et al. [112] utilized scRNA-seq on 10–15 weeks’ high-fat diet mouse, uncovering a notable resilience against obesity-induced oxidative stress in male pituitary cells compared to female ones, and pinpointing the pituitary cell populations most susceptible in females. Ruggiero-Ruff et al. [113] applied scRNA-seq on pituitary glands from male mice fed control and high-fat diet, emphasizing the diet-induced shifts in pituitary cell populations and gene expressions that contribute to the hormonal imbalances observed in obesity. Yan’s pioneering research delved into the pituitary gland’s response to systemic inflammation, utilizing scRNA-seq to demonstrate how all major pituitary hormone-producing cells engage in a cell type-specific response [114]. A key discovery was the upregulation of chemokine gene expression by hormone-producing cells, enhancing communication with immune cells and underscoring the pituitary’s complex role in modulating the body’s physiological response to inflammation.

The application of scRNA-seq in pituitary research represents a paradigm shift in our approach to understanding this crucial endocrine gland. Through detailed profiling of individual cells, scRNA-seq has illuminated the pituitary’s cellular diversity, revealing insights into the developmental trajectories, functional states, and pathological alterations of its constituent cells. Studies utilizing scRNA-seq have not only expanded our knowledge of PSC biology and the heterogeneity of PitNETs but have also highlighted the gland’s integral role in responding to systemic conditions. Despite the progress, challenges remain in translating these discoveries into clinical practice and developing targeted therapies for pituitary disorders. Future research, leveraging the depth and breadth of scRNA-seq, holds the promise of further unraveling the molecular underpinnings of pituitary function and disease, paving the way for precision medicine in endocrinology and oncology.

The authors have no conflicts of interest to declare.

This work was supported by the CAMS Innovation Fund for Medical Sciences (No. CIFMS 2021-I2M-1-003), the Beijing Municipal Natural Science Foundation (No. M22013), the Key-Area Research and Development Program of Guangdong Province (No. 2021B0101420005), the National High Level Hospital Clinical Research Funding (No. 2022-PUMCH-C-012), and the National Natural Science Foundation of China (82103302 to MC).

Shuangjian Yang conceived the review, collected and analyzed the literature, and completed the article. Congcong Deng, Changqin Pu, Xuexue Bai, and Chenxin Tian participated in the collection of literature and writing of the article. Mengqi Chang edited the language of the article and revised the manuscript. Ming Feng made the idea of the review, revised the manuscript, and took charge of the quality of the whole article.

Shuangjian Yang, Congcong Deng, and Changqin Pu contributed equally to this work.