Transcriptomic characterization of the pituitary under systemic inflammation

To explore the response of the pituitary cells to immune challenges, we first performed bulk RNA sequencing on pituitary tissues extracted from mice that were intraperitoneally (i.p.) injected with a medium-dose lipopolysaccharide (LPS; 5 mg/kg) or underwent mid-grade cecal ligation and puncture (CLP) (Fig 1A and S1 Table), both of which are commonly used to induce systemic inflammation related to bacterial infections [17]. Up-regulated and down-regulated genes exhibited a high degree of similarity between LPS and CLP treatment (S1A Fig). Principal component analysis (PCA) of the transcriptomic profiles revealed that both LPS and CLP treatments led to clusters distinct from sham and saline controls, with the LPS treatment producing more consistent responses (Figs 1B and S1B). Gene Set Enrichment Analysis (GSEA) revealed strong up-regulation in the expression of genes involved in the signaling pathways related to apoptosis, proinflammatory response, cytokine production, and myeloid leukocyte migration (Figs 1C and S1C). qPCR tests confirmed significant increases in the expression of several representative genes, such as Ripk1, Il6, Stat3, and Ccl2 that were found to be enriched in the gene sets in GSEA analysis (Fig 1D, top panel). We further explored the progression of inflammation in the pituitary by treating mice with a subseptic dose of LPS (0.5 mg/kg, i.p.) and examining gene expression using qPCR at various time points (6, 12, 24, and 48 h postinjection). We found that the pituitary gland responded dynamically to inflammatory processes, displaying intense disturbance at 6 h and gradually returning to the basal state from 12 to 48 h (Fig 1D). Taken together, these results indicate that the pituitary cells at the population level exhibit potent and dynamic responses to systemic inflammation.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. scRNA-seq reveals the transcriptional landscape of the pituitary under systemic inflammation.

(A) Schematic diagrams of the experimental setup. The representations of syringe and liquid drop were created with BioRender.com. (B) Volcano plot showing the pituitary gene expression from mice subjected to 5 mg/kg LPS for 6 h and the control groups treated with saline. Genes up-regulated or down-regulated by log₂(FoldChange) >|0.5| are shown in red and blue, respectively. (C) GSEA profiles showing significant enrichment of gene sets after LPS treatment in the pituitary. The transcriptome datasets used in (B) were used for GSEA analysis. (D) qPCR analysis of representative genes enriched in gene sets identified in (C) under 0.5 mg/kg and 5 mg/kg LPS treatments and at different time points (n = 3–4 mice). (E) UMAP plot showing pituitary cell populations from all treatment groups after Seurat v3 integration. Legends indicate cell types. (F) Scatterplots showing expression of canonical marker genes for HPCs, overlayed on the same embedding as in (E). (G) UMAP plot of HPCs from the saline- and LPS-treated groups showing different cell types and cell states based on HVGs. Dashed line: cell types; dot color: cell states. (H) UMAP plot of HPCs from the saline- and LPS-treated groups showing different cell types and cell states based on SCENIC regulon AUC scores. Dashed line: cell types; dot color: cell states. (I) Re-clustering of HPCs in (G) and (H). Dashed line: cell states. (J) Heatmap showing the representative binarized SCENIC regulon activity in HPCs with healthy and inflammatory cell states predicted in (I). Columns are cells and rows are regulons. White: not activated; black: activated. All data represent mean with SD. Statistical significance was determined by two-tailed Student’s t test, *p < 0.05, **p < 0.01, ***p < 0.001. The data underlying this figure can be found in S1–S3 Tables and S1 Data. GSEA, Gene Set Enrichment Analysis; Som, somatotropes; Lac, lactotropes; Cort, corticotropes; Mel, melanotropes; Gonad, gonadotropes; Thyro, thyrotropes; Pou1f1, proliferating pou1f1-positive cells; WBCs, white blood cells; RBCs, red blood cells; Endo, endothelial cells; Ambig, ambiguous; UMAP, uniform manifold approximation and projection; HVGs, highly variable genes; HPCs, hormone-producing cells; SCENIC, single-cell regulatory network inference and clustering; AUC, area under the curve; LPS, lipopolysaccharide; scRNA-seq, single-cell RNA-sequencing.

https://doi.org/10.1371/journal.pbio.3002403.g001

To map the transcriptomic responses with cell type-specificity, we performed scRNA-seq based on the modified single-cell tagged reverse transcription sequencing (STRT-seq) method on pituitary cells from mice treated with different doses of LPS (0.5, 1, 10, and 50 mg/kg) at different time points (3, 6, 24, 48 h and 3 weeks). We also challenged mice with polyinosinic-polycytidylic acid (Poly(I:C)) and tumor necrosis factor α (TNF-α) to investigate the potential effects of viral infection and proinflammatory cytokine, respectively, to broaden our understanding of the effects of systemic inflammation in the pituitary (Fig 1A). After conducting quality control checks on gene recovery, unique molecular identifiers (UMIs), and the percentage of ribosomal and mitochondrial RNA, we retained 5,506 high-quality cells for subsequent analysis (S2A–S2C Fig and S2 Table).

Utilizing the Seurat anchor method [18], we coalesced cell populations across treatments into a common space. Employing the unsupervised Louvain algorithm [19], we delineated 14 clusters, 13 of which exhibited distinct cell identities (Fig 1E). These clusters encompassed all 6 major types of secretory HPCs, as identified by the established markers, including hormone and hormone receptor genes (Gh and Ghrhr for somatotropes; Prl for lactotropes; Pomc for corticotropes and melanotropes; Cga, Lhb, and Fshb for gonadotropes; Cga and Tshb for thyrotropes), as well as a transcription factor gene (Pax7 for melanotropes) (Fig 1F). Additionally, we identified 7 auxiliary cell types by their characteristic marker expression, including white blood cells (WBCs, C1qb+), red blood cells (RBCs, Hbb-bt+), stem cells (Stem, Sox2+ and Cyp2f2+), proliferating Pou1f1 cells (Pou1f1, Pbk+ and Mki67+), endothelial cells (Endo, Pecam1+ and Emcn+), pericytes (Peri, Col1a1+ and Ogn+), and pituicytes—the pituitary glia cells (Pitui, Col25a1+ and Scn7a+) [16]. We also found a minor cluster containing a few single cells with ambiguous identity (cluster 14) (S2E Fig). Evaluating cell proportions, we observed that immune stimuli did not modify the pituitary cell type composition (S2D Fig).

We conducted more detailed analyses on HPCs, which are the principal cells responsible for hormone secretion in the pituitary [12]. To investigate the gene expression patterns of the HPCs from the saline- and LPS-treated groups, we extracted the top 2,000 highly variable genes (HVGs) (S3 Table). Using uniform manifold approximation and projection (UMAP) [20], we embedded the HPCs into an unintegrated, naïve, and low-dimensional space. In the UMAP-HVG embedding, pituitary cells could be segregated into 6 HPCs and 2 cell states, based on variations in their molecular signatures (Fig 1G). Re-clustering with the Louvain algorithm supported that each HPC can be subdivided into binary cell states (Fig 1I, top panel), which we named “healthy” or “inflammatory” states, respectively.

To confirm that the HPCs were indeed segregated into binary states, we applied single-cell regulatory network inference and clustering (SCENIC) to analyze the data at the regulon level [21]. SCENIC robustly aggregates gene expression at the transcription factor level and yields a regulon score (AUCell) that represents the regulon activity. Consistent with the UMAP-HVG analysis, the UMAP-SCENIC embedding clustered pituitary cells according to cell types and cell states (Fig 1H), with each cell type identified as 2 state-specific clusters (Fig 1I, bottom panel). The Pou1f1-dependent populations, including somatotropes, lactotropes, and thyrotropes [13,22,23], formed indiscernible boundaries, while corticotropes and melanotropes that are Pou1f1-independent populations with some shared marker genes, such as Pomc, clustered closer to each other and away from the Pou1f1-dependent populations (Fig 1H). These results indicate that the SCENIC analysis accurately recapitulates the healthy and inflammatory cell states in control and immune-challenged conditions, as well as the classical lineage relationships of pituitary HPCs. More importantly, the binary activity matrix of 356 SCENIC regulons revealed that HPCs in the healthy and inflammatory states exhibited both unique and shared features (S4 Table and S2F Fig), especially those related to highly enriched members of the Irf/Stat family, across different cell types (Fig 1J).

To investigate the contribution of HPCs to the pituitary inflammatory response process, we compared gene expression changes in the pituitary tissue from the RNA-seq dataset and HPCs from the scRNA-seq dataset after LPS treatment. Correlation analysis indicated that pituitary HPCs exhibited a highly similar transcriptomic signature to that of the entire pituitary gland (S2G Fig).

To explore the impact of Poly(I:C) and TNF-α on the pituitary, we employed Seurat label transfer to apply the “healthy” or “inflammatory” state designations, as established by the saline and LPS treatment groups, to the cells in the Poly(I:C) and TNF-α treatment groups (Figs 1G and S3A). UMAP and alluvial plots for the 6 HPCs demonstrated that the HPCs in the Poly(I:C) and TNF-α treatment groups also delineated into the binary “healthy” or “inflammatory” cell states, akin to those observed in the LPS treatment (S3B–S3E Fig).

Collectively, our high-dimensional pituitary single-cell atlas indicated that, at the transcriptional level, all pituitary HPCs respond strongly to systemic inflammation. The high-quality scRNA-seq dataset allowed us to further examine the properties of inflammatory responses by pituitary cells in a cell type-specific manner.

Corticotropes respond strongly to systemic inflammation

To investigate the distinct features and biological functions of HPCs under inflammation, we first performed differential expression analysis by calculating conserved and differentially expressed marker genes using healthy and inflammatory states as grouping factors (Fig 2A). We identified 418 conserved marker genes in HPCs that can serve as markers for identifying cell types and ensuring the quality of the sequencing data (S5 Table and Fig 2B, top panel). For example, Car10 was a conserved marker for somatotropes across healthy and inflammatory states, and so was Edil3 for lactotropes, Palla for corticotropes, Megf11 for melanotropes, Ttc24 for gonadotropes, and Pvalb for thyrotropes (S4A–S4F Fig). Next, we identified 969 differentially expressed genes (DEGs) across all HPCs, which can be used to study the specific cellular responses to inflammation (S6 Table and Fig 2B, bottom panel). UpSet plot visualization of the conserved marker genes and DEGs across the pituitary HPCs revealed that corticotropes had the largest number of unique DEGs among the HPCs, while the number of conserved genes did not show difference among cell types (S7 and S8 Tables and Fig 2C and 2D). Gene pathway enrichment analysis of the unique DEGs in each HPC revealed that aside from a consensus response in the inflammatory pathway, different cell types also exhibited unique responses in other pathways (S9 Table and S5 Fig). For example, somatotropes displayed a lymphocyte activation response, whereas melanotropes displayed an extrinsic apoptotic signaling response (S5G Fig). Overall, these results demonstrate that different types of HPCs exhibit distinct response features to systemic inflammation.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Differential expression and Augur analysis show corticotropes as the most responsive HPCs under systemic inflammation.

(A) Schematic diagrams of the conserved and differentially expressed markers in HPCs subjected to immune stimuli. (B) Heatmaps showing the conserved marker genes (top panel) and DEGs (bottom panel) for each cell type under healthy or inflammatory states (log₂(FoldChange) > 0.65 and p-value adjusted < 0.001). (C and D) Upset plots depicting the relationship of conserved markers (C) or DEGs (D) that are unique to or shared between cell types. Bars indicate the number of genes. Lines between cell types highlight shared conserved marker genes (C) or DEGs (D). (E–G) Augur analysis of HPCs in response to inflammation. Cell types with AUC scores closer to 1 responded more strongly to inflammation. (E) Scatter plot shows the Augur AUC intensity of individual cells. (F) Average Augur AUC for each major HPC cell type. (G) Plot of Augur AUC and the number of DEGs. The data underlying this figure can be found in S5–S8 Tables. DEG, differentially expressed gene; HPC, hormone-producing cell.

https://doi.org/10.1371/journal.pbio.3002403.g002

In agreement with the above findings, Augur AUC analysis [24] on 6 HPCs revealed that corticotropes responded most strongly to the inflammatory challenges (Fig 2E–2G). Gene Ontology (GO) analysis of corticotrope-specific DEGs revealed that genes involved in the pathways of protein modification and phosphorus metabolic process (representative genes: Sphk1, Gprc5a, Ptpn2, and Rnf180) were greatly up-regulated (S6A and S6C–S6F Fig and S10 Table), while the highly enriched pathways among the down-regulated DEGs primarily involve nucleoside triphosphate metabolic processes (representative genes: Gamt, Acaa2, Nme3, and Atp1a2) in the inflammatory state (S6B and S6G–S6J Fig and S10 Table). Taken together, corticotropes are the most perturbable cells in the pituitary during inflammation.

Systemic inflammation modulates the secretory functions of pituitary HPCs

The pituitary is best known for its functions of secreting peptide hormones. Under immune stress conditions, the pituitary corticotropes secrete ACTH, which promotes the serum levels of corticosterone [25], suggesting an essential function in pituitary secretion during inflammation. Consistent with this view, the transcription level of Pomc, a precursor polypeptide of ACTH, was up-regulated at a moderate but statistically significant level in corticotropes under inflammatory states compared to healthy states (S7A Fig). We also observed that the transcription levels of genes encoding GH, PRL, LH, and TSH underwent varying degrees of changes under inflammation (S7A Fig), thus indicating the potential effects of systemic inflammation on modulating the production of classic pituitary hormones.

We further asked whether inflammation might trigger the production of other secreting factors by the pituitary. GO analysis of 969 DEGs in pituitary HPCs revealed that many of them were enriched in the “extracellular region” and “cytokine-mediated signaling pathway” (S11 Table and Fig 3A and 3B). To validate our transcriptomic data, we performed proteomic analysis of the pituitary following LPS treatment and analyzed differentially abundant proteins using GO analysis (S12 Table and S7B Fig). The enriched proteins were also associated with the “extracellular space” and “extracellular region” GO terms (S7C Fig), which corroborated the transcriptomic results. The scRNA-seq dataset showed that the expression levels of myeloid migration-related chemokines, such as Ccl2, Cxcl1, and Cxcl10 [26–28], were strongly up-regulated in all HPCs under systemic inflammation (S7D Fig). These results suggest that pituitary HPCs may release non-hormone bioactive substances, such as cytokines, in response to inflammatory challenges.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Nptx2 is specifically released by corticotropes in response to systemic inflammation.

(A and B) GO analysis of DEGs in HPCs showing enriched GO terms in biological process (A) and cellular component (B) categories. (C) Dot plots showing the 13 candidate genes encoding secreted proteins in HPCs. Dot size and color represent expression abundance and the expression specificity of the gene, respectively. (D) qPCR screening of the 13 genes in multiple tissues from control and LPS-treated mice (n = 2–4 mice). (E) Representative images showing ISH of Nptx2 RNA (green) and IF of POMC (red) in the pituitary from mice treated with saline, 0.5 mg/kg or 40 mg/kg LPS. Scale bar, 50 μm. (F) Immunoblot analysis of Nptx2 in the pituitary from mice treated with saline or LPS (n = 3 mice). (G) Serum concentration of Nptx2 from mice treated with different doses of LPS (n = 3). (H) Serum concentration of Nptx2 from mice treated with the same dose of LPS for different durations (n = 3–6 mice). (I) Serum concentration of Nptx2 from mice subjected to AAV-SaCas9-U6-sgRNA-Nptx2 in the pituitary for 3 weeks, followed by LPS exposure (n = 5 or 9 mice per group). (J) FACS analysis of T cells, B cells, and neutrophils in PBMCs from mice pretreated with rNptx2, followed by saline or LPS treatment (n = 4–5 mice). (K) FACS analysis of T cells, B cells, and neutrophils in PBMCs from mice that received the same treatments as shown in (I) (n = 5-7 mice). (L) Analysis of the NBR from (J) LPS+vehicle and LPS+rNptx2 groups, and (K) sgRNA-LacZ and sgRNA-Nptx2 groups. The injection dose of LPS was 0.5 mg/kg in (D, H, I, J, and K). The time duration of LPS was 6 h in (D, E, F, G, I, J, and K). All data represent mean with SD. Statistical significance was determined by two-tailed Student’s t test (G, H, I, K, and L) or by one-way ANOVA with Tukey’s multiple-comparisons test (J), *p < 0.05, **p < 0.01, ****p < 0.0001. The data underlying this figure can be found in S11 Table and S1 Data. The original blot for this blot can be found in S1 Raw Images. DEG, differentially expressed gene; GO, Gene Ontology; ISH, in situ hybridization; IF, immunofluorescence; POMC, proopiomelanocortin; HPCs, hormone-producing cells; LPS, lipopolysaccharide; NBR, neutrophil-to-B cell ratio; PBMC, peripheral blood mononuclear cell.

https://doi.org/10.1371/journal.pbio.3002403.g003

Beyond genes encoding secreting factors, numerous DEGs involved in the maturation of secreted proteins or the regulation of secretion processes are transcriptionally perturbed. For example, Sult1a1, a gene encoding sulfotransferase that is responsible for catalyzing sulfate conjugation of hormone molecules to increase their solubilities [29], displayed pronounced up-regulation in somatotropes during systemic inflammation (S8A Fig). Similarly, Sphk1, the gene encoding a kinase that phosphorylates sphingosine to sphingosine-1-phosphate (S1P)—a sphingolipid involved in the regulation of inflammatory responses [30], was also up-regulated (S6C Fig). By contrast, Sstr5 and Crhr1, the 2 neuropeptide receptor genes for somatostatin and CRH from hypothalamic neurons to control pituitary hormone release [31,32], were markedly down-regulated (S8B–S8D Fig). Collectively, these results indicate that systemic inflammation strongly affects the secretory processes of pituitary HPCs.

Nptx2 is a corticotrope-specific factor that modulates systemic inflammation

Next, we investigated whether systemic inflammation might lead to the release of novel pituitary-specific factors to modulate immune responses. We chose a list of 13 candidate genes encoding secreted proteins that were induced in at least 1 type of HPC during inflammation (Fig 3C). Most of these candidate genes are involved in the immune or inflammation process, such as Fgg, Cartpt, Isg15, Lbp, Lgals9, Fndc4, Il1r2, and Serpina3n, according to previous studies [33–40]. Additionally, Nptx2, Cartpt, and Vgf are involved in nerve system processes [34,41–43], and Insl6 participates in the reproductive process [44].

To address whether induced expression of DEGs is specific to pituitary HPCs, we performed qPCR to examine the expression levels of the 13 candidate genes in multiple tissues, including the pituitary, all major peripheral organs, and 3 major brain areas, from saline- and LPS-treated mice. This screening identified 2 gene products, Neuronal pentraxin 2 (Nptx2) and Cocaine- and amphetamine-regulated transcript (Cartpt), that were robustly and specifically up-regulated in the pituitary gland during systemic inflammation (Fig 3D). The scRNA-seq data showed that Nptx2 was specifically up-regulated in corticotropes, while Cartpt was up-regulated in corticotropes and somatotropes (Figs 3C, S8E, and S8F). CART peptide (encoded by Cartpt) has an established role in immunomodulation [34,45], whereas our current understandings of Nptx2 functions are limited to synapse formation in normal brain functions and neurodegenerative diseases [41,46]. We thus investigated the potential roles of Nptx2 in immunomodulation.

In the pituitary, the gene expression level of Nptx2 showed a 20-fold increase at 6 h after LPS treatment and then gradually returned to basal levels 12 to 48 h later (S8G Fig). RNA in situ hybridization (ISH) analysis confirmed that intraperitoneal administration of LPS dose-dependently increased the percentage of Nptx2⁺ corticotropes (Figs 3E and S8H), a result that was further confirmed by western blotting analysis of Nptx2 protein (Fig 3F).

Consistently, serum Nptx2 protein levels displayed a dose-dependent increase 6 h after LPS treatment (Fig 3G), followed by a transient decline, with serum Nptx2 returning to basal levels 12 h post LPS treatment (Fig 3H). Through adeno-associated virus (AAV)-mediated CRISPR/Cas9 genome editing, we disrupted Nptx2 expression. Stereotaxic microinjection of AAV vectors into pituitary substantially reduced pituitary Nptx2 expression and serum Nptx2 levels during LPS treatment (Figs 3I and S8I). In unison, these findings indicate the release of Nptx2 from the pituitary into the circulatory system in response to inflammatory reactions.

LPS treatment induces strong inflammatory responses, resulting in the production of a diverse range of proinflammatory cytokines. To elucidate the regulatory mechanism of Nptx2 up-regulation within the pituitary, we investigated whether this effect was directly mediated by LPS or through the actions of proinflammatory cytokines. For this purpose, we employed mouse AtT-20 cells, a well-established model for ACTH-secreting corticotropes in the pituitary [47,48]. Treating AtT-20 cells with serum obtained from LPS-challenged mice or the proinflammatory cytokine IL-6, instead of LPS, IL-1α, IL-1β, or TNF-α, significantly up-regulated the expression levels of Nptx2 (S8J and S8K Fig). These results suggest that LPS-induced Nptx2 expression is mediated by LPS-triggered proinflammatory cytokines, such as IL-6.

A recent study highlights that the interaction between Nptx2 and complement C1q influences microglia activity within the brain [49]. We delved into the potential functional impact of Nptx2 on the peripheral immune system, specifically focusing on how it might affect the blood neutrophil-to-lymphocyte ratio (NLR), which is a marker of peripheral inflammation and immune system homeostasis [50,51]. We isolated peripheral blood mononuclear cells (PBMCs) from saline- and LPS-treated mice using CD45-based cell sorting and calculated the percentage of neutrophils and 2 major types of lymphocytes (T cells and B cells) within the PBMC population. We found that the intravenous injection of recombinant Nptx2 resulted in a similar reduction in the percentage of T cells in saline- and LPS-treated mice. Consistent with previous observations [52], LPS treatment significantly reduced the percentage of B cells and increased the percentage of neutrophils. More importantly, the intravenous injection of Nptx2 further enhanced this effect, leading to an increase in the neutrophil-to-B cell ratio (NBR) (Fig 3J and 3L).

To test whether Nptx2 in the pituitary contributes to immune modulation, we disrupt Nptx2 expression in the pituitary (S8I Fig), which resulted in the abolishment of the effect of LPS on reducing B cells and increasing neutrophils (Fig 3K and 3L). This suggests that the pituitary modulation of NBR depends on Nptx2. Taken together, these results indicate that Nptx2 acts as a novel pituitary hormone and may have a functional role in balancing systemic inflammation and immunity [53].

Crosstalks between the pituitary and immune cells during systemic inflammation

Our findings thus far suggest that pituitary HPCs respond profoundly to systemic inflammation and may release cytokines as well as other novel immunomodulators to affect the immune system (Figs 3B and S7D). To probe the potential bidirectional endocrine communications between the pituitary and the immune system, we performed cell–cell communication analysis with CellChat [54] using our pituitary scRNA-seq dataset and a previously published scRNA-seq dataset on the spleen [55]. Systemic inflammation was associated with more and overall stronger interactions within and between the pituitary and spleen populations (Fig 4A). More specifically, during systemic inflammation putative signaling increased between the pituitary and spleen populations and within the spleen, whereas it decreased within the pituitary (Fig 4B). These data suggest that inflammation increases the probability of endocrine signaling between the pituitary and the spleen.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Cell–cell communications between the pituitary and the immune cells under healthy or inflammatory states.

(A) Bar graph showing the total number (left) and strength (right) of possible interactions between pituitary cells and spleen cells under healthy or inflammatory state by CellChat analysis. (B) Circle plots showing the number of inferred interactions within and between pituitary and spleen populations. Red line: the interactions from the pituitary to the spleen; blue line: the interactions from the spleen to the pituitary. (C) Circle plots showing the number of statistically significant interactions for the CXCL pathways between HPCs in the pituitary and immune cells in the spleen. Each color represents 1 cell type; edges connecting circles represent significant intercellular signaling inferred between those cell types. (D) Inferred interactions from corticotropes to splenic immune cell populations under healthy or inflammatory states. The dot color and size represent the calculated communication probability and p-values; p-values are computed from one-sided permutation test. (E) Transwell index of monocyte, neutrophil, and macrophage recruited by PBS- or LPS-treated AtT-20 cells for 6 h (n = 3 replicates). (F and G) Flow cytometry analysis of macrophages (F) and neutrophils (G) in the pituitary from mice treated with saline or LPS (0.5 mg/kg; n = 3 mice). All data represent mean with SD. Statistical significance was determined by two-tailed Student’s t test, *p < 0.05, **p < 0.01, ***p < 0.001. The data underlying this figure can be found in S1 Data. HPC, hormone-producing cell; LPS, lipopolysaccharide; NK, natural killer cell; Macro, macrophage; Neut, neutrophil; pDCs, plasmacytoid dendritic cells; Mono, monocyte; T, T cell; B, B cell; Som, somatotropes; Lac, lactotropes; Cort, corticotropes; Mel, melanotropes; Gonad, gonadotropes; Thyro, thyrotropes.

https://doi.org/10.1371/journal.pbio.3002403.g004

Next, we explored the communication between pituitary HPCs and spleen immune cells in a cell type- and signaling pathway-specific manner. Systemic inflammation drastically increased the information flow in CXCL signaling between immune cells, especially neutrophils, and all major types of HPCs (Fig 4C). Corticotropes, the prominent responders to inflammation in HPCs, mediated the enhanced connections in CCL and CXCL signaling interactions with many immune cells, especially monocytes and neutrophils (Fig 4D). These results suggest that systemic inflammation promotes the cell–cell signaling between the pituitary cells and immune cell populations in a signaling pathway-specific manner.

Given the convenience of studying pituitary–immune cell–cell communications in vitro, we chose to functionally validate the communications using AtT-20 cells. Treating AtT-20 with the serum from LPS-challenged mice or with one of 5 inflammatory stimuli (LPS, IL-1α, IL-1β, TNF-α, and IL-6) significantly up-regulated the expression levels of multiple chemokine-encoding genes (S9A and S9B Fig), demonstrating that inflammatory cytokines could directly modulate the gene expression of corticotropes. Using Transwell migration assay, we found that after being treated with LPS, AtT-20 cells attracted neutrophils, monocytes, and bone marrow-derived macrophages (BMDMs) (Fig 4E), indicating that immune cells can be recruited by inflammatory corticotropes during inflammation in vitro.

At the in vivo level, fluorescence-activated cell sorting (FACS) results showed that the number of macrophages (CD45^hiCD11b⁺) [56] increased in the pituitary and returned to normal levels 48 h after LPS treatment, which were consistent with the macrophage marker gene expression level (Figs 4F, S9C–S9E, and S9H). The number of CD206⁺ M2-like macrophages, a population responsible for tissue repair [57,58], increased 24 h after LPS treatment and peaked at 48 h (S9I–S9L Fig). We also observed dynamic changes in the number of neutrophils in the pituitary (Figs 4G, S9C, S9F, and S9G). Taken together, these results corroborated the findings of the CellChat analysis and suggested that the pituitary is modulated by systemic inflammation to recruit immune cells into the pituitary likely through enhanced chemokine signaling.

Macrophages promote ACTH release under inflammation

Finally, we investigated the potential effects of immune cell activity on pituitary cells during inflammation. Since we only observed neutrophils adhering to the blood vessel walls and did not observe infiltration into the pituitary parenchyma (S9G Fig), we focused on the function of macrophages, which are located in the parenchyma (S9E Fig) and possess well-established methods of manipulation. When co-cultured with LPS-pretreated BMDMs, AtT-20 increased the expression of Pomc and Pcsk1 (Fig 5A), which are involved in the maturation of the ACTH peptide [59], and elevated ACTH concentration in the supernatant of the co-culture medium (Fig 5B), indicating that macrophages can affect hormone production in pituitary cells. Next, we treated mice with clodronate-containing liposomes to deplete macrophages [60] via stereotaxic microinjection into the pituitary (Fig 5C). After macrophage depletion, the concentrations of ACTH and corticosterone decreased in the serum of LPS-treated mice, whereas liposome-clodronate treatment did not affect baseline ACTH production (Fig 5D and 5E). This suggests that macrophages within the pituitary promoted the release of ACTH during inflammation. As CCL2 governs macrophage migration during inflammation [26], we investigated the in vivo impact of pituitary CCL2 overexpression. AAV-mediated CCL2 expression in the pituitary elevated serum concentrations of ACTH and corticosterone under both normal and immune-challenged conditions (Fig 5F–5H). Taken together, these results indicate that macrophages in the pituitary promote ACTH release and further support functional interactions between pituitary cells and immune cells under systemic inflammation.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Macrophages in the pituitary facilitate ACTH secretion under inflammation.

(A) Gene expression of Pomc and Pcsk1 in AtT-20 cells after co-culture with vehicle (Veh) or LPS-treated BMDMs for 6 h (n = 3 replicates). (B) ACTH concentration in the supernatant from AtT-20 cells after co-culture with Veh or LPS-treated BMDMs for 6 h (n = 3 replicates). (C) Representative images showing Iba1 (the marker of macrophage) expression in the pituitary from mice that received liposome-PBS (left) or liposome-Clodronate (right) directly to the pituitary for 24 h. Scale bar, 100 μm. (D) Serum concentrations of ACTH in saline- or LPS-treated (0.5 mg/kg LPS for 6 h) mice pretreated with liposome-PBS or liposome-Clodronate (n = 4–7 mice). (E) Serum concentrations of corticosterone in LPS-treated (0.5 mg/kg LPS for 6 h) mice pretreated with liposome-PBS or liposome-Clodronate (n = 6–7 mice). (F) qPCR validation of the Ccl2 transcript in mice subjected to AAV-GFP or CCL2 in the pituitary for 3 weeks (n = 5 mice). (G and H) Serum concentrations of ACTH (G) and corticosterone (H) before and after LPS treatment (0.5 mg/kg for 6 h) in pituitary GFP- or CCL2-overexpressing mice (n = 5 mice). All data represent mean with SD. Statistical significance was determined by two-tailed Student’s t test, *p < 0.05, **p < 0.01. The data underlying this figure can be found in S1 Data. AAV, adeno-associated virus; BMDM, bone marrow-derived macrophage; LPS, lipopolysaccharide.

https://doi.org/10.1371/journal.pbio.3002403.g005

Mice

Animal care and use were approved by the Animal Care and Use Committee of the National Institute of Biological Sciences, Beijing, China (Approval ID: NIBS2022M0036), in accordance with the Regulations for the Administration of Affairs Concerning Experimental Animals of China. Male C57BL/6 wild-type mice (8 to 12 weeks old) were obtained from the Beijing Vital River Laboratory Animal Technology. Mice were on a 12-h on, 12-h off-light cycle (lights off at 8 AM and on at 8 PM).

Cell lines

HEK 293T cells (CRL-3216, ATCC) were cultured in DMEM supplemented with 10% fetal bovine serum (FBS) (0099–141, Gibco) and 1% penicillin-streptomycin (P/S) (15140–122, Gibco). AtT-20 cells (CCL-89, ATCC) were cultured in F-12K medium supplemented with 15% horse serum (HS) (26050088, Thermo Fisher Scientific), 2.5% FBS, and 1% P/S. Cells were maintained at 37°C in a humidified atmosphere with 5% CO₂.

Constructs

Full-length mouse Ccl2 cDNA was reverse transcribed from RNA of mouse pituitary and subcloned into pAAV-CAG-3×Tag vector. The 3 sgRNAs targeting mouse Nptx2 were designed using the web tool Chopchop (https://chopchop.cbu.uib.no) and subsequently synthesized and cloned into the pX601 plasmid [70]. Primer sequences are summarized in S13 Table.

AAV packaging

The AAV vectors were packaged as previously described [71]. The AAV serotype used in this study was AAV-DJ. HEK 293T cells were co-transfected with AAV vectors and AAV helper plasmids. The cells were harvested 72 h after transfection. The virus was purified using cesium chloride density-gradient ultracentrifugation and dialyzed into Dulbecco’s phosphate-buffered saline (DPBS) solution. The viral titer was determined using RT-qPCR.

CLP

The mice were deeply anesthetized with 250 mg/kg of tribromoethanol via i.p. injection. The abdominal fur was carefully removed using a razor blade to expose the abdominal skin. The mice were positioned on a clean surgical operating table, and a clean surgical blade was used to make a midline incision in the abdominal skin, creating a 1 to 2 cm opening. The peritoneum was meticulously dissected to expose the cecum and the connected intestine. Using a medical suture, a ligature was applied approximately halfway up from the bottom of the cecum. Next, the cecum was punctured using a 21G needle, creating 2 punctures, and gently squeezed to expel a small amount of feces. Subsequently, the cecum and the adjacent intestine were placed back into the mouse’s abdominal cavity, and the peritoneum was sutured using a medical suture. Used 7 mm mouse wound clips to secure the abdominal skin. For the control group mice, the ligation and puncture steps were omitted.

Systemic immune stimulation

For medium-dose LPS treatment, mice were i.p. injected with 5 mg/kg LPS (L4130, Sigma-Aldrich). For bulk RNA sequencing experiment, mice were killed 6 h after the induction of inflammation (which corresponds to the peak time of Clinical Severity Score [17]). For scRNA-seq experiments, mice were exposed to different immune challenges. For LPS treatment, mice were i.p. injected with a lethal dose of 50 mg/kg LPS, a septic dose of 10 mg/kg, or subseptic doses of 0.5- and 1 mg/kg LPS [72]. For acute LPS treatment, pituitaries were collected at 3, 6, 24, or 48 h after LPS treatment. To study long-term effects, pituitaries were also collected at 3, 4, and 5 weeks after LPS treatment. For Poly(I:C) treatment, mice were i.p. injected with 10 or 20 mg/kg Poly(I:C) (P1530, Sigma-Aldrich) [73], and the pituitaries were collected at 3 or 6 h or 3, 4, and 5 weeks after inflammation onset. For TNF-α treatment, mice received 0.5 mg/kg TNF-α (50349-MNAE, Sino Biological), and the pituitaries were collected at 6 h after treatment.

Bulk RNA sequencing and bioinformatics analysis

Total RNA was extracted from the pituitary using TRIzol (15596018, Thermo Fisher Scientific) and subjected to single-end 75 bp high-throughput sequencing on an Illumina platform. RNA-seq reads were mapped to the mouse reference genome (mm10) using HISAT2 (v2.1.0, https://daehwankimlab.github.io/hisat2/). The SAM output of each HISAT2 run was sorted and converted into BAM using SAMtools (https://github.com/samtools/samtools/releases/). StringTie (http://ccb.jhu.edu/software/stringtie/; V1.3.5) was used to assemble and quantify read alignments obtained in the previous step. The statistical significance of the differential expression data was determined using DESeq2 (https://bioconductor.org/packages/release/bioc/html/DESeq2.html; V2.11.40.6). DEGs were subjected to GO (http://geneontology.org) and KEGG (https://www.genome.jp/kegg/) pathway enrichment analyses using the clusterProfiler R package (https://guangchuangyu.github.io/software/clusterProfiler/), with a p-value adjusted < 0.05 and log₂(FoldChange) > 0.5.

RT–qPCR analysis

Tissues or cells were collected at predetermined endpoints. Total RNA was isolated using the TRIzol reagent. Reverse transcription was performed using 5× All-In-One qPCR SuperMix (AE341-02, Transgen) following the manufacturer’s instructions. Differential expression of the genes of interest was quantified by RT-qPCR using 2× Taq Pro Universal SYBR qPCR Master Mix (LIN B1260LBB, Vazyme) on a CFX384 Real-Time System (Bio-Rad). The expression of target genes in all samples was calculated using the 2^(−ΔΔCT) method and normalized to the Gapdh housekeeping gene. Primer sequences are summarized in S13 Table.

Pituitary single-cell dissociation

Mice were deeply anesthetized and intracardially perfused with a pre-cold saline solution. The pituitaries were dissected, washed once with DPBS, and then digested with 1 mg/mL collagenase type II (C5138, Sigma-Aldrich) and type IV (C6885, Sigma-Aldrich) enzymes in Hank’s Balanced Salt Solution (HBSS) (pH 7.2 to 7.4, with Ca²⁺ and Mg²⁺) at room temperature for 25 min. DMEM supplemented with 10% FBS was added to stop the digestion, and the cells were filtered through a 40 μm nylon mesh. The cells were washed once with DPBS and resuspended in DPBS supplemented with 1% bovine serum albumin (BSA) (37525, Thermo Scientific) to obtain a single-cell suspension.

scRNA-seq library construction and sequencing

Single pituitary cells were randomly selected under a stereo microscope by mouth pipetting in DPBS supplemented with 1% BSA. The scRNA-seq library was constructed using a modified STRT-seq2 protocol as previously described [13,71]. Briefly, cells were lysed, and the mRNA was reverse transcribed into cDNA with an 8-nt cell barcode and unique molecular identifier (UMI) sequences in a cell lysis buffer. After cDNA pre-amplification, cells with different barcodes were pooled and labeled with a biotin modification index sequence at the 3′ end of the cDNA by 4 cycles of PCR. cDNA was broken into approximately 300 bp fragments randomly by sonication (Covaris). The 3′ end of the cDNA fragments was captured using Dynabeads MyOne Streptavidin C1 (65002, Invitrogen). Library construction was performed using a KAPA Hyper Prep Kit (KK8504, Kapa Biosystems). Finally, the cleaned libraries were sequenced on an Illumina platform with a 150 bp pair-end. Primer sequences are summarized in S13 Table.

RNA ISH

The protocol for in situ RNA hybridization has been previously described with minor differences [74]. Mice were deeply anesthetized with an overdose of pentobarbital and perfused intracardially with diethylpyrocarbonate (DEPC)-saline, followed by paraformaldehyde (PFA, 4% wt/vol in DEPC-PBS). The pituitaries were isolated, postfixed for 4 h at room temperature, and dehydrated in 30% sucrose solution overnight. The fixed pituitary was embedded in an OTC medium (Leica) and cryosectioned to produce 30-μm-thick sections on a cryostat microtome (CM1950, Leica). The sections were rinsed with DEPC-PBS and then permeabilized with DEPC-PTW (DEPC-PBS, 0.1% Tween 20) and 0.5% Triton in 2× SSC. After acetylation, the sections were incubated with prehybridization buffer (50% formamide, 5× SSC, 5 mM EDTA (pH 8.0), 0.1% Tween20, and 0.1% CHAPS) for 2 h at 65°C. Next, the sections were hybridized with digoxigenin-labeled antisense cRNA probe for Nptx2 or Crhr1 for 20 h at 65°C in hybridization buffer (50% formamide, 5× SSC, 5 mM EDTA (pH 8.0), 0.1% Tween 20, 0.1% CHAPS, 0.3 mg/mL transfer RNA, 1× Denhardt’s solution, and 1 μg/μL heparin). After electrophoresis to remove the unbound probe, the pituitary sections were incubated with anti-digoxigenin-POD (1:500 dilution, 11207733910, Roche) at 4°C for 36 h. The probe was detected using TSA Plus Cyanine 5 (1:100 dilution, NEL745001KT, PerkinElmer), and corticotropes were detected using a primary antibody, anti-POMC (1:1,000 dilution, H-029030, Phoenix Pharmaceuticals) at 4°C overnight. The samples were washed 3 times in PBS and incubated with fluorescent secondary antibodies (Goat anti-Rabbit Alexa Fluor 488, 1:500 dilution, 111-545-144, Jackson ImmunoResearch Labs) at room temperature for 2 h. The sections were then washed thrice with PBST (PBS, 0.3% Triton X-100). Pituitary sections were imaged using the Nikon A1 confocal microscope with a ×20 objective and processed using the QuPath software. Primer sequences are summarized in S13 Table.

Immunofluorescence

Mice were deeply anesthetized with an overdose of pentobarbital and intracardially perfused with a pre-cold saline solution, followed by 4% PFA (in PBS). Then, the pituitaries were isolated. After being postfixed and dehydrated as mentioned in the RNA ISH procedure, the pituitary sections were prepared using a cryostat microtome. The sections were permeabilized with PBST and blocked in 3% BSA in PBST at room temperature for 1 h. Subsequently, the sections were incubated overnight at 4°C with primary antibodies (anti-Iba1, 1:2,000 dilution, 019–19741, Wako; anti-Ly6G (FITC Conjugate), 1:400 dilution, 88876, Cell Signaling; anti-CD31, 2.5 μg/mL, AF3628, R&D Systems). Following that, the samples were washed 3 times in PBS and incubated with fluorescent secondary antibodies (Goat anti-rabbit-AF647, 1:500, 111-605-144, Jackson ImmunoResearch; Donkey anti-goat-AF647, 1:500, 705-605-003, Jackson ImmunoResearch) at room temperature for 2 h. Pituitary sections were imaged using the Olympus VS120 virtual microscopy slide scanning system with a ×10 objective and processed using the QuPath software.

Protein sample preparation for LC-MS/MS analysis

Mice were i.p. injected with saline or 50 mg/kg LPS, and the pituitary was collected at 6 h after LPS treatment. Ice-cold lysis buffer (500 μL 8 M Urea with 1× protease inhibitors cocktail) was used to extract protein from the indicated pituitary, and the tissue was manually homogenized with Dounce grinder and ultrasonication, then lysates were centrifuged at 16,000g for 10 min at 4°C. The supernatant was collected, and the total protein was determined by the BCA assay kit (Solarbio, PC0020). Protein was reduced in 5 mM dithiothreitol (DTT, Sigma-Aldrich) at 37°C for 1 h, followed by alkylation with 20 mM iodoacetamide (Sigma-Aldrich) in dark for 1 h, then added DTT again for another 15 min. Added 8 M urea to the sample up to 1 mL. After centrifugation, transferred the supernatant to a 10 KDa molecular weight cut-off (MWCO) filter tube (Millipore) and centrifuged for 10 min, then the concentrated protein in upper reservoir was washed once with 8 M urea and once with 50 mM NH₄HCO₃. Resuspended the sample with 200 μL 100 mM NH₄HCO₃ and added 100 μL 100 μg/mL sequencing-grade trypsin (Promega), then digested at 37°C for 12 h. After digestion, centrifuged at 14,000g for 10 min and collected the solution and dried in a vacuum centrifuge. Samples were resuspended in 100 μL of 0.1% formic acid (FA) before desalted with C18 stationary phase (Omix, Agilent). The samples dried down and resuspended in 5 μL of 0.01% FA before LC-MS/MS analysis.

LC-MS/MS analysis

The LC-MS/MS experiments were performed on Fusion03 (Thermo Scientific Orbitrap Fusion Lumos Tribrid Mass Spectrometer). The MS raw data were analyzed by PEAKS Studio v8.5 protein database containing 4,136 entries. Trypsin was set as an enzyme. The protein expression data were filtered to have at least 70% valid abundance values in the same treatment group. Log2(FoldChange) >1 or <-1.

BMDM primary culture

BMDM isolation and culture were performed as previously described [74]. Briefly, bone marrow was isolated from the femurs and tibias of 8- to 12-week-old male mice. The macrophage precursors purified by a Ficoll-Paque PLUS Media (17-1440-02, GE Healthcare) gradient were differentiated in DMEM supplemented with 10% FBS, 20 nM macrophage colony-stimulating factor (M-CSF) (315–02, PeproTech) and 1% P/S on bacteriological Petri dishes at 37°C in a humidified 5% CO₂ incubator. After 7 to 8 days, differentiated BMDMs were digested with 5 mM EDTA for 5 min and then plated in 6-well cell culture plates or 24-well Transwell plates in DMEM supplemented with 10% FBS and 1% P/S for further experiments.

Co-culture

BMDMs were stimulated with or without 100 ng/mL LPS for 6 h. After washing with DPBS 3 times, AtT-20 cells were added to the BMDMs and co-cultured for another 6 h. Suspended AtT-20 cells were collected for RNA extraction, and the co-culture medium was collected for ACTH detection using an ELISA kit (CSB-E06874m, Cusabio Biotech Co.) after centrifugation.

Inflamed serum treatment

Mice were i.p. injected with saline or 0.5 mg/kg LPS for 6 h. Control or inflamed serum was collected from mouse blood. AtT-20 cells were treated with 25 μL serum per mL cell medium for 6 h.

Neutrophils and monocytes isolation

Bone marrow was isolated from the femurs and tibias of 8- to 12-week-old male mice. The bone marrow cell suspension was then overlaid on top of the Histopaque-1077 (10771, Sigma) and Histopaque-1119 (11191, Sigma) double gradient. After centrifuge at 2,000 rpm for 30 min at room temperature, neutrophils were collected at the interface of the Histopaque-1119 and Histopaque-1077 layers, and monocytes were found in the upper part of Histopaque-1077. Both the neutrophils and monocytes were ready for use after washing with DPBS.

Transwell migration assay

Transwell migration assay was performed using 6.5 mm transwell plates with 5 μm pore inserts (CLS3421-48EA, Corning). AtT-20 cells were treated with or without 100 ng/mL LPS for 6 h. After washing 3 times, AtT-20 cells were added to the lower chamber. Simultaneously, neutrophils, monocytes, or BMDMs (10⁵ cells) suspended in 100 μL medium were added to the upper chamber. The cells were then allowed to migrate through the insert membrane for 6 h at 37°C in a 5% CO₂ atmosphere. For adherent BMDMs, the inserts were washed with DPBS for 3 times, and non-migrating BMDMs remaining on the upper surface were cleaned with a cotton swab. The migrated BMDMs on the insert were fixed with PFA, stained with DAPI, and counted under a microscope. Non-adherent neutrophils and monocytes were distinguished from AtT-20 cells and counted by FACS after surface marker staining (CD11b for monocytes and Ly6G for neutrophils).

FACS

The pituitary tissue was collected after treatment and digested into a single-cell suspension with collagenase type II and type IV enzymes. The digested cell suspension was centrifuged, and the pelleted cells were resuspended and filtered through a 75-μm cell strainer. After centrifugation, the cell pellets were resuspended in RBC lysis buffer (R1010, Solarbio) for 5 min. After washing with DPBS, the collected cell pellets were resuspended in Fixable Viability Dye-eFluor 506 (65-0866-14, Thermo Scientific). After blocking with anti-CD16/32 (101302, BioLegend), the cells were stained with antibodies diluted in FACS buffer (2 mM EDTA and 2% FBS in DPBS) for 20 min at 4°C. For CD206 intracellular staining, a Fixation/Permeabilization Kit (00-5523-00, Invitrogen) was used according to the manufacturer’s protocol. Cells were then subjected to flow cytometry analysis using FlowJo software (V10). For blood cell analysis, mice were given 1 μg rNptx2 (7816-NP-050, R&D system) intravenously and i.p. injected 0.5 mg/kg LPS, and whole blood was collected from the orbit. Blood (100 μL) was mixed with 50 μL 8 mM EDTA (in DPBS) and then resuspended in 1.5 mL RBC lysis buffer for 5 min. After washing with DPBS, the cell pellet was stained as described above. The following antibodies were from BioLegend and the dilutions were 1:200: anti-F4/80-PE (Cat:#123109), anti-CD45-APC/Cyanine7 (Cat:#103115), anti-CD11b-FITC (Cat:#101206), anti-Ly6C-Brilliant Violet 605 (Cat:#128035), anti-CD206 (MMR)-PE/Cyanine7 (Cat:#141719), anti-CD11c-APC/Cyanine7 (Cat:#117323), anti-Ly6G-PerCP/Cyanine5.5 (Cat:#127616), anti-CD3-APC (Cat:#103115), anti-CD45R/B220-Pacific Blue (Cat:#103230).

Immunoblot

Tissue extracts for protein analysis were homogenized using TGrinder (OSE-Y30, TIANGEN) in RIPA buffer (50 mM Tris-HCl (pH 7.5), 450 mM NaCl, 1% NP-40, 0.1% SDS, 0.5% deoxycholic, and cocktail protease inhibitors). Total protein was separated by SDS-PAGE on 10% acrylamide/bisacrylamide gels and transferred to PVDF membranes (1620177, Bio-Rad). Membranes blocked with protein blocking buffer were first incubated with the indicated primary antibodies, followed by horseradish peroxidase-conjugated anti-rabbit IgG or anti-mouse IgG secondary antibody. The following antibodies and dilutions were used: anti-Nptx2 (1:1,000 dilution, 10889-1-AP, Proteintech), anti-β-Actin (1:4,000 dilution, A5441, Sigma-Aldrich), goat anti-mouse IgG: HRP (1:30,000 dilution, 31430, Thermo Fisher Scientific), and goat anti-rabbit IgG: HRP (1:30,000 dilution, 31460, Thermo Fisher Scientific).

Macrophage depletion and virus injection in the pituitary

Mice were anesthetized with pentobarbital (80 mg/kg, i.p.) before surgery and then placed in a mouse stereotaxic instrument. Injections were performed using a microsyringe pump and a Micro4 controller (World Precision Instruments). For macrophage depletion, liposome-PBS or liposome-Clodronate (ClodronateLiposomes.com) was stereotaxically microinjected into the anterior pituitary (2.5 mm posterior from Bregma, 0.4 mm lateral, 6 mm below pia). The liposomes were delivered to the target site at a rate of 60 nL/min for 500 nL per site. Mice received saline or 0.5 mg/kg LPS 18 h after liposome delivery and were killed 6 h after inflammation was established. For CCL2 expression and Nptx2-KO, AAV was delivered directly into the pituitary. The injection site, rate, and volume were the same as those used for the liposome injection. Subsequent experiments were performed at least 3 weeks after virus injection.

Sequence data preprocessing, quality control, and analysis

UMI_tools were used for cell barcode and UMI extraction and demultiplexing. STAR (v2.7.9a) was used for sequence alignment. Subread featureCounts (v1.6.3) was used for feature assignment. Genome index was created with GRCm38 mm10 build, downloaded from RefSeq (https://www.ncbi.nlm.nih.gov/refseq/). For quality control, we computed library size, number of genes recovered, mitochondrial RNA percentage, and Ribosomal RNA percentage from the gene-by-cell expression matrix. The R package scater was used to detect and filter outliers based on the median absolute deviation (MAD) of these metrics.

For scRNA-seq data analysis, Seurat was used to perform normalization, dimensional reduction, integration, and label transfer on the dataset using default parameters. SCENIC was used to infer the regulon activity matrix. Differential expression analysis was performed using the MAST algorithm. Augur was used to prioritize HPC cell types based on their responses to immune challenges. The specificity of gene expression in HPCs was calculated with a previously described method [75]. GO analysis was performed using clusterProfiler. Cell–cell communication was analyzed with CellChat. Data visualization used Seurat, Scanpy, ggplot2, and UpSetR.