Hashimoto’s thyroiditis (HT) is a common organ-specific autoimmune disorder and the leading cause of hypothyroidism. It is characterized by dense lymphoid infiltration—including lymphocytes, plasma cells, macrophages, and dendritic cells (DCs)—into the thyroid gland, along with elevated serum levels of anti-thyroid peroxidase antibody (TPOAb) and anti-thyroglobulin antibody (TgAb). The global prevalence of HT is influenced by socioeconomic, environmental, and genetic factors, with an estimated prevalence of approximately 4.8%–25.8% in women and 0.9%–7.9% in men (1, 2). Additionally, HT is linked to increased risks of cardiovascular disease (3), malignancy (4), and pregnancy-related complications (5). The pathogenesis of HT involves genetic susceptibility, environmental triggers, and epigenetic dysregulation (6–8), which together disrupt immune homeostasis in the thyroid microenvironment.

In the process of thyroid immune microenvironment disruption, the initiation of the innate immune system involves multiple cell types. Among these, thyroid follicular cells (TFCs) have recently been recognized not merely as passive targets of autoimmune attack, but as active participants in inflammation. By sensing pathogen-associated molecular patterns (PAMPs) and damage-associated molecular patterns (DAMPs) through pattern recognition receptors (PRRs) such as Toll-like receptors (TLRs), TFCs can actively release inflammatory mediators and initiate local innate immune responses (9), thereby potentially priming the subsequent recruitment and activation of immune cells. Among the innate immune cells involved, dysregulated macrophage polarization plays a pivotal role. Studies indicate that excessive iodine intake promotes a shift toward the pro-inflammatory M1 phenotype via enhanced hexokinase 3 (HK3) expression (10). Furthermore, single-cell RNA sequencing has identified a distinct inflammatory macrophage subset in HT characterized by high interleukin-1β (IL-1β) expression, which together with DCs shapes a localized pro-inflammatory milieu (11). Although the precise mechanisms remain incompletely understood, TFCs may influence macrophage polarization via secreted factors or direct cellular interactions.

Moreover, a cytokine network driven by interleukin-17 (IL-17), interleukin-21 (IL-21), and inflammasomes (e.g., NOD-, LRR- and pyrin domain-containing protein 3 (NLRP3)) not only amplifies local inflammation but also directly damages follicular cells through processes such as pyroptosis (12–14). Thyroid stromal cells, such as C–C motif chemokine ligand 21-positive (CCL21⁺) fibroblasts, further contribute by promoting lymphocyte migration and the formation of tertiary lymphoid structures (11). This TFC-driven inflammatory microenvironment ultimately leads to pathological damage primarily mediated by adaptive immune dysregulation (15, 16), including T helper 1 (Th1) cell-mediated apoptosis coupled with regulatory T (Treg) cell functional deficiency (17), T helper 17 (Th17) cell hyperactivation (18), and cytotoxic CD8⁺ T cell activity (19). Nevertheless, the precise molecular mechanisms by which TFCs initiate and sustain this cascade—particularly their interplay with macrophage polarization—remain a crucial unresolved question requiring in-depth investigation.

This study aimed to systematically decipher the transcriptomic reprogramming of thyroid follicular cells in Hashimoto’s thyroiditis and its role in driving the immunodestructive process. Current knowledge has not yet fully elucidated the core transcriptional regulatory network through which TFCs actively initiate and maintain tissue-specific immune injury. We hypothesize that a set of hub genes plays a key role in HT pathogenesis by regulating the crosstalk between TFCs and immune cells, particularly M1 macrophages. To test this hypothesis, we employed an integrated strategy combining bioinformatics analysis with experimental validation. Our approach involved screening differentially expressed genes from transcriptomic data, constructing interaction networks to identify key hubs, analyzing their correlation with immune microenvironment features, and subsequently validating the findings in independent clinical samples. We sought to uncover the core molecular mechanisms by which TFCs mediate immune destruction, offering novel insights into the pathogenesis of HT.

We analyzed bulk RNA−seq data (GSE165724) and validated the findings using independent clinical specimens. The overall study design is summarized in Figure 1.

Hashimoto’s thyroiditis involves progressive destruction of TFCs and widespread immune infiltration, yet the intercellular regulatory networks remain poorly defined. Our study revealed a disrupted immune microenvironment in HT, characterized by abnormal activation of naïve B cells, Tfh cells, and M1 macrophages, along with up-regulated pro−inflammatory pathways. By integrating transcriptomics with clinical validation, we pinpointed CITED2 as a central hub gene that is consistently down−regulated in HT TFCs at both mRNA and protein levels. This deficiency strongly correlated with increased M1 macrophage infiltration and inversely with serological markers of disease severity (TPOAb, TgAb). Thus, our findings nominate CITED2 not merely as a candidate biomarker, but as a critical molecular nexus potentially bridging intrinsic TFC dysfunction to innate (M1 macrophage−driven) and adaptive immune dysregulation, thereby reinforcing the concept of TFC−initiated immune remodeling in HT pathogenesis.

Notably, our clinical validation demonstrated significant upregulation of both classical M1 (CD80, CD86) and M2 (CD163, MRC1) macrophage markers in HT tissues. This suggests the thyroid microenvironment in HT harbors a mixed population of activated macrophages, aligning with the known plasticity and phenotypic heterogeneity of macrophages in chronic inflammation (20). Critically, however, correlation analysis revealed distinct patterns: M1 markers (CD80/CD86) showed strong positive correlations with serum autoantibodies (TPOAb, TgAb) and IFNG expression, and a negative correlation with CITED2. In contrast, the M2 marker CD163 had weaker associations, and MRC1 showed no significant correlations with these parameters. This pattern underscores that the M1−like macrophage subset likely holds higher pathological relevance in driving HT immunopathology. M1 macrophages, known for secreting pro−inflammatory cytokines (e.g., TNF−α, IL−1β) and possessing antigen−presenting capacity, may directly participate in thyroid follicle destruction and amplify autoimmunity (11, 12). The concomitant increase in M2 cells may represent a compensatory regulatory or repair response, but their function in HT appears insufficient to counterbalance the inflammatory drive.

The down−regulation of CITED2 showed a distinct immune correlation profile, being negatively associated with M1 macrophages, naïve B cells, and memory B cells, but positively with monocytes and eosinophils. This complex signature raises key mechanistic questions: Does CITED2 modulate immune cells indirectly via paracrine signals from TFCs? Does its loss directly alter the immunogenic profile of TFCs? Could CITED2 contribute to immune dysregulation by rewiring TFC metabolism?

The capacity of CITED2 to link TFC dysfunction with immune dysregulation may stem from its evolutionarily conserved role as a transcriptional modulator. It fine−tunes gene expression by dynamically regulating the coactivators CREB−binding protein (CBP) and p300 (21, 22). Context dictates its activity: during development, CITED2 acts as a coactivator for factors such as transcription factor AP−2 (TFAP2) and suppressor of mothers against decapentaplegic homolog 2/3 (SMAD2/3) (23, 24), whereas in inflammatory settings it exerts potent transcriptional repression by competitively inhibiting NF−κB, hypoxia−inducible factor 1−alpha (HIF−1α), signal transducer and activator of transcription 1 (STAT1), and interferon regulatory factor 1 (IRF1) pathways (25–28). This anti−inflammatory role is supported by evidence that myeloid−specific Cited2knockout mice show increased susceptibility to endotoxin shock and atherosclerosis (27, 28). Therefore, CITED2 down−regulation in HT likely releases a “molecular brake” on pro−inflammatory pathways such as NF−κB and STAT1/IRF1, potentially creating a permissive environment for M1 macrophage polarization and subsequent immune activation.

Building on this molecular framework and our correlative findings, we propose several hypothetical yet plausible mechanistic models through which CITED2 deficiency could drive HT immunopathology. These models, which remain to be experimentally validated, aim to connect the observed CITED2 down-regulation with established pathogenic features of HT:

First, we speculate that CITED2−deficient TFCs may secrete specific chemokines that recruit and polarize monocytes toward an M1 phenotype. Chemokines such as C−X−C motif chemokine ligand 9 (CXCL9), CXCL10, and C−C motif chemokine ligand 2 (CCL2) are elevated in HT and correlate with disease activity (29, 30). As a known inhibitor of NF−κB and STAT1/IRF1 (25, 28), it is plausible that reduced CITED2 could enhance activation of these pathways, thereby elevating the expression of chemokines such as CXCL10. If experimentally confirmed in TFCs, this mechanism could establish a chemokine-driven loop that recruits monocytes and, in concert with local IFN−γ and TNF−α, promotes their M1 polarization, potentially sustaining an inflammatory microenvironment.

Second, down−regulation of CITED2 might directly remodel the immune−related gene network in TFCs. A testable hypothesis is that this could confer TFCs with aberrant antigen-presenting capacity. Spatial transcriptomics has shown elevated CD74 and macrophage migration inhibitory factor (MIF) expression in HT TFCs, colocalizing with immune activation (31). CITED2 may influence major histocompatibility complex class II (MHC−II) expression; thus, its loss might be permissive for the aberrant HLA−DR display on TFCs observed in HT—enabling autoantigen presentation to CD4⁺ T cells (19, 32). Concurrently, increased MIF might engage herpesvirus entry mediator (HVEM) to activate NF−κB and promote Th17 differentiation (33, 34), linking TFC−intrinsic changes to adaptive immune imbalance.

Third, given CITED2’s emerging role in metabolic regulation, its deficiency might induce “metabolic–immune” dysregulation in TFCs. Based on reports that Th1 cytokines can suppress sirtuin 1 (SIRT1), up−regulating HIF−1α and vascular endothelial growth factor A (VEGF−A) in TFCs (35). Since CITED2 antagonizes HIF−1α, its loss could trigger a similar metabolic shift, potentially creating a feed−forward loop that amplifies inflammatory and immune responses.

While these models implicate CITED2 deficiency in promoting immune dysregulation, they also highlight broader gaps in our understanding. First, the specific gene networks controlled by CITED2 in TFCs remain poorly defined. Beyond known inflammatory pathways, does it also regulate TFC homeostasis via peroxisome proliferator−activated receptor gamma (PPAR−γ) (36) or metabolic reprogramming akin to its role in hepatic gluconeogenesis (37)? Second, the upstream drivers of CITED2 down−regulation in HT are unknown. Potential contributors include genetic susceptibility, epigenetic modifications, or feedback inhibition by local inflammatory cytokines (e.g., IFN−γ, TNF−α). Third, CITED2 function exhibits marked cell−type specificity, acting as a metastasis promoter in some cancers (38–41) while being essential for hematopoietic stem cell homeostasis (42). Resolving these questions is crucial for fully elucidating CITED2’s role in HT.

Several limitations of this study should be considered. First, the immune cell proportions provided by CIBERSORT are bioinformatic estimates based on gene expression signatures, not direct cellular counts. Second, the discovery analysis relied on a public dataset where control tissues were from patients with laryngeal cancer, and tissues of interest were adjacent to PTC. Although we rigorously subgrouped samples based on lymphocytic thyroiditis, the potential influence of a PTC background is a recognized confounder. To address this, we established an independent validation cohort comprising patients definitively diagnosed with HT based on stringent serological, ultrasonographic, and histopathological criteria. The successful validation of CITED2 downregulation and its associations in this cohort strengthens the link of our findings to HT pathogenesis.

In summary, this study identifies CITED2 as a central molecular node linking TFC dysfunction to immune microenvironment dysregulation in HT. Future investigations should prioritize (1): using TFC−specific models to define CITED2’s cell−autonomous and paracrine effects on immune cells, testing the chemokine- and antigen-presentation-related hypotheses (2); employing genome−wide approaches to map its downstream target genes and upstream regulatory networks in TFCs; and (3) dissecting its potential role as a “metabolic–immune” integrator in TFCs. These efforts will clarify how follicular cell defects sustain immune pathology in HT and may reveal novel therapeutic targets.