背景：

甲状旁腺癌（Parathyroid carcinoma，PC）是一种少见的内分泌恶性肿瘤，临床表现为甲状旁腺激素（Parathyroid hormone，PTH）过度分泌和难以控制的高钙血症。根治性手术切除是当前唯一的治疗手段。因其高侵袭性和转移性，相当部分的PC患者在确诊时已经发生肿瘤远处转移，手术治疗效果有限，大部分患者术后会反复复发，预后较差。细胞分裂周期蛋白73（Cell division cycle protein 73，CDC73）的抑癌活性的缺失是PC最常见的分子发病机制。然而，国内外尚未建立标准化的实验模型，现有研究主要聚焦于甲状旁腺肿瘤的基因变异和临床特征，CDC73基因驱动的甲状旁腺肿瘤微环境（Tumor microenvironment，TME）变化及其调控的信号通路网络仍不明确，药物研发进展困难。

第一部分 基于空间单细胞转录组学分析
CDC73基因突变驱动的甲状旁腺肿瘤微环境探究

目的：

系统解析甲状旁腺肿瘤TME的细胞组成及空间分布特征，探讨甲状旁腺腺瘤（Parathyroid adenoma，PA）与PC及CDC73基因野生型（CDC73 Wild-Type，CDC73-WT）与CDC73基因突变型（CDC73 Mutant，CDC73-MUT）两种PC亚型在TME中的差异化特征，为解析PC发生路径以及CDC73突变驱动的肿瘤发展机制提供新依据。

方法：

收集于中国医学科学院北京协和医院基本外科接受甲状旁腺肿瘤切除术患者的病例资料及手术切除标本。研究共纳入3例PA（PA组）和12例PC（总PC组）标本。其中，总PC组包括4例CDC73-WT型PC（NPC组）及8例CDC73-MUT 型PC（PC组）标本。所有标本制作石蜡切片后构建组织微阵列，根据苏木素-伊红染色结果共计圈选400个扫描视野。使用CosMx^TM SMI平台对甲状旁腺肿瘤进行空间单细胞转录组学分析，通过单细胞聚类及注释、空间分布特征解析、细胞通讯分析、非负矩阵分解基因表达程序（Molecular Program，MP）分析和组间差异分析等方法，解析不同亚型的甲状旁腺肿瘤TME的差异特征。实验数据两组间比较采用 t检验。P<0.05表示差异具有统计学意义。

结果：

甲状旁腺肿瘤样本共聚类细胞26簇并注释为7类；

在15例甲状旁腺肿瘤样本中，共鉴定出26个细胞簇，通过Marker基因注释共识别出7种细胞类型，包括上皮细胞、内皮细胞、成纤维细胞、平滑肌细胞、T细胞、浆细胞和巨噬细胞。上皮细胞可以进一步分为内分泌上皮细胞-PA（Endocrine epithelial-PA，EP-PA），内分泌上皮细胞-NPC（Endocrine epithelial-NPC，EP-NPC）和内分泌上皮细胞-PC（Endocrine epithelial-PC，EP-PC）三种亚型。

甲状旁腺癌和CDC73-MUT型甲状旁腺癌中基因表达上调最显著的分别是H3C2和OPTN；

相较于PA组，EP-PC在总PC组中的比例显著增加，基因表达上调最显著的分别是H3C2。相较于CDC73-WT型PC中的EP-NPC，PC组中CDC73-MUT型PC的EP-PC细胞中共有2个基因表达上调，分别为OPTN和NRXN3，其中差异最显著的是OPTN（P=0.0003）。通路富集表明，PC中可能通过影响mRNA加工和剪接过程来调控EP-PC细胞高表达致癌基因的转录后修饰。

甲状旁腺癌中成纤维细胞比例增高，CDC73-MUT型PC中COL1A1基因高聚集表达；

相较于PA组，成纤维细胞在总PC组中比例显著增加，而且COL1A1_ITGB1的受配体互作表现出最强的相关性，富集提示与细胞外基质（Extracellular matrix，ECM）-上皮细胞的互作相关。对比CDC73-WT型PC，CDC73-MUT型PC中COL1A1基因高聚集分布表达，富集通路提示主要为ECM重塑。

甲状旁腺癌中巨噬细胞比例增高，且EP-PC细胞与巨噬细胞之间的空间接近程度显著增加；

相较于PA组，巨噬细胞在总PC组中比例显著增加，而且细胞通讯显示EP-PC细胞与巨噬细胞的空间接近程度显著增加。而EP-PA与巨噬细胞及T细胞之间的空间距离则显著增大。EP-NPC与内皮细胞以及平滑肌细胞的空间距离则较远。

基因表达程序的富集提示甲状旁腺癌与ECM受体密切相关；

甲状旁腺肿瘤中MP共分为8组，其中MP3与ECM受体相互作用相关，调控肿瘤细胞发育、分化和成熟过程。

结论：

PC中成纤维细胞比例增高，COL1A1_ITGB1互作提示ECM-上皮细胞互作关键；

PC中巨噬细胞比例增高，EP-PC与巨噬细胞共定位增强，靶向TAMs可能是有效免疫治疗策略；

CDC73-MUT型PC中COL1A1高表达提示ECM重塑可能是PC恶性转化的标志。

第二部分 基于糖酵解代谢调控
CDC73基因突变型甲状旁腺肿瘤的药物治疗探究

目的：

构建细胞、类器官及基因工程大鼠动物模型，通过高通量药物筛选CDC73-MUT型甲状旁腺肿瘤的候选治疗药物，并将从分子调控角度探究CDC73基因驱动甲状旁腺肿瘤发展及药物作用机制。

方法：

通过CDC73-MUT型甲状旁腺肿瘤研究模型，结合转录组学、代谢组学及蛋白组学结果，筛选CDC73调控的潜在靶点及参与的信号通路，检测CDC73在PC进展中的作用及机制。随后，利用CDC73突变型肿瘤细胞进行2863种小分子化合物的高通量筛选，选择候选治疗药物。通过体外细胞实验探究其对CDC73基因突变型甲状旁腺肿瘤的治疗作用机制。同时，通过脂质体包载运送技术，提高候选药物的运载效能及抗肿瘤作用。

结果：

细胞、类器官及基因工程大鼠动物模型能有效模拟甲状旁腺肿瘤特征；

通过慢病毒构建CDC73基因敲低（CDC73 knockdown，CDC73-KD）的甲状旁腺肿瘤细胞（Parathyroid neoplasm cell，PTNC），CCK8和划痕实验结果显示，细胞的增殖和迁移能力明显增强。RT-qPCR和Western Blot结果证明其抗凋亡能力增强。甲状旁腺肿瘤类器官模型能够持续分泌甲状旁腺激素，保留了甲状旁腺肿瘤的内分泌功能。基因工程大鼠的原位甲状旁腺组织体积增大；苏木素-伊红染色和免疫组化染色分别表明，细胞排列紊乱，呈肿瘤病变表现；CDC73调控的parafibromin蛋白表达部分缺失。三种模型均能有效模型甲状旁腺肿瘤特征。

CDC73-MUT型PTNC中MYC及HK2表达增加，糖酵解能力增强；

结合转录组学、代谢组学及蛋白组学结果，CDC73-MUT型PC中MYC表达升高，体内实验进一步证实CDC73-MUT型PTNC中MYC及糖酵解关键基因HK2表达上升。

高通量筛选出染料木黄酮（Genistein）通过抑制MYC/HK2轴，诱导CDC73-MUT型PTNC凋亡；

细胞增殖活性实验筛选得出5 µM Genistein作用48 h，细胞增殖率明显下降，凋亡率增加。MYC和糖酵解关键基因HK2表达下降。同时，丙酮酸、乳酸生成减少，糖酵解基础能力和储备能力下降。联合使用MYC抑制剂发现，上述抑制作用显著增强。

脂质体包载技术优化了Genistein的药物递送，对正常甲状旁腺细胞无明显毒副作用。

在相同的5 μM作用浓度下，处理48 h，脂质体-Genistein抑制肿瘤细胞增殖和侵袭的能力增强，同时提高了诱导肿瘤细胞凋亡能力。同时，脂质体-Genistein提高了对CDC73-MUT型PTNC的糖酵解抑制能力。此外，同样药物作用浓度下，脂质体-Genistein对正常甲状旁腺细胞的增殖活力没有影响。

结论：

成功构建的甲状旁腺肿瘤模型为分子机制及药物研究提供了有力工具；

Genistein通过抑制MYC/HK2轴调节糖酵解通路，增强CDC73突变型PTNC凋亡能力；

脂质体包载技术优化了Genistein的药物递送，显著增强其抗肿瘤效果。

Background：

Parathyroid carcinoma (PC) is a rare endocrine malignancy characterized by excessive secretion of parathyroid hormone (PTH) and refractory hypercalcemia. Radical surgical resection remains the only curative treatment. Due to its high invasiveness and metastatic potential, a considerable proportion of PC patients present with distant metastases at the time of diagnosis, limiting the efficacy of surgical outcomes. Most patients experience tumor recurrence postoperatively, resulting in poor prognosis. Loss of tumor suppressor function of cell division cycle protein 73 (CDC73) represents the most common molecular mechanism in PC. However, no standardized experimental models of PC have yet to be established. Current research primarily focuses on genetic alterations and clinical characteristics of parathyroid tumors, while the impact of the CDC73 mutation-driven tumor microenvironment (TME) and its related signaling pathways remain unclear, creating significant challenges for drug development.

Part I: High-plex Transcriptomic Analysis of CDC73 Mutation Parathyroid Tumor at Subcellular Level via Spatial Molecular Imaging

Objective：

This study aims to systematically characterize the cellular composition and spatial distribution of the parathyroid tumor microenvironment. It further investigates the distinct characteristics of parathyroid adenoma (PA) and PC, as well as the differences between CDC73 wild-type (CDC73-WT) and CDC73 mutant (CDC73-MUT) PC subtypes in the TME. The findings will provide new insights into the pathogenesis of PC and the role of CDC73 mutations in PC progression.

Methods：

Clinical and pathological data were collected, along with surgical specimens, from patients undergoing parathyroid tumor resection at Peking Union Medical College Hospital. The study included 3 PA samples (PA group) and 12 PC samples (total PC group), which were further subdivided into 4 CDC73-WT PC samples (NPC group) and 8 CDC73-MUT PC samples (PC group). Tissue microarrays were constructed from formalin-fixed, paraffin-embedded samples, and 400 fields of view were selected based on hematoxylin and eosin staining. High-plex transcriptomic analysis was conducted using the spatial molecular imaging. Single-cell clustering and annotation, spatial distribution analysis, cell-cell communication analysis, non-negative matrix factorization-based molecular program (MP) analysis, and intergroup differential analysis were performed to characterize TME differences among tumor subtypes. Statistical comparisons were performed using t-test, considering P values < 0.05 as statistically significant.

Results：

A total of 26 cell clusters were identified in parathyroid tumor samples and annotated into 7 major cell types.

A total of 26 cell clusters were identified across 15 parathyroid tumor samples and classified into seven major cell types, including epithelial cells, endothelial cells, fibroblasts, smooth muscle cells, T cells, plasma cells, and macrophages. Epithelial cells were further subclassified into endocrine epithelial cells associated with PA (EP-PA), NPC (EP-NPC), and PC (EP-PC).

In PC and CDC73-MUT PC, the most significantly upregulated genes were H3C2 and OPTN, respectively.

The proportion of EP-PC cells was significantly higher in the total PC group compared to the PA group. The most upregulated genes was H3C2 in the PC group. Compared with CDC73-WT PC, two upregulated genes were OPTN and NRXN3, with OPTN being the most significantly differentially expressed (P=0.0003). Pathway enrichment analysis indicated that mRNA processing and splicing regulation modulate oncogenic gene transcription in EP-PC cells.

In PC, the proportion of fibroblasts is increased, and COL1A1 gene exhibits high expression in CDC73-MUT PC.

Fibroblasts were significantly more abundant in the total PC group than in the PA group. The COL1A1_ITGB1 ligand-receptor interaction showed the strongest correlation, suggesting a crucial role for extracellular matrix (ECM)-epithelial cell interactions in PC. Notably, COL1A1 exhibited a highly localized expression pattern in CDC73-MUT PC, with enriched pathways suggesting ECM remodeling.

In PC, the proportion of macrophages is increased, and the spatial proximity between EP-PC cells and macrophages is significantly enhanced.

Compared to PA, the total PC group exhibited a higher proportion of macrophages. Cell-cell communication analysis revealed a significant increase in spatial proximity between EP-PC cells and macrophages, while EP-PA cells showed increased spatial separation from both macrophages and T cells. Additionally, EP-NPC cells demonstrated a greater spatial distance from endothelial and smooth muscle cells.

Enrichment of MPs suggests that PC is closely associated with ECM receptors.

Eight MPs were identified in parathyroid neoplasms, with MP3 being significantly associated with ECM receptors, potentially regulating tumor cell development, differentiation, and maturation.

Conclusions：

PC exhibits an increased proportion of fibroblasts, with COL1A1_ITGB1 interactions suggesting a critical role for ECM-epithelial cell crosstalk.

Macrophage enrichment and enhanced spatial colocalization between EP-PC cells and macrophages suggest that targeting tumor-associated macrophages may be a promising immunotherapeutic strategy.

High COL1A1 expression in CDC73-MUT PC may indicate ECM remodeling as a hallmark of malignant transformation.

Part II: Drug Therapy Targeting Glycolytic Metabolism in CDC73-Mutant Parathyroid Tumor

Objective：

To establish cell, organoid, and genetically engineered rat models for CDC73-MUT PC, perform high-throughput screening for candidate therapeutic compounds, and explore the molecular mechanisms of CDC73-driven tumor progression and drug action.

Methods：

CDC73-MUT PC models, combined with transcriptomics, metabolomics, and proteomics analyses, were employed to identify CDC73-regulated molecular targets and pathways. High-throughput screening of 2,863 small-molecule compounds was performed on CDC73-MUT tumor cells to select potential therapeutic agents. The effects and mechanisms of the selected drugs were evaluated in vitro. Liposomal encapsulation was used to enhance drug delivery and antitumor efficacy.

Results：

Cell, organoid, and rat animal models effectively simulate the characteristics of parathyroid tumors.

CDC73-knockdown (CDC73-KD) parathyroid neoplasm cells (PTNC) exhibited enhanced proliferation, migration, and anti-apoptotic properties. Organoid maintained endocrine functionality, while genetically engineered rats developed parathyroid gland hypertrophy, displaying histological features of malignancy and partial loss of parafibromin expression.

In CDC73-MUT PTNC, the expression of MYC and HK2 is increased, enhancing glycolytic capacity.

Multi-omics analysis revealed the upregulation of MYC in CDC73-MUT PC. Further validation confirmed increased MYC and HK2 expression in CDC73-MUT PTNC, indicating enhanced glycolytic activity.

High-throughput screening identified Genistein as a compound that induces apoptosis in CDC73-MUT PTNC by inhibiting the MYC/HK2 axis.

Genistein (5 µM, 48 h) significantly reduced cell proliferation and increased apoptosis. MYC and HK2 expression were decreased, along with reduced pyruvate and lactate production, leading to impaired glycolysis. The inhibitory effect was further enhanced with MYC inhibitor co-treatment.

Liposomal encapsulation technology optimized the drug delivery of Gen, with no significant toxicity to normal parathyroid cells.

Liposomal-genistein enhanced tumor cell suppression while sparing normal parathyroid cells, improving glycolysis inhibition and inducing apoptosis.

Conclusions：

The established PC models provide robust platforms for molecular and therapeutic research.

Genistein suppresses glycolysis and induces apoptosis in CDC73-MUT PTNC by inhibiting the MYC/HK2 axis.

Liposomal encapsulation enhances the antitumor efficacy of genistein, offering a promising therapeutic approach.