Skip to main content
NCBI home page
Scientific Reports logoLink to Scientific Reports
. 2026 Feb 26;16:8682. doi: 10.1038/s41598-026-41927-z

Visualizing malignant progression: in situ CD109-based spatial immunofluorescence assay delineates papillary to anaplastic thyroid carcinoma transformation within the tumor microenvironment

Tomoko Cohen 1,2, Keiji Suzuki 3, Katsuya Matsuda 1,4, Hirokazu Kurohama 1, Yuki Matsuoka 1, Mayu Ueda 1,2,4, Shinya Satoh 5, Hisakazu Shindo 5, Hiroyuki Yamashita 5, Atsushi Kawakami 2, Masahiro Nakashima 1,4,✉
PMCID: PMC12979697  PMID: 41748865

Abstract

Anaplastic thyroid carcinoma (ATC) is the rarest and most aggressive subtype of thyroid cancer, and considered to arise from differentiated thyroid carcinoma, however, the underlying molecular processes remain largely unknown. Using CD109 as a malignant marker, we analyzed an ATC case containing a papillary thyroid carcinoma (PTC) component. Our newly developed spatial immunofluorescence (SPI) assay, which enabled the differential expression of CD109 and CK8/18, a PTC marker, demonstrated gradual and completely opposed changes at the boundary between ATC and PTC components. Similar specific expression patterns were observed in E-cadherin, vimentin, PCNA, αSMA, Iba-1, collagen (COL) III/VI, TGFβ1-induced (TGFBI), active Yes-associated protein, periostin, and S100. The zonal and reciprocal transitions between PTC and ATC markers suggested that anaplastic transformation was not merely the result of clonal expansion, but rather it was executed by ATC-specific tumor microenvironment (ATC-TME) that recruited more Iba-1- and S100-positive macrophages, along with unique ATC-cancer-associated fibroblasts (CAFs), which deposited more COL III/VI and TGFBI. We proposed that expansion of ATC-CAFs caused extracellular matrix stiffening and compromised PTC cells, thereby inducing necroptosis and S100 release. This process simultaneously promoted the epithelial–mesenchymal transition in PTC cells and selected pre-existing PTC cells harboring additional gene mutations sufficient for anaplastic transformation.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-41927-z.

Keywords: Anaplastic thyroid cancer, Spatial immunofluorescence assay, CD109, Tumor microenvironment, Matrix stiffening

Subject terms: Biomarkers, Cancer, Cell biology, Oncology

Introduction

Anaplastic thyroid carcinoma (ATC) is the rarest subtype of thyroid cancer, accounting for 1.1 to 4% of all thyroid cancers1–4. The prognosis for ATC remains extremely poor and it has been reported that average survival is 3 months to 1 year, and 5-year relative survival rate is 11 to 14%3,5. Histopathological analyses have revealed that up to 70% of ATC cases involve coexisting regions of differentiated thyroid cancer (DTC), suggesting that ATC can arise from pre-existing DTC through anaplastic transformation6–9.

Although the BRAF and RAS gene mutations are key drivers in both DTC and ATC, ATC is characterized by a high prevalence of additional genetic changes8,10–12. For example, TP53 mutation was commonly reported with frequencies range from 48% to 73%13,14. Importantly, a recent study has demonstrated that in ATC cases with coexisting papillary thyroid carcinoma (PTC) components, TP53 mutations were detected with similar frequencies between the ATC and PTC components, suggesting that PTC does not always undergo anaplastic transformation even though mutation in the TP53 gene is taken place7. It should be emphasized that the shared mutational profiles of the ATC and DTC components completely unsupported the prevailing hypothesis that ATC arises through the clonal expansion of cells with additional driver mutations.

Considering that ATC obviously evolves from pre-existing DTC, we hypothesized that certain tumor microenvironment (TME) promotes the selection of pre-existing mutated PTC cells, followed by progressive dedifferentiation of PTC into ATC. In a recent transcriptomic analysis, ATC had a unique TME, characterized by high abundance of immune cells and fibroblasts15–18. Thus, ATC-specific TME is highly likely to promote the anaplastic transformation of PTC, and if so, a gradual and zonal transformation from PTC to ATC would be detectable within the same histological samples, and demonstrating this is the main objective of our study.

In order to detect anaplastic transformation of PTC, it is essential to apply a unique and sensitive marker. In the present study, we identified CD109, as a malignant marker. CD109 is a glycosylphosphatidylinositol (GPI)-anchored glycoprotein that has gained increasing attention in recent years. It is expressed in multiple cell types, and CD109 expression and tumor grades vary among cancer types, although, no reports have described CD109 expression in thyroid cancer so far19.

To demonstrate gradual and zonal transformation, it is indispensable to perform spatial analysis. So far, spatial transcriptomic (ST) analysis has been established, and recent study, using ST analysis, reported mRNA expression profiling in separated areas within a thyroid tissue20. Other ST studies have revealed distinct gene expression signatures and spatial heterogeneity among normal, PTC, and ATC regions, highlighting the transcriptional shifts that accompany anaplastic transformation21,22. However, these were analyzed using a transcriptome-based assay without any protein expression information within individual cells.

Therefore, we developed a novel and innovative spatial immunofluorescence (SPI) assay that enabled quantitative measurement of the fluorescence intensity of multiple markers in each defined regions of interest (ROIs) using image analyzing application. Our novel SPI assay provided clear evidence supporting a gradual, zonal transformation of PTC into ATC. From our results, we proposed a whole new model of anaplastic transformation of PTC, which was governed by an ATC-specific TME, characterized by the expansion of ATC-specific cancer-associated fibroblasts (CAFs), that reorganized matrix composition of TME enforcing matrix stiffness. It inevitably resulted in cell death in the PTC component. Thus, pre-existing mutated PTC cells were selected, and the epithelial–mesenchymal transition (EMT) of selected PTC cells was promoted through the inflammatory ATC-TME. Our study provides in situ visualization of protein expression dynamics with spatial resolution and mechanistic insights associated with anaplastic transformation.

Results

Histological findings and CD109 expression patterns

Hematoxylin and eosin (HE) staining of the tumor is shown in Fig. 1A. The ATC area (Fig. 1Aa-Ac and upper part of Ad) exhibited diverse cellular morphologies (Fig. 1A lower panel). Spindle-shaped tumor cells were arranged in a storiform pattern in Fig. 1Aa and Fig. 1Ab along with collagen fibers in the stroma, whereas plump tumor cells formed small solid nests (Fig. 1Ac). The lower part of Fig. 1Ad and Fig. 1Ae indicate the PTC area, together with the normal follicles (Fig. 1Af).

Fig. 1.

Fig. 1

Open in a new tab

Representative histopathological findings of a case of anaplastic thyroid carcinoma (ATC) with a papillary thyroid carcinoma (PTC) component. (A) Hematoxylin and eosin (HE) staining. (B) Immunohistochemistry of CD109 using diaminobenzidine (DAB). (a-c) ATC component. (d) Border between ATC (upper part) and PTC (lower part) components. (e) PTC component. f Normal thyroid follicles. Scale bars: 2000 μm in (A) and (B), 100 μm in (a-f).

Immunohistochemistry using anti-CD109 antibody with diaminobenzidine (DAB) detection (Fig. 1B) revealed cytoplasmic expression of CD109 in ATC cells, whereas the PTC component with papillary structure and normal thyroid follicles showed no detectable staining (Fig. 1 lower part of Bd, Fig. 1Be and Bf). It should be noted that the PTC component with a follicular structure showed CD109 expression on the apical membrane, despite the absence of cytoplasmic staining (Supplementary Fig. 1a, b).

CD109-based SPI assay reveals gradual transition from PTC to ATC

As shown in Fig. 2, a CK8/18-enriched PTC area and CD109-enriched ATC area were clearly distinguished. Notably, in the transitional zone (white dashed circle in Fig. 2), CD109 and CK8/18 were co-expressed, resulting in yellow to orange fluorescence signals.

Fig. 2.

Fig. 2

Open in a new tab

Multiplex immunofluorescence image showing staining for CD109 (red) and CK8/18 (green). The red circle marks the main anaplastic thyroid carcinoma (ATC) area, the green circle marks the papillary thyroid carcinoma (PTC) area, and the white dashed circle indicates the border between the two areas. Scale bar: 2000 μm.

A CD109-based SPI assay was performed where the mean fluorescence intensity (FL) value ratio of CD109 to CK8/18 (CD109 FL/CK8/18 FL) was calculated in each present ROI (Fig. 3). This normalization was applied to accurately evaluate CD109 expression per CK8/18-positive tumor cell, as the number and volume of tumor cell cross-sections varied across tissue slices. The main ATC area was marked by red circles, indicating the ratio was greater than 2.5. In contrast, the PTC area was defined by green circles, where the ratio was less than 0.4. It should be noted that in some PTC cells, CD109 was localized to the apical membrane (green circles with red borders), which represents the expression pattern of CD109 in some PTC (Supplementary Fig. 1b, c). Several transitional zones were simultaneously detected at the boundary between the PTC and ATC areas, marked by yellow, orange, tangerine and pink circles (FL ratio between 0.4 and 2.5). To further characterize this transition more in detail, a high-density SPI assay was performed in Region 3 (Fig. 4, Supplementary animation). While HE-stained image could not provide detailed molecular patterns, this analysis revealed a complex and gradual transition from green PTC to red ATC areas, through intermediate zones.

Fig. 3.

Fig. 3

Open in a new tab

Results of the standard spatial immunofluorescence (SPI) assay. (a) Assigned color marks are overlaid on the hematoxylin and eosin (HE) staining. (b) The assigned color marks are overlaid on the multiplex immunofluorescence image stained for CD109 (red) and CK8/18 (green). The left white circle marks the main anaplastic thyroid carcinoma (ATC) area, and the right white circle marks the papillary thyroid carcinoma (PTC) area. Green marks with red outlines indicate the PTC area in which CD109 is apically expressed, as shown in Supplementary Fig. 1. The assigned color marks are shown in the legend. FL: fluorescence. Scale bars: 2000 μm.

Fig. 4.

Fig. 4

Open in a new tab

Result of the high-density SPI assay in Region 3. (a) Hematoxylin and eosin (HE) staining of the analyzed area. (b) Multiplex immunofluorescence image stained for CD109 (red) and CK8/18 (green). c Results of the high-density SPI assay, with assigned color tiles overlaid on the multiplex immunofluorescence image shown in (b). Assigned color tiles are shown in the legend. FL: fluorescence. Scale bars: 500 μm.

Assignment of five regions

As the SPI assay revealed intratumoral heterogeneity in the expansion patterns of CD109 and CK8/18, we selected five distinct regions based on their immunofluorescence profiles (Supplementary Fig. 2). Region 1 is a typical PTC area (CD109 negative and CK8/18 positive). Region 2 is a PTC area but expresses CD109. Region 3 corresponds to the transitional zone, which is primarily represented by the yellow markings in the SPI assay. Region 4 is an ATC area with strong CD109 expression and little or no CK8/18 expression. Region 5 represents an ATC area characterized by strong CD109 expression without CK8/18 expression.

Multiplex immunofluorescence staining in region 3

Region 3 was selected to evaluate several molecular changes occurring at the interface of PTC and ATC, because it contains both PTC and ATC components, enabling in situ comparison of molecular changes associated with anaplastic transformation. As shown in Fig. 5a, CD109-positive ATC and CK8/18-positive PTC areas coexisted within a few hundred micrometers. Line plot analysis (Fig. 5b) confirmed reciprocal change in the expression of these markers at the zonal interface. Similar transitional changes were observed in several markers. For example, E-cadherin, another epithelial marker involved in cell-to-cell attachment, showed strong expression in the PTC area, whereas it was weakly expressed in the ATC area, with a dot-like expression pattern (Fig. 5c and Supplementary Fig. 3a-c). Although vimentin is frequently used as an EMT marker, it is expressed in thyroid follicular cells. Accordingly, PTC cells expressed vimentin predominantly at the basal area, whereas in the ATC area, vimentin levels increased and showed diffusely cytoplasmic distribution (Fig. 5d). Throughout the tissue, vimentin-positive cells showed a greater abundance in the ATC area than in the PTC area (Supplementary Fig. 5).

Fig. 5.

Fig. 5

Open in a new tab

Multiplex immunofluorescence images and comparative line plot analysis. (a) Multiplex immunofluorescence image showing staining for CK8/18 (red) and CD109 (green). (b) Line plot analysis of CK8/18 (red) and CD109 (green) was performed along the white line shown in (a). The x-axis represents the distance (pixels), and the y-axis represents the fluorescence intensity. (c-n) Multiplex immunofluorescence images stained for (c) E-cadherin (red) and CD109 (green), (d) CK8/18 (red) and Vimentin (green), (e) PCNA (red) and CD109 (green), (f) αSMA (red) and CK8 (green), (g) Active YAP (red) and CD109 (green), (h) Collagen III (COL III) (red) and CK8/18 (green), (i) Collagen VI (COL VI) (red) and CK8/18 (green), (j) Iba-1 (red) and CK8 (green), (k) S100 (red) and CK8/18 (green), (l) S100 (red), (m) S100 (red) and Iba-1 (green) and (n) Periostin (POSTN) (red) and CK8 (green). The circle in (d) indicates Vimentin expression in anaplastic thyroid carcinoma (ATC), whereas the white arrow indicates Vimentin expression in papillary thyroid carcinoma (PTC). The arrow in (g) indicates cells co-expressing active YAP and CD109. The circles in (g) indicate active YAP-positive cancer-associated fibroblasts (CAFs). PTC: papillary thyroid carcinoma region. ATC: anaplastic thyroid carcinoma region. Scale bars: 200 μm in (a) and (d), 50 μm in (c) and 100 μm in (e-n).

PCNA, a cell proliferation marker, -positive cells were detected more frequently in ATC than in PTC (Fig. 5e). Quantitative analysis of PCNA-positive cells in Supplementary Fig. 4, demonstrated that the percentage of PCNA-positive cells was 16.6% in CD109-CK8/18 + cells, whereas it was 57.6% and 57.7% in CD109 + CK8/18- and CD109 + CK8/18 + cells, respectively, indicating higher proliferation in the ATC area. Interestingly, 41.9% of CD109-CK8/18- tumor cells were also PCNA-positive, indicating that proliferation had already begun prior to CD109 expression.

A similar gradual change was detected in the expression of αSMA, which was markedly higher in the ATC than in the PTC area, indicating that myofibroblastic CAFs were more abundant in the ATC area (Fig. 5f). The frequency of active Yes-associated protein (YAP)-positive cells in the ATC area were higher compared with the PTC area (Fig. 5g). Furthermore, CD109- and CK8/18- double-positive cells in between the two ATC and PTC areas showed strong expression of active YAP (Supplementary Fig. 3d, e).

Area-dependent expression was also observed for collagen III (COL III), collagen VI (COL VI), Iba-1, and S100, detected using an antibody recognizing both S100A and S100B. All of them were more abundant in the ATC area (Fig. 5h-m). Notably, S100 and Iba-1 expressions were partially colocalized in the ATC area (Fig. 5m).

Periostin (POSTN), which was strongly expressed in the cytoplasm of PTC cells, showed decreased expression in the ATC area with a dot-like staining pattern (Fig. 5n).

Multiplex immunofluorescence staining and quantitative analysis of each marker in region 1–5

The results of quantitative analysis are demonstrated in violin plots, which represent the distribution and FL values of each marker, along with the results of statistical analysis using an analysis of variance (ANOVA) (Fig. 6). The results of Bonferroni’s post-hoc tests are shown in Supplementary Table 1. A gradual increasing trend from Region 1 to 5 was observed for CD109, vimentin, PCNA, αSMA, active YAP, and Iba-1, while a gradual decreasing trend was obvious in CK8/18, E-cadherin, and POSTN. COL III and VI showed increasing trends, albeit with some exceptions. As observed in the ATC area, some PTC cells also showed positivity for active YAP, with a significantly higher percentage of active YAP-positive cells among CK8/18 and/or CD109 positive cells. Notably, it was significantly higher in CAFs in Region 4 and 5 (Supplementary Table 2c). TGFβ1-induced (TGFBI) demonstrated a unique expression pattern, exhibiting significantly higher expression in Region 4 and 5 (Supplementary Fig. 6).

Fig. 6.

Fig. 6

Open in a new tab

Violin plots showing the distribution and the mean of fluorescence intensity (FL) values of each marker (open circles indicate the mean) in Region 1–5. Markers are (a) CD109, (b) CK8/18, (c) E-cadherin, (d) Vimentin, (e) PCNA, (f) αSMA, (g) Active YAP, (h) Collagen III (COL III), (i) Collagen VI (COL VI), (j) TGFβ1-induced (TGFBI), (k) Iba-1, (l) S100 and (m) Periostin (POSTN). The x-axis represents the region (R), and the y-axis represents the FL values. Statistical significance was assessed using analysis of variance (ANOVA) with Bonferroni’s post-hoc test (Supplementary Table 1). R1–5: Region 1–Region 5.

As the FL values of Iba-1 showed a gradually increasing trend from Region 3 and reached a maximum in Region 5, double immunofluorescence staining for CD80 and CD163, together with Iba-1, was performed to distinguish M1- and M2- type macrophages. As shown in Supplementary Table 2b, CD163-positive cells were more abundant in Region 5, whereas CD80-positive cells were more in Region 2 (Supplementary Fig. 7a-c). The frequency of CD3-positive T cells showed another increasing trend from Region 1 to 5 (Supplementary Table 2a, Supplementary Fig. 7d). The frequency of neutrophils was the highest in Region 4 (Supplementary Fig. 8, Supplementary Table 2a). The TUNEL assay in Region 2 showed negative results, indicating no apoptosis (Supplementary Fig. 9a). In contrast, phospho-H2AX signals, which represented reactive oxygen species (ROS)-associated DNA double-strand breaks, were frequently observed and were spatially associated with widespread morphological abnormalities in papillary structures (Supplementary Fig. 9b). Notably, cells with phospho-H2AX signals were most prominent in Region 5. The highest level of S100 expression was observed in Region 4 and the lowest level in Region 5 (Fig. 6l). In particular, S100- and Iba-1- co-expressing cells were detected in Region 4 (Supplementary Fig. 10).

Spatial variations in the expression of several markers and the frequency of marker-positive cells across Region 1–5 were summarized in a tiled immunofluorescence intensity map (Supplementary Fig. 11). Several markers showed systematic changes from Region 1 to Region 5. In particular, markers related to CD109, extracellular matrix (ECM) remodeling, inflammation, and proliferation exhibited increasing trends, whereas differentiation and cell adhesion markers showed decreasing trends.

Principal component analysis (PCA) and heatmap of region 1–5

To confirm that the five regions have independent properties, we performed PCA and hierarchical clustering based on the FL values of CK8/18, E-cadherin, PCNA, Iba-1, CD109, αSMA, TGFBI, COLVI, vimentin and active YAP (Fig. 7). The PCA plot revealed a distinct separation of regions along the first principal component (PC1), with Region 1 and Region 5 occupying opposing ends of the spectrum. This distribution pattern clearly indicates a gradual and continuous alteration in the molecular profile from the PTC (Region 1) to ATC (Region 5) areas.

Fig. 7.

Fig. 7

Open in a new tab

Multivariate analysis of marker expression across assigned regions. (a) Principal component analysis (PCA) of fluorescence intensity (FL) values in Region 1–5. The first principal component (PC1) accounted for 57.5% of the total variance, and the second principal component (PC2) accounted for 11.7%. Colored dots indicate regions of interest (ROIs) of each region: Region 1 (orange), Region 2 (purple), Region 3 (green), Region 4 (blue), and Region 5 (red). (b) Heatmap of FL values for CK8/18, E-cadherin, PCNA, Iba-1, CD109, αSMA, TGFβ1-induced (TGFBI), collagen VI (COL VI), Vimentin and active YAP, arranged by hierarchical clustering. FL values are displayed on a scale from − 1 to + 4. Regions are color coded as follows: Region 1 (orange), Region 2 (purple), Region 3 (green), Region 4 (blue), and Region 5 (red).

Consistent with the PCA results, a heatmap of the hierarchical clustering of marker expression patterns revealed region-specific profiles (Fig. 7b). CK8/18 and E-cadherin were predominantly expressed in Region 1, whereas PCNA, Iba-1, CD109, αSMA, TGFBI, COLVI, vimentin, and active YAP were upregulated toward Region 5. Notably, CD109 showed increased expression from Region 3 to 5, overlapping with zones of reduced CK8/18 and elevated vimentin. The spatially ordered changes in expression clearly indicated that cell growth, EMT, immune activation, CAF activation and stromal remodeling occurred as sequential processes during anaplastic transformation.

COL III-fiber orientation assay

To examine the organization of ECM fibers, DiameterJ analysis was performed. The results revealed that COL III fibers in the PTC area were relatively thin and exhibited a uniform orientation, whereas the ATC area was characterized by thicker fibers with disorganized and multidirectional alignment (Supplementary Fig. 12, Supplementary Table 3), indicating that matrix was qualitatively different between PTC and ATC areas.

Discussion

In this study, we established a CD109-based SPI assay to visualize protein expression dynamics with spatial resolution associated with anaplastic transformation within the same tissue, and demonstrated the gradual and reciprocal transition of CD109 and CK8/18 levels at the interface of the PTC and ATC components. Anaplastic transformation has generally been thought to result from clonal expansion of PTC cells with additional mutations10,11. However, if clonal expansion had occurred, concentric circle–like patterns in the ATC area would be expected. In the present case, no such pattern was observed. Although further studies are required to determine whether these findings can be generalized, our results suggest that anaplastic transformation may not simply reflect clonal outgrowth of mutated cells.

Based upon the results obtained in this study, we propose a new model illustrating a hypothetical mechanism of anaplastic transformation from PTC to ATC and highlighting progressive alterations in the TME (Fig. 8). Initially, as discussed previously, PTC secretes factors such as interleukin-6 (IL-6), TGF-β1, and ROS, which activates neighboring normal fibroblasts giving rise to PTC-associated CAFs (PTC-CAFs)23,24. In this early stage (Region 1), reciprocal interactions between PTC-TME including PTC-CAFs and PTC cells promote tumor proliferation. Then, as PTC proliferates the PTC-TME expands, and further activation or possible genetic mutations in PTC-CAFs may lead to the emergence of ATC-CAFs (Region 2–3), which might be characterized by the elevated secretion of cytokines and chemokines (Region 4). In fact, we observed a marked accumulation of CAFs with enhanced αSMA expression and nuclear YAP localization, together with increased infiltration of Iba-1- and S100-positive macrophages in TME associated with ATC. These Iba-1/S100-positive macrophages have been known to trigger inflammation, which results in the secretion of cytokines, chemokines, matrix remodeling factors, and ROS (Region 3–4)25. Thus, it is demonstrated that the ATC-TME possesses not only distinct stromal and immune features, but also distinct ECM component and structure compared with the PTC-TME. As CAFs are the primary source of ECM proteins26–29, we observed that ATC-TME showed both elevated levels of ECM components such as COL III/VI and TGFBI. TGFBI, also known as βig-H3, a secreted ECM protein regulated by TGF-β signaling, has been reported to contribute to matrix architecture through interactions with COL VI and participates in malignant transformation by altering cell adhesion and stimulating inflammation30–33. As shown in a recent single-cell transcriptomic study, TGFBI is a potential novel histological and functional marker of anaplastic transformation6, indicating that these unique ECMs constitute a distinct ATC-TME.

Fig. 8.

Fig. 8

Open in a new tab

Hypothetical mechanism of anaplastic transformation. Papillary thyroid carcinoma (PTC) secretes factors such as interleukin-6 (IL-6), TGF-β1, and reactive oxygen species (ROS), which activate normal fibroblasts giving rise to PTC-associated cancer-associated fibroblasts (PTC-CAFs). As the PTC-tumor microenvironment (TME) expands, further activation or possible genetic mutations in PTC-CAFs may lead to the emergence of anaplastic thyroid carcinoma-CAFs (ATC-CAFs). Factors secreted by ATC-CAFs induce the production of extracellular matrix (ECM) components, resulting in increased matrix stiffness and PTC cells undergo compression and are eliminated by necroptosis, transforming into ATC cells. In this stiffened ATC-TME, both tumor cells and CAFs activate YAP, promoting further matrix stiffening. The accumulation of ATC-CAFs and overproduction of ECM components drive an inflammatory cascade, contributing to dedifferentiation of mainly mutated cells via epithelial–mesenchymal transition (EMT) and ultimately leading to the emergence of ATC. R1–5: Region1–Region 5. PTC: papillary thyroid carcinoma. ATC: anaplastic thyroid carcinoma. CAF: cancer-associated fibroblast. TME: tumor microenvironment. IL-6: interleukin-6. FAP: fibroblast activation protein. ROS: reactive oxygen species. ECM: extracellular matrix. EMT: epithelial–mesenchymal transition.

In addition, elevated deposition of COL III/VI and TGFBI might also contribute to increasing the matrix stiffness of ATC-TME by lysyl oxidase (LOX), which mediates the cross-linking of collagen and elastin26,27,34. Accordingly, YAP, which is activated by matrix stiffness, was significantly activated in both ATC and ATC-CAFs35. COL III orientation analysis indicated that Region 4 and 5 showed much broader distribution of fiber orientations than Region 1 and 2, resulting in a mesh-like structure. Furthermore, the distribution of fiber diameters in ATC-TME demonstrated higher kurtosis and skewness (Supplementary Table 3), suggesting a more heterogeneous and aberrantly remodeled COL III architecture. Thus, despite being composed of the same COL III fibers, the complex mesh-like fiber arrangement in Region 4 and 5 should contribute to the increased matrix stiffness.

Regarding the potential role of matrix stiffness in anaplastic transformation, we observed that regions with increased ECM stiffness were spatially associated with widespread and extensive morphological anomality in papillary structures, and frequent cell death in PTC areas. Interestingly, driver mutations such as the p53 mutation are known to confer resistance against cell death in PTC cells; therefore, p53-mutated cells may be readily selected under “compression-driven cell death” in ATC-TME36. In fact, the abundant expression of neutrophils and Iba-1 suggests that cell death occurred frequently in the PTC-ATC transition area (Region 4). Phospho-H2AX signals caused by ROS production during necroptosis, were commonly observed in region 4, implying that necroptosis was possibly induced by increased matrix stiffness (Region 4–5)37.

Finally, in Region 4–5, the accumulation of ATC-CAFs and the overproduction of ECM components were assumed to drive an inflammatory cascade, as previous studies have indicated6,15,16,38. Importantly, S100 proteins released from necroptotic PTC may not only mediates cell death but also stimulates the inflammatory environment in ATC-TME39. We observed S100- and Iba-1- co-expressing cells in Region 4, suggesting widespread phagocytic activity by macrophages. Previous studies have reported that phagocytosed S100 proteins function as damage-associated molecular pattern proteins (DAMPs) that mediate inflammatory responses40,41. Consistent with previous reports demonstrating elevated S100A8 and S100A9 expression in ATC42, S100 may also facilitate EMT of p53-mutated PTC cells43,44. Altogether, an ATC-specific tumor immune microenvironment (ATC-TIME) could be established, which is indispensable for anaplastic transformation. Interestingly, our results demonstrated that Region 5 was enriched with Iba-1-positive cells, and some exhibited an M2-like phenotype, which are known to secrete immunosuppressive cytokines, matrix-degrading enzymes, and angiogenic factors26,36. This immune contexture contributes to the dedifferentiation of mainly mutated cells via EMT; such processes ultimately lead to the emergence of ATC (Region 5).

In conclusion, we developed a CD109-based SPI assay and provided clear evidence supporting the gradual, zonal transformation of PTC into ATC. We also examined the spatial expressions of cell proliferation, EMT and TME markers, which illustrated the sequential evolution of anaplastic transformation. Despite these promising discoveries, this study had some limitations. First, multiple immunofluorescence analyses were performed in only one ATC case with a PTC component, which requires confirmation in several other cases. Second, as the molecular consequences of the differential localization of CD109 in PTC and ATC remain unclear, the precise role of CD109 in anaplastic transformation has yet to be elucidated. Further studies are required to address these issues. Third, we hypothesized that necroptosis was induced by “compression-driven cell death”, which should be confirmed using specific markers. Similarly, we did not examine the expression of soluble factors, such as cytokines and chemokines, which requires further analysis using specific antibodies.

Future studies should enable us a comprehensive elucidation of the in situ molecular dynamics underlying anaplastic transformation. These results may offer novel therapeutic strategies for preventing or delaying anaplastic transformations in thyroid cancer.

Materials and methods

Sample collection

The ATC sample used for this study was FFPE tissue which was surgically resected at the Yamashita Thyroid Hospital in Fukuoka, Japan. Final diagnosis was histologically confirmed at the Department of Tumor and Diagnostic Pathology, Nagasaki University, following the diagnostic criteria of the WHO Classification of Tumors of Endocrine Organs (5th edition)1,4.

This study was conducted in accordance with the guidelines of the Declaration of Helsinki and was approved by the Ethics Committee of Nagasaki University (Approval Date: April 28, 2023; No.: #15062617-6). Informed consent was obtained from the patient in this study through an opt-out process.

Immunohistochemical staining using DAB

Following deparaffinization, antigen retrieval was performed using a 30 min microwave treatment in a target retrieval solution (pH 9.0; Agilent Technologies, Santa Clara, CA, USA). Protein blocking and endogenous peroxidase blocking were performed using a NovoLink Polymer Detection System (RE7140-K; Leica Biosystems, Newcastle, UK) according to the manufacturer’s instructions. Tissue section was then incubated with anti-CD109 mouse monoclonal antibody (1:750; sc-271085; Santa Cruz Biotechnology, Dallas, TX, USA) for 1 h at 20℃ before DAB staining using the NovoLink system.

Multiplex immunofluorescence staining

Following deparaffinization, antigen retrieval was performed in a water bath at 95℃ for 30 min in the ethylenediamine tetraacetic acid (Tris-EDTA) buffer (pH 9.0; Nichirei Bioscience, Tokyo, Japan). Tissue sections were pre-incubated with serum-free Dako protein block (DakoCytomation, Glostrup, Denmark) for 15 min or in TBS-DT (20 mM Tris-HCl, 137 mM NaCl, pH 7.6, containing 50 mg/ml skim milk and 0.1% Tween-20, skim milk, 232100; BD Biosciences, Tokyo, Japan). Primary antibodies were incubated at room temperature for 2 h, followed by secondary antibodies incubation for 1 h. Antibodies used in this study are listed in Supplementary Table 4. A TUNEL assay was performed to evaluate DNA fragmentation associated with apoptotic cell death. In contrast, phospho-H2AX immunostaining was used to assess DNA double-strand breaks. The slides were mounted with phosphate-buffered saline (PBS) containing 10% glycerol and 1 µg/mL diamidino-2-phenylindole (DAPI) for nuclear staining. All images, including those of DAB and immunofluorescence staining, were acquired using a high-standard all-in-one Fluorescence Microscope (Biorevo BZ-X710; KEYENCE Japan, Osaka, Japan).

Spatial immunofluorescence (SPI) assay

For the standard SPI assay covering 20 mm × 15 mm area, multiplex immunofluorescence-stained images (1400 μm × 1000 μm) with anti-CD109 and CK8/18 antibodies were acquired at 10× magnification using a fluorescence microscope and divided into 35 sections (each 200 μm × 200 μm) (Supplementary Fig. 13a).

For the high-density SPI assay covering 4400 μm × 4000 μm area, similarly stained images (700 μm × 500 μm) were acquired at 20× magnification and divided into sections (each 90 μm × 90 μm) (Supplementary Fig. 13b). The FL values of each RGB channel in each section were quantified using Fiji (version 2.3.0/1.53q; National Institutes of Health, Bethesda, MD, USA). Relative fluorescence intensity (CD109 fluorescence/CK8/18 fluorescence) was calculated and visualized as a color-coded spatial map overlaid on the image. The FL values obtained from the high-density SPI assay were used to generate an animation using Microsoft Excel (version 16.0; Microsoft Corporation, Redmond, WA, USA) and Microsoft PowerPoint (version 16.0; Microsoft Corporation).

Multiple immunofluorescence image analysis

Line plot analysis was performed using multiplex immunofluorescence-stained images (1400 μm × 1000 μm) with anti-CD109 and CK8/18 antibodies acquired at 10× magnification. The FL values of each RGB channel along the defined line were quantified using Fiji, and line plots were generated using Microsoft Excel.

Multiplex immunofluorescence-stained images (700 μm × 500 μm) were acquired at 20× magnification using a fluorescence microscope. The FL values of each RGB channel were quantified using Fiji, and three to five images were analyzed per designated region. Three-dimensional graphs were created using Graphing Calculator 3D software (version 10.6; Runiter, www.runiter.com). Violin plots were generated using PlotsOfData, an online data visualization tool (https://huygens.science.uva.nl/PlotsOfData/)45. PCA and heatmap visualization were performed using ClustVis, an online tool for multivariate data analysis (https://biit.cs.ut.ee/clustvis/)46.

COL III-fiber orientation assay

Immunofluorescence-stained images using anti-COL III antibody were used. Three to four images were acquired at 20× magnification using a fluorescence microscope. Quantitative analysis of the fiber morphology was then conducted using DiameterJ, an ImageJ/Fiji plugin specifically designed for analyzing fiber networks47. COL III fiber parameters, such as the fiber diameter, the fiber length, density, and orientation, were obtained.

Statistical analysis

Differences in continuous variables among multiple characteristic regions were assessed using ANOVA with Bonferroni’s post-hoc test, which was performed using Statisty (https://statisty.app/one-way-anova-calculator). Welch’s t-test was used to compare protein expression frequencies between two selected regions. All statistical tests were two-sided, and p < 0.05 was considered statistically significant. Analyses were performed using Microsoft Excel (version 16.0; Microsoft Corporation, Redmond, WA, USA) and Statisty.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary Material 1 (27.9MB, mp4)
Supplementary Material 2 (7.3MB, pdf)

Author contributions

Conceptualization: T.C., K. S., K. M., M. N.; Methodology: T. C., K. S.; Pathological diagnosis: H. K., Y. M., M. N.; Formal analysis and investigation: T. C., K. S.; Writing-original draft preparation: T. C.; Writing-review and editing: K. S., K. M., H. K., Y. M., M. U., A. K., M. N,; Funding acquisition: K. M., M. N.; Resources: S. S., H. S., H. Y.; Supervision: M. N.

Funding information

This work was funded by the Atomic Bomb Disease Institute, Nagasaki University, a Grant-in-Aid for Scientific Research from the Japanese Ministry of Education, Science, Sports and Culture grant numbers [22K06982 and 23K06465], and the Program of the Network-Type Joint Usage/Research Center for Radiation Disaster Medical Science.

Data availability

All data generated or analyzed during this study are included in this published article and its Supplementary Information files.

Declarations

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

  • 1.Basolo, F., Macerola, E., Poma, A. M. & Torregrossa, L. The 5th edition of WHO classification of tumors of endocrine organs: changes in the diagnosis of follicular-derived thyroid carcinoma. Endocrine80, 470–476 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Lim, H., Devesa, S. S., Sosa, J. A., Check, D. & Kitahara, C. M. Trends in thyroid cancer incidence and mortality in the United States, 1974–2013. JAMA317, 1338–1348 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Dal Maso, L. et al. Survival of 86,690 patients with thyroid cancer: A population-based study in 29 European countries from EUROCARE-5. Eur. J. Cancer. 77, 140–152 (2017). [DOI] [PubMed] [Google Scholar]
  • 4.Baloch, Z. W. et al. Overview of the 2022 WHO classification of thyroid neoplasms. Endocr. Pathol.33, 27–63 (2022). [DOI] [PubMed] [Google Scholar]
  • 5.Califano, I., Smulever, A., Jerkovich, F. & Pitoia, F. Advances in the management of anaplastic thyroid carcinoma: transforming a life-threatening condition into a potentially treatable disease. Rev. Endocr. Metab. Disord. 25, 123–147 (2024). [DOI] [PubMed] [Google Scholar]
  • 6.Lu, L. et al. Anaplastic transformation in thyroid cancer revealed by single-cell transcriptomics. J. Clin. Invest.133, e169653 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Zeng, P. Y. F. et al. The genomic and evolutionary landscapes of anaplastic thyroid carcinoma. Cell. Rep.43, 113826 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Karimi, A. H. et al. Novel insights in advanced thyroid carcinoma: from mechanisms to treatments: molecular insights into the origin, biology, and treatment of anaplastic thyroid carcinoma. Eur. Thyroid J.14, e250057 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Kamma, H. et al. Pathological diagnosis of general rules for the description of thyroid cancer by Japanese Society of Thyroid Pathology and Japan Association of Endocrine Surgery. Endocr. J.69, 139–154 (2022). [DOI] [PubMed] [Google Scholar]
  • 10.Pozdeyev, N. et al. Genetic analysis of 779 advanced differentiated and anaplastic thyroid cancers. Clin. Cancer Res.24, 3059–3068 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Ngo, T. N. M. et al. Primary versus secondary anaplastic thyroid carcinoma: perspectives from multi-institutional and population-level data. Endocr. Pathol.32, 489–500 (2021). [DOI] [PubMed] [Google Scholar]
  • 12.Lacka, K. et al. Rationale for testing TP53 mutations in thyroid cancer-original data and meta-analysis. Int. J. Mol. Sci.26, 1035 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Saito, Y. et al. Comprehensive genomic profiling from C-CAT database unveiled over 80% presence of oncogenic drivers in anaplastic thyroid carcinoma including BRAF, RAS family, NF1, and FGFR1. Clin. Endocrinol. (Oxf). 101, 170–179 (2024). [DOI] [PubMed] [Google Scholar]
  • 14.Pan, Z. et al. Integrative proteogenomic characterization reveals therapeutic targets in poorly differentiated and anaplastic thyroid cancers. Nat. Commun.16, 3601 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Byun, H., Lee, H. S., Song, Y. S. & Park, Y. J. Transcriptome of anaplastic thyroid cancer reveals two molecular subtypes with distinct tumor microenvironment and prognosis. Thyroid35, 367–378 (2025). [DOI] [PubMed] [Google Scholar]
  • 16.Feng, C. et al. Integrative single-cell transcriptome analysis reveals immune suppressive landscape in the anaplastic thyroid cancer. Cancer Gene Ther.30, 1598–1609 (2023). [DOI] [PubMed] [Google Scholar]
  • 17.Ruan, X. et al. Persistent pre-exhausted CD8 + T cells shape the tumor immune microenvironment in anaplastic thyroid cancer. Endocr. Relat. Cancer. 32, e240169 (2025). [DOI] [PubMed] [Google Scholar]
  • 18.Zhou, K., Zhang, S., Shang, J. & Lan, X. Exploring immune gene expression and potential regulatory mechanisms in anaplastic thyroid carcinoma using a combination of single-cell and bulk RNA sequencing data. Comput. Biol. Chem.115, 108311 (2025). [DOI] [PubMed] [Google Scholar]
  • 19.Mii, S. et al. CD109: a multifunctional GPI-anchored protein with key roles in tumor progression and physiological homeostasis. Pathol. Int.69, 249–259 (2019). [DOI] [PubMed] [Google Scholar]
  • 20.Condello, V., Paulsson, J. O., Zedenius, J., Näsman, A. & Juhlin, C. C. Spatial transcriptomics in a case of follicular thyroid carcinoma reveals clone-specific dysregulation of genes regulating extracellular matrix in the invading front. Endocr. Pathol.35, 122–133 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Liao, T. et al. A spatially resolved transcriptome landscape during thyroid cancer progression. Cell. Rep. Med.6, 102043 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Loberg, M. A. et al. An integrated single-cell and spatial transcriptomic atlas of thyroid cancer progression identifies prognostic fibroblast subpopulations. bioRxiv, 01.08.631962 (2025). [DOI] [PMC free article] [PubMed]
  • 23.Fozzatti, L. & Cheng, S. Y. Tumor cells and cancer-associated fibroblasts: a synergistic crosstalk to promote thyroid cancer. Endocrinol. Metab. (Seoul). 35, 673–680 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Shin, E. & Koo, J. S. Cell component and function of tumor microenvironment in thyroid cancer. Int. J. Mol. Sci.23, 12578 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Singh, P. & Ali, S. A. Multifunctional role of S100 protein family in the immune system: an update. Cells11, 2274 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Mai, Z., Lin, Y., Lin, P., Zhao, X. & Cui, L. Modulating extracellular matrix stiffness: a strategic approach to boost cancer immunotherapy. Cell. Death Dis.15, 307 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Zhang, M. & Zhang, B. Extracellular matrix stiffness: mechanisms in tumor progression and therapeutic potential in cancer. Exp. Hematol. Oncol.14, 54 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Yu, S., Wang, S., Wang, X. & Xu, X. The axis of tumor-associated macrophages, extracellular matrix proteins, and cancer-associated fibroblasts in oncogenesis. Cancer Cell. Int.24, 335 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Avagliano, A. et al. Thyroid cancer and fibroblasts. Cancers (Basel). 14, 4172 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Ween, M. P., Oehler, M. K. & Ricciardelli, C. Transforming growth factor-beta-induced protein (TGFBI)/(βig-H3): a matrix protein with dual functions in ovarian cancer. Int. J. Mol. Sci.13, 10461–10477 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Hou, W. Role of TGFβ-activated cancer-associated fibroblasts in the resistance to checkpoint blockade immunotherapy. Front. Oncol.15, 1602452 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Lecker, L. S. M. et al. TGFBI production by macrophages contributes to an immunosuppressive microenvironment in ovarian cancer. Cancer Res.81, 5706–5719 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Hanssen, E., Reinboth, B. & Gibson, M. A. Covalent and non-covalent interactions of betaig-h3 with collagen VI. Beta ig-h3 is covalently attached to the amino-terminal region of collagen VI in tissue microfibrils. J. Biol. Chem.278, 24334–24341 (2003). [DOI] [PubMed] [Google Scholar]
  • 34.Löser, R. et al. Lysyl oxidases as targets for cancer therapy and diagnostic imaging. ChemMedChem18, e202300331 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Ishihara, S. & Haga, H. Matrix stiffness contributes to cancer progression by regulating transcription factors. Cancers (Basel). 14, 1049 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Pakkianathan, J. et al. Mutational landmarks in anaplastic thyroid cancer: a perspective of a new treatment strategy. J. Clin. Med.14, 2898 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Schütz, C. S., Stope, M. B. & Bekeschus, S. H2A.X phosphorylation in oxidative stress and risk assessment in plasma medicine. Oxid. Med. Cell. Longev. 2060986 (2021). [DOI] [PMC free article] [PubMed]
  • 38.Xia, J., Shi, Y. & Chen, X. New insights into the mechanisms of the extracellular matrix and its therapeutic potential in anaplastic thyroid carcinoma. Sci. Rep.14, 20977 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Wang, Y., Kang, X., Kang, X. & Yang, F. S100A6: molecular function and biomarker role. Biomark. Res.11, 78 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Foell, D., Wittkowski, H., Vogl, T. & Roth, J. S100 proteins expressed in phagocytes: a novel group of damage-associated molecular pattern molecules. J. Leukoc. Biol.81, 28–37 (2007). [DOI] [PubMed] [Google Scholar]
  • 41.Chaouhan, H. S. et al. Necroptosis: a pathogenic negotiator in human diseases. Int. J. Mol. Sci.23, 12714 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Reeb, A. N. et al. S100A8 is a novel therapeutic target for anaplastic thyroid carcinoma. J. Clin. Endocrinol. Metab.100, E232–E242 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Allgöwer, C. et al. Friend or foe: S100 proteins in cancer. Cancers (Basel) 12, (2020). [DOI] [PMC free article] [PubMed]
  • 44.Shakib, H. et al. Epithelial-to-mesenchymal transition in thyroid cancer: a comprehensive review. Endocrine66, 435–455 (2019). [DOI] [PubMed] [Google Scholar]
  • 45.Postma, M. & Goedhart, J. PlotsOfData-A web app for visualizing data together with their summaries. PLOS Biol.17, e3000202 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Metsalu, T. & Vilo, J. ClustVis: a web tool for visualizing clustering of multivariate data using principal component analysis and heatmap. Nucleic Acids Res.43, W566–W570 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Hotaling, N. A., Bharti, K., Kriel, H. & Simon, C. G. DiameterJ: A validated open source nanofiber diameter measurement tool. Biomaterials61, 327–338 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material 1 (27.9MB, mp4)
Supplementary Material 2 (7.3MB, pdf)

Data Availability Statement

All data generated or analyzed during this study are included in this published article and its Supplementary Information files.

Articles from Scientific Reports are provided here courtesy of Nature Publishing Group

RESOURCES