TYPE Original Research PUBLISHED 18 November 2022 DOI 10.3389/fcell.2022.1052098 Integrating digital pathology with OPEN ACCESS transcriptomic and epigenomic EDITED BY Alessandro Arcucci, University of Naples Federico II, Italy tools for predicting metastatic REVIEWED BY Zhizhan Gu, uterine tumor aggressiveness American Type Culture Collection, United States Irina R. Matei, Giorgia Sonzini 1†, Sofia Granados-Aparici 1†, Sabina Sanegre 1,2, Cornell University, United States Angel Diaz-Lagares 2,3, Juan Diaz-Martin 2,4, Carlos de Andrea 2,5, *CORRESPONDENCE Núria Eritja 2,6, Aida Bao-Caamano 3,7, Nicolás Costa-Fraga 3,7, Xavier Matias-Guiu, xmatias@bellvitgehospital.ca David García-Ros 5, Carmen Salguero-Aranda 2,4, Rosa Noguera, rnoguera@uv.es Ben Davidson 8,9, Rafael López-López 2,10, Ignacio Melero 2,5, These authors share first authorship † Samuel Navarro 1,2, Santiago Ramon y Cajal 2,11, SPECIALTY SECTION Enrique de Alava 2,4, Xavier Matias-Guiu 2,6* and This article was submitted to Cancer Cell Biology, Rosa Noguera 1,2* a section of the journal 1 Department of Pathology, Medical School, University of Valencia-INCLIVA, Valencia, Spain, 2Cancer Frontiers in Cell and Developmental CIBER (CIBERONC), Madrid, Spain, 3Epigenomics Unit, Cancer Epigenomics, Translational Medical Biology Oncology Group (Oncomet), Health Research Institute of Santiago (IDIS), University Clinical Hospital of RECEIVED 23 September 2022 Santiago (CHUS/SERGAS), Santiago de Compostela, Spain, 4Institute of Biomedicine of Sevilla (IBiS), ACCEPTED 04 November 2022 Virgen del Rocio University Hospital/CSIC/University of Sevilla, Seville, Spain, 5Clínica Universidad de PUBLISHED 18 November 2022 Navarra, University of Navarra, Pamplona, Spain, 6Institut de Recerca Biomèdica de LLeida (IRBLLEIDA), Institut d’Investigació Biomèdica de Bellvitge (IDIBELL), Department of Pathology, Hospital U Arnau de CITATION Vilanova and Hospital U de Bellvitge, University of Lleida - University of Barcelona, Barcelona, Spain, 7 Sonzini G, Granados-Aparici S, Universidad de Santiago de Compostela (USC), Santiago de Compostela, Spain, 8Institute of Clinical Sanegre S, Diaz-Lagares A, Medicine, Faculty of Medicine, University of Oslo, Oslo, Norway, 9Department of Pathology, Oslo Diaz-Martin J, de Andrea C, Eritja N, University Hospital, Norwegian Radium Hospital, Oslo, Norway, 10Roche-Chus Joint Unit, Translational Bao-Caamano A, Costa-Fraga N, Medical Oncology Group (Oncomet), Health Research Institute of Santiago (IDIS), Santiago de García-Ros D, Salguero-Aranda C, Compostela, Spain, 11Department of Pathology, Vall d’Hebron University Hospital, Autonoma Davidson B, López-López R, Melero I, University of Barcelona, Barcelona, Spain Navarro S, Ramon y Cajal S, de Alava E, Matias-Guiu X and Noguera R (2022), Integrating digital pathology with transcriptomic and epigenomic tools for The incidence of new cancer cases is expected to increase significantly in the predicting metastatic uterine future, posing a worldwide problem. In this regard, precision oncology and its tumor aggressiveness. Front. Cell Dev. Biol. 10:1052098. diagnostic tools are essential for developing personalized cancer treatments. doi: 10.3389/fcell.2022.1052098 Digital pathology (DP) is a particularly key strategy to study the interactions of COPYRIGHT tumor cells and the tumor microenvironment (TME), which play a crucial role in © 2022 Sonzini, Granados-Aparici, tumor initiation, progression and metastasis. The purpose of this study was to Sanegre, Diaz-Lagares, Diaz-Martin, de integrate data on the digital patterns of reticulin fiber scaffolding and the Andrea, Eritja, Bao-Caamano, Costa-Fraga, García-Ros, Salguero-Aranda, Davidson, immune cell infiltrate, transcriptomic and epigenetic profiles in aggressive López-López, Melero, Navarro, Ramon y uterine adenocarcinoma (uADC), uterine leiomyosarcoma (uLMS) and their Cajal, de Alava, Matias-Guiu and Noguera. This is an open-access article distributed respective lung metastases, with the aim of obtaining key TME biomarkers under the terms of the Creative Commons that can help improve metastatic prediction and shed light on potential Attribution License (CC BY). The use, therapeutic targets. Automatized algorithms were used to analyze reticulin distribution or reproduction in other forums is permitted, provided the original fiber architecture and immune infiltration in colocalized regions of interest author(s) and the copyright owner(s) are (ROIs) of 133 invasive tumor front (ITF), 89 tumor niches and 70 target tissues in credited and that the original publication in this journal is cited, in accordance with a total of six paired samples of uADC and nine of uLMS. Microdissected tissue accepted academic practice. No use, from the ITF was employed for transcriptomic and epigenetic studies in primary distribution or reproduction is permitted and metastatic tumors. Reticulin fiber scaffolding was characterized by a large which does not comply with these terms. and loose reticular fiber network in uADC, while dense bundles were found in Frontiers in Cell and Developmental Biology 01 frontiersin.org Sonzini et al. 10.3389/fcell.2022.1052098 uLMS. Notably, more similarities between reticulin fibers were observed in paired uLMS then paired uADCs. Transcriptomic and multiplex immunofluorescence-based immune profiling showed a higher abundance of T and B cells in primary tumor and in metastatic uADC than uLMS. Moreover, the epigenetic signature of paired samples in uADCs showed more differences than paired samples in uLMS. Some epigenetic variation was also found between the ITF of metastatic uADC and uLMS. Altogether, our data suggest a correlation between morphological and molecular changes at the ITF and the degree of aggressiveness. The use of DP tools for characterizing reticulin scaffolding and immune cell infiltration at the ITF in paired samples together with information provided by omics analyses in a large cohort will hopefully help validate novel biomarkers of tumor aggressiveness, develop new drugs and improve patient quality of life in a much more efficient way. KEYWORDS uterine adenocarcinoma, uterine leiomyosarcoma, lung metastasis, digital pathology, invasive tumor front, multiplex immunofluorescence, transcriptomics, methylomics Introduction number of cases develop metastatic disease, particularly at the lung (Roberts, Aynardi and Chu, 2018). The premise of precision oncology is to develop The last few decades have witnessed growing interest in treatments targeting the molecular characteristics of patient understanding the characteristics underlying tumor invasion, tumors, drawing upon extensive databases that curate genetic with research centered on type-specific new stroma generated and molecular features gathered from genome-wide at the site of active tumor invasion [the invasive tumor front, association studies and tumor-sequencing efforts ITF] (Antonicelli et al., 2012). Stromatogenesis, the formation of (Schwartzberg et al., 2017). However, clinicians will be new stroma at the ITF, disrupts the continuity of normal tissues, confronted with a vast amount of complex molecular facilitating the invasion process. This newly formed tumor- information derived from omics, drug biomarker testing, associated stroma is composed of stromal cells and the digital pathology and electronic health records (Alami extracellular matrix (ECM), which is biochemically distinct in et al., 2020), intensifying the gap between technological composition (Sivridis, Giatromanolaki and Koukourakis, 2004; capacity, clinical knowledge, and the ability to translate Sanegre et al., 2021). Specifically, the new ECM is characterized molecular and digital pathology data into effective medical by increased deposition of fibrillar collagen and cross-links, decision-making (Tamborero et al., 2020). Fortunately, leading to increased tissue stiffness and causing desmoplasia, multidisciplinary tumor boards are playing a key role in typical of the tumor (Eble and Niland, 2019). The newly addressing this disparity, via emerging frameworks utilizing deposited collagen fibers are reoriented and remodeled to automated data capture, interpretation and reporting systems, generate larger and stiffer collagen bundles that contribute to from which researchers may subsequently share and harness the distinctive features of the ECM surrounding the tumor results (Tamborero et al., 2020). (Paolillo and Schinelli, 2019). Uterine endometrial adenocarcinoma (uADC) is the most Among the different fibrillar collagens, type III collagen common cancer of the female reproductive organs in the constitutes the fundamental protein of the reticulin fibers. developed world. uADC can present an aggressive form in a Reticulin fibers are thin fibers of less than 2 µm in diameter substantial number of cases, leading to poor prognosis (Huvila that originate primarily from mesenchymal cells and et al., 2021). Among the routes of metastasis in uADC patients undifferentiated fibroblasts. As fibroblasts differentiate, are intra-abdominal spread, lymph node involvement and distant production of fibers and also type I collagen is increased. organ metastasis, and although very rare, this last route has a Reticulin fibers form an extensive branched network, with a dramatic impact on prognosis and survival in uADC patients. In mesh-like pattern, forming part of many organs, mainly as part of this context, in a recent study the lung was reported as the most the basement membrane in epithelial tissue, in the basal lamina common distant metastatic organ in patients with uADC (Mao of adipose tissue and in locations with high cellular content, such et al., 2020). Uterine leiomyosarcoma (uLMS) is a rare tumor, as the endocrine glands, liver and bone marrow (Ushiki, 2002). accounting for 1%–2% of all uterine malignancies. Although They are found in high abundance in several disorders and cancer in patients suffering from this disease is often confined to tumors such as Castleman disease (Carbone et al., 2021), the uterus, or with locoregional spread to the pelvis, a high myeloproliferative neoplasms (Gleitz, Benabid and Schneider, Frontiers in Cell and Developmental Biology 02 frontiersin.org Sonzini et al. 10.3389/fcell.2022.1052098 2021) and aggressive primary uADC and uLMS (Sanegre et al., expression and cell adhesion-related DNA methylation 2021). Under the microscope, reticular fibers are signatures (Sanegre et al., 2021). In this study, therefore, we indistinguishable from the other types in conventional stains, used the same digital analysis, transcriptomic and epigenomic but they can be observed stained black with Gomori’s silver tools to investigate whether these changes are maintained in impregnation method, a technique routinely employed for metastatic uADC and uLMS samples, and also compared them visualization and characterization in many pathology with the primary tumors. Overall, this study aims to enhance departments (Ushiki, 2002). Along with the other ECM prediction of metastatic lung behavior and pinpoint possible new components, reticular fibers undergo changes during tumor diagnostic, prognostic and/or therapeutic targets, which together stromatogenesis that affect their biotensegrity and thus have a with expert pathologist supervision and validation, could functional impact on TME elements, such as the migratory optimize routine pathological tasks in the future. potential of tumoral cells or the infiltration capacity of immune cells. Therefore, histomorphometric characterization of reticular fibers using digital analysis tools may add value Materials and methods for precision diagnostics in cancer (Montironi et al., 2020). Changes in tissue composition that favor the growth of overt Patients and case selection metastases are collectively known as the premetastatic niche (Kai, Drain and Weaver, 2019) which represents an abnormal A total of 24 uADC, 11 uLMS, 6 uADC lung metastases and microenvironment supportive of tumor growth (Paolillo and 9 uLMS lung metastases (LM-uADC and LM-uLMS) were Schinelli, 2019). Secreted proteins from the primary tumor, such included in the study. We used formalin-fixed and paraffin- as proinflammatory cytokines and exosomes, can both directly embedded (FFPE) tissue samples from uADC and uLMS primary activate tissue resident cells to synthesize and remodel the ECM in tumors (previously described in Sanegre et al. (2021), with a distant tissues and also stimulate recruitment of ECM-modifying particular focus on paired cases (primary tumors with their cell populations (Kai, Drain and Weaver, 2019). The lung is the respective lung metastases). Tumors were classified according second most frequent site of metastatic focus. Normal lungs possess to the latest WHO criteria and were surgically staged and graded anatomical and cellular features that act as a defensive barrier following the International Federation of Gynecology and against foreign pathogens and particles. However, under specific Obstetrics (FIGO) classification and staging systems. The conditions, such as inflammatory state, the lung microenvironment study was approved by the local research ethics committee, displays features that may support carcinogenesis. The TME and specific informed consent was given. Whole slide FFPE contains tumor-infiltrating immune cells that are closely linked tissue sections of 5 μm of selected uADC, uLMS and their to clinical prognosis in cancer patients (Xiao and Yu, 2021). respective lung metastases were stained with H&E Although malignant cells are able to evade the immune (hematoxylin and eosin) and examined by a centralized expert response, promoting tumor progression, tumor-infiltrating pathologist group to select representative areas to include in the immune cells often initiate an immune response that aims to study. Sections of FFPE tissue samples were scanned using a eliminate cancer cells. In this regard, the immune contexture digital slide scanner [Pannoramic 250 Flash II 2.0 (3D Histec)] and T cell abundance, functional activity, and spatial distribution prior to and following microdissection. Whole sections including in the TME are crucial prognostic and predictive factors (Galli et al., the ITF were used for morphometric analysis. Serial uADC and 2020). B cells, which are involved in humoral immunity of the uLMS whole slides were used for histomorphometric analysis of adaptive immune system, can infiltrate the TME at the ITF. In reticulin fibers (Gomori’s stain) and multiplex mouse models, B cells have been observed to have an antitumor role immunofluorescence-based immune profiling. ITF regions by increasing T-cell function (Zhang and Zhang, 2020). were identified in stained tissue by extrapolation of previous With the emergence of precision oncology, automated H&E-selected regions. A 5 × 4 mm region of interest (ROI) for analysis of an ever-growing number of omics-generated each sample was identified in the ITF-stained area. These 5 × biomarkers and the development of deep learning digital 4 mm regions were used to establish a correlation between the pathology models for their measurement in spatial context, all results of transcriptomic, epigenomic and morphometric studies. require methodological approaches for integrating the resulting To achieve in-depth characterization of the reticular fibers and data, which will bring insight into the molecular mechanisms immune infiltrate of the ITF, the 5 × 4 mm ROIs were further leading to different pathways of original and metastatic tumor broken down into 1 × 1 mm ROIs, representing the following characterization in each patient (Mungenast et al., 2021). We categories: 1) tumor (uADC or uLMS or LM-uADC or LM-uLMS recently characterized and compared the ITF of aggressive with absence of target tissue), 2) target tissue (myometrium for primary uADC and uLMS in a multicenter study, finding that primary tumors and lung tissue for the respective metastases, despite morphometric resemblance between reticular fibers and excluding tumor infiltrated target tissue from the analysis), and high presence of macrophages, some variance could be observed 3) balanced representation of the invasion front (containing in other immune cell populations, as well as distinctive gene approximately 50% tumor and 50% target tissue). Cases and Frontiers in Cell and Developmental Biology 03 frontiersin.org Sonzini et al. 10.3389/fcell.2022.1052098 ROIs in which the algorithm failed due to insufficient or excessive was performed using Target Retrieval Solution High staining were excluded from analysis, as were samples with pH (Dako-Agilent) at 97° for 30 min prior to incubation unsatisfactory segmentation. Supplementary Table S1 indicates with a mouse monoclonal anti-TTF1 antibody (clone: the number of 5 × 4 mm and 1 × 1 mm ROIs of each case in SPT24, Gennova) for 30 min. EnVisionTM Flex Mouse Gomori and multiplex immunofluorescence stained tissue. The Linker (Dako) was added for 10 min followed by EnVision interface between tumor tissue and adjacent myometrium or FLEX/HRP (Dako) for 20 min and incubation with lung tissue was microdissected under the microscope for primary diaminobenzidine (DAB) chromogen for 5 min. QuPath tumors and metastatic tumors, respectively. Microdissected v0.2.3 open-source digital pathology software was used for tissue from the ITF was employed for transcriptomic and TTF1 quantification. A positive cell detection algorithm was epigenomic studies. The amount of ITF microdissected tissue employed to detect every nucleus that expressed TTF1. The varied from case to case, but the median width was 5 mm number of positive cell nuclei per mm2 (density) at the ITF (±1.66 mm) and median length 15 mm (±4.91 mm). regions (1 × 1 mm) was calculated for each ROI in LM-uADC and LM-uLMS (Supplementary Table S1) and statistical differences were obtained by Student’s t-Test. Histomorphometric analysis of reticulin fiber Multiplex immunofluorescence staining The architecture of reticulin fibers stained using Gomori’s and tissue imaging method was studied in uADC, uLMS, LM-uADC and LM-uLMS. All samples were digitalized with the whole-slide scanner Ventana A validated and standardized multiplex immunofluorescence iScan HT (Roche) at ×20 with a resolution of 0.46 μm/pixel. We protocol was used to characterize immune cell infiltrate in 1 × used QuPath open-source digital pathology software for sample 1 mm ROIs of LM-ADCs and LM-LMS. This multiplex panel visualization and identification of ROIs (Bankhead et al., 2017). 1 × enabled simultaneous visualization and quantification of several 1 mm ROIs were exported to ImageJ (Schneider, Rasband and cellular markers in a single FFPE tissue section. Briefly, whole- Eliceiri, 2012), where they were saved as TIFF for image analysis. slide tissue sections were deparaffinized and subjected to The advanced morphometric methodology selected for this study sequential rounds of antibody staining, as previously described was Gomoripath, an algorithm based on a probabilistic method for (Sanegre et al., 2021). Antigen retrieval was performed using automatic reticulin fiber segmentation, as described in previous Dako PT-Link heat-induced antigen retrieval with low pH (pH6) studies (Sanegre et al., 2021). Fifteen morphometric parameters or high pH (pH9) target retrieval solution (Dako). The antibody defining the histological organization of reticular fibers were panel included a mouse monoclonal anti-CD68 antibody (clone: calculated for each fiber detected and the mean for each sample PG-M1, ready-to-use; Agilent), a rabbit polyclonal anti-CD3 was calculated. Area, width, height and perimeter were used to antibody (IgG, ready-to-use; Agilent), a mouse monoclonal determine the size of the fibers, while shape was defined by anti-CD8 antibody (clone: C8/144B, ready-to-use; Agilent), a roundness, aspect ratio, perimeter ratio, shape factor, vertices mouse monoclonal anti-CD20 antibody (IgG2α, clone: L26, and fractal dimension of the fibers (Tadeo et al., 2016). An ready-to-use; Agilent), a mouse monoclonal anti-CD66b additional size parameter defined in this study was deformity, antibody (clone: 80H3, dilution: 1:200; LS-Bio, Seattle, WA, corresponding to the degree of spatial dispersion of reticulin fibers United States) and DAPI. Tyramide signal amplification (TSA) relative to its center. The algorithm also measured the stained area with fluorophores Opal 650, 540, 570, 690, and 520 was used for of the tissue analyzed (excluding holes and damaged tissue), visualization of CD3, CD8, CD20, CD68, and CD66b, allowing us to determine the number of fibers per mm2 respectively. Thereafter, nuclei were counterstained with (density) and the percentage of fiber-stained area [%SA] spectral DAPI (Akoya Bioscience). Sections were then (considering the sum of the areas of all fibers). Means and mounted with Faramount Aqueous Mounting Medium standard deviations were calculated for each morphometric (Dako). Multiplexed immunofluorescence slides were scanned parameter of uADC, uLMS and their respective lung metastases on a Vectra-Polaris Automated Quantitative Pathology Imaging ROIs, and statistical comparison between the different case studies System (Akoya Biosciences). Spectral unmixing was performed was performed by Student’s t-Test. using inForm software (version 2.4.8, Akoya Biosciences), as described (Abengozar-Muela et al., 2020). Image analysis was performed using QuPath and Fiji/ImageJ. Briefly, cells were Thyroid transcription factor segmented based on nuclear detection using the StarDist 2D 1 quantification in LM-uADC and LM-uLMS algorithm. A random trees algorithm classifier was trained for each cell marker. Cells were then subclassified as CD68+, CD3+, Immunostaining for TTF1 was obtained in LM-uADC and CD8+, CD66b+, and CD20+ cells. CD4+ T-cells were defined as LM-uLMS using a Dako Omnis platform. Antigen retrieval CD3+ CD8−. Cells negative for these markers were defined as Frontiers in Cell and Developmental Biology 04 frontiersin.org Sonzini et al. 10.3389/fcell.2022.1052098 TABLE 1 Comparison of histomorphometric characteristics of reticulin fibers between each tumor area in 1 × 1 mm ROIs. I II. III. IV. uADC uLMS LM uADC LM uLMS T vs. TT T vs. TT vs. T vs. TT T vs. TT vs. T vs. TT T vs. TT vs. T vs. TT T vs. TT vs. ITF ITF ITF ITF ITF ITF ITF ITF Quantity Fiber per mm2 TT** ITF* ITF* ITF* %SA TT* ITF* Size Area T* T* T* TT* TT** ITF* Width T** T*** TT** TT*** TT** Height ITF* TT* ITF* Perimeter T* TT** ITF* Deformity T* T* TT* TT* Shape Roundness T** T** Aspect T* Perimeter ratio T*** Shape T* ITF* Vertices T** T*** TT** TT** TT*** ITF** TT*** Fractal dimension T* T** TT* TT* TT** TT* Evaluation in primary uterine adenocarcinoma and uterine leiomyosarcoma metastasizing to lung (uADC and uLMS, respectively). Assessment in lung metastasis (LM uADC and LM uLMS, respectively). Comparisons between tumor (T), target tissue (TT) and invasive tumor front (ITF). Statistically significant differences in quantity, size and shape parameters of reticular fibers are shown (p values, * <0.05, **<0.01, ***<0.001) and the tissues with higher values are noted. LM: tumor lung metastasis; %SA: percentage of stained area. Area in μm2; width, length and perimeter in μm. “other cell types”. Measurements were calculated as cell densities htgmolecular.com/assets/htg/publications/HTG202015_WP- (cells/mm2). Immunopheno-04final_1.pdf), which uses xCell algorithm as a reference for gene enrichment analysis (Aran, Hu and Butte, 2017). Four of the 23 signatures measure stromal cells, including Transcriptomic profiling adipocytes, endothelial cells, epithelial cells, and cancer-associated fibroblasts. The remaining 19 signatures measure the relative Total RNA for gene expression assays was prepared from 5 µm abundance of different immune cells, including different subtypes FFPE tissue sections of microdissected ITF from LM-uADC and of T-lymphocytes, B-lymphocytes, phagocytes, and granulocytes. LM-uLMS using the Agencourt FormaPure kit (A33341; Beckman iDEP v0.951 (http://bioinformatics.sdstate.edu/idep/) was used for Coulter, Indianapolis, IN, United States) and following the pathway analysis of expression values from RNA-Seq data. manufacturers’ instructions. RNA concentration was determined Expression data from ITF of primary tumors generated in our with Qubit 4 fluorometer and Qubit® RNA HS Reagent (Thermo former study (Sanegre et al., 2021) were used for comparisons with Fisher Scientific, Waltham, MA, United States). We selected RNA the transcriptional profile of the lung metastatic samples. samples from lung metastasis that had passed quality control evaluation (uADC = 6, uLMS = 7). Transcriptomic profiling was performed as previously described (Sanegre et al., 2021) using HTG DNA methylation profiling by Infinium EdgeSeq Precision Immuno-Oncology Panel. HTG EdgeSeq host Methylation EPIC array software performed the alignment of the FASTQ files to the probe list, then results were parsed, and the output obtained as a read Total genomic DNA (gDNA) from 10-μm FFPE tissue sections counts matrix. The relative abundance of immune and stromal of the ITF from primary tumors and LM-uADC (n = 6) and LM- cellular content (TME score) and specific cell types were estimated uLMS (n = 9) was isolated using the AllPrep DNA/RNA FFPE Kit by analyzing parsed data files using the 23 HTG EdgeSeq Reveal ™ (Qiagen) according to the manufacturer’s instructions. DNA samples Immunophenotyping Signatures. The signatures were implemented were quantified by fluorometry with the Qubit 1 × dsDNA HS (High- ™ in HTG EdgeSeq Reveal statistical analysis software (https://www. Sensitivity) Assay Kit (Thermo Fisher). The suitability of the DNA for Frontiers in Cell and Developmental Biology 05 frontiersin.org Sonzini et al. 10.3389/fcell.2022.1052098 FFPE restoration was tested following the Infinium HD FFPE QC 1AI−III). These results suggest that the fibers at the ITF region and Assay (Illumina). All DNA samples (100–250 ng) were selected for myometrium of uADC have a greater extent in space and present a bisulfite conversion using the EZ DNA Methylation kit (Zymo looser crosslinked network at the ITF region than those in uLMS. Research). The FFPE restoration step was followed by Infinium We next analyzed reticulin scaffolding in the three regions in HD FFPE methylation assay using Infinium MethylationEPIC each metastatic tumor. The tumor area of LM-uADC and LM- BeadChips, which cover over 850,000 CpG sites along the human uLMS showed many differences in size and shape relative to their genome (Moran, Arribas and Esteller, 2016). Whole-genome ITF and lung regions, as well as different ITF to the other regions in amplification and hybridization were performed on the BeadChips both metastases (Table 1I, II). Comparing reticulin scaffolding in followed by single-base extension and analysis on a HiScan (Illumina) the three regions between metastatic tumors revealed less fiber to assess the methylation status of cytosines. Quality control and deposition at the tumor area in LM-uADC than LM-uLMS preprocessing of methylation data were carried out using the R (Supplementary Table S2, II). Interestingly, reticulin fibers in the package RnBeads (Assenov et al., 2014). First, the Greedycut lung regions of LM-uADC showed higher deformity, as observed in algorithm was used to filter out probes and/or samples. Next, the myometrium of the primary tumors, and increased area probes overlapping with single nucleotide polymorphisms and compared to those in LM-uLMS. Finally, significant differences probes whose sequence maps to multiple genomic locations in roundness, width and vertices were observed comparing reticulin (cross-reactive) were removed. Raw intensities obtained in the fibers of ITF regions in LM-uADC and LM-uLMS (Figures array were normalized using the BMIQ method. For each CpG 1BI−III). These results suggest that like the pattern observed in site we obtained a specific β value by calculating the ratio of primary tumors, the reticulin fibers at the ITF are large and form a fluorescent signal between methylated (M) probe relative to the loose plexus in LM-uADC, whereas a dense fiber network is formed sum of the M and unmethylated (U) probes (β = M/(M + U)). β in LM-uLMS. To test whether these results were due to alveolar values ranged from 0 (no methylation) to 1 (completely methylated). collapse with a difference in the number of TTF1-positive cells in Finally, hierarchical linear models were used to obtain between-group LM-uLMS, we analyzed the percentage of cells expressing TTF1. No differences. p values were corrected for multiple testing (False differences were observed between LM-uADC and LM-uLMS at the Discovery Rate, FDR) using Benjamini–Hochberg method and a ITF regions (Supplementary Figure S1), suggesting that these threshold of p < 0.05 was selected for significance. All statistical findings could be due to reticulin fiber architectural changes analyses were performed in the R statistical environment (v.3.6.1). without TTF1-related hyperplasia. Enrichment analysis of biological pathways for the methylation profiles was evaluated by gene ontology (GO) using Genecodis (Tabas-Madrid, Nogales-Cadenas and Pascual-Montano, 2012). ITFs and tumor areas of paired uLMS present reticulin fibers of similar size Results We next compared paired tumors of uADC and uLMS (primary tumors with their respective lung metastases). Reticular scaffolding was ITFs and target tissues of uADC present a different in parameters of size and shape between tumor areas in paired large and loose reticular network uADC, whereas no differences were observed in paired uLMS (Supplementary Table S2, III and IV). As expected, fibers in the We first studied the histological architecture of reticulin fibers in target tissue (myometrium and lung regions) presented many size- the tumor, myometrium and ITF, including more primary uADC and and shape-related differences in both types of tumor (Supplementary uLMS ROIs metastasizing to lung than were selected in Sanegre et al., Table S2, III and IV). Finally, comparing reticular fibers at the ITF 2021 to increase statistical robustness (Supplementary Table S1). region between paired uADC, width and vertices values were significant Comparing reticulin scaffolding at the tumor area in uADC, higher in LM-uADCs, suggesting that an even looser reticular network reduced fiber deposition and multiple differences in fiber size and is formed in metastatic samples (Figures 2AI−III). When reticular fibers shape were observed relative to the myometrium and ITF, whose at the ITF region were compared in paired uLMS, no differences in size fibers did not differ significantly from each other (Table 1I). In were observed, and significant differences in shape resulted in dense contrast, the reticulin scaffolding in uLMS was similar in the three fiber bundles in LM-uLMS (Figures 2BI−III). regions analyzed (Table 1II). We also investigated differences between uADC and uLMS. Comparison between tumor areas of uADC and uLMS showed decreased deposition of fibers in uADC, in addition to ITFs in metastatic and paired tumors differences in size and shape parameters (Supplementary Table S2, I). contain increased immune cell infiltrate in Moreover, reticulin fibers of the myometrium and ITF regions showed uADC a significant increase in size parameters (deformity and area, respectively) in uADC compared to uLMS, contrasting with a Multiplex immunofluorescence was performed for CD20, marked decrease in roundness at the ITF region in uADC (Figures CD3, CD8, CD66b, CD68 to estimate five different immune Frontiers in Cell and Developmental Biology 06 frontiersin.org Sonzini et al. 10.3389/fcell.2022.1052098 FIGURE 1 Histomorphometric characteristics of reticulin fibers between primary and metastatic tumors at the tumor area, ITF and target tissue in 1 × 1 mm ROIs. Representative images of reticulin fiber segmentation and statistically significant results of morphometric parameters. Reticulin fibers segmented using Gomori’s stain (in red) are shown in the histological images. Comparison between primary tumors metastasizing to lung (A) and lung metastasis from both uterine tumors (B). Statistically significant differences in quantity, size and shape parameters of reticular fibers are shown (p values, * <0.05, **<0.01, ***<0.001). a. T, tumor, b. TT, target tissue and c. ITF, invasive tumor front. LM, lung metastasis. Frontiers in Cell and Developmental Biology 07 frontiersin.org Sonzini et al. 10.3389/fcell.2022.1052098 FIGURE 2 Histomorphometric characteristics of reticulin fibers from analysis of paired tumors at the tumor area, ITF and target tissue in 1 × 1 mm ROIs. Representative images of reticulin fiber segmentation and statistically significant results of morphometric parameters. Reticulin fibers segmented using Gomori’s stain (in red) are shown in the histological images. Comparison between primary uterine adenocarcinomas (uADC) and lung metastasis (A), and primary uterine leiomyosarcoma (uLMS) and lung metastasis (B). Statistically significant differences in quantity, size and shape parameters of reticular fibers are shown (p values, * <0.05, **<0.01, ***<0.001). a T, tumor, b. TT, target tissue and c. ITF, invasive tumor front. LM: lung metastasis. Frontiers in Cell and Developmental Biology 08 frontiersin.org Sonzini et al. 10.3389/fcell.2022.1052098 FIGURE 3 Immune infiltrate distribution in metastatic tumors. (A,B, and C) metastatic uterine adenocarcinoma) and (D,E, and F) metastatic uterine leiomyosarcoma). Representative multiplex immunofluorescence images in 1 × 1 mm ROIs in LM-uADC (A–C) and LM-uLMS (D–F). Color code as follows: DAPI (blue), CD20 (yellow), CD68 (green), CD8 (red), CD3 (white), CD66 (cyan). Dot plots show the mean of the densities (number of cells/ mm2) for each cell marker in each region for LM-uADC and LM-uLMS. p values and significance (p < 0.05, pp < 0.01) for statistical comparison between the means of LM-uADC and LM-uLMS at the tumor area, invasive front (ITF) and lung (TT) are displayed. Scale bars represent 50 µm. Frontiers in Cell and Developmental Biology 09 frontiersin.org Sonzini et al. 10.3389/fcell.2022.1052098 infiltrate subpopulations in LM-uADC and LM-uLMS. The 1 × 1 mm ROIs previously selected for reticulin fiber studies (annotated as tumor, target tissue or ITF) were used to estimate immune cell density. We observed diverse cellular compositions of the immune environments among the lesions (Figure 3). Within LM-uADC, higher densities of CD20+ B lymphocytes were found in ITF regions when compared to the tumor area and target tissue [p = 0.002] (Supplementary Table S3) which were more significant between the ITF and the target tissue [p = 0.0071] (Figure 3). No differences were found for the other immune cell infiltrates. Within LM-uLMS, higher densities of CD20+ B lymphocytes were also found in ITF regions when compared to the tumor area and target tissue [p = 0.012] (Supplementary Table S3). Similarly, in the LM-uLMS, a significantly higher infiltration of CD20+ B lymphocytes were found at the ITF when compared to the tumor area (p = 0.0089). No differences were found for the other immune cell infiltrates. When comparing the immune cell infiltrates of LM-uADC and LM-uLMS, higher infiltration of CD20+ B lymphocytes was found at the ITF of LM-uADC [p = 0.006] (Figure 3). The immune and stroma scores of the ITF region in metastatic and primary tumors were obtained using expression data from the HTG EdgeSeq Precision Immuno- Oncology Panel (Figure 4A). In this region we observed higher immune scores for LM-uADC, whereas in LM-uLMS the stroma score was increased, possibly due to its intrinsic mesenchymal features. We next used the same expression data to determine the relative abundance of immune cell types. Regarding metastatic samples, major differences were observed in B cells/plasma cells, and CD4-T cell populations were also different, with increased abundance in LM-uADC compared with LM-uLMS metastasis (Figure 4B). We did not observe significant differences in CD8+ T cells (susceptible to exhaustion) in metastatic samples, and no notable differences were found between key T-cell exhaustion and activation markers (Supplementary Figure S2). We also performed the same analysis to compare the ITF of paired tumors of uADC FIGURE 4 Evaluation of immunophenotyping signatures at the ITF of and uLMS. We found a higher abundance of mononuclear metastatic and paired tumors. (A) Estimation of immune and phagocytes and B cells/plasma cells at the ITF of LM-uADC stromal cellular content. The TME (Tumor Microenvironment) than that of primary uADC (Figure 4C), whereas the ITF of score is the sum of the immune and stromal score (B–D) Left panels, hierarchical clustering of tumor samples and immune/ LM-uLMS were enriched in neutrophils, monocytes and stromal cell types based on the relative abundance score macrophages and presented less abundance of dendritic estimated with HTG EdgeSeq Reveal Immunophenotyping Signatures. Right panels, radar charts displaying the cells than primary uLMS (Figure 4D). immunophenotyping signatures which showed the major differences in each comparison (*, denotes statistically significant differences in scores for relative abundance). Transcriptomic profiling at the ITF reveals upregulation of genes involved in immune response in uADCs and double-strand break repair in uLMS fold change ≥1). Interestingly, we found that antimicrobial peptides (LCN2, DEFB1) were upregulated at the ITF of LM- Differential expression analyses between ITF of LM-uADC uADC, as was observed in primary tumors in our previous and LM-uLMS identified 242 genes (FDR cutoff <0.1, log2- study (Sanegre et al., 2021) [Supplementary Table S4]. Both Frontiers in Cell and Developmental Biology 10 frontiersin.org Sonzini et al. 10.3389/fcell.2022.1052098 FIGURE 5 Pathway analyses of transcriptional profiling at the ITF of metastatic and paired tumors. Comparisons are indicated in each panel (A–C). Left, red and blue indicate activated and suppressed GO cellular processes, respectively. Right, network of related pathways (nodes). Darker nodes are more significantly enriched gene sets. Bigger nodes represent larger gene sets. Thicker edges represent more overlapped genes. paired uLMS and uADC comparisons showed upregulation of as well as adaptive immune response at the ITF of LM-uADC, in lung markers in metastasis (NKX2.1 (TTF1) and AGER), likely accordance with the higher relative abundance of B cells, plasma reflecting lung alveolar cells trapped in the tumoral mass cells and CD4-T cells. Moreover, LM-uADC exhibited expression (Supplementary Tables S5, S6). signatures related to somatic recombination of immunoglobulin Parametric gene set enrichment analysis using GO biological genes. In contrast, the ITF of LM-uLMS exhibited upregulation process terms showed lymphocyte activation and differentiation of genes involved in double-strand break repair (Figure 5A). Frontiers in Cell and Developmental Biology 11 frontiersin.org Sonzini et al. 10.3389/fcell.2022.1052098 FIGURE 6 Genome-wide DNA methylation analysis of the ITF in paired tumors of uADC patients. (A) Schematic flowchart used to identify significant differentially methylated CpGs at the ITF. (B) Principal component analysis (PCA) of DNA methylation data at the ITF. (C) Scatter plot representing mean normalized levels of DNA methylation (β-values) at the ITF of primary tumors and lung metastasis. Dots in red show significant differentially methylated CpGs. (D) Genomic distribution of the 44,085 significant differentially methylated CpGs at the ITF by CpG context and gene region. (E) Hierarchical clustering of the 10,000 most differentially methylated CpGs (FDR <5%) in the ITF of primary and lung metastasis of uADC patients. (F) Gene ontology (GO) analysis of the biological process categories for the 10,000 most differentially methylated CpGs in the ITF of primary and lung metastasis of uADC. DMCpGs, differentially methylated CpGs; Meth, DNA methylation; P, Primary tumors; LM, lung metastasis, uADC, uterine adenocarcinomas. Pathway analysis of the ITF in paired uLMS denoted activation of (Figure 5B). When comparing the ITF of paired uADC, the several GO processes in metastasis such as lipid metabolism, most distinctive feature was the activation of granulocyte MAPK cascade, cell-cell adhesion and lung morphogenesis chemotaxis and migration in metastatic specimens (Figure 5C). Frontiers in Cell and Developmental Biology 12 frontiersin.org Sonzini et al. 10.3389/fcell.2022.1052098 TABLE 2 Top 25 significant differentially methylated CpGs from promoter CGIs/shore regions belonging to the epigenetic signature that differentiates the ITF of paired tumors in uADC patients. TargetIDa Chrb Position Gene Accession number Gene region CpG Δβc name density cg10576280 10 124133822 PLEKHA1 NM_001001974 TSS1500 Shore -0.65 cg02833127 4 177116733 SPATA4 NM_144644; NM_144644 1stExon; 5′UTR Island 0.60 cg12307484 7 19158721 TWIST1 NM_000474 TSS1500 Shore -0.57 cg21424940 7 19158666 TWIST1 NM_000474 TSS1500 Shore -0.56 cg26250747 20 57225100 STX16 NM_001001433; NM_001134772; NM_003763; NM_001134773; TSS1500 Island -0.56 NM_001204868NR_037943; NR_037942; NR_037941 cg02408153 2 177028655 HOXD3 NM_006898 TSS200 Shore -0.56 cg03109841 17 73007443 ICT1 NM_001545; NM_001303265 TSS1500 Shore 0.55 cg14391419 7 19158647 TWIST1 NM_000474 TSS1500 Shore -0.55 cg09258804 17 59532564 TBX4 NM_018488 TSS1500 Island -0.52 cg03370588 15 52821290 MYO5A NM_000259; NM_001142495 TSS20 Island 0.51 cg04481096 7 27142100 HOXA2 NM_006735 1stExon Shore -0.50 cg00886996 11 47599133 NDUFS3 NM_004551 TSS1500 Shore 0.50 cg16703737 11 47599085 NDUFS3 NM_004551 TSS1500 Shore 0.49 cg06653878 9 98080618 FANCC NM_001243743; NM_001243744; NM_000136 TSS1500 Shore 0.49 cg07011913 2 177014626 HOXD4; NM_014621; NR_029609 TSS1500 Shore -0.48 MIR10B cg08717880 2 177014849 HOXD4; NM_014621; NR_029609 TSS1500; TSS200 Shore -0.48 MIR10B cg09386073 17 73007747 ICT1 NM_001545 TSS1500 Shore 0.48 cg10624122 7 19158747 TWIST1 NM_000474 TSS1500 Shore -0.48 cg19692584 10 96161889 TBC1D12 NM_015188 TSS1500 Shore 0.48 cg06055873 7 27142204 HOXA2 NM_006735; NM_006735 1stExon; 5′UTR Shore -0.47 cg10126205 7 19158664 TWIST1 NM_000474 TSS1500 Shore -0.47 cg22445940 20 57224919 STX16 NM_001134773; NM_003763; NM_001134772; NM_001001433 TSS1500 Island -0.47 cg15543523 6 152127812 ESR1 NM_000125; NM_001122741; NM_001122740; NM_001122742 TSS1500; 5′UTR; Shore 0.47 TSS1500; 5′UTR cg24736345 19 3093246 GNA11 NM_002067 TSS1500 Shore 0.47 cg09583532 2 177015999 HOXD4 NM_014621 TSS200 Shore -0.46 a Identification of CpG in EPIC array. b Chromosome. c Δβ-values (β-value LM - β-value P). A positive Δβ-value indicates hypermethylation in LM relative to P, and a negative Δβ-value indicates hypomethylation in LM relative to P. P, primary tumor; LM, lung metastasis. The CpGs are order according to the absolute Δβ-value. All the CpGs showed FDR <0.05. Genome-wide DNA methylation analysis shore regions. Analysis of the 10,000 most identified DMCpGs in at the ITF shows a different epigenetic our study revealed an epigenetic signature able to clearly pattern in paired uADC differentiate the ITF of paired tumors in uADC patients. GO analysis of this epigenetic signature showed enrichment of The DNA methylation profile of the ITF in paired uADC differentially methylated genes involved in relevant biological patients was analyzed by Infinium Methylation EPIC array [EPIC processes related to signal transduction, regulation of array] (Figure 6A). Principal component analysis (PCA) showed transcription, cell differentiation and cell adhesion, among a different methylation pattern in the ITF of paired tumors others (Figure 6F). Table 2 shows the 25 most identified (Figure 6B). Importantly, we identified 44,085 significant DMCpGs of the epigenetic signature that are located at CGIs differentially methylated CpGs (DMCpGs) (p-value < 0.05; or CGI shore regions of gene promoters. FDR <0.05) between the ITF of paired tumors (DS). These Next, we analyzed the DNA methylation profile of the ITF in DMCpGs were distributed across the genome, mainly in the paired tumors of uLMS patients (Figure 7A). The PCA showed body and promoter region of genes (Figure 6D). According to the that the paired tumors had a similar DNA methylation pattern at density of CpGs, most differences were located in regions of DNA the ITF (Figure 7B). This concurs with the high similarity with low density of CpGs (open sea) and in CpG island (CGI) between the ITF of paired tumors showing the scatter plot in Frontiers in Cell and Developmental Biology 13 frontiersin.org Sonzini et al. 10.3389/fcell.2022.1052098 FIGURE 7 Genome-wide DNA methylation analysis of the ITF in paired tumors of uLMS patients. (A) Schematic flowchart used to identify significant differentially methylated CpGs at the ITF. (B) Principal component analysis (PCA) of DNA methylation data at the ITF. (C) Scatter plot representing mean normalized levels of DNA methylation (β-values) at the ITF of primary tumors and lung metastasis. Dots in red show significant differentially methylated CpGs. (D) Hierarchical clustering of the six differentially methylated CpGs (FDR <5%) found at the ITF of primary and lung metastasis of uLMS patients. DMCpGs, differentially methylated CpGs; Meth, DNA methylation; P, Primary tumors; LM, lung metastasis; uADC, uterine adenocarcinomas; uLMS, uterine leiomyosarcomas. Figure 7C. Although the majority of CpGs analysed (735775 valid were located mainly in the body and promoter region of genes CpGs) were similar, we found an epigenetic signature of six (Figure 8D). Evaluating the density of CpGs, most differences significant DMCpGs (p-value < 0.05; FDR <0.05) which were were distributed between open sea regions and CGIs. able to differentiate the ITF of paired tumors in uLMS patients Most of the 2,318 observed DMCpGs defined an epigenetic (Figure 7D). The six CpGs (cg03567028, cg02833127, signature that clearly differentiated the ITF of patients with LM- cg03370588, cg03885684, cg04821107, cg00566642) of this uADC vs LM-uLMS (Figure 8E). GO analysis of this epigenetic epigenetic signature are indicated in Table 3. signature revealed enrichment of genes involved in important biological processes such as signal transduction, transcription regulation and cell adhesion, among others (Figure 8F). Table 4 ITFs of metastatic tumors exhibit some shows the 25 most DMCpGs of the epigenetic signature found epigenetic differences that are located at CGIs or CGI shore regions of gene promoters. The DNA methylation pattern of the ITF was compared in LM-uADC and LM-uLMS patients using the EPIC array Discussion (Figure 8A). DNA methylation analysis by PCA showed some differences between metastases (Figure 8B). In particular, we Critical for metastasis formation in distant organs, tumor identified 2,318 DMCpGs (p-value < 0.05; FDR <0.05) between invasion is the most important prognostic factor and the main the ITF of LM-uADC and LM-uLMS (Figures 8A–C), which actual cause of death in cancer patients (Kunita et al., 2018). Frontiers in Cell and Developmental Biology 14 frontiersin.org Sonzini et al. 10.3389/fcell.2022.1052098 TABLE 3 All identified significant differentially methylated CpGs of the epigenetic signature that differentiate the ITF of paired tumors in uLMS patients. TargetIDa Chrb Position Gene name Accession number Gene region CpG density Δβc cg03567028 13 44544293 Intergenic Island -0.71 cg02833127 4 177116733 SPATA4 NM_144644; NM_144644 1stExon; 5′UTR Island 0.69 cg03370588 15 52821290 MYO5A NM_000259; NM_001142495 TSS200; TSS200 Island 0.40 cg03885684 2 120770471 EPB41L5 NM_020909 TSS200 Island 0.29 cg04821107 4 170947779 MFAP3L NM_021647 TSS1500 Island 0.19 cg00566642 1 156825288 NTRK1; INSRR NM_001007792; NM_014215 Body; Body Shelf -0.11 a Identification of CpG in EPIC array. b Chromosome. c Δβ-values (β-value LM - β-value P). A positive Δβ-value indicates hypermethylation in LM relative to P, and a negative Δβ-value indicates hypomethylation in LM relative to P. P, primary tumor; LM, lung metastasis. The CpGs are ordered according to the absolute Δβ-value. All the CpGs showed FDR <0.05. Tumor cells that spread from the primary tumor are dependent This linearized phenotype mirrors the pattern observed for on the TME, and presence of necessary stroma at secondary sites collagen I fibers in breast cancer, which has been related to is critical in ensuring their activity and survival (Barcus and increased aggressiveness, and was accompanied by a dense Longmore, 2021). In this study, we analyzed the ITF of uADC cancer-associated ECM with aligned collagen highways (Amos and uLMS paired tumors (uterine primary tumors and their and Choi, 2021) that tumor cells use to facilitate invasion (Cox respective lung metastases) based on the morphological and Erler, 2011). Moreover, second harmonic imaging parameters of the stroma component of the TME, reticulin microscopy of tissue and tumor slices revealed that collagen fibers, and their transcriptomic and epigenetic profiles, surrounding normal epithelia or early tumors is curly, smooth, looking for their potential predictive value in pulmonary and aligned parallel to the tumor surface, whereas invasive metastases. Given the importance of immune cell infiltration tumors present increased deposition of thickened, linearized, in tumor invasion and cancer treatment, we also sought to and arranged collagen fibers perpendicular to the tumor surface, characterize ITF-infiltrating immune cells at lung metastases. increasing the stiffness of the ECM. These observed changes Active invasion of cancer cells requires a process called prompted the development of a prognostic tool named the stromatogenesis, defined as the formation of specific stroma at tumor-associated collagen signature [TACS] (Provenzano the ITF after peritumoral fibroblast proliferation et al., 2006). According to these and other studies (Conklin (Giatromanolaki, Sivridis and Koukourakis, 2007). Tumor et al., 2011; Acerbi et al., 2015; Zhou et al., 2017; Jia et al., 2019; stroma consists of ECM, non-malignant cells and the Seo et al., 2020), increased linearization and reorientation of signaling molecules they produce. It is an integral and vital reticulin fibers at the ITF in uLMS as compared to uADC could component of primary tumors, evolving progressively as they indicate higher degree of aggressiveness. These results contrast develop, and is crucial in determining the growth characteristics, with the similarities previously found in the reticulin scaffolding morphology, and invasion of the tumor (Sangaletti et al., 2017). at the ITFs in primary tumors, regardless of where they The most abundant proteins within the ECM are fibrillar metastasized (Sanegre et al., 2021). The differences observed collagens I and III, whose quantity increases in many tumors in the present study could therefore be related to the crucial (Barcus and Longmore, 2021). The presence of excess collagen I role of the ECM in driving tumor progression at the primary site, fibers and altered reticulin fiber architecture in the stromal differences that are only evident when the fate of lung metastases microenvironment has been reported in oral squamous cell is considered. In addition, primary tumor-mediated ECM carcinoma, where differences were observed in quantity, remodeling is implicated in the systematic appropriation of distribution, and morphology at the ITF (Yinti et al., 2015). pre-metastatic niches (Cox and Erler, 2011), in which cells Another study in the same tumor type showed a change in length spread and colonize a secondary site, turning the TME into a and thickness of reticulin fibers at the ITF, which correlated with mechanically complex niche due to changes in ECM rigidity. In degree of malignancy (Walke and Bhagat, 2017). Increased fiber morphometric study of reticulin scaffolding, the loose fiber deposition was observed at the ITF in uLMS compared to uADC, plexus predominant in uADC could be associated with the and these fibers always appeared smaller in size, suggesting less pattern of ECM remodeling during the metastatic process spatial expansion. Furthermore, the same results were obtained deriving from its epithelial cell origin and the pulmonary when the ITF of the respective lung metastases were compared niche. In contrast, the dense and linearized fiber bundle without hyperplasia of type 2 pneumocytes. These differences formation in uLMS may be due to their mesenchymal origin could indicate organization of uLMS fibers into dense bundles. and subsequent development of a microenvironment with Frontiers in Cell and Developmental Biology 15 frontiersin.org Sonzini et al. 10.3389/fcell.2022.1052098 FIGURE 8 Genome-wide DNA methylation analysis of the ITF between metastatic tumors. (A) Schematic flowchart followed to obtain significant differentially methylated CpGs at the ITF. (B) Principal component analysis (PCA) of DNA methylation data at the ITF. (C) Scatter plot representing mean normalized levels of DNA methylation (β-values) at the ITF. Dots in red show significant differentially methylated CpGs. (D) Genomic distribution of the 2,318 significant differentially methylated CpGs in ITF by CpG context and gene region. (E) Hierarchical clustering of the 2,318 differentially methylated CpGs (FDR <5%) in the ITF of LM-uADC and LM-uLMS patients. (F) Gene ontology (GO) analysis of the biological process categories for the 2,318 differentially methylated CpGs at the ITF of LM-uADC and LM-uLMS. DMCpGs, differentially methylated CpGs; Meth, DNA methylation; LM, lung metastasis; uADC, uterine adenocarcinomas; uLMS, uterine leiomyosarcomas. Frontiers in Cell and Developmental Biology 16 frontiersin.org Sonzini et al. 10.3389/fcell.2022.1052098 TABLE 4 Top 25 significant differentially methylated CpGs from promoter CGIs/shore regions belonging to the epigenetic signature that differentiates the ITF of lung metastatic uADC and uLMS. TargetIDa Chrb Position Gene name Accession number Gene region CpG Δβc density cg20574490 chr8 102504666 GRHL2 NM_024915 TSS200 Island -0.52 cg04060128 chr16 31488852 TGFB1I1 NM_001042454; NM_015927; NM_001164719 Body; Body; Body Island -0.50 cg19814116 chr1 6086448 KCNAB2 NM_172130; NM_003636; NM_003636; 1stExon; 1stExon; 5′UTR; Island -0.49 NM_172130 5′UTR cg11867697 chr12 122356394 WDR66 NM_144668; NM_001178003 TSS200; TSS200 Island -0.49 cg03780851 chr8 102504564 GRHL2 NM_024915 TSS200 Island -0.49 cg07637175 chr8 102504256 GRHL2 NM_024915 TSS1500 Shore -0.48 cg05984244 chr16 31488238 TGFB1I1 NM_001042454; NM_015927; NM_001164719 Body; Body; Body Island -0.48 cg01970383 chr5 177729224 COL23A1 NM_173465 Body Island -0.48 cg02047049 chr9 35790836 Island 0.47 cg10191210 chr5 42992128 Island -0.46 cg05230013 chr6 43212262 TTBK1 NM_032538 5′UTR Shore 0.46 cg25658463 chr19 14543048 PKN1 NM_002741 TSS1500 Island -0.45 cg01121275 chr5 115696282 Shore 0.45 cg07359545 chr22 19711327 GP1BB; NM_000407; NM_002688 Body; 3′UTR Island -0.45 SEPT5 cg26643631 chr8 102504903 GRHL2 NM_024915; NM_024915 5′UTR; 1stExon Shore -0.45 cg10207240 chr12 122356781 WDR66 NM_144668 5′UTR Island -0.44 cg18465694 chr22 36806185 Island -0.43 cg26054842 chr7 155915712 Island 0.43 cg19769613 chr1 182028259 ZNF648 NM_001009992 5′UTR Shore 0.43 cg17315219 chr19 48833559 EMP3 NM_001425 Body Island -0.43 cg12258785 chr3 24537407 THRB NM_001128177; NM_001128176; NM_000461 TSS1500; TSS1500; TSS1500 Island -0.42 cg14202477 chr8 102504859 GRHL2 NM_024915; NM_024915 1stExon; 5′UTR Shore -0.42 cg08334760 chr8 102504562 GRHL2 NM_024915 TSS200 Island -0.42 cg14056864 chr3 146187210 PLSCR2 NM_001199978; NM_001199979; NM_020359 TSS200; Body; 5′UTR Island -0.42 cg26523005 chr3 42947267 ZNF662 NM_001134656; NM_207404 TSS1500; TSS200 Shore 0.42 a Identification of CpG in EPIC array. b Chromosome. c Δβ-values (β-value LM-uADC - β-value LM-uLMS). A positive Δβ-value. Indicates hypermethylation in LM-uADC relative to uLMS, and a negative Δβ-value indicates hypomethylation in LM-uADC relative. To LM-uLMS. LM, lung metastasis. The CpGs are ordered according to the absolute Δβ-value. All the CpGs showed FDR <0.05. cellular and non-cellular components different to those found in dictate the migratory trajectory of the T cells, preventing the lung (Saggioro et al., 2020). infiltration into the tumor. Consistently, a less dense matrix Besides its role in tumor invasion, collagen fibers from the increased the ability of T cells to meet tumor cells, promoting TME affect T-cell differentiation and spatial distribution to immune infiltration into the tumor (Salmon et al., 2012). In the modulate antitumorigenic immunity (Zhou et al., 2022). For tumor cohort from this study, the results from the multiplex example, the number of T cells present within a tumor and their immunofluorescence-based immune profiling showed no ability to migrate and reach tumor cells are critical to achieving differences in the spatial infiltration pattern of T cells, an effective antitumor response. It has been observed that in neutrophils, and macrophages. However, CD20+ B human lung and ovarian cancers, T cells preferentially lymphocytes were found in significant numbers at the ITF, accumulate in stromal regions that exhibit loose matrix with a higher density in LM-uADC. Interestingly, an important architecture, rather than in dense regions. Therefore, role in the adaptive immune response is played by B remodeling and stiffening of the ECM and its components lymphocytes, often characterized as a favorable prognostic could be a barrier to T lymphocyte migration and infiltration biomarker in many cancer types. In fact, a study in (Nicolas-Boluda et al., 2021). Furthermore, aligned fibers in and colorectal cancer (CRC) revealed that patients with tumors around the perivascular regions of the tumor epithelial cells highly infiltrated by CD20+ B lymphocytes were found to have Frontiers in Cell and Developmental Biology 17 frontiersin.org Sonzini et al. 10.3389/fcell.2022.1052098 prognostic advantage (Edin et al., 2019) and non-smoker 2018). In uterine neoplasms, regulation of cell adhesion patients with non-small-cell-lung cancer (NSCLC) showed processes has shown a key association with invasion/ better survival outcomes when tumors contained high migration and with the immune/inflammatory milieu (Abal number of tumor-infiltrating CD20+ B cells (Kinoshita et al., et al., 2007; Rutherford, Hill and Hopkins, 2018), indicating 2016). In line with the immune infiltration pattern, our findings that the differences in methylation profile observed at the ITF from the transcriptomic analyses revealed increased B cell and of uADC and uLMS patients could contribute to explaining CD4+ T cell populations and activation of adaptive immune the differing aggressiveness of lung metastasis. response and lymphocytes at the ITF of LM-uADC in Our focus on the morphometric parameters of reticulin comparison with LM-uLMS. Indeed, activation of somatic fibers revealed loose plexus scaffolding in LM-uADCs recombination and diversification of immunoglobulins in compared with the linearized fiber pattern in LM-uLMS. ITF of LM-uADC probably reflect immunoglobulin These differences, together with the increased quantity of production by B cells. When considering the morphometric immune cell infiltrates in LM-uADCs, suggest a lower degree parameters of reticulin fibers in the same context, these results of aggressiveness in LM-uADCs. In this context, LM-uADCs may suggest a correlation between the selective accessibility of showed an increased methylation pattern of genes involved in immune cells and reticulin fiber scaffolding. The large and cell adhesion, supporting the key role of this epigenetic isolated fibers with loose plexus at the ITFs in LM-uADCs could mechanism in regulating the ECM and tumor immune facilitate the passage of immune infiltrates such as CD4+ T and milieu of LM-ADCs. Furthermore, transcriptomic analyses CD20+ B cells, whereas the dense bundled fiber network of LM- revealed increased immune infiltrate in LM-uADCs, uLMS may hinder immune infiltration. In line with this corresponding mainly to B cell and CD4+ T cell lineages, hypothesis, greater immune infiltrates might be indicative of with concomitant activation of an adaptive immune better prognosis in LM-uADC patients than in LM-uLMS response. Finally, upregulation of antimicrobial response patients. genes plus activation of chemotaxis could also contribute to Transcriptomic profiling revealed that antimicrobial better prognosis in LM-uADCs, as observed in our previous peptides (LCN2, DEFB1) were upregulated at the ITF of LM- study. We therefore propose an integrative approach uADC compared to LM-uLMS, as reported at the ITF of combining these methodologies as a prime target for analysis primary tumors in our recent study (Sanegre et al., 2021). of tumorigenesis and development in primary and lung Moreover, activation of granulocyte chemotaxis and metastases at the ITF, which may improve precision migration was also observed in LM-uADC, which may diagnostics and therapeutics in patients suffering from further potentiate this antimicrobial response. Thus, an aggressive uterine tumors. impaired antimicrobial humoral response could be implicated in distant spread of uLMS. Interestingly, the ITF in LM-uLMS showed activation of DNA repair processes that Data availability statement could be a consequence of intense genotoxic treatment regimens. Furthermore, comparing activation of processes in The datasets presented in this study can be found in online uLMS metastasis vs. primary tumors, events such as lung repositories. The names of the repository/repositories and morphogenesis, cell-cell adhesion and epithelial cell accession number(s) can be found below. For primary tumors differentiation may correspond to the pulmonary of uADC and uLMS, DNA methylation and expression profiling microenvironment at the metastatic location. data are deposited with accession numbers GSE171142 (https:// Epigenetic profile analysis revealed more DNA www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE171142) and methylation differences in paired samples in uADC GSE214779 (GSE214779 https://www.ncbi.nlm.nih.gov/geo/ compared to paired samples in uLMS. In addition, a query/acc.cgi?acc=GSE214779), respectively. For metastatic different methylation profile was found at the ITF of LM- tumors, accession numbers are GSE216313 (https://www.ncbi. uADC and LM-uLMS. These variations in epigenetic profile nlm.nih.gov/geo/query/acc.cgi?acc=GSE216313) and GSE172043 were in partly related to changes in the methylation levels of (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE172043), genes involved in cell adhesion processes. These results are in respectively. line with findings from our previous study (Sanegre et al., 2021) where we observed a different methylation profile linked to cell adhesion in the ITF of primary tumors from Ethics statement uADC and uLMS patients. Cell adhesion is a biological process involved in maintaining the cell architecture The studies involving human participants were reviewed and (Ladoux and Mège, 2017) and regulating diverse approved by Comité Coordinador de Ética de la Investigación homeostatic processes, including angiogenesis, migration Biomédica de Andalucía: Deciphering the site-specific tumor and barrier function (Ebnet, 2017; Ford and Rajagopalan, microenvironment of advanced uterine tumors (Biobank code: Frontiers in Cell and Developmental Biology 18 frontiersin.org Sonzini et al. 10.3389/fcell.2022.1052098 S1800086) and Comité de Ética de la Investigación with the codes by a “Juan Rodés” contract from the ISCIII (JR17/00016). The CEIC-1892, CEIC-2083, and CEIC-1858. Written informed consent funders had no involvement in the research process nor in the for participation was not required for this study in accordance with preparation and submission of the article. national legislation and institutional requirements. Acknowledgments Author contributions The authors thank Kathryn Davies for English corrections. RN and XM-G contributed to study conception and design. GS and SGA wrote the first draft of the manuscript. GS, SG-A, NE, CA, JD-M, and AD-L wrote sections of the manuscript. BD, RL-L, SR, EA, Conflict of interest and XM-G provided the clinicopathological data. NE provided the collaborative consortium with microdissected samples. GS, SG-A, SS, The authors declare that the research was conducted in the and RN designed and performed morphometric characterization of absence of any commercial or financial relationships that could reticulin fibers. CA, DG-R, and IM performed multiplex staining and be construed as a potential conflict of interest. analysis. JD-M, CS-A, and EA evaluated the transcriptomic profile. AB-C, AD-L, and NC-F performed genome-wide DNA methylation analysis. All authors contributed to manuscript revision, read, and Publisher’s note approved the submitted version. All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated Funding organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or This research was supported by grants from the ISCIII and claim that may be made by its manufacturer, is not guaranteed or ERDF (PI17/01558), by the CIBERONC (contracts CB16/12/ endorsed by the publisher. 00484, CB16/12/0328, CB16/12/00363, CB16/12/00364, CB16/ 12/00481, and CB16/12/00231) and Grupos Coordinados Estables from the Asociación Española Contra el Cáncer Supplementary material (AECC). AB-C was funded by a predoctoral contract PFIS from ISCIII (FI19/00240) co-funded by “Fondo Social The Supplementary Material for this article can be found Europeo” (FSE). NC-F was funded by a predoctoral contract online at: https://www.frontiersin.org/articles/10.3389/fcell.2022. from “Xunta de Galicia” (IN606A-2020/004). AD-L was funded 1052098/full#supplementary-material References Abal, M., Llauradó, M., Doll, A., Monge, M., Colas, E., González, M., et al. (2007). Bankhead, P., Loughrey, M. B., Fernández, J. A., Dombrowski, Y., McArt, D. G., Molecular determinants of invasion in endometrial cancer. Clin. Transl. Oncol. 9 Dunne, P. D., et al. (2017). QuPath: Open source software for digital pathology (5), 272–277. doi:10.1007/s12094-007-0054-z image analysis. Sci. Rep. 7 (1), 16878. doi:10.1038/s41598-017-17204-5 Abengozar-Muela, M., Esparza, M. V., Garcia-Ros, D., Vásquez, C. E., Echeveste, Barcus, C. E., and Longmore, G. D. (2021). Collagen linearization within tumors. J. I., Idoate, M. A., et al. (2020). Diverse immune environments in human lung Cancer Res. 81 (22), 5611–5612. doi:10.1158/0008-5472.CAN-21-2939 tuberculosis granulomas assessed by quantitative multiplexed immunofluorescence. Carbone, A., Borok, M., Damania, B., Gloghini, A., Polizzotto, M. N., Jayanthan, Mod. Pathol. 33 (12), 2507–2519. R. K., et al. (2021). Castleman disease. Nat. Rev. Dis. Prim. 7 (1), 84. doi:10.1038/ Acerbi, I., Cassereau, L., Dean, I., Shi, Q., Au, A., Park, C., et al. (2015). Human s41572-021-00317-7 breast cancer invasion and aggression correlates with ECM stiffening and immune Conklin, M. W., Eickhoff, J. C., Riching, K. M., Pehlke, C. A., Eliceiri, K. W., cell infiltration. Integr. Biol. 7 (10), 1120–1134. doi:10.1039/c5ib00040h Provenzano, P. P., et al. (2011). Aligned collagen is a prognostic signature for Alami, H., Lehoux, P., Gagnon, M. P., Fortin, J. P., Fleet, R., and Ag Ahmed, M. A. survival in human breast carcinoma. Am. J. Pathol. 178 (3), 1221–1232. doi:10.1016/ (2020). Rethinking the electronic health record through the quadruple aim: Time to j.ajpath.2010.11.076 align its value with the health system. BMC Med. Inf. Decis. Mak. 20 (1), 32. doi:10. Cox, T. R., and Erler, J. T. (2011). Remodeling and homeostasis of the 1186/s12911-020-1048-9 extracellular matrix: Implications for fibrotic diseases and cancer. Dis. Model. Amos, S. E., and Choi, Y. S. (2021). The cancer microenvironment: Mechanical Mech. 4 (2), 165–178. doi:10.1242/dmm.004077 challenges of the metastatic cascade. Front. Bioeng. Biotechnol. 9, 625859. doi:10. Eble, J. A., and Niland, S. (2019). The extracellular matrix in tumor progression and 3389/fbioe.2021.625859 metastasis. Clin. Exp. Metastasis 36 (3), 171–198. doi:10.1007/s10585-019-09966-1 Aran, D., Hu, Z., and Butte, A. J. (2017). xCell: digitally portraying the tissue Ebnet, K. (2017). Junctional adhesion molecules (JAMs): Cell adhesion receptors cellular heterogeneity landscape. Genome Biol. 18 (1), 220. doi:10.1186/s13059-017- with pleiotropic functions in cell physiology and development. Physiol. Rev. 97 (4), 1349-1 1529–1554. doi:10.1152/physrev.00004.2017 Assenov, Y., Müller, F., Lutsik, P., Walter, J., Lengauer, T., and Bock, C. (2014). Edin, S., Kaprio, T., Hagström, J., Larsson, P., Mustonen, H., Böckelman, C., et al. Comprehensive analysis of DNA methylation data with RnBeads. Nat. Methods 11 (2019). The prognostic importance of CD20(+) B lymphocytes in colorectal cancer (11), 1138–1140. doi:10.1038/nmeth.3115 Frontiers in Cell and Developmental Biology 19 frontiersin.org Sonzini et al. 10.3389/fcell.2022.1052098 and the relation to other immune cell subsets. Sci. Rep. 9 (1), 19997. doi:10.1038/ Saggioro, M., D’Angelo, E., Bisogno, G., Agostini, M., and Pozzobon, M. (2020). s41598-019-56441-8 Carcinoma and sarcoma microenvironment at a glance: Where we are. Front. Oncol. 10, 76. doi:10.3389/fonc.2020.00076 Ford, A. J., and Rajagopalan, P. (2018). Extracellular matrix remodeling in 3D: Implications in tissue homeostasis and disease progression. Wiley Interdiscip. Rev. Salmon, H., Franciszkiewicz, K., Damotte, D., Dieu-Nosjean, M. C., Validire, P., Nanomed. Nanobiotechnol. 10 (4), e1503. doi:10.1002/wnan.1503 Trautmann, A., et al. (2012). Matrix architecture defines the preferential localization and migration of T cells into the stroma of human lung tumors. J. Clin. Invest. 122 Galli, F., Aguilera, J. V., Palermo, B., Markovic, S. N., Nisticò, P., and Signore, A. (2020). (3), 899–910. doi:10.1172/JCI45817 Relevance of immune cell and tumor microenvironment imaging in the new era of immunotherapy. J. Exp. Clin. Cancer Res. 39 (1), 89. doi:10.1186/s13046-020-01586-y Sanegre, S., Eritja, N., de Andrea, C., Diaz-Martin, J., Diaz-Lagares, A., Jacome, M. A., et al. (2021). Characterizing the invasive tumor front of aggressive uterine Giatromanolaki, A., Sivridis, E., and Koukourakis, M. I. (2007). The pathology of adenocarcinoma and leiomyosarcoma. Front. Cell Dev. Biol. 9, 670185. doi:10.3389/ tumor stromatogenesis. Cancer Biol. Ther. 6 (5), 639–645. doi:10.4161/cbt.6.5.4198 fcell.2021.670185 Gleitz, H. F. E., Benabid, A., and Schneider, R. K. (2021). Still a burning question: Sangaletti, S., Chiodoni, C., Tripodo, C., and Colombo, M. P. (2017). The good The interplay between inflammation and fibrosis in myeloproliferative neoplasms. and bad of targeting cancer-associated extracellular matrix. Curr. Opin. Pharmacol. Curr. Opin. Hematol. 28 (5), 364–371. doi:10.1097/moh.0000000000000669 35, 75–82. doi:10.1016/j.coph.2017.06.003 Huvila, J., Pors, J., Thompson, E. F., and Gilks, C. B. (2021). Endometrial Schneider, C. A., Rasband, W. S., and Eliceiri, K. W. (2012). NIH image to ImageJ: carcinoma: Molecular subtypes, precursors and the role of pathology in early 25 years of image analysis. Nat. Methods 9 (7), 671–675. doi:10.1038/nmeth.2089 diagnosis. J. Pathol. 253 (4), 355–365. doi:10.1002/path.5608 Schwartzberg, L., Kim, E. S., Liu, D., and Schrag, D. (2017). Precision oncology: Jia, H., Janjanam, J., Wu, S. C., Wang, R., Pano, G., Celestine, M., et al. (2019). The Who, how, what, when, and when not? Am. Soc. Clin. Oncol. Educ. Book. 37, tumor cell-secreted matricellular protein WISP1 drives pro-metastatic collagen 160–169. doi:10.1200/EDBK_174176 linearization. Embo J. 38 (16), e101302. doi:10.15252/embj.2018101302 Seo, B. R., Chen, X., Ling, L., Song, Y. H., Shimpi, A. A., Choi, S., et al. (2020). Kai, F., Drain, A. P., and Weaver, V. M. (2019). The extracellular matrix modulates Collagen microarchitecture mechanically controls myofibroblast the metastatic journey. Dev. Cell 49 (3), 332–346. doi:10.1016/j.devcel.2019.03.026 differentiation. Proc. Natl. Acad. Sci. U. S. A. 117 (21), 11387–11398. doi:10. Kinoshita, T., Muramatsu, R., Fujita, T., Nagumo, H., Sakurai, T., Noji, S., et al. 1073/pnas.1919394117 (2016). Prognostic value of tumor-infiltrating lymphocytes differs depending on Sivridis, E., Giatromanolaki, A., and Koukourakis, M. I. (2004). Stromatogenesis" histological type and smoking habit in completely resected non-small-cell lung and tumor progression. Int. J. Surg. Pathol. 12 (1), 1–9. doi:10.1177/ cancer. Ann. Oncol. 27 (11), 2117–2123. doi:10.1093/annonc/mdw319 106689690401200101 Kunita, A., Baeriswyl, V., Meda, C., Cabuy, E., Takeshita, K., Giraudo, E., et al. Tabas-Madrid, D., Nogales-Cadenas, R., and Pascual-Montano, A. (2012). (2018). Inflammatory cytokines induce podoplanin expression at the tumor GeneCodis3: A non-redundant and modular enrichment analysis tool for invasive front. Am. J. Pathol. 188 (5), 1276–1288. doi:10.1016/j.ajpath.2018.01.016 functional genomics. Nucleic Acids Res. 40, W478–W483. (Web Server issue). Ladoux, B., and Mège, R. M. (2017). Mechanobiology of collective cell behaviours. doi:10.1093/nar/gks402 Nat. Rev. Mol. Cell Biol. 18 (12), 743–757. doi:10.1038/nrm.2017.98 Tadeo, I., Berbegall, A. P., Castel, V., García-Miguel, P., Callaghan, R., Påhlman, Mao, W., Wei, S., Yang, H., Yu, Q., Xu, M., Guo, J., et al. (2020). S., et al. (2016). Extracellular matrix composition defines an ultra-high-risk group of Clinicopathological study of organ metastasis in endometrial cancer. Future neuroblastoma within the high-risk patient cohort. Br. J. Cancer 115 (4), 480–489. Oncol. 16 (10), 525–540. doi:10.2217/fon-2020-0017 doi:10.1038/bjc.2016.210 Montironi, R., Cimadamore, A., Scarpelli, M., Cheng, L., Lopez-Beltran, A., Tamborero, D., Dienstmann, R., Rachid, M. H., Boekel, J., Baird, R., Braña, I., Mikuz, G., et al. (2020). Pathology without microscope: From a projection screen to et al. (2020)., 7. United States, 992–994. doi:10.1038/s41591-020-0969-2Support a virtual slide. Pathol. Res. Pract. 216 (11), 153196. doi:10.1016/j.prp.2020.153196 systems to guide clinical decision-making in precision oncology: The cancer core europe molecular tumor board portalNat. Med. Moran, S., Arribas, C., and Esteller, M. (2016). Validation of a DNA methylation microarray for 850, 000 CpG sites of the human genome enriched in enhancer Ushiki, T. (2002). Collagen fibers, reticular fibers and elastic fibers. A sequences. Epigenomics 8 (3), 389–399. doi:10.2217/epi.15.114 comprehensive understanding from a morphological viewpoint. Arch. Histol. Cytol. 65 (2), 109–126. doi:10.1679/aohc.65.109 Mungenast, F., Fernando, A., Nica, R., Boghiu, B., Lungu, B., Batra, J., et al. (2021). Next-generation digital histopathology of the tumor microenvironment. Genes Xiao, Y., and Yu, D. (2021). Tumor microenvironment as a therapeutic target in (Basel) 12 (4), 538. doi:10.3390/genes12040538 cancer. Pharmacol. Ther. 221, 107753. doi:10.1016/j.pharmthera.2020.107753 Paolillo, M., and Schinelli, S. (2019). Extracellular matrix alterations in metastatic Zhang, Y., and Zhang, Z. (2020). The history and advances in cancer processes. Int. J. Mol. Sci. 20 (19), E4947. doi:10.3390/ijms20194947 immunotherapy: Understanding the characteristics of tumor-infiltrating immune cells and their therapeutic implications. Cell. Mol. Immunol. 17 (8), 807–821. doi:10. Provenzano, P. P., Eliceiri, K. W., Campbell, J. M., Inman, D. R., White, J. G., and 1038/s41423-020-0488-6 Keely, P. J. (2006). Collagen reorganization at the tumor-stromal interface facilitates local invasion. BMC Med. 4 (1), 38. doi:10.1186/1741-7015-4-38 Zhou, Y., Tian, Q., Gao, H., Zhu, L., Zhang, Y., Zhang, C., et al. (2022). Immunity and extracellular matrix characteristics of breast cancer subtypes based on Roberts, M. E., Aynardi, J. T., and Chu, C. S. (2018). Uterine leiomyosarcoma: A identification by T helper cells profiling. Front. Immunol. 13, 859581. doi:10. review of the literature and update on management options. Gynecol. Oncol. 151 (3), 3389/fimmu.2022.859581 562–572. doi:10.1016/j.ygyno.2018.09.010 Zhou, Z. H., Ji, C. D., Xiao, H. L., Zhao, H. B., Cui, Y. H., and Bian, X. W. Rutherford, E. J., Hill, A. D. K., and Hopkins, A. M. (2018). Adhesion in physiological, (2017). Reorganized collagen in the tumor microenvironment of gastric cancer benign and malignant proliferative states of the endometrium: Microenvironment and and its association with prognosis. J. Cancer 8 (8), 1466–1476. doi:10.7150/jca. the clinical big picture. Cells 7 (5), E43. doi:10.3390/cells7050043 18466 Frontiers in Cell and Developmental Biology 20 frontiersin.org