Biotech

Spatially resolved gene expression profiling of tumor microenvironment reveals key steps of lung adenocarcinoma development

Time Stamp: December 5, 2024 7:00 PM
Source Node: 786606

Overview of study design

In this study, spatial omics analyses were performed on 30 lung adenocarcinoma cases (eight IA and 22 AIS/MIA cases). The datasets used in this study are listed in Supplementary Table 1. Genetic and clinico-pathological information for IA cases is summarized in Table 1. For the AIS/MIA cases, we published genomic statuses and clinico-pathological information recently¹⁶ (also see Supplementary Table S1).

Table 1 General information of the eight IA cases

Full size table

The overall workflow for this study is depicted in Fig. 1. Spatial transcriptome sequencing¹⁰ Visium was performed on 16 tissue sections from eight IA cases with EGFR or KRAS driver mutations (Supplementary Table 2). We then analyzed Visium datasets from 28 sections of 22 AIS/MIA cases, 25 of which were newly obtained (Supplementary Table S3). Both fresh frozen (FF) and formalin-fixed and paraffin-embedded (FFPE) specimens were utilized depending on the samples. For the higher resolution analysis, PhenoCycler multiplexed immunostaining was performed on representative cases (Supplementary Tables S4 and S5). Finally, spatial expression profiling Xenium was conducted, and 17 sections from five IA and 12 AIS/MIA cases (two from our previous study¹⁶) were analyzed (Supplementary Tables S6 and S7). Using Visium data from IA cases (Supplementary Fig. S1; Supplementary Fig. S2 for PhenoCycler data), we first examined the observed diverse expression profiles for several individual cases. More specifically, we focused on LUAD No. 2 and 3, which harbored KRAS or EGFR driver mutations, respectively. EGFR and KRAS are two of the most important driver genes in lung adenocarcinoma, accounting for a large proportion of cases. We used data from FFPE sections that included regions with transcriptomic/histological features of both well-differentiated and malignant regions, which is critical for understanding tumor cell and microenvironmental characteristics in boundary regions. Then, we attempted to generalize the factors that would determine the fate of the cancers.

Spatial expansion patterns of LUAD No. 2 (KRAS, mucinous cancer)

First, we examined one case, LUAD No. 2, as an example, before proceeding more general analysis. This case was selected as a representative IA case because it harbors a common driver KRAS mutation (Table 1). In this case, Visium spots were divided into 11 clusters (FFPE section C, Fig. 2a; information on all sections of LUAD No. 2 is depicted in Supplementary Fig. S3). Normal or well-differentiated regions (clusters 1, 3, 4, and 5) were identified by generally high expression of NAPSA and surfactant genes^17,18, indicating that these regions possessed normal alveolar epithelial cell characteristics (Fig. 2b and Supplementary Fig. S4b). Notably, cluster 3 showed high DUOX1 expression (Supplementary Fig. S4c), indicating that tumor cells in these regions are subjected to oxidative stress¹⁹. Cluster 6 contained an inflammatory region with high levels of IFITM1 expression. Mucinous regions exhibited high MUC5AC expression²⁰ in clusters 0 and 8. Clusters 4 and 9 were immune cell-rich regions, whereas clusters 2, 9, and 10 had COL1A1, ACTA2, and SPARC expression, which are fibroblast and CAF markers (Supplementary Fig. S4c and d).

**Fig. 2: Spatial transcriptome analysis of LUAD No. 2.**

We attempted to identify the key factors that influence these patterns. We examined changes in gene expression at the cluster boundaries. In this case, as well as other study cases described below, tumor cells showed drastic changes in gene expression patterns as they crossed the borders defined by regions of immune cells, which appeared as a “stream.”

In LUAD No. 2, pathological examination roughly divided the cancer into two parts: 1) the left side, which represented normal or less malignant well-differentiated tumor cells, and 2) the right side, which represented more malignant or mucinous regions (Supplementary Fig. S4a). Accordingly, the Visium data revealed high expression of typical differentiated markers, such as NAPSA, on the left side of the clusters (Fig. 2b). In this region, a vital transcription factor of alveolar epithelium lineages, NKX2-1, was active (left, Fig. 2c). In contrast, in the right region, NKX2-1 expression was lost, while another transcription factor, HNF4A, was active (right, Fig. 2c). HNF4A is a key transcription factor in gastrointestinal epithelium, which suggests that tumor cells may transdifferentiate in this region^21,22. Consistently, mucin production began, as evidenced by the expressions CDX2 and MUC2 (region 3, Supplementary Fig. S4a).

A closer look at this boundary region revealed the presence of a cluster labeled “active inflammatory reactions” (cluster 6). Here, typical inflammatory response genes, such as IFITM1 and IFI6, were found to be highly activated (Fig. 2d) and associated with interferon signaling (Fig. 2e). Adjacent to this region, we discovered an immune cell-rich cluster (cluster 4) near a lymphoid follicle structure (region 5, Supplementary Fig. S4a), implying extensive immune cell infiltration and attack in this region as a result or cause of the tumor cell’s drastic transition. In this boundary region, tumor cells activated IDO1 (cluster 6; log2 fold change = 0.86, adjusted p-value = 2.0e-65). IDO1 activation was demonstrated at the protein level (Fig. 2f). IDO1 is induced by IFN-γ (type II interferon) and suppresses effector T and NK cells while activating regulatory T cells (Tregs) and myeloid-derived suppressor cells²³.

As they progressed to the right end region, tumor cells developed new transcriptional and phenotypic characteristics. In the end region, protective mucin expression was no longer detected (Fig. 2g), and tumor cells exhibited high proliferative markers. In their surrounding stroma, CAF markers ACTA2 and SPARC were found to be highly expressed (Fig. 2h and i). When we looked at the boundary between the mucinous region and the CAF-rich invasive region (cluster 2), we discovered high expression of genes associated with extracellular matrix remodeling and anoikis resistance, such as SPINK1^24,25 and MMP7^26,27,28 (Fig. 2g and i).

Collectively, it was proposed that tumor cells in the initial boundary region (cluster 6 between clusters 4 and 0) had begun to develop inhibitory responses to immune cells. It is possible that the mucin expression was induced to protect tumor cells from immune cell attacks. Furthermore, it was discovered that tumor cells in this region began to evolve into a more malignant state. In the second boundary (clusters 0 and 2), tumor cells were surrounded by a fibroblast blanket, and immune cell attacks were significantly reduced. Furthermore, CXCL14, an invasion-associated factor^29,30, with an unknown role in the TME³¹, was overexpressed in this region. At this stage, protection from mucin may not be as beneficial to tumor cells as it was in the less-invasive stages, particularly for expansion.

Spatial expression of LUAD No. 3 (EGFR, non-mucinous cancer)

We looked LUAD No. 3 (FFPE section B) for another example (Fig. 3a and b). This case was chosen as an example of cases having another common EGFR mutation (Table 1). The pathological diagnosis revealed a more complicated regional structure, so this was chosen as the case. Pathologically, this tissue predominantly showed a papillary pattern. The tumor cells are thought to have originated in clusters 1 or 9. During their expansion from this region to other regions, such as clusters 6 and 3 regions, the tumor cells exhibited a more malignant expression pattern, such as high expression levels of malignant markers, such as TNC and TGFBI (Supplementary Fig. S5a and b), as well as active proliferation markers, such as FOS and WEE1 (Supplementary Fig. S5c). EMT markers like RHOB and VIM, were also upregulated in cluster 3 (Fig. 3c). Between clusters 6 and 3, a large immune cell-rich region (cluster 5) was discovered (Fig. 3d). The region contains various immune cell types, such as B cells, T cells, and macrophages, in this region. However, we discovered that resolution in Visium analysis was insufficient to analyze the detailed distribution patterns of individual immune cells and their interactions with tumor cells.

**Fig. 3: Local transitions to invasive phenotypes in LUAD No. 3.**

To obtain finer-resolution images, we used multiplexed immunostaining analysis with PhenoCycler at least for genes of interest. Because the vertically consecutive sections from Visium analysis needed to be used for PhenoCycler multiplexed immunostaining analysis, we created a bioinformatics pipeline to superimpose images from Visium and PhenoCycler (Supplementary Fig. S6; detailed description is provided in the “Methods” section). The PhenoCycler analysis enabled highly resolution spatial analysis (Fig. 3e and Supplementary Fig. S6).

By combining Visium and PhenoCycler data, we were able to investigate the immune cell distribution pattern in cluster 5. We discovered that the characteristic immune cells did not always to have a uniform distribution even in this region. The expression of exhausted T cells increased in the left peripheral region of cluster 5 (“Downstream” in Fig. 3e), which is consistent with the findings of the PhenoCycler analysis of FOXP3 + CD4 T cells (Supplementary Fig. S5e). This lends support to the notion that the TME of the corresponding part of cluster 5 should reflect more exhausted features of immune statuses. In contrast, in central (“upstream”) regions of the left side, the distribution of active cytotoxic immune cell populations (CTLs, plasma cells, etc.) remained relevant. Consistently, tumor cell presence was more relevant in downstream regions than in upstream areas.

CXCL13+ immune cells co-localized with CXCR5+ cells in cluster 6, which spread from cluster 5’s immune cell region (Fig. 3f). Tumor-associated macrophages (TAMs) that express MMP9 and APOE were also present in this cluster. Toward the nearest region to cluster 3, CTL markers, such as PRF1, GZMA, and GZMK, were highly expressed. However, they did not invade cluster 3’s internal region (Supplementary Fig. S5d). Macrophages with high SPP1 expression, which reportedly play an important role as anti-inflammatory TAMs involved in angiogenesis³², were found in the region of cluster 3 (Fig. 3g and Supplementary Fig. S5f and h). These macrophages expressed both the M2 macrophage marker CD163 and the alveolar macrophage marker MARCO. SPP1+ macrophages may interact with CAFs³³ and induce EMT³⁴. Consistently, fibroblast markers, such as ACTA2, were elevated in cluster 3 (Fig. 3g).

Taken together, we believe the following scenarios could shape the cancer landscape. First, the tumor cells crossed the barrier at the downstream part of cluster 5 and spread into the region of cluster 3. Prior to this potential expansion, in the region of cluster 6, immune cell subsets, such as CXCL13+ lymphocytes and TAMs, were important. These cells are thought to be associated with the pro-inflammatory microenvironment, which may trigger an effective response to PD-L1 blockade³⁵. Therefore, tumor cells and TME with immune cell infiltration in cluster 6 should continue to respond to immune checkpoint blockade (ICB). Once cancer cells had spread to the regions of cluster 3, interactions between SPP1+ macrophages and fibroblasts were observed, which may aid in avoiding cytotoxicity by immune cells and, as a result, EMT in tumor cells themselves. When cancer tissue reaches this stage, it is more likely that ICB efficacy will be limited. Such heterogeneity in tumor cell status and TME may contribute to patients’ overall ineffective response and resistance to immunotherapies. The inferred molecular features of TMEs in clusters 3, 5, and 6, are summarized in Fig. 3h.

Diverse TMEs and tumor cells in the same block of LUAD No. 3

On the other side of the same section in LUAD No. 3, we discovered even more diverse microenvironment statuses developed. The region of cluster 11 did not contain any tumor cells. Predominant naïve immune cell markers, such as CCR7, were found instead (Supplementary Fig. S5g). Gaps were discovered in several locations where tumor cells were likely to have originated, implying that cancer cells that migrated in this direction were killed by immune cells.

Cluster 10 contained the cancer’s most invasive and potentially fatal component. Similar to clusters 3 and 6, a stream of immune cells (cluster 5) was found in the peripheral region of cluster 10. In this case, there was no clear separation between exhausted immune cells (Supplementary Fig. S5g). ACTA2 was highly expressed in this region, indicating that the tumor cells were closely linked to myofibroblasts and CAFs (Fig. 3i). Furthermore, we found high levels of a number of relevant matrix metalloproteinases, including MMP7 and MMP11, in the cluster 10 region (Fig. 3i and j). Immune cells were almost completely excluded from this area. For this region, we believe that when tumor cells spread across the fibroblast zone, they may have developed the ability to control interactions with fibroblasts. This ability would be further used to exclude immune cells from tumor cells, potentially allowing the cancer to spread further.

Among all regions, the most notable discussion should be as follows. In the current scenario, the chemical therapeutic treatment for this patient is either a tyrosine kinase inhibitor or an ICB. However, this decision is being made, primarily based on the genomic mutation information, despite the fact that their expected effects vary even within this small section (6.5 mm square in width). To develop therapeutic strategies for such cases, it may be better to consider the transcriptomic features and their heterogeneity.

Cross-case or section scoring of tumor cells and surrounding cells in IA cases

After examining several individual cases, we sought to generalize spatially distinctive gene expression profiles and their mutual association from a broader perspective. Because upregulated individual genes can vary by patient, we focused on gene groups commonly used to represent transcriptomic statuses, such as differentiation, proliferation, invasion, and immune cell activation or repression. Genes were chosen based on differences in expression patterns between the clusters and other studies^17,18. The spatial expression patterns of the selected genes were converted to activity scores using PAGE analysis of Giotto³⁶ (Supplementary Table S8). The results were manually checked, and the overall consistency of the annotations obtained from Visium clustering analysis was confirmed. Figure 4a depicts some example statuses in FFPE section C of LUAD No. 2 (refer to Supplementary Fig. S7 for full images). Using this scoring method, mutual comparisons were carried out between different specimens.

**Fig. 4: Cross-case or section TME scoring of local expression features.**

Based on the calculated scores, each spot was classified for mutual comparison. For IA cases, the areas of a given status could be depicted more simply than the expression profiles themselves (Fig. 4b). By comparing each score, status degrees between specimens can be compared. Based on the obtained data, we could consistently compare the width of the area with a given status between specimens. For example, the FF section of LUAD No. 2 had the largest area of well-differentiated cells (an adjacent region located on the left side of FFPE section C of LUAD No. 2, which was evaluated above). For immune cells (including both B and T cells), the most prominent area was in FFPE section A of LUAD No. 3. The score distribution for each status for each section is presented in Fig. 4c. All the results were consistent with results of the manual inspection by the pathologists, including the inter-case comparisons.

We concentrated on invasive areas (red color) as the most malignant feature in each section. In LUAD No. 14, the largest invasive area had the most poorly differentiated phenotype (Fig. 4d). FFPE section B of LUAD No. 3 had an invasive area on the right side of the section (Fig. 4e). In this region, cells other than CAFs expressed invasion-related genes. Notably, no clear histological changes were observed in this region, implying that transcriptomic profile-based scoring can be more useful for dissecting features that are obscured by limited histological or morphological information used for pathological examination.

Using the generated scoring scheme for local profiling, we attempted to determine how each profile might be related to the other. Before systematically comparing the profiles in the following section, we manually examined the possible association between the features in various regions of several specimens (Fig. 4b). We discovered several possible associations. For example, we found that the M2 macrophage profile was associated with the regions of tumor cells, particularly those with an aggressive/proliferative profile. Certain types of macrophages, such as anti-inflammatory macrophages, may have helped to create a favorable microenvironment for tumor cell proliferation and expansion while also excluding other types of immune cells, such as B cells and CTLs.

Mutual associations of the local profiles

To better understand the molecular signatures of tumor cells adapting to their respective microenvironments, we incorporated the inferred route of cancer progression, identified through Visium trajectory analysis, into the above TME scoring (Fig. 5a and b; also see Supplementary Fig. S8a). We determined how the cancer should have moved across the TME landscape (Fig. 5c). Furthermore, we identified profiles that showed mutually positive or negative correlations with changes in tumor cell profiles. For the representative cases with heterogeneous of tumor cell clusters, a positive correlation was found between malignant and CAF profiles (Fig. 5d). The negative association between invasion and immune cells was the most significant (Fig. 5d). Relevant changes in tumor cells affecting profile landscapes, such as drastic transformations of tumor cells or changes in pathological phenotypes, occurred in the overlapping area of the peak region of immune cell activity. The interaction was clear in at least four cases (eight transcriptome trajectories), depending on the threshold (Fig. 5e).

**Fig. 5: Tumor progression with distinct TME landscapes.**

Given all of the data generated and analyzed for various IA specimens, we highly believe that immune cells may act as one of the most significant barriers to tumor cell expansion. These tumor cells can only grow if they undergo phenotypic changes in response to an immune cell attack.

Validation analyses of TME at the single-cell level in IA cases

We aimed to validate expression profiles, also known as profile landscapes, and their mutual interactions at the single-cell level. For this purpose, we performed in situ gene expression analysis with Xenium (Fig. 6) on five specimens dissected from the same tissue blocks used in Visium analysis. Although data were only obtained from 302 designated genes, single-cell resolution data were obtained for all individual cells in the section, with an average of 264,710 cells per section. Using clustering analysis of the obtained Xenium data, we could classify each cell distinguishing between stromal and immune cells and tumor cells (Fig. 6a and Supplementary Fig. S9b). Furthermore, Xenium single-cell expression patterns allowed us to decompose Visium data at 55 μm resolution (Supplementary Fig. S9c).

**Fig. 6: Spatial and single-cell characterization of TME in LUAD No. 2.**

We used the obtained data to validate the transcriptomic characteristics of tumor cells, with a focus on cell lineage and differentiation markers. The Xenium data clearly showed a representative profile change––from NKX2-1– to HNF4A-positive tumor cells in LUAD No. 2 (Fig. 6b). Xenium’s finer resolution analysis revealed that cells expressing NKX2-1 or HNF4A were adjacent to each other in a mutually exclusive way (Fig. 6c). Furthermore, the immune response and high IDO1 expression were found to be relevant. We also found CCL22-positive cells, which could be DC or T cells, in this region. This feature was first discovered by deconvolution analysis of Visium data using Xenium data (Fig. 6d). CCL22 is a chemokine that attracts Tregs and promotes immune suppression³⁷. In conjunction with IDO1 expression on the tumor cell side, CCL22-positive cells create local environments that protect tumor cells from immune cell attacks. Consistently, single-cell resolution Xenium data revealed that at least several CD8 + T cells were infiltrating and interacting, potentially attacking tumor cells in this region. Tumor cells respond to immune cell attacks by expressing IDO1 and recruiting CCL22-positive cells (Fig. 6e). Notably, even in an immunosuppressive environment, some CD8 + T cells still expressed GZMB and PRF1, indicating that T-cell cytotoxicity was active. Thus, activating the remaining immune cells with ICBs could result in elimination of malignant-transformed mucinous tumor cells.

We investigated novel factors for which gene expression was activated by immune cell attacks in the boundary region. To complement the limited number of genes detected in Xenium analysis, we performed reference-free cell-type deconvolution³⁸ on Visium data (Supplementary Fig. S10a). We extracted 15 cell types. Cell-type X9 was found along the mucinous boundary. Cell-type X9 highly expressed immune response-related genes, such as IDO1 (log2fc = 2.5). Furthermore, RHOV (log2fc = 3.1) was highly expressed in this cell type (Supplementary Fig. S10c). RHOV is linked to cell proliferation, migration, and metastasis in lung adenocarcinoma^39,40. This observation could be an example of erroneous de- or re-differentiated cells beginning to express a series of genes that promote cancer development.

We assessed the association between tumor cells and CAFs using the improved spatial resolution of Xenium. We investigated the CAF-rich tumor invasion region in LUAD No. 2. In this region, Visium data did not distinguish between cells expressing key genes. For example, MMP7 and CXCL14, which play important roles in tumor cell invasion and inflammation, respectively, Xenium data clearly indicated that tumor cells were responsible for their expression (Fig. 6f). These findings are critical when considering pharmacological interventions targeting key molecules involved in tumor cell invasion, such as MMP7 and CXCL14^41,42. On the CAF side, we found expression of other matrix metalloproteinase MMP11 (Supplementary Fig. S9d). Although we could not identify specific interaction factors between tumor cells and CAFs, we hypothesized that both tumor cells and CAFs (and possibly their engagement) would play a role in ECM remodeling, resulting in the exclusion of anti-tumor immune cells^43,44,45 from this region.

From the viewpoint of immune environments in invasive regions, anti-inflammatory macrophages are infiltrated, while other cytotoxic immune cells are often excluded. In general, alveolar macrophages expressing MARCO are found in lung adenocarcinoma. We compared macrophage profiles to the obtained landscape profiles and found that SPP1-expressing macrophages became dominant in moderately differentiated and invasive tumor regions, which is consistent with previous studies^12,16,46,47. Notably, in LUAD No. 14, macrophages with high SPP1 expression increased in a region characterized by de-differentiated and invasive tumor cells with acinar and solid histological patterns (Supplementary Fig. S9e). These features were first revealed through single-cell resolution analysis with Xenium.

Investigation of TME statuses at the earlier phase of tumor progression

Having identified distinct changes in microenvironment statuses and tumor progression in IAs, we sought to investigate whether these changes occurred earlier in tumor development. Therefore, we conducted a similar analysis on very early lung adenocarcinoma cases, such as AISs and MIAs.

Using Visium data (Supplementary Fig. S11), we first assessed TME scoring in local tissue regions of AIS/MIA cases (Fig. 7a and Supplementary Fig. S12), as shown in IA cases. Most of the tumors were “well-differentiated.” The immune cell regions were not significantly lower than IAs, indicating that immune responses had already occurred at AIS and MIAs. In contrast, “malignant” regions (the sum of “proliferative” and “invasive” regions) were significantly smaller. Notably, these regions appeared as sparsely distributed “spots.” Because the cancer cell trajectory observed in IAs had not yet begun, we had to modify the previously described analytical scheme to characterize the spots in AISs/MIAs. We defined “possibly malignant” regions as those in which a number of proliferative or invasive spots were enriched with a specific enrichment score (“possibly malignant-invasive” or “possibly malignant-proliferative” regions). Overall, the degree of enrichment was greater in IAs for both proliferative and invasive spots, indicating that these regions are more densely packed in IAs (Fig. 7b). Nonetheless, we could depict “possibly malignant” regions from AIS and MIA cases (Supplementary Fig. S13a). In these “possibly malignant” regions, we attempted to characterize TME landscapes.

**Fig. 7: TME characterization of very early cases.**

A total of 25 regions in 11 cases were identified as “possibly malignant” regions in AIS/MIA cases (Supplementary Fig. S13b–d), indicating that AIS/MIAs do contain “possibly malignant” regions, albeit they are typically smaller and sparser. We found that in the “possibly malignant” regions of both AIS/MIA, similar to IA cases, pathways associated with response to various stresses, such as oxidative stress, unfolded protein, growth factors, cytokines, hormone stimulation, and apoptosis, were upregulated, as well as those of invasive phenotypes, such as cell motility and stromal development (Supplementary Fig. S14a–c). Vasculature development signaling was also increased in “possibly malignant” regions in both AIS/MIA and IA cases (Supplementary Fig. S14a), indicating that endothelial signaling is activated for tumor cell proliferation and invasion beginning in the early stages. These findings indicate that even in AIS/MIAs, core gene expression changes had already begun to progress to IAs.

Nonetheless, distinct gene expressions were found in AIS/MIA versus IA cases. Particularly, genes associated with the existence of stromal and immune cells were highly expressed in some cases (Supplementary Fig. S14a–c). For example, the fibrosis/elastosis-associated genes (CCN2, TIMP3, MFAP4, and LTBP4) were found to be overexpressed in “possibly malignant” regions of AIS/MIAs. Among them, the most distinguishing feature was that inflammatory lymphocyte- and/or macrophage-related gene expression was significantly more relevant in the “possibly malignant” regions of AIS/MIAs than in IAs. Immune cells likely infiltrated “possibly malignant” regions in AIS/MIAs more aggressively than IAs (Supplementary Fig. S14d).

A thorough examination of several cases revealed that the more active interaction of immune cells in AIS/MIAs was the true cause. For example, in the “possibly malignant” region of TSU-33, we found inflammatory immune cells, such as B cells (MS4A1 and CXCR4) as well as high levels of the pro-inflammatory chemokine CCL19 (Fig. 7c). This “possibly malignant-invasive” region contained several lymphocyte-enriched structures, which we confirmed using PhenoCycler analysis (see our report¹⁶). Similarly, M1-like alveolar macrophages were found in the possibly malignant-invasive regions of TSU-19 and TSU-27 (Supplementary Fig. S14e, f). In TSU-30, FCGR3A (CD16) was highly expressed in the “possibly malignant-invasive” region, indicating a fibrotic feature associated with high levels of COMP and COL15A1 (Fig. 7d). FCGR3A-expressing cells, most likely macrophages, appeared to recruit cytotoxic T lymphocytes. These findings indicate that in early tumors, tumor cells in the interior of “possibly malignant” regions are exposed to inflammatory stimuli from immune cells, which are absent in IAs.

We characterized the immune cell profiles more precisely. As partly described in our previous paper¹⁶, we investigated and found that in the possibly malignant regions of AIS/MIAs, co-localization of FABP4+ and SPP1+ macrophages was especially important (Supplementary Fig. S14g). Macrophages with high FABP4 expression have been identified as pro-inflammatory, and they are primarily found among normal-like alveolar macrophages^46,48. These findings collectively suggest that these possibly malignant regions are the regions where tumor cells have just begun to break through the barrier of the immune cells by first transforming their phenotypic appearance.

We then investigated the behaviors of immune cells at a finer resolution using PhenoCycler and Xenium (data summary for all AIS/MIA in Supplementary Figs. S15 and S16; analysis for stromal cells in Supplementary Fig. S17). We found co-localization of high MARCO-expressing alveolar macrophages and SPP1-expressing macrophages in the “possibly malignant” regions, such as TSU-27 (Fig. 7e). In TSU-30, several FABP4-expressing macrophages were found to co-localize with SPP1+ macrophages, even within the same alveolar space (Fig. 7f). Furthermore, we confirmed that lymphocyte infiltration existed in several “possibly malignant” regions, including TSU-33. TLS-like structures with B and T-cell aggregation were observed in “possibly malignant” regions expressing MMP7 (Fig. 7g). Immunosuppressive cells, such as CCL22– and IDO1-expressing cells, were found at the boundary between well-differentiated tumors (SFTPC) and more invasive tumors (MMP7-expressing cells). All of the results support Visium’s hypothesis, which is that, in those “possibly malignant” regions, the interaction between tumor cells and immune cells is changing.

Taken together, in early adenocarcinomas, inflammatory immune subsets can still infiltrate malignant regions (Fig. 7h), where immune cells effectively suppress tumor cell proliferation and expansion. Tumor cells that pass through this initial barrier continue to expand outward into empty space, where they face harsher immune cell attacks, such as those seen in IA (Fig. 7i). At both early and invasive stages, by inducing drastic changes in cancer cell gene expression, interactions with immune cells play a critical role in determining the fate of cancer cells.

SEO Powered Content & PR Distribution. Get Amplified Today.
PlatoData.Network Vertical Generative Ai. Empower Yourself. Access Here.
PlatoAiStream. Web3 Intelligence. Knowledge Amplified. Access Here.
PlatoESG. Carbon, CleanTech, Energy, Environment, Solar, Waste Management. Access Here.
PlatoHealth. Biotech and Clinical Trials Intelligence. Access Here.
Source: https://www.nature.com/articles/s41467-024-54671-7

Spatially resolved gene expression profiling of tumor microenvironment reveals key steps of lung adenocarcinoma development

Overview of study design

Spatial expansion patterns of LUAD No. 2 (KRAS, mucinous cancer)

Spatial expression of LUAD No. 3 (EGFR, non-mucinous cancer)

Diverse TMEs and tumor cells in the same block of LUAD No. 3

Cross-case or section scoring of tumor cells and surrounding cells in IA cases

Mutual associations of the local profiles

Validation analyses of TME at the single-cell level in IA cases

Investigation of TME statuses at the earlier phase of tumor progression

Latest Intelligence

Antag Therapeutics Secures €80 Million to Advance GIPR-Based Obesity Therapies

Ini Situs Toto Slot Togel Bet Terendah Terpercaya Bukan Abal Abal Boy!

Cherry Biotech Publishes Research on Advanced Skin-on-Chip Model in Microfluidics

Putting patients first: Workers suffer when their employers lack critical information about PBMs

With conflicts of interest in focus, Trump’s pick to run FDA could face scrutiny of his own industry ties

Lilly adds to obesity drug production push with $3B investment

Lilly’s $3 billion investment in Wisconsin

Rep. Buddy Carter calls for vote on PBM reform before end of year

Donating embryos for research is surprisingly complex

Cracking the code of biotech valuations