Effects of Tumor Microenvironment on Lung Cancer Stem Cells: Bidirectional Regulatory Mechanisms.

1
Background
In recent years, people have gained a better understanding of lung cancer, including its occurrence, development, treatment and prevention. Lung cancer is a malignant tumor that originates in the bronchi and alveoli and is aggressive and heterogeneous. It is histopathologically classified into two major categories: non‐small cell lung cancer (NSCLC) and small cell lung cancer (SCLC). NSCLC accounts for approximately 85% of cases and can be further subdivided into major histological subtypes, including adenocarcinoma, squamous cell carcinoma, and large cell carcinoma [1, 2]. In contrast, SCLC represents only 10%–15% of cases but is characterized by high invasiveness, early metastatic potential, and a distinct neuroendocrine phenotype [3]. It proliferates significantly faster than NSCLC, exhibits initial sensitivity to chemoradiotherapy but readily develops resistance and recurrence, and possesses a TME typified by profound immunosuppression [4, 5]. According to the American Cancer Society, lung cancer is the second most diagnosed and most deadly type of cancer in the United States [6]. The causative factors of lung cancer include tobacco and non‐tobacco factors, with the former being the main causative factor and the latter including chronic pneumonia, occupational exposures, and genetic factors [7]. Lung cancer treatment varies by stage, with stage I and II NSCLC usually treated with surgical resection; the management of advanced disease is highly stage‐dependent and personalized. For selected patients with stage III disease, the goal is often cure, frequently achieved with concurrent chemoradiotherapy. In contrast, stage IV disease is managed with systemic therapies, which may include targeted therapy, immunotherapy, chemotherapy, or combinations thereof, based on the tumor's molecular characteristics [8, 9] (Figure 1). As conventional treatments often lead to tumor recurrence as well as the development of drug resistance, the five‐year survival rate of NSCLC patients is generally low at 25%, and even lower in metastatic patients, dropping to 8.2% [6]. This suggests a need to explore new therapeutic strategies.
LCSCs, a key oncology concept, are self‐renewing cells that drive tumor diversity and expansion and are central to lung cancer recurrence [10]. These cells are highly heterogeneous and plastic, often quiescent and insensitive to radiotherapy and chemotherapy, but can rapidly proliferate in response to stimulatory factors, causing drug resistance and recurrence [11]. The TME is a complex ecosystem composed of cancer cells, immune cells, stromal cells. Traditional views often depicted the TME as a supportive background that promotes tumor survival and development [12, 13]. Meanwhile, the TME creates hypoxic, acidic, and nutrient‐deficient conditions, which boost cancer cells' stemness and enhance their invasiveness and metastatic potential [12]. However, emerging evidence reveals a sophisticated bidirectional dialog between LCSCs and the TME. In this review, We aim to move beyond reviewed the bidirectional regulatory relationship between TME and LCSCs. Furthermore, this review will critically discuss how targeting these bidirectional regulatory mechanisms offers novel therapeutic strategies to disrupt these vicious cycles, ultimately aiming to improve lung cancer treatment therapies (Figure 2).

2
Characteristics of LCSCs
2.1
Definitions and Markers
2.1.1
Definitions
It is well known that lung cancer cells are heterogeneous and that not all cancer cells are capable of maintaining and proliferating tumors and generating metastasis. Based on this view, the LCSCs theory has emerged, stating that LCSCs with stem cell properties mainly drive lung cancer proliferation, drug resistance, and metastasis [10]. These cancer stem cells can self‐replicate and differentiate into various cancer cells, surviving conventional treatment and causing recurrence. Meanwhile, specific markers expressed by LCSCs play a key role in cancer progression, resistance to drugs and metastasis [10, 14]. They not only help to identify lung cancer cells with stem cell properties but also provide potential biological targets for targeted therapy. Therefore, identifying these markers of LCSCs is crucial to understand their properties.

2.1.2
Markers
2.1.2.1
CD133
CD133, a prominin family glycoprotein with five transmembrane segments, is widely expressed in solid tumors like lung cancer and is crucial for LCSCs' self‐renewal, differentiation, tumor angiogenesis, hypoxia adaptation, and epithelial‐mesenchymal transition (EMT). Eramo et al. cultured CD133 positive cells by serum‐free culture technology, which showed strong tumor‐forming ability in a nude mouse model, establishing CD133 as a marker for LCSCs isolation [15].

2.1.2.2
CD44
CD44 is a multifunctional transmembrane glycoprotein belonging to the extracellular matrix receptor family that exhibits remarkable heterogeneity [16]. CD44 also enhances cancer cell invasion by regulating cell motility and adhesion to extracellular matrix [17, 18]. And specific variants of CD44, such as CD44v6, may play a key role in invasion and metastasis, participating in signaling and promoting EMT [19].

2.1.2.3
ALDH1
ALDH1 (aldehyde dehydrogenase 1) is an enzyme that plays a key role in cellular redox reactions and contributes to the maintenance of intracellular environmental homeostasis. Jiang's experiments have shown that high expression of ALDH1 is closely associated with lung cancer progression and cancer metastasis [20]. Previous studies had analyzed ALDH1 expression levels by immunohistochemistry in 92 patients with lung adenocarcinoma, and multivariate analysis found that negative ALDH1 expression was independently associated with higher survival and longer disease‐free survival, revealing the importance of ALDH1 positivity in influencing prognosis in lung adenocarcinoma by affecting the activity of LCSCs to maintain their stemness [21].

2.1.2.4
SOX2
SOX2, a member of the SRY‐related HMG box family, is a key transcription factor involved in cancer cell proliferation, EMT, and maintenance of CSC‐like properties [22]. Qiu's team revealed the central role of OCT4 and SOX2 in promoting the expression of salivary acidified cancer IgG (SIA‐cIgG), which further enhances SOX2 activity by activating the c‐Met/Akt/ERK signaling pathway, forming a positive feedback loop that maintains LCSCs properties [23].

2.1.2.5
OCT4/POU5F1
OCT4/POU5F1 is a core member of the POU family, which plays a key role in cell pluripotency and fate regulation [23]. In TME, OCT4 promotes the polarization of M2‐type tumor‐associated macrophages (TAMs) by regulating the expression of macrophage colony‐stimulating factor (M‐CSF), and these polarized macrophages secrete vascular endothelial growth factor (VEGF) and hepatocyte growth factor (HGF), which provide a favorable microenvironment for LCSCs and enhance their self‐renewal and drug resistance [24].
In addition, other cellular markers also play key roles in lung cancer development, drug resistance and metastasis (Figure 3).

2.1.3
Stem Cell Characterization
Compared to normal stem cells, LCSCs are significantly different in several ways, which are mainly related to their biological properties, function, and behavior in tissues(Table S1), such as self‐renewal and pluripotency, drug resistance, and tumor recurrence potential [10, 25, 26].

2.2
Composition and Pro‐Tumorigenic Functions of the TME
2.2.1
Composition
LCSCs are maintained within the specialized TME, a complex network supported by cancer associated fibroblasts (CAFs), immune cells, and an abundant extracellular matrix together with its macromolecules [27]. This niche not only preserves LCSCs' characterization but also drives their proliferation, invasion, and metastasis. Moreover, a supportive TME is often established before overt metastasis occurs [28], providing the requisite conditions for disseminated cancer cells to survive, migrate, and colonize distant sites.

2.2.2
Pro‐Tumorigenic Functions
The TME can promote tumor growth and help tumor invasion and metastasis. The proliferation of cancer cells is dependent on an appropriate TME and adequate nutrient supply. Under the influence of the interaction between cancer cells and pericellular microenvironment components, the pericellular microenvironment is gradually transformed into the TME that supports tumor growth, characterized by cells like CAF and TAM. Several studies have shown that cells from the TME not only promote tumor growth and proliferation and support tumor development through their secretome and extracellular vesicles (EV) [24, 25, 26] and enhance tumor growth by activating related pathways [29, 30, 31]. At the same time, the TME vascular component and stromal component also have important roles. There is a consensus that tumor growth relies on blood vessels for nutrient supply [32]. Similarly, the generation and remodeling of extracellular matrix (ECM) can also promote tumor proliferation and growth. By recruiting other cells to reach the cancerous tissue [33, 34], which in turn exerts enhanced effects on the TME's tumor‐promoting effects.

3
The Immunosuppressive Niche of LCSCs Within the TME
3.1
TME‐Derived Signals That Drive LCSCs Immune Evasion
The TME serves as a protective niche for LCSCs. One of the key molecules in this process is TGF‐β. It interacts with Wnt and Notch signaling to augment their invasive potential [35, 36]. Critically, TGF‐β is a master regulator of immune suppression. It inhibits the activation and function of dendritic cells (DCs), which are crucial for initiating T‐cell responses [37]. However, the role of TGF‐β is not unidirectional [38]. In the early stages, TGF‐β triggers cell‐cycle arrest and apoptosis, thereby maintaining tissue homeostasis by restraining cellular proliferation and malignant transformation [39]. This includes, but is not limited to, the induction of cyclin‐dependent kinase (CDK) inhibitors such as P15 and P21 [40, 41], and the activation of Smad2/3 to guide damaged or abnormal cells into programmed cell death [42]. Recent studies have revealed that specific regulatory factors in the cellular context, especially the binding partners of the transcriptional modulators Smads, play a crucial role in determining the shift in TGF‐β responses, switching its effects from pro‐apoptotic to pro‐metastatic [43]. The opposite functions of TGF‐β signaling during cancer progression make it a challenging target to develop anticancer interventions, as indiscriminate inhibition risks unleashing early tumor growth while sparing late‐stage evasion mechanisms. The other related molecules are listed in the Table S2.

3.2
LCSC‐Mediated Reprogramming of Immune Cells
A feature of CSCs communication with the TME is the regulation mediated through their secretome [44, 45]. At the same time, LCSCs also actively shape an immunosuppressive niche through the secretion of cytokines and exosomes.
3.2.1
Cytokine‐Mediated Immunosuppression
LCSCs can contribute to immunosuppression through the secretion of cytokines, including TGF‐β, IL‐4, IL‐6, IL‐8, IL‐10, GM‐CSF, and FOXP3 [46, 47, 48]. For instance, IL‐10 targets IRF transcription factors (e.g., IRF1 and IRF5) to inhibit inflammatory and interferon‐stimulated gene expression via epigenetic mechanisms, thereby suppressing antitumor immunity [49]. Furthermore, LCSCs recruit TAMs by secreting CCL2, thereby indirectly evading immunosurveillance [50]. Additionally, LCSCs can secrete the inflammatory factor SAA to drive type 2 immunity, thereby suppressing anti‐tumor immunity [51].

3.2.2
Exosomal Communication
By transferring immunomodulatory molecules, exosomes can shape immune cell behavior within the TME, thereby shaping an immunosuppressive niche that allows cancer cells to escape immune surveillance and sustain their progression [29, 52, 53]. A key mechanism involves exosomal PD‐L1 (ExoPD‐L1). LCSCs release exosomes expressing PD‐L1 on their surface, which can suppress T‐cell activity, exhaust CD8+ T cells, or induce dendritic cells to inhibit indirect anti‐tumor immune effects [54, 55, 56].
Kim et al. evaluated the immunosuppressive role of exosomal PD‐L1 derived from NSCLC patient samples. They demonstrated that these exosomes suppress CD8+T cells function in a dose‐dependent manner by reducing IL‐2 and IFN‐γ production [55]. The interaction between PD‐1 and PD‐L1 leads to the downregulation of casein kinase II (CK2), which normally acts as an inhibitor of PTEN [57, 58]. This downregulation subsequently enhances PTEN activity, resulting in reduced activation of the PI3K–Akt pathway in CD8+ T cells [59], thereby inducing T‐cell dysfunction through apoptosis, anergy, and exhaustion [60, 61]. In addition, Ge et al. found that β‐TrCP‐containing exosomes markedly decreased the protein level of the transcription factor YAP1 in CD8+ T cells, disrupting amino acid metabolism and inducing mTOR inactivation, which ultimately contributed to CD8+ T cells exhaustion (Tex) [62]. This may be associated with the dysregulation of multiple exosome‐related genes in T cells, including SLC7A5, SLC38A1, SLC2A1, and TUBB6, which could influence the prognosis of lung cancer by affecting immune cell function and infiltration [62, 63]. Interestingly, previous studies have shown that non‐glycosylated PD‐L1, once phosphorylated by GSK3β (e.g., at Tyr180/Ser184), can bind to β‐TrCP, triggering its ubiquitination and degradation—a process associated with tumor‐suppressive effects [64]. However, LCSCs appear to inhibit PD‐L1 degradation by promoting PD‐L1 deubiquitination or glycosylation, thereby maintaining its protein stability [65]. This observation seems contradictory to the aforementioned findings. Considering the wide range of β‐TrCP substrates, including both tumor suppressors and oncoproteins, it remains challenging to determine whether β‐TrCP functions as an oncogene or a tumor suppressor in specific disease contexts [66].

3.2.3
Direct Modulation of Immune Cell Function
Co‐culturing CSCs with TAMs could promote the polarization of TAMs toward the tumor‐promoting M2 phenotype [67]. In specific contexts, such as the NEUROD1‐high subtype of SCLC, LCSCs alter macrophage phenotype and pro‐tumor behavior by modulating the expression of immunosuppressive lectin receptors on monocyte‐derived macrophages [68]. However, the mechanisms by which it regulates M2 macrophage polarization and subsequently forms an immunosuppressive TME still require further investigation in future studies [69]. Current views suggest that KLF4 is an environment‐dependent transcription factor and a key regulator of the STAT6 signaling pathway. The JAK1‐STAT6 pathway is one of the critical pathways involved in tumor‐promoting macrophage formation and is closely associated with the IL‐13 signaling pathway, where IL‐13 levels are crucial for M2 phenotype polarization [70, 71]. Additionally, it has been reported that DNMT3A (DNA methyltransferase 3A) methylates SLIT2, thereby weakening SLIT2‐mediated inhibition of cell polarization, suggesting that epigenetic regulation may also play an important role [72].
LCSCs also directly impair T‐cell function. The secretion of Tenascin‐C (TNC) by CSCs binds to the α5β1 integrin receptor on T cells, downregulating downstream pathways and impairing T‐cell proliferation [73, 74]. They inhibit the infiltration of CD8+ T cells to counteract their antitumor effects [75]. Moreover, CSCs recruit regulatory T cells (Tregs) to suppress T‐cell activity [73, 76]. The frequency of PD‐L1+ LCSCs has been shown to alter the frequency and phenotype of T cells, weakening their immune function [77]. While evidence for LCSC‐mediated immune reprogramming is growing, many findings rely on in vitro models. The relative contribution of LCSCs versus bulk cancer cells in shaping the immune landscape in vivo requires further validation.

3.3
Key Roles of HH Pathways in the Immune Crosstalk
In healthy tissues, the Hh pathway is strictly regulated, while its aberrant activation is frequently observed in several cancers [78], including but not limited to prostate cancer [79], colon cancer [80], pancreatic cancer [81] and lung cancer. In lung cancer, this pathway facilitates a multi‐directional dialog. Shh secreted by lung cancer cells activates the Hh‐Gli1‐Klf4/STAT3 signaling cascade, promoting the polarization of TAMs toward the M2 phenotype. This process upregulates PD‐L1 expression and suppresses TAM production of CXCL9 and CXCL1, reducing the number and function of CD8+ T cells and consequently weakening the antitumor immune response [82, 83]. Furthermore, polarized M2 TAMs, via the Hh pathway, increase the expression of stemness‐related genes (e.g., SOX2, NANOG, and OCT4), significantly enhancing the chemoresistance of LCSCs [84, 85]. In summary, the Hh signaling pathway regulates a complex interaction network between the lung cancer microenvironment and CSCs through soluble mediators.

4
Hypoxia‐Induced Metabolic Reprogramming and Angiogenic Crosstalk Between LCSCs and TME
The TME vascular component not merely deliver nutrients, it constitutes a dynamic and specialized vascular niche that engages in bidirectional regulation with LCSCs [86]. Hypoxia within this microenvironment not only activates angiogenic signaling but also drives metabolic reprogramming that sustains LCSCs' stemness, survival, and adaptability. In turn, LCSCs remodel the vascular niche through pro‐angiogenic and metabolic feedback, reinforcing tumor progression and metastasis.
4.1
Hypoxia‐Driven Metabolic Reprogramming and LCSCs Maintenance
Solid tumors, including lung cancer, are characterized by regions of profound hypoxia, a primary driver of angiogenesis and a key element of the vascular niche. Hypoxia significantly affects LCSCs by activating hypoxia‐inducible factors (HIFs) that support LCSCs [87, 88](Table S3). Among them, HIF‐1α is a typical example. The hypoxic environment can also affect LCSCs through metabolism. HIF‐1α can help LCSCs adapt to the harsh microenvironment through metabolic reprogramming [89]. A prominent example of this phenomenon is the upregulation of glycolysis under aerobic conditions, known as the Warburg effect, wherein cancer cells preferentially generate energy through glycolysis despite adequate oxygen availability, enabling rapid biosynthesis and maintaining an optimal ATP/ADP ratio [89, 90]. Compared with non‐stem cancer cells, LCSCs isolated from NSCLC cells exhibit significantly higher glycolytic activity [91]. A study found that upon binding to the MH2 domain of phosphorylated Smad3, HIF‐1α activates the expression of c‐Myc and thereby influences metabolic reprogramming [92]. Meanwhile, hypoxia boosts LCSCs' uptake of glucose and glutamine, regulates metabolism via HIF‐1α, and maintains high metabolic activity in nutrient‐poor environments [93], which ensures their dominance in nutrient‐deprived environments. Specifically, when HIF‐1α enters the nucleus, it upregulates the expression of GLUT1 and GLUT3 through the SLC2A1 and SLC2A3 genes [94]. The primary cellular function of GLUT is to facilitate the entry of glucose molecules into the cell, and its translocation to the cancer cell membrane is an important factor influencing the rate of cellular energy production [95]. In addition, ERCC6L, a DNA helicase essential for mitosis, is markedly upregulated in lung cancer. By competing with HIF‐1α for binding to VHL (a key E3 ubiquitin ligase), ERCC6L prevents VHL‐mediated ubiquitination and degradation of HIF‐1α, thereby enhancing tumor glycolysis and promoting stem cell–like traits in LUAD cells [96].

4.2
LCSC‐Driven Angiogenesis and Pre‐Metastatic Niche Formation
LCSCs are not only beneficiaries of the vascular niche, but also active architects that shape it. CSCs secrete a wide array of pro‐angiogenic factors, including VEGF, HIF1, IL‐8, SDF‐1, and TGF‐β, to stimulate new blood vessel growth [45, 73]. This pro‐angiogenic secretome can be enhanced through intracellular pathways such as PI3K/AKT. Other mechanisms involve CSC‐derived exosomes, which act as molecular messengers delivering pro‐angiogenic cargo (VEGF, MMPs, microRNAs like miR‐141 and miR‐23a) to TME cells like endothelial cells under hypoxic conditions, thereby promoting angiogenesis [44, 97, 98].
This LCSCs‐induced angiogenesis establishes a vicious cycle: the newly formed vessels supply oxygen and nutrients to support LCSCs and tumor growth, which in turn prompts continuous angiogenic stimulus from LCSCs, thereby driving tumor progression [32]. Furthermore, LCSCs take advantage of this function to participate in pre‐metastatic niche formation, ensuring a supportive environment for disseminated tumor cells.

4.3
Convergent Signaling Pathway
In response to TME hypoxia stimulation, Notch signaling enhances the recruitment of HIF‐1α in the lysyl oxidase (LOX) promoter region and activates the expression of LOX, thus indirectly regulating Snail, a key EMT regulator, thereby promoting lung cancer cell invasion and metastasis [99, 100]. Studies have shown that the high expression of Notch receptor 1 is closely related to the poor prognosis of lung adenocarcinoma, and targeting the Jagged1/Notch1 signaling pathway effectively suppresses the LCSCs phenotype and tumor growth [101, 102].
Meanwhile, LCSCs regulate the endothelial cell component of the TME through secretion of VEGF in synergy with the Notch pathway [103, 104]. Under VEGF stimulation, certain endothelial cells differentiate into “tip cells,” located at the forefront of sprouting vessels. These cells exhibit elevated VEGF receptor 2 (VEGFR2) activity and reduced Notch signaling, guiding neovascularization and infiltrating the tumor [105]. Adjacent endothelial cells, upon receiving the ligand DLL4 from tip cells, activate Notch signaling. This process downregulates VEGFR2 expression, reducing their responsiveness to VEGF and promoting a “stalk cell” phenotype, thereby preventing their conversion into tip cells [105]. Conversely, ligands Jagged1 and low levels of DLL4 expressed by stalk cells bind to Notch receptors on tip cells, exerting a competitive antagonistic effect, resulting in reduced Notch activity in tip cells [106].
The interplay between these two cell types is fundamental to normal vascular development and function. Disruption of this balance, such as through clinically used DLL4/Notch pathway inhibitors, may result in uneven tumor angiogenesis, incomplete vessel structures, and increased permeability. These poorly perfused abnormal vessels exacerbate hypoxia in TME, leading to an expanded lung cancer stem cell subpopulation [107, 108] (Figure 4).

5
The Remodeling of Stromal Cells and the ECM
5.1
LCSCs Activate and Reprogram CAFs
The TME is an active ecosystem shaped by its resident cells. LCSCs paly an important role in this process, with a key mechanism being the reprogramming of resident fibroblasts into CAFs. This “education” process is mediated by a set of signaling molecules secreted by LCSCs [109, 110].
As illustrated in Figure 5, key cytokines are critical in activating resting fibroblasts and inducing a CAF phenotype [45, 111, 112]. Other factors like basic fibroblast growth factor (bFGF) and platelet‐derived growth factor (PDGF) also contribute to the program [111]. Furthermore, the Hedgehog ligand Shh, secreted by LCSCs, plays a significant role in regulating CAF function [113].

5.2
Activated CAFs Reinforce Cancer Cell Stemness Through Soluble Signaling Factors
Once activated, CAFs in turn create a niche that aggressively sustains and expands the LCSCs population. A primary mechanism is the secretion of a cocktail of soluble factors that target lung cancer cells to activate critical intracellular pathways, most notably the Wnt/β‐catenin signaling pathway, thereby promoting LCSCs expansion and self‐renewal [112, 114, 115].
Su et al. showed that IL‐6 derived from CD10+GPR77+ CAFs maintains cancer stemness and promotes lung cancer progression and chemotherapy resistance [116]. It activates the IL‐6/STAT3 pathway in lung cancer cells to maintainβ‐catenin stabilization by promoting the degradation of INGF(Inhibitor of Growth Family) [117]. The growth factor midkine enhances Wnt signaling, upregulating c‐Myc expression, which in turn initiates HK2 kinase transcription, promoting glycolysis and augmenting lung cancer cell stemness [118]. Besides, the chemokine SDF‐1 promotes LCSCs formation by binding to the CXCR4 receptor on the surface of cancer cells in order to activate the pathway [119, 120]. CXCR4, as an important marker of lung cancer, is closely associated with cancer cell stemness and drug resistance [121]. Also, there is a complex crosstalk between HGF derived from CAFs and the Wnt pathway, and this interaction likewise supports the de‐differentiation of lung cancer cells to form LCSCs [122]. Finally, the common outcome of these CAF‐derived factors is the inhibition of GSK‐3β phosphorylation, leading to β‐catenin stabilization, its nuclear translocation, and the transcription of stemness‐related genes(Figure 6).

5.3
ECM Remodeling by CAFs Creates a Pro‐Stemness Mechanical Niche
The ECM is a key component of the TME and is able to influence the differentiation process of LCSCs by regulating cellular stiffness [123]. Collagen deposition and remodeling in the ECM activates signaling pathways such as YAP/TAZ through mechanical tension, enhances the viability of LCSCs, promotes their adaptation to environmental stimuli, and increases tumor invasion and metastasis [124]. Stiff ECM can maintain the stemness of LCSCs through enhanced cell‐matrix signaling and enhance their anti‐apoptotic capacity through integrin signaling, and in lung cancer, integrin β1 can promote drug resistance in LCSCs through activation of PI3K/AKT and MAPK signaling pathways [125]. In addition, increased ECM hardness in the TME enhances LCSCs migration and invasion via integrin‐mediated mechanosensing pathways, while concomitant ECM remodeling—often accompanied by MMP overexpression—plays a dual role: it degrades the matrix to release sequestered growth factors and active molecules, thereby further promoting LCSCs proliferation in a synergistic feedback loop [126]. Examining malignant tumor dynamics provides a compelling lens for dissecting MMPs' dual potentials in tissue remodeling; regenerative tissues often emulate their protumor behaviors in expansion and metastasis, although debates persist regarding their roles in niche priming versus cell migration [127].

6
Bidirectional Regulatory Mechanisms: Signaling and Epigenetic Regulation
6.1
Regulatory Mechanisms of Signaling Pathways
As previously described, Wnt, Hh, and Notch are the primary regulatory pathways for LCSCs. Dysregulation of these pathways leads to oncogenesis, invasiveness, metastasis, and recurrence. Overall, they are interconnected and crosstalk through signaling molecules, inducing the transcription of various target genes associated with cell proliferation (Myc, PDGFR), angiogenesis (VEGF, Ang1/2), EMT (Snail, MMP), and CSCs self‐renewal, which maintain LCSCs and drive lung cancer progression [128] (Figure 7). This process is multifactorial and complex. It is now widely accepted that inhibiting these signaling pathways can reduce LCSCs and enhance chemoresensitivity to various drugs.
Also, these signaling pathways are commonly upregulated in tumors, and research on the role of Notch signaling in tumorigenesis has primarily focused on its oncogenic function. In contrast, Notch can also function as a tumor suppressor. Transcriptome analysis of SCLC has shown that Notch family genes are functionally mutated in both human and mouse SCLC models, and overexpression of the transcriptionally active intracellular domain of Notch1 or Notch2 can attenuate the in vitro proliferation of SCLC cells [129, 130]. This strongly supports the above conclusion. Notably, in SCLC stem cells, Notch signaling is involved in early stem cell differentiation [131]. The downstream targets of Notch signaling, HES‐1 and HEY‐1, act as key transcription factors for neuroendocrine (NE) differentiation in pulmonary epithelial cells, transcriptionally repressing the expression of ASCL1 and promoting differentiation into pulmonary neuroendocrine cells (PNECs) [132]. Cancer cells originating from PNECs, however, maintain ASCL1 levels through low Notch activity, thereby promoting SCLC proliferation and the NE differentiation phenotype of SCLC [133].

6.2
Gene and Epigenetic Regulation
6.2.1
MicroRNA
Studies have confirmed that miRNAs involved in epigenetic regulation can be categorized into oncomiRNAs and tumor suppressor miRNAs [134]. Under TME‐induced hypoxia, circ_0000977 acts as a molecular sponge to downregulate the expression of the tumor suppressor miR‐153, resulting in enhanced expression of its downstream target, HIF‐1α [135]. And HIF‐1α directly binds to the upstream promoter of oncomiR‐1275, in turn inducing its overexpression [136]. Abnormally expressed oncomiR‐1275 targets and inhibits several suppressors of the Wnt pathway (including DKK3, SFRP1, GSK3β, and RUNX3) as well as the Notch pathway antagonist NUMB, thereby activating these pathways in cancer cells and maintaining a stem cell‐like phenotype [136]. Wang et al. [137] found that exosomal miR‐210‐3p derived from LCSCs targets FGFRL1, modulating the expression of EMT‐related proteins in the TME. This results in upregulation of N‐cadherin, vimentin, MMP‐9, and MMP‐1, and downregulation of E‐cadherin, promoting cancer cell migration and invasion [137].

6.2.2
LncRNA
It is now confirmed that lncRNA plays a regulatory role between LCSCs and TME mainly through (1) regulation at the post‐transcriptional level, including acting as miRNA regulators or targeting mRNAs [138]; (2) interacting with target proteins [139]; (3) inducing regulation of gene expression at the transcriptional level [140] as shown in the table below (Table S4).

6.2.3
Epigenetic Regulation‐Methylation
N6‐methyladenosine (m6A) modification is the most common internal modification of eukaryotic mRNA and plays a crucial role in post‐transcriptional gene regulation, its modification holds significant prognostic value in lung cancer [141]. It is well‐established that m6A modification is dynamically regulated by methyltransferases, known as “writers” (METTL3, METTL16), and demethylases, referred to as “erasers” (FTO, ALKBH5). Once m6A is recognized by various “readers,” it regulates downstream target genes to promote cancer cell stemness [142, 143]. For example, YTH N6‐methyladenosine RNA binding protein F2 (YTHDF2), driven by histone lactylation, recognizes the m6A modification on secreted frizzled‐related protein 2 (SFRP2), thereby promoting glycolysis and stemness in NSCLC cells [144]. Another m6A “reader,” insulin‐like growth factor 2 mRNA binding protein 1 (IGF2BP1), manipulates the stability of BUB1 mitotic checkpoint serine/threonine kinase B (BUB1B), thereby regulating the tumor immune microenvironment and influencing the malignant behavior and immune tolerance of LCSCs [145]. Additionally, studies have shown that protein arginine methyltransferase 5 (PRMT5) can methylate STAT3, activating downstream signaling pathways. This not only reprograms normal cells into LCSCs but also leads to the overexpression of DNA methyltransferase 1 (DNMT1), which silences the expression of tumor suppressor genes P53 and P21, thus promoting lung cancer progression [146, 147]. In contrast, TGF‐β‐induced EMT and enhanced cancer cell stemness have been shown to be associated with the demethylation activation of the gene promoters of the EMT marker Slug and the stem cell marker CD87 [148]. TGF‐β promotes the demethylation of the PD‐L1 promoter and enhances its expression by reducing DNMT1 levels, in cooperation with TNF‐α activation of the NF‐κB pathway [149]. This process increases the binding of PD‐L1 to PD‐1 on the surface of T cells, thereby inhibiting T cell activity and helping LCSCs to evade host immune surveillance [149]. In summary, different signals in the TME can regulate the growth, differentiation, and stem cell properties of LCSCs through opposing modification patterns, namely DNA methylation and demethylation, reflecting the complexity of epigenetic regulation.

7
Clinical Applications and Research Perspectives
7.1
Therapeutic Implications and Challenges in Immunotherapy
Previously, researchers have made significant efforts in understanding immune mechanisms, leading to the development of corresponding therapeutic strategies that have been applied clinically. Current immunotherapies include oncolytic virus therapy, cancer vaccines, cytokine therapy, adoptive cell transfer, and immune checkpoint inhibitors (ICIs) [150, 151]. However, despite some success in immunotherapy, many patients remain unable to benefit due to factors such as drug resistance [152, 153]. While the efficacy of PD‐1(L) inhibitors has been extensively demonstrated, resistance remains a significant concern. During ICI treatment, over half of patients develop acquired resistance [154, 155]. Studies reveal that tumor tissues from patients with acquired resistance exhibit IFN‐stimulated gene upregulation, IFN‐γ signaling defects, and immune features of CD8+ T cell exhaustion [154, 155]. In lung cancer, adoptive cell therapy (ACT) remains in its early stages [156]. Furthermore, the lack of rational target antigens in lung cancer often leads to off‐target toxicity [151]. Cytokine release syndrome and immunosuppressive MTE also pose barriers to ACT [151]. These promising strategies face significant challenges in clinical translation that cannot be overlooked.
To overcome these challenges, future research may focus on bidirectional regulation between LCSCs and the TME, translating this into novel therapeutic strategies. Within immune‐related bidirectional regulatory mechanisms, targeting key signaling molecules involved in immune suppression during LCSCs‐TME communication is paramount. LCSCs actively orchestrate immune evasion through high plasticity, thereby generating multidimensional drug resistance. The continuous secretion of the LCSC secretome persistently constructs an immunosuppressive TME. Simultaneously, LCSCs implement long‐distance signaling through mechanisms such as exosomes. Furthermore, components within the TME can also achieve immunosuppression via various cytokines, providing a favorable environment for LCSCs. Beyond key signaling molecules, targeting the direct interactions between LCSCs and immune cells is also vital. Following this line of reasoning, existing immunotherapies may be improved and supplemented. Cytokine therapy represents an emerging approach in lung cancer immunotherapy [151]. Its potential is currently under evaluation, with some studies exhibiting small sample sizes that fail to provide conclusive evidence or even demonstrate poor therapeutic outcomes [151, 157]. Nevertheless, cytokine therapy warrants further investigation. Fully considering the role of LCSCs may enhance cytokine therapy, facilitating translation from mechanism to therapeutic strategy. ACT therapy can also incorporate the interactions between LCSCs and immune cells. Researchers also have proposed other various approaches, TILs represent a highly personalized therapeutic approach, often requiring selection based on reactivity to tumor antigens [156]. It is also crucial to account for the negative regulation of immune cells by LCSCs and mitigate the impact of the immunosuppressive TME. Furthermore, cell‐based immunotherapies represent a major avenue [158], such as selecting CAR‐T cells targeting LCSC markers [159]. However, its efficacy remains inferior to that achieved in hematological malignancies, largely due to unique barriers posed by solid tumors [160, 161]. Major obstacles include target antigen heterogeneity or loss, impaired antigen processing and presentation, poor T‐cell trafficking, and T‐cell exhaustion within the suppressive TME [160, 161, 162]. In NSCLC, established targets—including EGFR, MUC1, and mesothelin—are under active investigation, while exploration of stemness‐associated antigens continues; notably, intratumoral heterogeneity fuels antigen escape and disease relapse [163, 164]. To overcome these challenges, innovative engineering solutions are being pursued. These include multi‐specific or logic‐gated CARs (e.g., SynNotch systems) and adapter platforms that enhance specificity while improving safety profiles [161, 163]. Tumor targeting is augmented through chemokine receptor engineering and remodeling of physical barriers.

7.2
Anti‐Angiogenic Therapy
Currently, anti‐angiogenic therapy (AAT) targeting blood vessels has achieved significant results. The Chinese anti‐angiogenic inhibitor anlotinib has been approved for cancer treatment [165], indicating that anti‐angiogenic therapy is gradually maturing. Current clinical AATs predominantly target VEGF. However, emerging evidence suggests VEGF inhibition may trigger compensatory mechanisms, including elevated levels or activation of other proangiogenic factors, leading to drug resistance [166]. Furthermore, AATs are often combined with other therapies, such as chemotherapy or immunotherapy, to achieve optimal efficacy [167]. For example, bevacizumab, an anti‐angiogenic agent widely used clinically, has demonstrated significant improvement in overall survival when combined with atezolizumab and chemotherapy (carboplatin and paclitaxel) in the ABCP regimen [168]. However, combination therapy with anti‐angiogenic inhibitors increases the risk of rare severe adverse events (AEs), such as bleeding and neutropenia [169]. More intriguingly, studies indicate that while existing EGFR‐TKI therapies have advanced treatment for EGFR‐mutant lung cancer, they primarily kill conventional cancer cells, allowing LCSCs to survive and drive disease recurrence and progression [170].
As mentioned earlier, targeting VEGF alone often leads to drug resistance, and exploring new targets may offer fresh insights. Within the bidirectional regulatory mechanism between LCSCs and the TME, LCSCs promote angiogenesis within the TME by bypassing VEGF blockade through upregulating other proangiogenic factors (such as FGF and IL‐8) and exosome‐mediated signaling. Therefore, simultaneously targeting these proangiogenic factors may further alleviate resistance to AAT. However, studies also indicate that multi‐targeted anti‐angiogenic drugs may not necessarily outperform single‐target anti‐VEGF agents in cancer treatment [167], a topic requiring further exploration and consideration. Furthermore, AAT‐induced TME hypoxia can further activate HIF‐1α, thereby sustaining drug‐resistant LCSCs and often mediating resistance [171]. Concurrently, conventional AAT combined with other therapies often fails to effectively eliminate LCSCs. Therefore, combining AAT with targeted LCSC therapies appears to be a promising new direction. Targeting LCSCs that overexpress angiogenesis markers may effectively mitigate AAT resistance, thereby achieving superior therapeutic outcomes.

7.3
Targeting CAFs and Stromal Metabolism
The crosstalk between LCSCs and CAFs forms a potent, self‐sustaining vicious cycle that serves as a primary driver of tumor progression [172]. LCSCs secrete factors that convert fibroblasts into CAFs. CAFs activate Wnt/β‐catenin, stiffen ECM, and sustain LCSCs' self‐renewal, invasion, and therapy resistance—forming a self‐reinforcing loop. However, therapeutic interventions targeting CAFs and the matrix remain underdeveloped.
As we all know, integrins serve as the bridge connecting LCSCs to the TME. LCSCs can receive various biochemical and mechanical signals through integrins, and these signals maintain LCSC stemness, ultimately leading to drug resistance [173]. For example, lung cancer resistance to the EGFR inhibitor erlotinib is associated with αvβ3 integrin‐mediated cancer stemness [174]. Clinically, attempts have been made to address this phenomenon [175]. In lung cancer treatment, the chimeric monoclonal antibody Etaracizumab, which targets αvβ3 integrin and disrupts its interaction with fibronectin, has entered clinical trials [176]. Concurrently, αvβ3‐specific CAR‐T cells are under development [177].
Furthermore, CAFs can provide suitable niches for LCSCs, thereby mediating drug resistance [116]. Research targeting CAFs and their crosstalk with LCSCs is ongoing. However, many current approaches, such as CAF depletion therapy, have demonstrated limited efficacy [115]. CAR‐T cell therapies targeting CAFs have also exhibited lethal toxicity [115]. Further research and optimization may be required in the future. The matrix is a crucial component of the TME, mediating diverse functions through integrins. Anti‐angiogenic therapies often induce hypoxia, leading to matrix stiffening that impedes treatment [178, 179]. Future modulation of ECM stiffness will be essential for improved therapeutic outcomes. Notably, regulating ECM stiffness cannot be a simple unidirectional process. Research indicates that high matrix stiffness may lock the ECM conformation, preventing the exposure of binding sites such as integrins and thereby reducing responsiveness to the ECM [180]. Simultaneously, low matrix stiffness may also promote tumor metastasis [181]. These factors warrant further consideration.

8
Discussion
The bidirectional regulation between LCSCs and the TME is not static; both the tumor and its surroundings may evolve over time. In studies, it has been found that the gene expression patterns of colon cancer cells are constantly changing over time, as evidenced by functional changes in the cancer cells [182]. It is suspected that similar changes exist in lung cancer, with neuro‐endocrine‐immune components in patients varying stage‐dependently over time [183]. As an important component of TME, changes in the immune system also suggest that TME may change over time. Is it questionable whether this is related to the function of LCSCs at different stages of cancer development? During tumor development, LCSCs continue to influence all aspects of surrounding cells over time, such as behavior, secretion, and genetic expression, ultimately realizing the education of the TME, making the TME lose its original natural inhibitory function, or even making it suitable for LCSCs to survive, becoming a breeding ground for LCSCs to support their survival. We therefore propose that at distinct stages of lung cancer progression, LCSCs secrete unique “education” signals that reprogram the TME in a stage‐specific manner. Collectively, these interactions drive temporal remodeling of the TME, underscoring the need for further mechanistic investigation to enable more precisely targeted therapies.
In the pathway of bidirectional regulation, besides influencing cellular secretion and behavior, the impact on cellular metabolism cannot be ignored. LCSCs not only inhibit anti‐tumor immune cells in the TME but also reprogram the metabolism of immune cells to reverse their behavior, shifting them to support LCSCs. As mentioned earlier, M1‐type TAMs possess anti‐tumor functions, while M2‐type TAMs exert pro‐tumor effects. Studies have found that metabolic reprogramming is one of the important factors affecting the polarization of TAMs. Within TAMs, α‐ketoglutarate, a product of glutamine metabolism, is associated with the activation of M2‐type TAM genes, which is regarded as a key factor promoting tumor progression. Meanwhile, under starvation conditions, both the expression level and activity of glutamine synthetase (GS) in M2‐type TAMs show an upward trend [184]. Therefore, in our opinion, M2‐type TAMs may share similar characteristics with LCSCs. Promoting the conversion of M2‐type to M1‐type—for example, by inhibiting the activity of GS—may indirectly suppress the invasion and metastasis of LCSCs. In the future, investigating the metabolic shifts of cells in the TME may help uncover more secrets of the TME and bring greater benefits to patients.
Based on this dynamic bidirectional regulation, numerous innovative approaches may effectively benefit patients. Through single‐cell and transcriptome sequencing of cancer tissues, collecting proteomic, radiomic, and metabolomic characteristics, AI can be utilized to analyze cancer subtypes, thereby helping patients quickly identify suitable treatment regimens. Spatial transcriptomics and single‐cell multi‐omics analysis can reveal the in situ interaction network between LCSCs and specific TME cells as well as their temporal evolution rules with unprecedented resolution. Meanwhile, integrating these high‐dimensional omics data, clinical imaging information, and patient prognosis data using AI algorithms is expected to construct dynamic models capable of predicting tumor progression trajectories and treatment responses, rather than merely performing static subtyping. The key challenge in the future lies in how to accurately define and obtain the key characteristics of LCSCs and TME for modeling, thereby developing clinical guidance tools with high predictive power.

9
Conclusion
In summary, our study reviewed the mechanisms by which the TME and LCSCs regulate each other and suggested a series of potential therapeutic targets, including targeting the TME and LCSCs, as well as personalized therapies, etc. We hope that our study will provide guidance for researchers.

Author Contributions
Conceptualization: Hongyi Lai, Ruihang Luo, Wei Zou, Xuyang Liu, Mingshan Liu, Jingtao Zhang. Writing of manuscript: Hongyi Lai, Ruihang Luo, Wei Zou, Xuyang Liu. Reviewing of manuscript: All authors. Study supervision: Mingshan Liu, Jingtao Zhang. Both authors read and approved the final manuscript.

Funding
This work is supported by the National Key R&D Program of China (2023YFC2508604), Health Commission of Jiangxi Province 202210336 and Youth Project of the First Affiliated Hospital of Nanchang University YFYPY202036.

Ethics Statement
The authors have nothing to report.

Consent
The authors have nothing to report.

Conflicts of Interest
The authors declare no conflicts of interest.

Supporting information

Table S1: cam471868‐sup‐0001‐TableS1.docx.

Table S2: cam471868‐sup‐0002‐TableS2.docx.

Table S3: cam471868‐sup‐0003‐TableS3.docx.

Table S4: cam471868‐sup‐0004‐TableS4.docx.