Wnts comprise a family of nineteen glycoproteins, which act as ligands (Komiya and Habas, 2008). The Wnt ligand binds to one of the ten types of Frizzled (Fz) extracellular receptors, which is categorized as a seven-transmembrane G-protein coupled receptors (GPCRs) (Komiya and Habas, 2008). There are two main Wnt pathways: 1) the canonical pathway, which is dependent on a low-density-lipoprotein-related protein5/6 (LRP5/6) co-receptor, and 2) the non-canonical β-catenin-independent pathway, which features ROR1/ROR2/RYK co-receptors, which can be further subdivided into the Planar Cell Polarity (Wnt/PCP) and the Wnt/Ca²⁺ pathways (Komiya and Habas, 2008; Ackers and Malgor, 2018). Most notably, the canonical Wnt pathway features a central element, β-catenin, which upon translocating to the nucleus, regulates gene expression by recruiting CREB-binding protein (CBP) to form complexes with the transcription factors, T-cell/lymphoid enhancer (TCF/Lef) (Wiese et al., 2017). If the Wnt ligands fail to activate the receptor, β-catenin is directed to a degradation complex composed of the tumor suppressor Adenomatous Polyposis Coli (APC), AXIN1/2, and kinases CK1α/GSK-3β (Duchartre et al., 2016). Tankyrase (TNKS) is an enzyme that mediates the ubiquitination and degradation of AXIN1/2; most notably tankyrase inhibitors will upregulate the function of the β-catenin degradation complex, serving as a promising target for therapeutics (Wang et al., 2021). Under normal conditions, the Wnt pathway serves a vital role in embryogenesis and the cell cycle, in which β-catenin promotes differentiation and development of specific T-cells, dendritic cells, and tissue systems (Kaldis and Pagano, 2009; Zhan et al., 2017). However, CSCs will “hijack” the Wnt pathway, resulting in uncontrolled proliferation of cancer cells (Fang et al., 2016). Over time, Wnt remains constitutively active due to mutations in tumor suppressor genes and oncogenes in select cancer subtypes (Fang et al., 2016). The large number of Wnt ligand-receptor complexes trigger multiple complex cascades, which play a large role in various types of cancer. For example, Wnt pathway activation results in reduced survival rates in over 50% of breast cancer patients, Wnt signaling modulates gastrointestinal cancers, and the balance between canonical and non-canonical Wnt signaling is linked to melanoma progression (Zhan et al., 2017). However, due to the complexity of the cascade, current Wnt targeted therapeutics are typically halted at the preclinical stage or phase I/II stages of clinical trials (Katoh, 2017). Regardless, further exploration into combined Wnt therapeutics and targeted treatments will provide valuable insight on the intricacies of CSCs.

Wnt pathways exhibit crosstalk with the Ras/MAP kinase, PI3/Akt, PLCγ, Notch, Hh and TGFβ/BMP pathways through various feedback loops (Katoh, 2017). Specifically, TGF-β pathway upregulates Wnt2B and Wnt3 expressions in turn activating the canonical Wnt pathway, while Wnt5A and Wnt11 genes are upstream activators of the non-canonical Wnt pathway (Katoh, 2017; Lecarpentier et al., 2019). Both Wnt signaling pathways require activation by the endoplasmic reticulum acyltransferase Porcupine5-7 (PORCN) (Liu et al., 2022). Specifically, PORCN is a membrane-bound O-acyltransferase that is required for the covalent attachment of fatty acids to Wnt ligands in a palmitoylation process, a vital step in Wnt ligand-receptor activation (Guan and Fierke, 2011; Liu et al., 2013a). Upon receptor activation, Dishevelled (Dsh/Dvl) is the first downstream activated intracellular signaling protein, which is involved in each branch of the Wnt cascade (Komiya and Habas, 2008). Activation of the canonical Wnt cascade upregulates the Hh cascade in rodent breast CSCs downstream, while the non-canonical Wnt cascade triggers PI3K-Akt activation promoting CSC survival (Katoh, 2017). Activation of Dsh also leads to β-catenin accumulation in the cytosol, which will then translocate to the nucleus to bind the TCF/Lef to induce cyclin D1, cMYC, monocarboxylate transporter 1 (MCT1), pyruvate dehydrogenase kinase (PDK), and fibronectin target genes. T cell factor 1 (TCF1) is also a downstream transcription factor of canonical Wnt signaling and plays a key role in the development on CD8⁺ memory and effector T cells (Lecarpentier et al., 2019).

Many human cancers are associated with mutation or loss-of-function in both canonical and non-canonical Wnt genes: 1) abnormal catenin beta 1 gene (CTNNB1), a downstream co-activator of TCF/Lef, is associated with human breast cancer (van Schie and van Amerongen, 2020) 2) tumor suppressor APC mutations are mainly associated with colorectal cancers (Aghabozorgi et al., 2019), and 3) mutated AXIN1/2 genes, which encode the AXIN protein, a key component of the β-catenin degradation complex, are associated with gastrointestinal cancers (Mazzoni and Fearon, 2014; Zhan et al., 2017; Lecarpentier et al., 2019). The hallmark of canonical Wnt associated cancers is an upregulation of cyclin-D1 and C-Myc genes, which are associated with the cell cycle (Lecarpentier et al., 2019). Interestingly, the mutations in the canonical Wnt pathway are commonly accompanied by mutations in the circadian genes: CLOCK, BMAL1, PER (Lecarpentier et al., 2019). Notable canonical Wnt pathway associated CSC surface markers include: 1) LGR5, a Wnt target gene that encodes the receptor for the R-spondin (RSPO) ligand, which is associated with colorectal, pancreatic, endometrial, and intestinal CSCs (Barker and Clevers, 2010; Lin et al., 2015; Tomita et al., 2020), 2) epithelial cell adhesion molecule (EPCAM) upregulation associated with colon cancer (Zhou et al., 2015), 3) CD44+/CD133+ CSCs are positive regulators of the Wnt pathway associated with prostate (Acikgoz et al., 2021) and colorectal cancer (Tsuchiya and Shiota, 2021), and 5) CD44v6 associated with colorectal CSCs (Lin et al., 2015; Duchartre et al., 2016; Katoh, 2017; Zhan et al., 2017). TERT genes, which maintain CSC’s long telomeres, enhance binding of β-catenin to its promoter, thereby highlighting the correlation between telomerase activity and the Wnt pathway (Zhan et al., 2017). Overall genes associated for the canonical Wnt pathway initiate the EMT of CSCs, while non-canonical Wnt genes are responsible for the persistence and metastasis of CSCs (Katoh, 2017). CSCs undergo EMT to transition from their previous static epithelial cell state to gain migratory and anti-apoptotic abilities (Pattabiraman and Weinberg, 2014). Also of interest, the key genes, AXIN2, APC downregualted-1 gene (APCDD1), and Dickkopf Wnt pathway inhibitor 1 (DKK1) (Koinuma et al., 2006; Sato et al., 2007; de Sousa and Vermeulen, 2016), are negative regulators of the Wnt pathway, so silencing these genes will inhibit CSC expansion. Combined therapeutics should target both the canonical and non-canonical Wnt pathways to irradicate CSCs.

The Wnt signaling pathway plays a critical role in stem cell biology and is implicated in CSC/tumor initiating cell population with aberrant Wnt signaling associated with tumor formation, suggesting that arresting Wnt signaling may block CSC maintenance (Kim and Kahn, 2014). Therapeutics which target the Wnt pathway have four general categories: 1) ligand/receptor-targeted drugs (ex. cirmtuzumab, rosmantuzumab and vantictumab): are currently in clinical trials and show anti-CSC effects, 2) PORCN inhibitors are under clinical trials targeting the small molecule PORCN, an O-acyltransferase that is required for Wnt ligand palmitoylation (ex. as IWP-2, WNT974, and ETC-159) (Liu et al., 2013b), 3) Tankyrase inhibitors, such as AZ1366, G007-LK, and JW55, upregulate the destruction complex via AXIN1/2 and are undergoing preclinical testing, 4) β-catenin inhibitors block CSC motility, most notably ICG-001 (and the clinical equivalent PRI-724) are in phase I/II clinical trials (Katoh, 2017; Wang et al., 2021). β-catenin inhibitors are challenging therapeutics due to the complexity of the Wnt pathway, but specific small molecule and monoclonal antibody treatments for select Wnt pathway proteins may also prove successful. For example, ICG-001, a small molecule inhibitor, specifically targets the CBP to arrest tumor growth in both animal and human cancer cell lines (Pattabiraman and Weinberg, 2014; Deng et al., 2020; Wiese et al., 2020). LF3 another small molecule inhibitor of the β-catenin/TCF4 interaction, which was found in an in vitro colon cancer mouse cell line, was shown to block CSC’s self-renewal, inhibit migration, and trigger differentiation into a benign tumor state, while not interfering with any other signaling pathway (Fang et al., 2016). Anti-FZD mAb, anti-ROR1 mAb, anti-RSPO3, and anti-LGR5 mAb therapeutics are currently in preclinical or clinical trials (Katoh, 2017). In another study, knockdown of mutated Wnt pathway members in chronic lymphocytic leukemia (CLL) cells showed that leukemia initiating cells (LIC) are dependent on the Wnt pathway for survival (Zhan et al., 2017). Studying patient specific CSC’s for their EMT and MET, immune editing, and metabolism through a genome sequencing process referred to as “omics monitoring” is vital for Wnt targeted therapy (Katoh, 2017). Through omics monitoring, Wnt pathway specific therapeutics can be optimized with combination therapy to target CSCs.

Despite extensive research on the Wnt signaling pathway, as of March 2023 there is still no FDA approved drug to treat cancer that specifically targets Wnt signaling. Notwithstanding the number of Wnt inhibitors in phase I/II clinical trials [NCT02413853, NCT02278133], overall safety and effectiveness remains a challenge as generalized Wnt inhibitors have detrimental effects on embryogenesis and overall cell homeostasis (Kaldis and Pagano, 2009; Zhan et al., 2017). For instance, the PORCN inhibitor WNT974 (aka LGK974) not only resulted in tumor growth inhibition in head and neck cancer cell lines but also decreased intestinal epithelium when administered at high doses (Liu et al., 2013a). Moreover, in human clinical trials, WNT974 treatment was generally well tolerated to treat advanced solid tumors with the most common side effect being dysgeusia, altered taste perception, in 50% of patients [NCT01351103], (Rodon et al., 2021). However, higher grade adverse events are expected to occur in clinical trials involving OMP-54F28, a Wnt recombinant fusion protein, as bone density loss had to be supplemented with vitamin D3 and calcium carbonate [NCT01608867]. Overall, concerns regarding Wnt pathway inhibition include effects on intestinal cells, bone density, dysgeusia, decreased appetite, fatigue, muscle spasms, and overall cell homeostasis (Jimeno et al., 2017). The questions as to whether the benefit of Wnt inhibitors outweighs the risks remains. However, despite diverging opinions regarding Wnt targeted signaling in cancer, as our understanding of the Wnt pathway increases, so does the growing enthusiasm for novel Wnt therapeutics (Figure 2).

The Notch signaling pathway not only plays a critical role in embryogenesis, but also in regulating CSC proliferation, maintenance, and differentiation, with notable contribution to angiogenesis (Venkatesh et al., 2018). Classically, Notch serves a dual role acting as both a tumor suppressor and an oncogene, at times leading to tumorigenesis depending on tissue type or genetic mutation (Meisel et al., 2020). Depending on the microenvironment, Notch is typically downregulated in prostate, skin, lung, liver, and some breast cancers, while Notch is upregulated in gastric, colon, pancreatic, and some breast cancers (Yang et al., 2020). Under hypoxic conditions, the Notch pathway undergoes EMT, therefore the consensus is that CSCs typically upregulate the Notch pathway to enhance their stemness properties (Meisel et al., 2020). Notch signaling initiates when Delta/Serrate/Lag2 (DSL) transmembrane Notch ligands (i.e., Delta-like 1/3/4, or Jagged 1/2) bind to a Notch receptor (Notch1–4) in a juxtracrine fashion, triggering cleavage via ADAM-10 protease, followed by gamma-secretase (γ-secretase) intramembrane protease, which releases the intracellular Notch portion from the plasma membrane (Anders et al., 2006; Venkatesh et al., 2018). This cleavage results in Notch’s intracellular domain (NICD) translocating to the nucleus and binding to a CBF1/Suppressor of Hairless/Lag-1 (CSL) transcription factor, ultimately activating target genes (Venkatesh et al., 2018; Yang et al., 2020). Since activation of the Notch signal transduction pathway in CSCs, treatment directed at this pathway may control CSC replication, survival, and differentiation, thus regulating CSC renewal and modulating CSC-mediated tumor formation and its recurrence (Venkatesh et al., 2018).

The following upstream activators of the Notch pathway contribute to the self-renewal capacity of CSCs: 1) nitric oxide (NO) synthase increases the stemness of liver CSCs, and NO targeted therapy increases tamoxifen potency in breast CSCs (López-Sánchez et al., 2021), and 2) MAP17, a 17 kDa non-glycosylated membrane protein that is associated with an increase in reactive oxygen species (ROS), is upregulated in cervical, breast, colon, and lung CSCs (Guijarro et al., 2007; Garcia-Heredia et al., 2017; Yang et al., 2020). Hyperactivation of Notch can typically occur through a genetic mutation resulting in either increased expression of the cleaved portion of Notch’s intracellular domain, or ligand-independent receptor activation (Meisel et al., 2020). However, genetically independent receptor activation via ligand abundance or increased ligand-receptor affinity could also result in carcinogenesis (Meisel et al., 2020). For example, an increase in the expression of Notch2 and Jag1 are correlated with an increased incidence of medulloblastoma and prostate cancer, respectively (Meisel et al., 2020). Downstream, Notch communicates with a variety of immune cells in the TME, such as myeloid derived suppressor cells (MDSCs), tumor associated macrophages (TAMs), and T regulatory cells (Tregs) (Venkatesh et al., 2018). MDSCs are exceedingly dynamic cells that further promote CSC stemness via NO secretion via both the STAT-3 (Müller et al., 2020) and Notch pathway (Chen et al., 2021a). Notch signaling also regulates the downstream proto-oncogene C-Myc, cell cycle regulators (ex. cyclin, D1, and Cdkn1) and can act as a positive or negative regulator on itself via Deltex E3 Ubiquitin Ligase 1 (DTX1), a protein coding gene (Matsuno et al., 1995; Meisel et al., 2020). The Hh pathway also induces C-myc expression, which acts in parallel with the Notch pathway (Meisel et al., 2020). The Notch pathway also exhibits crosstalk with STAT3 signaling to promote EMT transition (Shan et al., 2021).

The Notch signaling pathway significantly contributes to CSC maintenance in several cancer types, paving the way for development of Notch inhibitors as an anti-cancer strategy (Purow, 2009). The following markers and molecules are known to enhance and maintain the stemness properties of CSCs via Notch signaling: 1) delta-like ligand 4 (DLL4) maintains gastric CSCs and regulates tumorigenesis (Segami et al., 2021), 2) tumor necrosis factor-α (TNFα) inhibition in liver CSCs decreases cancer metastasis (Renz et al., 2018), 3) BMP4 has a dual role - it is upregulated in breast CSCs (Bach et al., 2018) while BMP4 inhibited hepatic CSC self-renewal (Zhang et al., 2012a), 4) Jagged 2 is upregulated on breast and lung CSCs under hypoxic conditions, and 5) blocking VEGFR2 causes skin CSC pool size reduction (Beck et al., 2011; Takebe et al., 2014; Yang et al., 2020). Similarly, the following genes aid in CSC expression via Notch: 1) Gli3 in oral squamous cell carcinoma (SCC) CSCs, 2) Notch1 in ovarian CSCs under hypoxic conditions, 3) BRCA1 in breast CSCs, 4) Hairy enhancer of split genes (Hes1-7) in medulloblastoma CSCs (de Antonellis et al., 2013), 5) Hey (Hey1, Hey2 and HeyL) in HNSCC (Moon et al., 2019), 6) Notch-regulated ankyrin-repeat protein (NRARP) associates with non-small lung cancer (Liao et al., 2018), 7) cyclin D1 silencing suppresses liver CSC differentiation (Zhang, 2020), 8) DVL1 gene is upregulated on glioblastoma CSCs (Hsu et al., 2021), and 9) ADAM19 is elevated on breast CSCs (Takebe et al., 2014; Meisel et al., 2020; Yang et al., 2020). In a microarray study, the following genes: Notch1, Hes4/5, Hey1/L, and NRARP were inversely proportional to concentrations of the Notch inhibitors (i.e., MK-0752), confirming their effectiveness as biomarkers (Takebe et al., 2014). Strategies that target Notch genes, such as Notch 1, may prove beneficial for cancer treatment (Takebe et al., 2015; Gharaibeh et al., 2020). The following molecules are known to inhibit Notch signaling: 1) microRNA-34a is a potent tumor suppressor known to inhibit gastric CSCs (Jang et al., 2016), 2) PER3 polymorphisms are associated with colorectal CSCs, 3) miR-200b-3p expression is associated with pancreatic CSCs and downregulates colorectal cancer (CRC) cells (Feifei et al., 2019), and 4) miR-26a expression is found on osteosarcoma CSCs and inhibits ovarian (Gao et al., 2020) and colorectal cancers (Yang et al., 2020; Chen et al., 2021b). Inhibition of these Notch signaling molecules could also serve as candidate targets for developing potential cancer therapeutics (Zhang et al., 2017a; Zhang et al., 2017b).

The overall premise in targeting Notch signaling pathway is that combining a specific notch inhibitor with a traditional chemotherapeutic will result in a more potent anti-cancer treatment plan. For example, Notch1 inhibition in combination with a chemotherapeutic drug reduced CSC self-renewal in HNSCC CSCs in vitro and in vivo, confirming Notch1 as an ideal target for cancer treatments (Fukusumi and Califano, 2018; Venkatesh et al., 2018). Clinical agents which target Notch signaling fall under two categories: 1) γ-secretase inhibitors (GSIs) (ex. peptide isosteres, azepines, and sulfonamides) inhibit cleavage of Notch resulting in suppression of angiogenesis and apoptosis of tumor cells, and 2) monoclonal antibodies (mABs) that interfere with Notch ligand-receptor bonding or prevent the conformational change required for cleavage (Venkatesh et al., 2018; Yang et al., 2020). The following GSIs have emerged as promising therapeutics for inhibiting CSCs specifically by modulating the Notch pathway: 1) RO4929097 combined with bevacizumab decreased glioma CSCs in phase I clinical trials, and other RO4929097 combinations are known to combat advanced solid tumors [ NCT0113123], 2) PF-03084014 combined with gemcitabine decreased pancreatic CSCs in vivo and decreased desmoid tumors in phase II studies, 3) MRK003 decreased glioma CSCs in vivo, and 4) MK-0752 combined with docetaxel decreased metastatic breast cancer CSCs in vitro, MK-0752 combined with cisplatin treated ovarian cancer in vivo, MK-0752 combined with gemcitabine treated ductal adenocarcinoma of the pancreas in preclinical trials, and MK-0752 showed inhibition of pediatric central nervous system tumors in phase I clinical trials (Venkatesh et al., 2018; Yang et al., 2020). Interestingly, GSIs proved ineffective against in triple negative (ER, PgR and HER-2) breast cancer, which displayed elevated Notch signaling, emphasizing the unique complexity of CSCs and tumor types (Meisel et al., 2020).

In parallel, preclinical studies and clinical trials which feature mAbs that target the Notch pathway and inhibit CSCs have also shown encouraging results: 1) Tarextumab (OMP-59R5) decreased breast, small-cell lung, ovarian, and pancreatic CSCs by 40% after chemotherapy termination in vivo and effectively decreased solid tumors in phase Ib/II clinical trials by targeting Notch2/3 [NCT01277146], and 2) enoticumab (ex. REGN421, SAR153192) partially treated ovarian cancer and solid tumors by inhibiting DLL4 (Chiorean et al., 2015; Venkatesh et al., 2018; Smith et al., 2019) [NCT0087155]. Interestingly, through a combination of both a GSI and a mAb in early-stage cancers, there is a marked increase in chemotherapeutic effect, but a particular side-effect that commonly arises is gastrointestinal toxicities (Takebe et al., 2014; Venkatesh et al., 2018). Through identification of patient-specific Notch pathway pharmacodynamic biomarkers, the patient can be matched to the appropriate Notch inhibitor to create an appropriate treatment plan (Takebe et al., 2014; Venkatesh et al., 2018).

The discovery of effective, safe Notch inhibitors that predict a positive clinical outcome is essential. However, the Notch signaling pathway is not only vital for CSCs self-renewal but is also necessary for embryogenesis and angiogenesis (Venkatesh et al., 2018), posing as a hurdle for selective Notch-targeted therapies. Research is currently focused on Notch-selective GSIs, which inhibit angiogenesis and trigger tumor cell apoptosis, as well as inhibitory monoclonal antibodies that target Notch pathway molecules (Venkatesh et al., 2018). For example, in an in vivo study Notch pathway inhibition via delivery of a GSI reduced the percentage of CSCs within the tumor, while activation of the Notch pathway increased the percentage of CSCs within the tumor (Abel et al., 2014). Moreover, in a phase I clinical trial of Tarextumab, a mAb Notch2/3 inhibitor that targets solid tumors, GI toxicity was the most common adverse event, with patients experiencing diarrhea (81%), fatigue (48%), nausea (45%), anorexia (38%), vomiting (38%), abdominal pain (24%), and constipation (24%) [NCT01277146] (Smith et al., 2019). Tarextumab was also tolerated better at low doses <2.5 mg weekly [NCT01277146] (Smith et al., 2019). Despite intestinal toxicity being the most common side effect of Notch inhibitors, in vivo testing confirms that GI toxicity can be reduced by implementing a Notch3-selective antibody rather than Notch1/2 inhibitors (Wu et al., 2010). Therefore, in order to limit the toxicity of GSIs and mAbs doses must be selective, moderated, and taken intermittently. In conclusion, while drugs targeting the Notch pathway are typically halted at phase I/II clinical trials, an in-depth evaluation of Notch signaling has paved the way for novel CSC therapeutics with the ultimate goal of curing cancer (Figure 3).

The JAK-STAT signaling pathway is integral to the survival, self-renewal, maintenance, and metastasis of CSCs (Yang et al., 2020). The binding of ligands, typically cytokines or interferons (IFNs), to the cytokine cell-surface receptor results in receptor dimerization, which induces both positive and negative regulatory pathways (Kisseleva et al., 2002). After receptor-ligand engagement, the tyrosine kinase JAK (i.e., JAK1-3, and Tyk2), which is composed of seven domains, will phosphorylate the receptor, creating binding sites for proteins possessing an SH2 domain, such as STAT proteins (Kisseleva et al., 2002; Yang et al., 2020). STAT proteins (i.e., STAT1-4, STAT5a/b, STAT6) are transcription activators composed of N and C terminals, a DNA-binding region, and SH2/3 domains (Yang et al., 2020). Upon STAT binding to the receptor, it also undergoes tyrosine phosphorylation, causing the STAT to dissociate from its corresponding receptor, form an anti-parallel dimer, and translocate to the nucleus to affect target downstream molecules (Schindler et al., 2007; Yang et al., 2020).

Specifically, JAK-STAT pathway constitutive activation and mutation are associated with many tumors (Yang et al., 2020). The following JAK-STAT upstream activators contribute to the self-renewal capacity of CSCs: 1) IL-6 activates the JAK1/STAT3 pathway in endometrial CSCs, induces EMT in breast and colorectal CSCs (Zhang et al., 2018a), 2) IL-10, an immunosuppressive cytokine, induces stemness properties in lung CSCs (Yang et al., 2019), 3) GM-CSF will create a positive feedback loop between CSCs and TAMs within the TME (Kokubu et al., 2016), 4) PDGF increased stemness and metastatic potential in ovarian CSCs (Raghavan et al., 2020). Downstream, JAK-STAT upregulates PI3-Akt signaling and promotes activation of the MAPK/ERK pathway. The PI3K protein (similar to STAT) contains an SH2 domain, so it can bind to JAK phosphorylated tyrosine receptors and activate the PI3-Akt pathway (Rawlings et al., 2004). In a similar fashion Grb2, an integral protein for MAPK/ERK signaling, also features a SH2 domain, which allows it to bind JAK phosphorylated receptors, and activate the MAPK/ERK pathway (Rawlings et al., 2004). JAK-STAT signaling can integrate multiple signaling pathways, contributing to its complexity and therapeutic potential.

JAK-STAT signaling behaves as a double-edged sword, whereby STAT3 is commonly linked to CSC’s immunosuppression capacity and aberrant TME, while STAT 1/2 activates an anti-tumor immune response via interferons, (i.e., IFN I/II) (Owen et al., 2019; Yang et al., 2020). Therefore, therapeutic avenues should navigate the complexity of JAK-STAT signaling with caution. The following genes, markers, and molecules are known to enrich the stemness properties of CSCs via JAK-STAT signaling: 1) Oct4 promotes ovarian CSCs, and lung CSCs through M2 macrophage polarization (Lu et al., 2020); 2) erythropoietin (Epo) upregulates breast and colorectal CSCs, as well as human gliomas via Epo-dependent constitutive activation of STAT-5 (Kondyli et al., 2010); 3) retinol-binding protein 4 (RBP4) promotes colon CSCs self-renewal via the JAK2-STAT3 pathway (Karunanithi et al., 2017); 4) Hypoxia inducible factor-1 alpha (HIF-1α), a key transcription factor in cancer progression (Rashid et al., 2021), is associated with glioma CSC upregulation (Yang et al., 2020); 5) miR-500a-3p in hepatocellular carcinoma CSCs leads to STAT3 constitutive activation (Yang et al., 2020); 6) TH17, an immunosuppressive cytokine, upregulates CSCs via the secretion of IL-17 via a STAT3-dependent signaling pathway in ovarian, colorectal, and gastric cancers (Chen et al., 2021a). In addition to STAT3, the MAPK (Mitogen Activated Protein Kinase) signaling pathway appears to also solicit the link between TH17 and CSCs in ovarian and pancreatic cancers (Chen et al., 2021a). The following molecules are known to inhibit JAK-STAT signaling: 1) Mir-218 inhibits the JAK-STAT3 pathway resulting in downregulation of lung CSC’s self-renewal capacity (Yang et al., 2017); 2) Ajuba, a LIM domain-containing scaffolding protein, promotes the proliferation of colorectal CSCs through suppression of JAK1/STAT1 (Jia et al., 2017); 3) Von Hippel–Lindau (VHL) cell surface proteins act via the JAK2/STAT3 pathway to suppresses self-renewal ability of glioma CSCs (Yang et al., 2020). Recognizing the dual properties of JAK-STAT, the activation and inhibition of these JAK-STAT signaling regulators represent potential candidate targets for cancer therapy.

Due to the integral contribution of the JAK-STAT signaling pathway in CSC preservation, JAK-STAT inhibitors serve as a notable approach to anti-cancer treatment. Targeted JAK-STAT pathway therapeutics fall into one of three categories: 1) cytokine or receptor antibodies, 2) JAK inhibitors, 3) STAT inhibitors (Hu et al., 2021).

In reference to the first category, antibody-cytokine novel fusion proteins (immunocytokines) exert an anti-cancer effect, as shown in reference to IL-2 in phase I/II clinical trials (Mortara et al., 2018). Ruxolitinib in combination tocilizumab, antibodies against IL-6, show improved survival ovarian cancer tumors in vivo (Qureshy et al., 2020). Ruxolitinib enhances cancer treatment in HNSCC, pancreatic, and glioblastoma (Qureshy et al., 2020).

Under the second category, JAK inhibitors typically result in immunosuppressive effects through decreasing proinflammatory cytokines. For example, pacritinib is a JAK2 inhibitor currently under clinical trials to treat AML, prostate, colon, rectal, and non-small cell lung cancer, but no response was observed in colorectal cancer patients (Hu et al., 2021). Cerdulatinib, a JAK1/2 inhibitor, is undergoing clinical trials for treatment of non-Hodgkin’s lymphoma [NCT04757259].

In the third category, STAT inhibitors include SMIs, peptide inhibitors, STAT-targeting small interfering RNAs (siRNA), ASOs that interfere with STAT mRNA, and decoy oligonucleotides (ODNs) (Hu et al., 2021). STAT3 inhibition can occur via the SMI drug, Napabucasin (Chen et al., 2021a). Napabucasin suppresses MDSC immunosuppressive capacity in melanoma-bearing mice (Bitsch et al., 2022) and is in phase I clinical trials for assessment of safety, tolerability, and pharmacokinetics in healthy volunteers (Dai et al., 2021) and patients with metastatic colorectal cancer (Taniguchi et al., 2021). A notable peptide inhibitor is PY*LKTK, which was shown to disrupt STAT3:STAT3 dimerization in vitro (Yue and Turkson, 2009). siRNA targeting STAT 5b enhances the chemosensitivity of gastric cancer cells to gefitinib in clinical trials (Sun et al., 2015). AZD9150, a STAT3 ASO, is undergoing phase 1 clinical trials for patients with advanced hepatocellular carcinoma (HCC) [NCT01839604] as well as advanced solid malignancies [NCT03394144]. ODNs bind to the DNA binding domain preventing STAT proteins from reacting with their appropriate DNA response element (Furqan et al., 2013). Specific ODN targeting STAT3/STAT5 (ex. K562, U251, A172, etc.) were studied in vitro and in vivo on a mice xenograft lung cancer model (Furqan et al., 2013).

JAK-STAT pathway dysregulation is a major contributor to cancer progression. Within the TME, JAK-STAT signaling controls cytokine secretion, inflammatory cascades, and regulates CSCs maintenance and proliferation via upstream and downstream mechanisms. In most cancer subtypes JAK-STAT pathway inhibition serves as a potential chemotherapeutic treatment, but selective targeting remains a challenge due to variabilities in patient genetics, epigenetics, and variations in tumor subtypes. Therefore, JAK-STAT pathway combined therapy targeted drugs should not only inhibit overactivation of the pathway but also delve into the intricacies of JAK-STAT pathway crosstalk as a potential means for novel therapeutics. Current JAK-STAT pathway therapeutics, which include antibody therapy, JAK-inhibitors, and STAT-inhibitors, are typically in phase I/II clinical trials and are focused on administering the drug alone or in combination with chemotherapy, rather than targeting multiple major CSC signaling pathways [NCT03421353, NCT04021082]. JAK-STAT is a complex, non-linear signaling pathway and in some cases crosstalk with JAK-STAT and downstream signaling pathways, such as MAPK/Erk and PI3K/Akt, can contribute to chemotherapy resistance. For example, in a colon cancer cell line, IL-6 secretion mediated the activation of JAK-STAT3 and MAPK/Erk pathways, which increased anti-apoptotic proteins, Bad and Bcl-2, thus leading to chemoresistance to 5-Fluorouracil (Li et al., 2018a). Moreover, JAK1/2 inhibitors, namely AZD1480 for the treatment of solid tumors and Momelotinib that treats non-small cell lung cancer, resulted in clinical trial termination due to neurotoxicity and neutropenia, respectively [NCT01112397, NCT02206763]. Conversely, Ruxolitinib showed improved overall survival rates in a subgroup of pancreatic ductal adenocarcinoma patients with inflammation [NCT01423604]. Despite a clinical trial depicting the benefit of Ruxolitinib in a select patient population, the majority of clinical trials testing the effects of Ruxolitinib resulted in termination, perhaps due to JAK inhibition resulting in decreased immune response, which hinders overall anti-tumor effects [NCT02117479, NCT02119663, NCT01562873], (Schwartz et al., 2017). Despite potential drawbacks in research, JAK-STAT targeted CSC therapy is a novel research avenue with the future direction of curating personalized medicine treatment plans based on variations in JAK-STAT signaling among different tumor subtypes (Figure 4).

The hedgehog family is comprised of three homologs: Sonic hedgehog (Shh), Indian hedgehog (Ihh), and Desert hedgehog (Dhh), of which Shh is the best understood (Sasai et al., 2019). Under normal conditions, the hedgehog signaling pathway plays a unique role in embryonic development and cell differentiation (Yang et al., 2020). However, abnormal Hh pathway activation is linked to a variety of cancer types (Schulenburg et al., 2015), and CSCs uniquely express high levels of Hh activation (Yang et al., 2020). When extracellular Hh ligands bind to its downstream twelve-pass transmembrane receptor, Patched (PTCH), such engagement triggers signal transduction, most notably activating downstream GLI transcription factors (i.e., Gli1/2/3) (Pietrobono et al., 2019; Yang et al., 2020). In the absence of ligand-receptor activation, PTCH acts a negative regulator that will inhibit the seven-pass transmembrane GPCR Smoothened (SMO), preventing signal transduction (Pietrobono et al., 2019; Yang et al., 2020). Similarly, in the absence of a ligand, suppressor of fused (SUFU) binds to GLIs, anchors them in the cytoplasm, and prevents them from activating GLI target genes, thereby inhibiting Hh signaling (Rimkus et al., 2016). Selective inhibition of Hh signaling is an effective strategy for impeding cancer progression and halting CSC metastasis.

The following Hh upstream activators contribute to the maintenance, self-renewal, and regenerative capacity of CSCs: 1) IL-6 activates non-small-cell lung CSCs (Wei, 2019), 2) Gli1 initiates Hh transcription and notably promotes EMT and metastasis of non-small cell lung carcinoma (Jiang et al., 2020), and 3) Gli2 mainly inhibits transcription but in some instances also triggers Hh transcription due to variation in levels of phosphorylation (Pietrobono et al., 2019; Yang et al., 2020). Downstream, the Hh pathway also activates Gli3, a known transcription inhibitor, and Gli3 mutation has been linked to pancreatic adenocarcinoma (Yang et al., 2020). After translation of the Hh protein, the membrane-bound Hedgehog acyltransferase (Hhat) catalyzes a Hh-palmitate linkage and recycles the Hh ligand, so it may again bind to PTCH1 (Rimkus et al., 2016). Hh signaling also contributes to the metastasis of CSCs through upregulation of the following downstream markers: 1) ALDH1 a biomarker associated with early-stage non-small cell lung cancer via Hh signaling (Raz et al., 2012), 2) CD44 a biomarker associated with hepatocellular carcinoma via Hh signaling (Wang et al., 2020), 3) Twist1, Snail, and ZEB1 contribute to CSCs EMT capacity (Rimkus et al., 2016; Li et al., 2021), 4) C-Myc and cyclin-D1 are transcription factors involved in CSC proliferation and cell cycle progression (Rimkus et al., 2016; Elbadawy et al., 2019), 5) Nanog, Oct4, and Sox2 are transcription factors that contribute to CSC self-renewal (Shan et al., 2012; Yin et al., 2015; Rimkus et al., 2016; Yang et al., 2020) 6) PDGFR upregulation on glioblastoma CSCs (Schulenburg et al., 2015; Yang et al., 2020). The Hh pathway also displays crosstalk with the Wnt and Notch pathways via Hh upregulation of Wnt2 and Jagged1, respectively (Ding and Wang, 2017; Yang et al., 2020).

Increased Hh signaling is observed in CSCs to support their maintenance, growth, and metastasis (Yang et al., 2020). The following genes, markers, and molecules are known to enrich the stemness properties of CSCs via Hh signaling: 1) CK2α, a protein kinase, supports lung CSCs via upregulation of Gli1 (Zhang et al., 2012b), 2) retinoic acid receptor α2 (RARα2) upregulates Hh and Wnt pathways to maintain myeloma CSCs (Yang et al., 2013), 3) PPM1D, a protein phosphatase, increased Hh signaling via target protein Gli1 in medulloblastoma CSCs (Wen et al., 2016), and 4) lncHDAC2 promotes liver CSCs through Hh signaling (Yang et al., 2020; Jahangiri et al., 2021). Speckle-type POZ (SPOP) is a unique protein in that it at times inhibits Hh signaling in gastric cancer through increasing the degradation of Gli2 (Zeng et al., 2014), while SPOP can also upregulate Hh signaling, decreasing CRC CSC’s rate of apoptosis (Zhi et al., 2016). Similarly, Vasohibin 2 (VASH2) protein suppresses pancreatic CSCs through decreasing SMO and Gli1/2 (Yang et al., 2020), while also functioning as a tumor promoter, stimulating resistance to doxorubicin (DOX) (Mirzaei et al., 2021) upregulating pancreatic CSCs via the Hh and Notch pathways (Liang et al., 2021a). The following molecules are known to inhibit Hh signaling thereby hindering the metastasis of CSCs: 1) BCL6, a transcriptional repressor, inhibits the positive Hh effectors, Gli1 and Gli2, in medulloblastoma CSCs (Yang et al., 2020), 2) RUNX3, a transcription factor, suppresses colorectal CSCs metastasis by blocking Gli1 (Kim et al., 2020), 3) miR-361-3p hinders retinoblastoma CSC self-renewal, by Gli1/3 inhibition and regulates liver CSCs through directly targeting SOX1 (Pietrobono et al., 2019; Qu et al., 2021), 4) miR-326, a scaffolding protein, impedes Gli2 and regulates SMO, thus creating a negative feedback loop for modulating Hh signal transduction (Jiang et al., 2014; Pietrobono et al., 2019), 5) β-arrestin1 (Arrb1), a scaffolding protein, acts as another negative modulator of Hh signaling through acetylation of Gli1, specifically identified in relation to medulloblastoma CSCs (Miele et al., 2017). Probing Hh-associated genes, markers, and molecules may reveal novel pathways or refine existing signaling pathways associated with CSC metastasis, cancer onset, and progression, paving the way for Hh-based cancer therapeutics.

The following molecules and proteins have emerged as promising therapeutics for inhibiting CSCs by controlling the Hh pathway. Targeted Hh pathway therapeutics fall into one of three categories: 1) SMO inhibitors, 2) Gli inhibitors, and 3) ligand/enzyme inhibitors (Rimkus et al., 2016).

Under the first category, SMO inhibitors prevent Gli activation downstream, leading to target gene inhibition and CSC suppression (Rimkus et al., 2016). LDE225 (also known as Erismodegib/Sonidegib) is a SMO inhibitor as evidenced in vitro and in vivo as a SMO antagonist (D’Amato et al., 2014), and is currently marketed as Odomzo® to treat advanced stage basal cell carcinoma (BCC) [NCT04066504]. The same SMO inhibitor, Sonidegib, also acts synergistically with JAK2 inhibitors (ex. ruxolitinib) to inhibit HER2-positive triple-negative breast CSCs in vitro (Doheny et al., 2020). In the same token, the Hh antagonist Vismodegib, a cyclopamine derivative that binds SMO preventing Gli activation and macrophage recruitment modifier, is a clinically approved therapy to treat BCC and is undergoing phase II clinical trials for medulloblastoma and other cancers (Gampala et al., 2021) [NCT00833417]. Another SMO inhibitor, steroidal alkaloid cyclopamine, is a natural plant product that directly antagonizes Hh in vivo (Incardona et al., 1998; Tremblay et al., 2009) and is currently under phase II clinical trials for medulloblastoma [ NCT01878617]. Due to the integral role of SMO in Hh signaling, it is the main target of many therapeutics.

Under the second category, Gli antagonists (GANTs) act downstream of the Hh signaling pathway and, therefore can be activated via Hh-independent pathways (Rimkus et al., 2016). Arsenic trioxide (ATO), a Gli1/2 suppressor, is the first FDA approved drug that targeted the Hh pathway to treat acute promyelocytic leukemia ATO and currently undergoing preclinical testing for a variety of cancers, including prostate and colon cancer (Rimkus et al., 2016; Carpenter and Ray, 2019). GANT-61 inhibits both Gli1/2 downstream effectors and decreases tumor growth in vivo and in vitro in a variety of cancers, including prostate and ovarian cancers (Rimkus et al., 2016; Carpenter and Ray, 2019). Interestingly, Balanophora polyandra Griff (BPPs) plant extract polysaccharides also suppress the Hh pathway through EMT inhibition downstream, in vivo and in vitro, but the specific mechanism remains largely unknown (Li et al., 2021). Combined SMO and Gli-targeted therapies are currently undergoing preclinical testing (Rimkus et al., 2016), as finding the balance between upstream and downstream regulation could be the key to Hh targeted therapeutics.

Under the third category, Shh ligand is the most abundant of the hedgehog ligands, and therefore is a promising therapeutic target (Rimkus et al., 2016). Shh monoclonal antibody 5E1 has been implicated to inhibit medulloblastoma (Coon et al., 2010) and to reduce pancreatic tumor size (Chang et al., 2013) in vivo, but 5E1 has not yet reached human trials (Carpenter and Ray, 2019). RU-SKI 43 inhibits the downstream enzyme Hhat and reduces proliferation of breast cancer and pancreatic cancer cells, as shown in preclinical trials (Carpenter and Ray, 2019). Through both upstream and downstream inhibitors, Hh targeted therapeutics serve as effective strategies to decrease tumor growth and metastasis.

The Hh pathway is necessary for the maintenance of CSCs in multiple cancer subtypes. Current research targeting the Hh pathway is focused on SMO inhibitors, GANTs, and ligand/enzyme inhibitors, of which select inhibitors are FDA approved to treat BCC while others are still undergoing pre-clinical and clinical trials. For example, GDC-0449 (Vismodegib), a SMO antagonist that suppresses the activation of downstream Hh target genes, displays anti-tumor activity in locally advanced and metastatic BCC [NCT00607724]. Moreover, Vismodegib, displayed good activity when treating BCC over long-term regimens (>3 months) (Ally et al., 2014; Dréno et al., 2017). However, a phase II clinical trial administering Vismodegib to treat BCC resulted in poor efficacy and multiple adverse events among patients, including muscle spasms (76%), alopecia (58%), and dysgeusia (50%) (Sofen et al., 2015). While there are several unresolved cases regarding the specific role of the Hh signaling pathway, it is known that Hh signaling plays a critical role in cancer progression, and a better understanding of Hh interactions in each tumor sub-type is vital. It is also important to highlight that the Hh pathway may be activated in response to chemotherapy (Merchant and Matsui, 2010) and can overcome chemotherapy resistance in medulloblastomas (Bar et al., 2007) and pancreatic cancer (Olive et al., 2009). Therefore, Hh inhibitors may be better served to combat early-stage tumors or should be administered following surgical removal of the tumor mass. Overall, Hh inhibitor administration during end-stage tumor progression typically correlates with poorer prognosis. In association with this claim, one established paradox in regard to CSC treatment states that positive patient response rates do necessarily correlate to patient survival rates, and the key to novel CSC treatments should be to develop new statistical methods to better asses changes in CSCs at multiple timepoints (Huff et al., 2006). Despite the vast array of clinical knowledge surrounding Hh signaling pathway (Figure 5), a greater understanding of Hh pathway’s role in CSCs is essential for designing combination therapy that utilizes selective Hh-inhibitors, traditional chemotherapeutics, and/or other major CSC signaling pathway inhibitors.

VEGF is not only necessary for angiogenesis and tumor progression but also acts independently of its angiogenic role to maintain CSCs via paracrine and autocrine signaling (Mercurio, 2019a; Müller et al., 2020). Inhibitors of VEGF may regulate CSCs by decreasing VEGF and subsequently suppressing tumor progression (Müller et al., 2020). VEGF ligands A-E and placental growth factor (PIGF) ligand each activate varying vascular endothelial growth factor receptors (VEGFRs), which each possess seven immunoglobulin-like extracellular domains and an intracellular tyrosine kinase domain (Shibuya, 2011; Takahashi, 2011). There are 3 types of VGFRs: VEGFR-1 (Flt-1), VEGFR-2 (Flk-1/KDR), and VEGFR-3 (Flt-4) (Shibuya, 2011). For instance, VEGF-A regulates angiogenesis through activating VEGFR-1/2, while ligands VEGF-C/D activate VEGFR-3 to initiate lymphangiogenesis, the formation of new lymphatic vessels from pre-existing ones (Shibuya, 2011). Therefore, VEGF-A and VEGFR-1/2 are the classical targets for anti-angiogenic therapeutics (Takahashi, 2011). Upon ligand-receptor engagement, the cell surface receptors will become activated and dimerized through transphosphorylation, resulting in a downstream signaling cascade (Holmes et al., 2007).

VEGF not only promotes CSC stemness by stimulating angiogenesis through paracrine signaling but also creates a microenvironment for CSCs through neuropilin1 (NRP1) via an autocrine signaling pathway (Beck et al., 2011). VEGF ligands can also act independently of the VEGFRs via neuropilins1/2 (NRPs), transmembrane glycoprotein receptors largely expressed on tumor cells, which function in tumor initiation and maintenance (Mercurio, 2019a). VEGF-NRP2 triggers the extracellular matrix (ECM) glycoprotein laminin’s engagement with α6β1 integrins on the cell membrane (Goel et al., 2012). Laminin-integrin engagement activates the FAK/Ras pathway, stimulates Hh target gene Gli1, and ultimately enhances the expression of a stem cell factor, BMI-1 (Mercurio, 2019a). The VEGF pathway also activates the Rac1GTPase, which in turn inhibits the Hippo kinase LATS, thus ultimately promoting the activation of TAZ, which can reprogram cancer cells into cancer stem cells (Piccolo et al., 2014; Elaimy et al., 2018). While the classically activated VEGF-VEGFR pathway contributes to angiogenesis and lymphangiogenesis to support the maintenance of the tumor, the VEGF-NRP loop facilitates the self-renewal and persistence of CSCs (Müller et al., 2020; Yang et al., 2020). Most notably, the VEGF-NRP loop results in a positive feedback loop with the Hh pathway, since Gli1 upregulates NRP2 expression (Mercurio, 2019a). Similarly, the crosstalk between the VEGF-NRP loop and Hh pathway is also crucial to Hh signaling that engages the PIGF ligand in the onset and progression of medulloblastomas (Snuderl et al., 2013).

The following genes, markers, and molecules enhance the stemness properties of CSCs via VEGF signaling: 1) C-Myc and Sox2 genes show a strong correlation with VEGFR-2 through activation of JAK2/STAT3 signaling in breast CSCs in vivo (Zhao et al., 2015); 2) Hypoxia-inducible factor-1α α (HIF-1α) is a known target gene of VEGF that interacts via the PI3K/Akt pathway (Choi et al., 2011); 3) CD133, a transmembrane protein biomarker found on CSCs, promotes tumor stemness and reoccurrence of HCC through a mechanism dependent on VEGFR2 and Nanog (Liu et al., 2017); 4) COUP transcription factor II (COUP-TFII) is a unique protein that stimulates lymphangiogenesis and tumorigenesis in prostate cancer, enhancing NRP-2 expression (Lin et al., 2010; Qin et al., 2013); and 5) Heparin and NRP1 are involved in VEGFR activation via VEGF-A mechanisms (Takahashi, 2011). Interestingly, some Notch ligands (ex. DLL4) display unique properties in that they may suppress VEGFR2 and NRP1 expression, while other Notch ligands, such as DLL1, stimulate VEGF signaling (Mercurio, 2019a). The following molecules are known to naturally inhibit VEGF signaling, thereby impeding CSCs metastasis: 1) Vitamin C inhibits VEGF signaling through degrading HIF-1α (Zhao et al., 2020); and 2) Melatonin (MLT), a natural pineal gland hormone, regulates neoangiogenesis and inhibits lung and breast cancer cells through the downregulation of HIF-1α/ROS/VEGF (Cheng et al., 2019). While VEGF activates the Raf/MEK/ERK and PI3K/Akt pathways downstream (Ferrara et al., 2003), additional studies are warranted to explore the distinct crosstalk between VEGF and its associated genes and molecules, such as Notch and JAK-STAT signaling components, to unravel novel CSC-targeted therapeutics.

The following VEGF targeted therapeutics are promising therapeutics for inhibiting CSCs. Targeted VEGF pathway therapies fall into one of four categories: 1) VEGF ligand inhibitors, 2) receptor inhibitors, 3) VEGF decoy receptors, and 4) ribozymes targeting VEGF (Cardones and Banez, 2006).

Under the first category, bevacizumab (Avastin) is an anti-VEGFA ligand inhibitor approved by the FDA to treat metastatic CRC and renal cell carcinomas (RCC) in combination with traditional therapeutics (Cardones and Banez, 2006). Bevacizumab is currently undergoing phase II clinical trials to treat advanced stage prostate cancer in combination with traditional chemotherapeutic Docetaxel [NCT00574769]. Despite bevacizumab being the most common anti-VEGF drug by blocking VEGFR tyrosine kinases it also enhances VEGF/NRP signaling, thereby resulting in a “VEGF paradox”: bevacizumab cannot fully eradicate prostate CSCs since the drug will enrich CSCs while primarily targeting non-CSCs (Ferrara, 2005; Mercurio, 2019a). To this end, VEGFR therapy combined with NRP2 inhibitors is more effective than the stand-alone drug to treat prostate cancer (Mercurio, 2019a). Similar to bevacizumab, microRNA-140-5p is also an anti-VEGFA ligand inhibitor that hinders breast CSC tumor metastasis and angiogenesis both in vitro and in vivo (Lu et al., 2017). The most common anti-VEGF strategies have focused on molecules and ligands aimed at diminishing VEGFR action, but the downside to this approach is that VEGFRs bind to multiple ligands within their super-family (Ferrara, 2005). Ligand targeted therapeutics should not only diminish non-CSC tumor cells but also eradicate CSCs by hindering the VEGF-NRP loop.

Under the second category, VEGFR tyrosine kinase (TRK) inhibitors hinder angiogenesis and tumor growth (Takahashi, 2011). Sunitinib, a VEGFR tyrosine kinase inhibitor, is FDA approved in an oral tablet form to treat RCCs and gastrointestinal tumors (Takahashi, 2011). Sorafenib is also a tyrosine kinase VEFGR2/3 inhibitor and is FDA approved to treat advanced renal carcinoma and HCC (Takahashi, 2011). Similarly, ramucirumab (Cyramza) is a monoclonal antibody blocks VEGFR2 and is FDA approved to treat multiple solid cancers (Itatani et al., 2018). Vatalanib blocks kinase VEGFR activity and has completed phase II clinical trials with promising results to treat progressive meningioma [NCT00348790] and generalized advanced stage cancers [NCT00171587]. Furthermore, an in vivo study of CSC skin papillomas highlighted how anti-VEGFR2 antibodies in combination with NRP1 deletion prevented tumor metastasis and decreased tumor size through impairing CSC stemness properties (Beck et al., 2011). VEGFR inhibitors show promising effects in both clinical and non-clinical studies.

Under the third category of VEGF decoy receptors, Aflibercept (Zaltrap®) is a soluble recombinant fusion protein that acts as a decoy receptor by binding to VEGF-A ligand with high affinity, preventing VEGFR1/2 activation (Ciombor and Berlin, 2014). Aflibercept has been clinically approved to treat metastatic colorectal cancer in combination with traditional therapeutics (Ciombor and Berlin, 2014), and is currently undergoing phase II clinical trials to treat esophageal and gastric cancers [NCT01747551]. Although effective in the short-term, anti-VEGF therapy is not stable, thus new approaches involving VEGF splicing (Lee et al., 2018) and co-adjunctive treatment with anti-vascular mimics have been explored (Liang et al., 2021b).

Under the fourth category, Angiozyme, a synthetic ribozyme-based VEGFR1 inhibitor cleaves site-specific RNA molecules, which downregulates Flt-1 and KDR, paving the way for a new class of VEGF therapeutics (Weng and Usman, 2001; Weng et al., 2005). A phase I dose escalation study of Angiozyme on patients with refractory tumors showed promising results (Weng et al., 2005), and a phase II clinical trial on metastatic kidney cancer patients was also well tolerated [NCT00021021]. In in vitro studies, anti-VEGF₁₆₅ ribozyme showed 90.7% efficacy in anti-angiogenesis gene therapy for the treatment of tumors (Gu et al., 2004). More research should be conducted to better understand the promising effects of this novel class of anti-VEGF ribozyme therapeutics.

Both the classical VEGF pathway and VEGF-NRP pathway promote the progression of CSCs within multiple cancer subtypes. Targeting VEGF signaling in combination with chemotherapeutic drugs that target non-CSC pathways can potentially overcome therapy resistance. While VEGF pathway inhibitors are typically well tolerated via multiple routes of administration, common toxicities include fatigue, hypertension, proteinuria, hypothyroidism, and deficient wound healing (Kieran et al., 2012). Further studies are ongoing to address the more serious adverse events that can occur as a result of anti-VEGF related therapies [NCT05108519, NCT04788381], namely thrombosis, encephalopathy, deficient growth plate formation, wound healing deficiencies, and death (Kieran et al., 2012). Current research targeting the VEGF pathway is focused on ligand inhibitors, receptor inhibitors, VEGF decoy receptors, and ribozymes targeting VEGF, of which select inhibitors are marketed to treat CRC, RCC, HCC, solid tumors, and gastrointestinal tumors. Clinical trials also highlight how VEGF combination therapy (i.e., anti-VEGF drug in combination with traditional chemotherapy) is more effective than standard therapy. For example, the most studied FDA approved anti-VEGF inhibitor, bevacizumab, is currently in phase III trials in combination with capecitabine to treat metastatic breast cancer and showed increased response rate from 9.1% to 9.8% (p = 0.001) (Miller et al., 2005). Similarly, another phase III study on bevacizumab and paclitaxel for the treatment of breast cancer provided further supporting evidence of improved objective response rates (Ho and Kuo, 2007). However, positive response rates to combination therapy may not necessarily transfer to overall survival rates of the patient. Moreover, although bevacizumab is widely used to treat CRC and RCC, as of 2010 bevacizumab’s breast cancer indication has been withdrawn (Sasich and Sukkari, 2012). The FDA withdrawing bevacizumab’s indication highlights the major hurdle of anti-angiogenic therapies at treating an array of heterogenic cancer subtypes. As stated above, perhaps the reason for bevacizumab’s inefficacy at treating breast cancer is the VEGF paradox; the paradox states that anti-VEGF drugs commonly target the classical pathway while simultaneously upregulating the VEGF-NRP pathway, which supports CSC maintenance. In accordance with this paradox, it may be more beneficial to treat early tumors with anti-angiogenic drugs that inhibit tumor vascularization, rather than to treat highly vascularized, late-stage tumors, but research on early-stage anti-VEGF treatments is currently limited. Clinical trials on the VEGF signaling pathway (Figure 6) should be expanded upon to encompass the unique signaling mechanisms involved in CSC function.

Under normal conditions the TGF-β signaling pathway plays a vital role in cell differentiation, homeostasis, and organism development (Yang et al., 2020). However, when downstream signaling proteins are mutated, TGF-β suppresses the development of immune cells and allows for the metastasis and growth of prostate, lung, and breast cancer CSCs (Bellomo et al., 2016; Futakuchi et al., 2019). Upon ligand binding, a type II TGF-β ligand bound receptor will phosphorylate a type I receptor, and the TβRII–TβRI receptor complex will ultimately phosphorylate receptor-regulated Smads (R-Smads) to transcribe targeted genes in the nucleus (Yang et al., 2020). There are three categories of Smads: 1) receptor-regulated Smads are either activated by activin, a beta subunit dimer of the TGF-β protein superfamily, termed activin TGF-β (i.e., Smads 2/3) or are activated by BMP (i.e. Smads1/5/8/9), 2) coSmad (i.e. Smad4), is common to multiple TGF-β signaling pathways, and 3) i-Smads (i.e. Smad6/7) will inhibit the signal transduction of the TGF-β family (Yang et al., 2020). While the canonical TGF-β signaling pathway utilizes Smad2/3 to regulate gene transcription, the non-canonical TGF-β signaling pathways involves Smad 1/5 and can lead to the downstream activation of MAPK/ERK pathway proteins and transcription factors, which are involved in fibrosis and tumorigenesis (Clayton et al., 2020; Finnson et al., 2020; Guo et al., 2020). The non-canonical pathway also leads to the activation of the PI3K/Akt pathway, which is one of the most frequently over-activated downstream intracellular pathways that is involved in several human cancers (Rascio et al., 2021). Other molecules activated by the non-canonical TGF-β signaling pathways include NF-κB, Rho/Rac1, Cdc42, focal adhesion kinase (FAK), Src, and Abl (Neuzillet et al., 2013). The TβRII–TβRI receptor complex will also activate other non-Smad proteins signals, such as ubiquitin ligases, or small GTPases, resulting in a diverse molecular cascade (Bellomo et al., 2016). The TGFβ pathway is a potential avenue for cancer therapeutics due to its vital role in homoeostasis, induction of inflammatory cytokines in the ECM and regulation of T and B lymphocytes (Bellomo et al., 2016). Non-cellular components are mainly defined by the ECM, which make up the bulk of the stem cell niche (Pattabiraman and Weinberg, 2014). In this way, the TGF-β pathway monitors CSCs, and aberrant TGF-β signaling results in improper function of the immune response.

It is important to highlight TGFβ’s dual function—at times inhibiting and at other times promoting CSCs progression. Interestingly, in the early stages of cancer, TGF-β suppresses tumors, but in the later stages, TGF-β promotes tumor growth and survival (Futakuchi et al., 2019). On one hand, in the bone microenvironment, TGF-β leads to liver, gastric, melanoma, prostate, renal, glioblastoma, leukemia and bladder cancer proliferation via the downstream activation of the MAPK/ERK and PI3K/Akt, and induces breast cancer metastasis via activation of the Wnt pathway (Bellomo et al., 2016; Futakuchi et al., 2019). TGFβ features an autocrine loop via its downstream targets Sox4, Oct4, and Sox2, and disruption of the TGFβ pathway would hinder the progression of glioma CSCs (Ikushima et al., 2009). TGFβ is also implicated in cancer progression by inducing EMT and attenuating the anti-tumorigenic effects of dendritic cells, natural killer (NK) cells, CD8⁺ T cells, and CD4⁺ T cells, which can be further subdivided into pro-inflammatory TH1 and anti-inflammatory TH2 cells (Akhurst and Hata, 2012; Kim et al., 2021a; Kim et al., 2021b). On the other hand, TGFβ also inhibits ALDH1 on CSCs, thereby limiting their self-renewal capacity and halting tumor progression (Bellomo et al., 2016). Along these lines, TGFβ may act as a double-edged sword, wielding both pro- and anti-inflammatory effects by driving TH17 via secretion of IL-6 (Akhurst and Hata, 2012).

Moreover, Nodal and Activin are TGFβ ligands can both signal through the same receptors and in many cases the effects are indistinguishable from each other in that both Activin and Nodal and share the downstream effectors Smad2/3 (Pauklin and Vallier, 2015). Under normal conditions, Nodal ligand plays a vital role in embryogenesis, namely inducing epiblast implantation and guiding left-right axis neurulation (Pauklin and Vallier, 2015). Consistent with Nodal’s role in embryogenesis, Activin/Nodal signaling maintains pluripotency in stem cells through induction of Nanog and other cell cycle factors (Pauklin and Vallier, 2015). Cripto-1 is similar to Nodal in that it is an embryonic protein involved in TGF-β signaling (Arboretto et al., 2021). Both Cripto-1 and Nodal have minimal expression in terminally differentiated cells, but are often re-expressed in CSCs and in a dose-dependent manner can trigger more aggressive phenotypes and worsened prognosis for several cancer-subtypes (Arboretto et al., 2021).

Due to the complexity of the TGFβ’s highly context-dependent downstream mechanisms, developing therapeutics should first identify the stage of cancer progression optimal for a personalized, patient-specific treatment plan, which balances TGFβ’s pro-inflammatory and anti-inflammatory effects.

The following genes and markers are known to upregulate the stemness properties of CSCs via TGF-β/Smad signaling: 1) TGF-β1 in low concentration over short time periods upregulate breast CSCs (Hariyanto et al., 2021), 2) cancer upregulated gene (CUG) 2 promoters associated with YAP1 increase EMT in lung CSCs (Kaowinn et al., 2019), 3) cyclin D1 activation of Smad2/3/4 increases liver CSCs self-renewal and stemness (Xia et al., 2017), 4) CD51 increased migratory and invasive potential in colorectal CSCs (Wang et al., 2017), 5) CD133 acts via a Smad-dependent transcriptional mechanism in HCC, melanoma (Korn et al., 2021), and ovarian CSCs (Ikram et al., 2021), and 6) CD44/CD44v integrates with TME signals to upregulate HCC CSCs (Yan et al., 2015; Bellomo et al., 2016; Yang et al., 2020). Targeting these known upstream molecules, will result in inhibition of the TGF-β/Smad signaling pathway and could serve as potential therapeutics. Within the bone microenvironment, TGF-β is also known to induce the parathyroid thyroid hormone-related peptide (PTHrP) downstream, which stimulates osteoclast activity to destroy bone and promote tumor growth and metastasis (Futakuchi et al., 2019). Currently, the following molecules are known to inhibit TGF-β/Smad signaling: 1) miR-106 targeting Smad7 inhibits gastric CSCs and is implicated as a potential biomarker in CRC, and 2) Dkk-3 inhibits matrix metallopeptidases 9/13, which are downstream TGF-β-induced enzymes, implicated in preventing prostate CSC metastasis (Peng et al., 2020; Yang et al., 2020). Further investigation will uncover the distinct crosstalk between TGF-β and the TME, leading to breakthrough immunotherapeutics.

TGF-β is a valuable target in oncology that should be regulated accordingly. Current drugs on the market that inhibit TGFβ signaling include angiotensin type II receptor inhibitors (i.e. losartan and candesartan) (Akhurst and Hata, 2012). Drugs targeting the TGF-β pathway under clinical trials include anti-ligand antisense oligonucleotides (ASOs), competitive ligand/receptors, antibodies, kinase inhibitors, and small molecule inhibitors (SMIs) (Akhurst and Hata, 2012).

TGFβ antibody therapy has proven to be very effective; since TGFβ can induce an inflammatory TME thereby lowering the efficacy of cancer and radiotherapy (RT) treatments, inhibiting the TGFβ pathway is likely to directly enhance the efficacy of RT (Chen et al., 2021a). Indeed, TGFβ receptor 2 (TGFβR2)-neutralizing antibody MT1 and the small molecule TGFβR1 inhibitor LY3200882 in combination with RT results in increased anti-tumor efficacy against murine orthotopic models in HNSCC (Hamon et al., 2022). The TME is composed of an array of different immune cells and proteins that support the development of cancer cells, including CSCs (Arneth, 2019). The mechanism through which TGFβ interacts with the TME is closely connected to CSCs, in part providing the inflammatory microenvironment, which plays a critical role in tumorigenesis, tumor progression, and metastasis (Zhang et al., 2018b).

In preclinical studies, TGF-β can be directly inhibited. For example, through receptor kinase inhibitors, TGF-β signaling plays a vital role in the bone microenvironment, which harbors significantly more SOX2, CD44, and CD166 positive CSCs than those in the subcutaneous (subQ) microenvironment (Futakuchi et al., 2019). Treatment with TGF-β R1 kinase inhibitor (R1-Ki) in mouse mammary tumor cells reduced tumor volume in the bone microenvironment via TGF-β activation of ERK 1/2 and Akt pathways, but BMP signaling did not contribute to tumor growth (Futakuchi et al., 2019). Similarly, another R1-Ki, SD-208, suppressed the development of melanoma bone metastasis by blocking TGF-β induction of Smad3 phosphorylation (Mohammed et al., 2016). Other studies, which not only block the TGF-β Nodal/Activin Alk4/7 receptor but also simultaneously inhibit the Hh pathway, abolish CSC’s self-renewal capacity and render CSCs more susceptible to gemcitabine in vivo (Lonardo et al., 2011). Monoclonal antibodies that interfere with TGF-β proteins Cripto-1 or Nodal in vitro have also shown promising effects at abrogating CSCs (Arboretto et al., 2021). Moreover, SB431542, a SMI, partially reversed Nodal-induced chemoresistance in melanoma CSCs, in vitro (Li et al., 2018b). Therefore, the majority of TGF-β cancer therapeutics work to downregulate rather than upregulate TGF-β.

Rather than inhibiting TGF-β, treatment with TGF-β resulted in a significant decrease in proliferative CSC cytokines (i.e., cytokeratin-14, frizzled-7) and saw an increase in markers for slow proliferation (i.e., mucin-1 and cytokeratin-18) (Bellomo et al., 2016). Similarly, in vivo TGFβ’s activation of Smad2/3 inhibited ABCG2, a chemotherapy efflux transporter, leading to reduction of gastric CSCs and overall reduction in tumor size (Ehata et al., 2011). Therefore, due to the dual role of TGF-β in the immune response, at times benefiting and at times hampering the growth of CSCs, we must turn to personalized medicine.

In the clinic, ample consideration of the patient’s genetics and specific biomarkers may guide the development of effective TGF-β treatments. While many TGF-β treatments aim to discover patient-specific anti-TGF-β therapy, there is growing interest in assessing patients based on their unique TGF-β tumor response and relevant oncogenic pathways in cancer cells (Chen et al., 2021c). Such strategy requires probing germline genetic variation between individuals, which should further guide the identification of patients likely to respond to anti-TGF-β therapy (Chen et al., 2021c). Besides looking into patient genetics, TGF-β biomarkers include circulating levels of TGFβ, levels of P-SMAD2 levels in the peripheral blood mononuclear cells (PMNCs), and USP15 in glioblastoma breast and ovarian cancer (Akhurst and Hata, 2012; Eichhorn et al., 2012). Biomarkers serve as a non-invasive way of providing targeted therapeutic treatment. Moving forward, the goal of TGFβ drug development should be to enhance the tumor-suppressing effects of TGFβ while inhibiting TGFβ’s role in CSC progression.

The development of cancer therapeutics that target the TGF-β pathway have demonstrated satisfactory safety and efficacy in cancer patients. However, TGF-β’s dual role in cancer progression should be further explored to develop superior patient-specific anti-TGF-β treatments. Moreover, since TGF-β is ubiquitously expressed in multiple cell types to regulate cell homeostasis and organism development, there is an inherent physiological barrier to treatment. Current anti-TGF-β treatments including ASOs, competitive ligand/receptors, antibodies, kinase inhibitors, and SMIs are limited to pre-clinical and phase I-II clinical trials. Due to the limited dual role of TGF-β functioning as both a tumor promoter and tumor suppressor, the development of safe and effective TGF-β antagonists remains a challenge. For example, in a phase II study the SMI, galunisertib, in combination with the chemotherapeutic lomustine, failed to efficiently treat glioma compared to the placebo [NCT01582269], and galunisertib must be closely monitored to avoid cardiac toxicity (Kovacs et al., 2015). Furthermore, selective TGF-β biomarkers are only in the initial stages of clinical development. While TGF-β treatment alone is limited, combination therapy has proven to be more effective at targeting not only non-CSCs but also ablating CSCs. For example, Vactosertib, a TGF-β kinase inhibitor, in combination with chemotherapy has proven to be well tolerated in phase I-II clinical trials to treat solid tumors [NCT02160106], metastatic colorectal or gastric cancers [NCT03724851, NCT03698825], non-small cell lung cancer [NCT03732274], and urothelial cancer [NCT04064190]. In conclusion, drugs that target the TGF-β pathway are highly valuable to the cancer research field, as they boast improved patient response when compared to classical cancer treatments. While there are still many hurdles that TGF-β therapies must overcome before reaching the market, expanding knowledge of targeted CSC treatments are enhanced by continued TGF-β research (Figure 7).