Biomarkers and Targeted Therapy for Cancer Stem Cells

Abstract

Cancer stem cells (CSCs) are a small subpopulation of cancer cells with capabilities of self-renewal, differentiation, and tumorigenicity, and play a critical role in driving tumor heterogeneity that evolves insensitivity to therapeutics. For these reasons, extensive effort has been made to identify and target CSCs to potentially improve the antitumor efficacy of therapeutics. While progress has been made to uncover certain CSC-associated biomarkers, the identification of CSC-specific markers, especially the targetable ones, remains a significant challenge. Here we provide an overview for the unique signaling and metabolic pathways of CSCs, summarize existing CSC biomarkers and CSC-targeted therapies, and discuss strategies to further differentiate CSCs from non-stem cancer cells and healthy cells for the development of enhanced CSC-targeted therapies.

Cancer stem cells (CSCs) and CSC-targeted cancer therapy

Cancer is still a leading cause of human death, with over 10 million lives taken away by cancer annually in the world. While a variety of cancer therapies have been developed over the past decades, chemotherapy and radiation therapy remain the mainstream treatment for nearly all types of cancers [1, 2]. However, the therapeutic efficacy of chemotherapy and radiation therapy is still limited in general and is often accompanied by severe side effects, largely due to the lack of effective cancer-targeted therapies and the resistance of tumors to drugs and immune attack [3, 4]. Among the various factors contributing to the intractability of tumors, the presence of cancer stem cells (CSCs), a small subpopulation of cancer cells with capabilities of self-renewal, differentiation, and tumorigenicity (Fig. 1a–b), has been a notorious one [5–8]. Historically, CSCs were first identified from the blood as a subpopulation of cancer cells that possess infinite proliferative ability and were initially named as leukemia stem cells [6]. Further research uncovered the presence of CSCs in various types of solid tumors, including breast cancer, brain cancer, liver cancer, head and neck cancer, colorectal cancer, melanoma, and others [9–14]. These CSCs are characterized by their stem cell-like differentiation potential [7], often take up less than 1% of tumor cells [8], and are able to maintain their high proliferative ability in the tumor microenvironment [15, 16]. Indeed, a small amount of CSCs could often lead to the growth of a significant tumor mass [16, 17], indicating the immense aggressiveness and toughness of CSCs [15, 16].

Figure 1. Characteristics of cancer stem cells (CSCs) and their relation to cancer.

(a) Ability of CSCs to self-renew and differentiate into non-stem cancer cells. CSCs can self-renew to new CSCs while maintaining the stem likeness. They can also differentiate into non-CSCs and further proliferate into tumor. (b) CSCs with a small number can re-form the tumor mass. The eradication of CSCs can effectively reduce the regression of tumors while only removing non-CSCs will result in tumor relapse and the enrichment of CSCs. (c) CSCs outperforms non-stem cancer cells in the formation of metastatic tumors in vivo. CSCs that migrate into remote tissues (e.g., lungs) via the blood stream more likely result in the formation of a new tumor, in comparison with non-stem cancer cells. (d) CSCs contribute to the resistance of tumors to conventional chemotherapy and radiation therapy. CSCs tend to be more resistant to conventional chemotherapy and radiation therapy due to their stem-like property and can be enriched during these therapies, resulting in the relapse of tumors with drug resistance.

While CSCs possess infinite proliferation and differentiation potential, they are often quiescent in the core region of tumor masses and are more resistant to treatments such as chemotherapy and radiation therapy than the non-CSC population [17–20]. The surviving CSCs, even at a small number, can rebuild a highly aggressive tumor resistant to existing therapeutics within days to months (Fig. 1c–d) [18–20]. Indeed, the failure to destroy CSCs has been one of the key factors responsible for the overall low efficacy of numerous therapeutics. CSCs also play a critical role in driving tumor heterogeneity and develop insensitivity to therapeutics, dramatically increasing the chance of tumor recurrence and metastases [19–21]. For these reasons, the identification and targeted treatment of CSCs, which albeit remains challenging, is critical for achieving a better therapeutic outcome. The development of CSC-targeted therapies inevitably lies in the ability to identify CSC-specific surface receptors or biomarkers [5, 7, 8], and necessitates a comprehensive understanding of the molecular and metabolic differences between CSCs and non-stem cancer cells or normal cells.

In this Review, we will provide an overview of the altered signaling and metabolic pathways of CSCs that enable the identification of CSC-associated biomarkers, summarize the overexpressed molecules in CSCs that could serve as potential CSC biomarkers, and discuss CSC-targeted therapies including chemotherapy and immunotherapy (see glossary). As conventional endogenous markers are hardly specific to CSCs, efforts have also been made to introduce artificial biomarkers to the surface of CSCs for subsequent development of CSC-targeted therapies, which will also be discussed in this Review. We will also provide insights on future directions for the identification or metabolic tagging of CSCs, and further development of CSC-targeted cancer therapies with potent antitumor efficacy and minimal off-target effects.

Altered signaling and metabolic pathways in CSCs

Accompanied with the unique stemness of CSCs, various intracellular pathways of CSCs, including the Wnt pathway, Notch pathway, Hedgehog pathway, PI3K/Akt pathway, NF-κB pathway, and JAK/STAT pathway, are altered in comparison with non-stem cancer cells (Fig. 2) [22–25]. These alterations in genetic and epigenetic regulations could be essential for the initiation and development of CSCs, and confer the infinite proliferative ability and differentiation potential to CSCs [7, 8, 15, 16]. They were also shown to play a critical role in shaping the drug resistance and metabolic activities of CSCs [17–21]. As a special type of stem cell-like population residing within the tumor mass, CSCs also manage to adapt their metabolic programs for the tumor microenvironment in comparison with conventional stem cells [26–28]. For example, different from conventional stem cells that are biased towards the aerobic glycolysis instead of oxidative phosphorylation (OXPHOS) to minimize the production of reactive oxygen species (ROS) which could impair the state of stemness, CSCs heavily rely on glucose consumption for energy production to sustain the growth of tumors while maintaining their stemness (Fig. 3a) [27–29]. CSCs also undergo active mitochondrial oxidative metabolisms to match the upregulated oxygen consumption within the tumor microenvironment [29, 30]. Compared to non-stem cancer cells, CSCs also exhibit altered lipid metabolic activities (Fig. 3b). In particular, the lipid desaturation pathways were reported to be upregulated in CSCs to maintain their stem cell-like properties [31, 32]. Iron metabolisms and ferroptosis in CSCs are also upregulated to sustain the intracellular levels of iron and transferrin [33]. Also, the ceramide signaling, which is important for regulating the immune status of the tumor microenvironment and the programmed cell death, is upregulated in CSCs to facilitate the immune evasion and progression of tumors [34]. These aberrantly altered intracellular pathways of CSCs, in comparison with non-stem cancer cells and normal cells, provide molecular candidates for CSC biomarkers. However, when narrowing down to specific molecules, the differences between CSCs and other subpopulations are often non-significant, posing difficulty in identifying unique CSC markers for the development of CSC-targeted therapies.

Figure 2. Representative upregulated signal pathways that contribute to the stemness, survival, self-renewal, and invasion of CSCs.

(1) Wnt/β-catenin pathway: Wnt protein binds to its receptors Frizzled and LRP to induce the activation of Dishevelled, which leads to the activation of β-catenin that regulates gene expression for actin cytoskeleton remodeling, pluripotency regulation and cell differentiation. (2) Notch pathway: the binding of the jagged delta-like ligand to the Notch receptor on the cell surface induces the cleavage of the two proteolytic sites of the receptor to release Notch-IC (intracellular domain). Notch-IC can then mediate the gene expression related to cell survival, self-renewal, metastasis, and apoptosis. (3) Hedgehog pathway: the binding between ligand (red circle) and transmembrane protein PTCH results in the inhibition of the transmembrane protein SMO, resulting in the activation of transcription factors and downstream gene expression. (4) PI3K/Akt pathway: the binding of a ligand (e.g. growth factor) to RTK can induce the activation of PI3K and thus the phosphorylation of Akt. The activated Akt can then regulate the expression of genes related to the survival, proliferation, and invasion of CSCs. (5) JAK/STAT pathway: The binding of a ligand (blue) to cytokine receptors can induce the phosphorylation of STAT3, for subsequent regulation of genes related to stemness, survival, and invasion of CSCs.

Figure 3. CSCs exhibit aberrant sugar and lipid metabolisms.

(a) Carbohydrate metabolisms of CSCs and non-stem cancer cells. Non-stem cancer cells largely rely on glycolysis for ATP production even in the presence of sufficient oxygen, which refers to the Warburg effect. In addition to glycolysis, CSCs also leverage OXPHOS, with the help of HIF1α, for energy production. The tendency towards glycolysis or OXPHOS energy production pathways in CSCs is partially regulated by the PDC/PDK axis. (b) CSCs show upregulated lipid desaturation, lipogenesis, lipid storage, fatty acid oxidation (FAO) pathways in comparison with non-stem cancer cells. For lipogenesis, multiple genes encoding transporters and enzymes, including SREBP1, ACLY, ACC1 and FASN, are upregulated. Lipid droplets (LDs) are upregulated to facilitate the lipid storage. SCD1 is overexpressed to facilitate the lipid desaturation. Transcription factor NANOG is upregulated to enhance the FAO. LDH: lactate dehydrogenase; OXPHOS: oxidative phosphorylation; HIF1α: hypoxia Inducible Factor 1 Subunit Alpha; PDC: pyruvate dehydrogenase complex; PDK: pyruvate dehydrogenase kinases; SREBP1: sterol regulatory element-binding protein 1; ACLY: ATP citrate lyase; ACC1: acetyl coenzyme A carboxylase 1; FASN: fatty acid synthase; SCD1: stearoyl-CoA Desaturase1.

Conventional CSC biomarkers and CSC-targeting strategies

The potential CSC biomarkers identified based on their aberrant signaling and metabolic pathways can be divided into two major categories: cell surface markers and intracellular markers (Fig. 4). Cell-surface markers, most of which are transporters and signaling receptors, are often a more attractive target, as they can be directly utilized for targeted delivery of diagnostic and therapeutic agents to CSCs. For example, ABCB5, a cell-surface transporter under the ATP-binding cassette sub-family, is a commonly used marker for melanoma CSCs [35]. ABCG2, another ATP-binding cassette sub-family transporter, has been identified as a biomarker for lung cancer, pancreatic cancer, liver cancer, breast cancer, and ovarian cancer CSCs [36]. Some surface markers including ABCG2 are shared by different types of CSCs, while others are more specific to certain types of CSCs. For example, CD133, a type of cell-surface glycoprotein that can bind to cholesterol, is conserved across breast cancer, liver cancer, lung cancer, and ovarian cancer CSCs [37]. In contrast, CD34 is a reliable marker for identifying and targeting leukemia CSCs instead of solid tumor CSCs [8, 35]. Cell-surface markers commonly used for characterizing conventional stem cells have also been utilized to identify CSCs. One notable example is CD44, a cell surface adhesion receptor shared by nearly all types of stem cells or adult cells with some levels of stemness. Indeed, CD44 is widely used as a biomarker for breast cancer, colon cancer, and glioblastoma CSCs [7, 8, 15, 16]. These cell-surface markers have enabled the development of antibody- or aptamer-based targeting technologies for the diagnosis and targeted treatment of CSCs in various preclinical studies. Therapeutic agents, including small molecules, proteins, nucleic acids, and nanoparticles, can be functionalized with the corresponding antibodies or other targeting ligands for the enhanced binding to CSCs [5, 7–9, 38]. One central challenge for leveraging the cell-surface markers of CSCs, though, is their poor specificity and low abundance. All of the identified surface markers are hardly specific to any type of CSCs, as they are often also expressed by non-stem cancer cells or healthy cells, albeit at a lower abundance.

Figure 4. Common CSC-associated biomarkers for targeting purposes.

CSC-associated biomarkers, either surface or intracellular markers, are identified from their aberrant signaling and metabolic pathways. Cell-surface markers include CD133, CD44, and ABC family markers. Among them, CD133 is involved in the Wnt and PI3K-Akt signaling pathways. CD44 is a receptor for hyaluronic acid and other ECM components and is often overexpressed by CSCs. ABC family markers are closely related to the drug transportation pathways and drug resistance of CSCs. Intracellular markers include ALDH1 and transcription factors such as OCT3/4, SOX2, and BMI-1. ALDH1, especially the isoform ALDH1A1, is over-expressed by CSCs to facilitate the oxidization of retinal and aliphatic aldehydes. OCT3/4, SOX2 and BMI-1 are transcription factors that maintain the stemness of CSCs.

While a large library of intracellular molecules might show varied concentrations between CSCs and other populations, intracellular enzymes that are overexpressed by CSCs have been the main type of molecules explored for CSC targeting purposes. Various CSC-overexpressing enzymes enable the design of prodrugs that can be activated in the presence of the specific enzyme and thus preferentially kill CSCs [39–41]. For example, aldehyde dehydrogenase 1 (ALDH1) is overexpressed by breast cancer, prostate cancer, colon cancer, lung cancer, and ovarian cancer CSCs, and has been utilized to design CSC-activatable prodrugs [42]. ALDH1 has also been actively explored as an intracellular marker for the detection and sorting of CSCs. In addition to leveraging the enzymatic activity, various types of enzymes involved in essential intracellular metabolic pathways of CSCs have been the target of inhibitor designs. Similarly, transcription factors that dictate the proliferation and differentiation processes of CSCs, including BMI-1 and c-MYC, have allowed for the design of inhibitors that induce the apoptosis of CSCs [43, 44]. Other CSC-associated transcription factors of interest include OCT3/4 and SOX2 that are essential to maintain the tumorigenicity and stemness of CSCs, respectively [45, 46]. One concern, though, is that intracellular transcription factors are not specific to CSCs as most signaling and metabolic pathways are shared by CSCs, non-stem cancer cells, and healthy cells.

CSC-targeted chemotherapy and immunotherapy

Cancer therapies, including chemotherapy and immunotherapy, that aim to eradicate CSCs have been actively explored. CSC-targeted chemotherapies either aim to disrupt or orchestrate the essential intracellular pathways of CSCs or utilize the surface receptors to deliver chemotherapeutic agents into CSCs, in order to effectively kill CSCs. As the immunosuppressive tumor microenvironment plays a critical role in maintaining the stem-like property of CSCs, a variety of immunotherapies that can reprogram the tumor microenvironment are also shown to effectively eliminate CSCs.

Chemotherapy remains the first-line treatment for the vast majority of cancer patients. Various types of chemotherapeutic drugs have been designed to preferentially kill CSCs, with a goal of improving the overall therapeutic benefit while reducing the off-target side effects. One common strategy is to design drugs to inhibit the intracellular proteins that play a critical role in the proliferation, differentiation, or metabolisms of CSCs. For example, the inhibitors of ALDH1 were able to suppress the expression of stemness-associated factors and induce the apoptosis of CSCs [42]. Several kinase inhibitors were developed to disrupt the essential intracellular pathways of CSCs such as the Wnt/β-catenin signaling pathway [40, 41, 47–49]. Napabucasin, an STAT3 inhibitor, was able to induce the effective killing of lung cancer CSCs, reduce the metastasis, and prolong the survival of patients in clinic trials [49]. GDC-0449, an inhibitor of the hedgehog pathway designed to target glioblastoma CSCs, is also being evaluated for treating glioblastoma in a clinical trial [50]. In addition to the intracellular pathways, cell-surface receptors such as CD133 and CD44 have also been utilized to deliver chemotherapeutic drugs into CSCs. For example, the CD133 antibody-toxin conjugate (see glossary) was developed for the treatment of ovarian cancer in a mouse model [38]. Gemcitabine-incorporating nanoparticles functionalized with CD44 antibody was able to target CD44-positive breast cancer CSCs and improve the overall therapeutic efficacy [51].

CSCs play a critical role in shaping the immunosuppressive tumor microenvironment, especially for poorly immunogenic solid tumors such as breast cancer, glioblastoma, lung cancer, liver cancer, and pancreatic cancer [52–54]. CSCs can suppress the activity of effector T cells, antigen presenting cells, and natural killer cells by expressing high levels of inhibitory receptors (e.g., PD-L1), secreting non-inflammatory cytokines including IL-4, IL-10, IL-13, and TGF-β, expressing a low level of major histocompatibility complex I (MHC I), and preventing the tumor infiltration of immune cells [55, 56]. Existing immunotherapies such as immune checkpoint blockades, chimeric antigen receptor (CAR) T cell therapy, and therapeutic cancer vaccines were also able to reprogram CSCs, as a part of the mechanism to reprogram the immunosuppressive tumor microenvironment (Fig. 5). For example, natural killer cell therapy was shown to effectively kill glioblastoma CSCs [57, 58]. CSC-based therapeutic cancer vaccines were also able to induce potent CSC-specific cytotoxic T lymphocyte (CTL) response and robust antitumor efficacy against murine D5 melanoma and SCC7 squamous cell tumors [59]. Silencing the Nanog gene of tumor cells, a characteristic gene of CSCs, further improved the CSC-specific CTL response and therapeutic efficacy [60]. In addition, similar to chemotherapeutics and radioisotopes, the surface receptors of CSCs have also been utilized for targeted delivery of immunomodulatory agents [52, 53].

Figure 5. CSC-targeted immunotherapy.

(1) CSCs overexpress PD-L1, as part of the mechanism to suppress T cells. Immune checkpoint blockades such as anti-PD-1 and anti-PD-L1 can reactivate effector T cells in the tumor microenvironment for improved killing of CSCs. (2) Chimeric antigen receptor (CAR) T cell therapy that targets surface markers of CSCs such as CD133 and CD44 can be designed. (3) CSC-based cancer vaccines can also induce the presentation of CSC antigens by dendritic cells, for subsequent priming of CSC-specific effector CD8⁺ T cells. (4) As CSCs express high levels of ligands for NK activation receptors (NKG2D, DNAM1 and NKp30), NK cell therapy has also been explored for effective killing of CSCs.

Installation of artificial markers on CSCs

While endogenous surface receptors of CSCs are promising for the development of CSC-targeted therapies, none of the known surface markers are specific to CSCs. The difference in the expression levels of these biomarkers between CSCs, non-stem cancer cells, and healthy cells is typically small, limiting the achievable CSC-targeting efficiency [61–63]. In view of this, effort has been made to install artificial markers to the surface of CSCs for subsequent targeting applications. Metabolic glycan labeling of unnatural sugars provides a facile yet powerful strategy to introduce unique chemical tags (e.g., azide, alkyne, alkene, ketone, etc.) onto the membrane of cells [64, 65]. These unnatural sugars can be taken up by cells including CSCs, undergo the metabolic glycoengineering processes, become conjugated to glycoproteins and glycolipids, and are eventually expressed on the cell membrane [66–68]. Recently, a controlled metabolic glycan labeling strategy, i.e., selective labeling of cells of interest in a trigger responsive manner, has also been developed (Fig. 6a–b) [69, 70]. Only in the presence of a specific trigger, which could be ultraviolet light, hydrogen peroxide, and enzymes, the unnatural sugars can be reactivated and metabolized by the cells. One such azido-sugar, histone deacetylase (HDAC)/cathepsin L (CTSL)-responsive acetylated azidomannose (DCL-AAM), could preferentially label HDAC/CTSL-overexpressing cancer cells with azido groups. The cell-surface azido groups can serve as artificial receptors to mediate targeted conjugation of dibenzocyclooctyne (DBCO)-bearing therapeutics via efficient and bioorthogonal click chemistry between azide and DBCO [69]. Recently, an ALDH1A1-responsive sugar precursor named as AAMCHO was developed for selective metabolic labeling of ALDH1A1-overexpressing CSCs (Fig. 6a–b). It was demonstrated that AAMCHO can metabolically label colon cancer and lung cancer CSCs with azido groups in vitro and in vivo, enabling subsequent targeted conjugation of DBCO-chemotherapeutics to CSCs via efficient click chemistry. Compared to the conventional CD44 marker, AAMCHO-mediated metabolic tagging could better distinguish CSCs from non-stem cancer cells, and provides an unprecedented approach to targeting CSCs [70].

Figure 6. ALDH1A1-responsive AAMCHO can selectively label CSCs with azido groups.

(a) Schematic illustration of AAMCHO-mediated metabolic labeling of CSCs with azido groups and subsequent targeting of therapeutics via click chemistry. AAMCHO is designed to have a ALDH1A1 responsive moiety and a self-immolative linker. Only in the presence of ALDH1A1, AAMCHO can be reactivated and metabolized via the metabolic glycoengineering pathways. As CSCs overexpress ALDH1A1, AAMCHO can preferentially metabolically label CSCs with azido groups. The azido-labeled CSCs then enable targeted conjugation of therapeutics via efficient and bioorthogonal click chemistry. (b) Metabolic glycoengineering process of ALDH1A1-responsive AAMCHO. AAMCHO can be oxidated by ALDH1A1 to form AM-CHO, which is further converted to AM via the self-immolative cleavage of the linker and hydrolysis. AM can then participate in the metabolic glycoengineering pathways and become expressed on the cell membrane in the form of glycoproteins. Figure a and b are adapted from Ref. 70.

Concluding Remarks and Future Perspectives

In summary, the past decades has witnessed significant progress in the elucidation of aberrant intracellular signaling and metabolic pathways of CSCs and the identification of CSC biomarkers. These biomarkers include surface receptors, enzymes, transcription factors, and other molecules upregulated in CSCs [8, 71, 72]. Cell-surface markers of CSCs such as CD133 and CD44 enable the targeted delivery of therapeutics such as chemotherapeutics, radioisotopes, and immunomodulators [73, 74]. CSC-overexpressing enzymes can be utilized to design enzyme-activatable prodrugs that preferentially kill CSCs, and also allow for the design of enzyme inhibitors that disrupt the essential pathways and thus induce the apoptosis of CSCs [75, 76]. Other intracellular molecules involved in key intracellular pathways of CSCs, including transcription factors, have also inspired the design of inhibitors as CSC-targeted chemotherapeutics [77–79]. However, none of the biomarkers are exclusively specific to CSCs, with most of them showing slight difference between CSCs and non-stem cancer cells or normal cells. These issues motivate the identification of new endogenous CSC markers or the development of novel strategies to differentiate CSCs from other cell populations.

In addition to the endogenous biomarkers, strategies to introduce artificial biomarkers to CSCs have also been actively explored. Metabolic glycoengineering of unnatural sugars provides a facile tool to introduce artificial chemical receptors to cell membranes for subsequent targeting purposes. However, the metabolic glycoengineering processes occur in both cancer cells and healthy cells. The recently developed controlled labeling strategy, i.e., modification of the carbon-1 site of azido-mannoses with a trigger-cleavable protecting ether bond, enables the metabolic labeling of cells in a trigger-dependent manner [69, 70]. This trigger could be external (e.g., ultraviolet light) or internal (e.g., enzymes and hydrogen peroxide). One such azido-sugar, AAMCHO, can metabolically label cells in a ALDH1A1-activatable manner, and thus can preferentially label ALDH1A1-overexpressing CSCs with azido groups [70]. This approach converts the intracellular ALDH1A1 to the cell-surface clickable tags, providing a new avenue to developing CSC-targeting technologies. In principle, unnatural sugars that can reactivated by other CSC-overexpressing enzymes can also be developed for CSC labeling and targeting. One concern, though, is the relatively low labeling efficiency of AAMCHO, as the installation of an ALDH1A1-responsive moiety is at the sacrifice of the overall metabolic labeling efficiency. A fine balance between the labeling efficiency and selectivity needs to reach. This issue can be partially addressed by deploying strategies to improve the delivery of enzyme-activatable unnatural sugars to tumors, such as the use of nanoparticles [81, 82].

Moreover, metabolic lipid labeling could be an alternative approach to achieving chemical tagging of CSCs [83–85]. Considering the aberrant lipid metabolisms in CSCs, the rational design of unnatural lipids based on either the structural lipids (e.g., dipalmitoylphosphatidylcholine (DOPC)) or signaling lipids (e.g., ceramide and phosphatidylinositol lipid) could potentially achieve preferential tagging of CSCs over non-stem cancer cells or healthy cells [86]. The chemical tags (e.g., azido groups) then enable targeted delivery of therapeutics to CSCs. Compared to metabolic glycoengineering pathways that involve numerous reactions, lipid metabolic pathways involve less steps and potentially lead to a higher labeling efficiency [87]. Similar to metabolic glycan labeling, strategies that can improve the delivery of unnatural lipids to tumors and the uptake of unnatural lipids by CSCs could further increase the labeling efficiency and subsequent CSC targeting effects [88].

In addition, efforts to make the best use of existing CSC markers are also needed. While it remains challenging to differentiate between CSCs and non-stem cancer cells, the differences between these cancer cells and healthy cells are often more significant. This allows for the preferential delivery of therapeutics to CSCs and non-stem cancer cells or the activation of prodrugs in the tumor tissues. The synergistic attack at CSCs and non-stem cancer cells could potentially lead to favorable antitumor efficacy [89–91]. Indeed, simultaneous targeting of CSCs and non-stem cancer cells could be essential for amplifying the immunomodulation of the tumor microenvironment, tumor antigen spreading, and systemic antitumor T cell responses [92–94].

Outstanding questions Box.

How to identify new CSC-specific biomarkers, either cell-surface or intracellular ones?

How to design CSC-specific unnatural sugars that can selectively label CSCs with clickable chemical tags in vivo?

How to make the best use of existing CSC-associated markers for cancer treatment?

How to better integrated CSC-targeted cancer therapies with conventional chemotherapy, radiation therapy, and immunotherapy?

Highlights.

Cancer stem cells (CSCs) are a small subpopulation of cancer cells with capabilities of selfrenewal, differentiation, and tumorigenicity, and play a critical role in driving tumor heterogeneity drug resistance.

CSCs that survive chemotherapy or radiation therapy, even at a small number, can rebuild a highly aggressive tumor resistant to existing therapeutics within days to months.

The identification and targeted treatment of CSCs, which albeit remains challenging, is critical for achieving a better therapeutic outcome.

CSCs exhibit distinct signaling and metabolic pathways from non-stem cancer cells and healthy cells, enabling the identification of cell-surface and intracellular biomarkers.

Metabolic glycan labeling provides an emerging method to install artificial chemical receptors on the membrane of CSCs, for subsequent conjugation of therapeutics.

Glossary

Antibody-toxin conjugate

The conjugate of a toxin molecule with an antibody that can bind to a specific ligand

Click chemistry

A type of quantitative and rapid conjugation reaction between two functional chemical groups

Cancer vaccine

A type of immunotherapy that functions by delivering tumor antigens and adjuvants to antigen presenting cells such as dendritic cells

Immunotherapy

Therapy that boosts the body’s immune responses against diseased cells such as cancer cells