Mechanisms involved in cancer stem cell resistance in head and neck squamous cell carcinoma

INTRODUCTION

Head and neck malignancies are now the seventh most common type of cancer worldwide^[1]. More than 90% of head and neck tumors are derived from mucosa epithelium and are diagnosed as squamous cell carcinoma (HNSCC). Although sharing identical histological subtypes, HNSCC can be divided into at least two genetic subclasses based on the absence or participation of human papillomavirus (HPV) in carcinogenesis^[2]. The oral cavity represents the main subsite for HPV-negative tumors and the oropharynx for HPV-positive ones^[3]. Moreover, these subgroups also differ in clinical profile, tumor behavior, survival rates, and prognoses^[4].

The mainstay treatment for HNSCC consists of surgery with adjuvant or neoadjuvant chemotherapy and radiotherapy. More recently, immunotherapy with checkpoint inhibitors has been indicated for recurrent and metastatic HNSCC with promising results, although only a subset of patients with HNSCC has shown a response to this therapy^[5]. TNM stage of the disease and anatomic subsites influence therapeutic options for HNSCC. While radical surgeries are the first choice for locally advanced oral cancer, the main treatment for oropharyngeal tumors is chemoradiotherapy, regardless of HPV status. Nowadays, transoral surgeries (robotics and laser microsurgery) have also been performed in the oropharynx region^[6].

Despite the advances in current therapy, the prognosis of HNSCC remains poor. More than half of patients die from the disease or complications within a short period, varying from a few months to five years^[7]. The primary cause of mortality is related to resistance to therapy which leads to local recurrence, cervical lymph node metastasis, and occasionally, distant organ metastasis^[6]. Tumor heterogeneity and cancer stem cells (CSCs) are known to enhance metastatic dissemination and therapeutic resistance, contributing to lethality^[8].

CSCs represent a small but critical subpopulation of cells in the tumor capable of self-renewal and multilineage differentiation and regenerating a tumor when serially transplanted into mice models^[9]. Since tumors can regrow from a single CSC, cancer treatment success may be attributed to the complete eradication of CSCs populations^[8].

Besides, CSCs also demonstrate cellular plasticity; they can reversibly switch between different stem cell phenotypes and between a stem and non-stem cell state^[10]. CSCs activity is modulated by different signals and cellular interactions provided by the tumor microenvironment, allowing CSCs to achieve highly invasive and aggressive behavior or resist conventional therapies. Thus, activating the Epithelial-to-Mesenchymal Transition (EMT) program by the CSCs represents a valuable strategy to promote invasion, metastasis, and treatment resistance^[10-12].

CSCs may originate from adult stem cells or progenitor cells in which the accumulation of mutations over time leads to the activation of transcriptional gene signatures and signaling pathways related to the maintenance of stem cell phenotype and malignant transformation^[13-15]. Moreover, differentiated cells can also acquire stemness traits due to genetic instability throughout their division process and dedifferentiate, acquiring stem cell properties^[16,17]. It is essential to highlight that malignant cells can dedifferentiate and acquire stem cell characteristics under challenging situations, including exposure to chemotherapy and radiotherapy^[18].

In HNSCC, Prince et al. first described the presence of a small fraction of CD44-positive cells capable of generating new tumors when inoculated in immunocompromised mice and re-establishing original tumor heterogeneity^[19]. Moreover, this subpopulation expressed the Bmi1 gene, a stemness marker involved in tumorigenesis and self-renewal^[19]. Since then, other common HNSCC CSC markers, such as CD44, ALDH1, CD133, c-Met, and Bmi-1, have been described^[20-22]. ALDH1 is considered a highly specific CSC marker, mainly when evaluated with CD44^[20]. Moreover, based on CD44 and EpCAM expression levels, CSCs in oral squamous cell carcinoma (OSCC) seem to switch between two distinct phenotypes. First, CD44^high/EpCAM^high presents an epithelial morphology and colony formation capability, and second, CD44^high/EpCAM^low has a mesenchymal morphology (EMT profile) with high invasive potential, metastasis and radioresistance ability^[10,23]. More recently, LIN28A and LIN28B proteins, located in the cytoplasm and nucleus/nucleoli, respectively, were identified as reprogramming factors that can lead to the de-differentiation of malignant oral squamous cancer cells into CSCs and contribute to their immune evasion^[24].

For other types of cancers, distinct CSCs can be identified and isolated by fluorescence-activated cell sorting (FACS) using phenotypic surface markers alone or in combination. More than 40 surface markers are known to identify CSCs in solid tumors, and the majority are derived from embryonic or adult stem cells^[25]. In general, high positivity of CD44, CD24, CD133, CD90, EpCAM, and Aldehyde Dehydrogenase 1 (ALDH1), and elimination of Hoechst 3334 dye via ABC transporters are the most used markers^[26]. The isolated CSCs can be propagated in vitro as spheroids or used in organoid cultures. Moreover, spheroid cultures are CSCs enriched, show self-renewal ability in vitro and in vivo, and generate tumors that resemble the original tumor heterogeneity and differentiation^[27].

More recently, in addition to the conventional 2D cell culture, 3D culture models have been used to represent tumor microenvironment heterogeneities properly and reproduce patients’ tumor behavior. Engelmann L. et al. developed a 3D Organotypic Co-Culture (3D-OTCs) utilizing HNSCC fresh tissue (non-HPV driven and HPV-driven) placed on top of dermal equivalents (human fibroblasts cultured on a viscose fiber fabric) and analyzed samples’ behavior^[28]. All non-HPV-driven 3D-OTCs were capable of proliferating cancer cells for up to 21 days and exhibited a heterogeneous, invasive, and expansive growth pattern^[28]. In the same context, Miserocchi G. et al. developed a 3D culture using HPV-positive and HPV-negative HNSCC cells in a collagen-based scaffold. They suggested that the 3D model might induce more mesenchymal phenotypes than 2D cultures^[29]. Also, in this study, HPV-negative cells presented an upregulation of FLT1 and ABCA3 when seeded in scaffolds, overexpressed EMT-related genes, and increased migration ability compared to HPV-positive cells^[29]. Based on these findings, collagen-based scaffolds seem to activate drug-resistance mechanisms reassuring the ability of 3D scaffolds to reproduce HNSCC tumor microenvironment impeded by other in vitro systems. Accordingly, regarding response to treatment analyses, 3D culture is promising in the future of HNSCC and CSC research.

Several associations between clinicopathological characteristics and CSCs have been appointed in HNSCC, including tumor size, regional and distant metastases, perineural invasion, radiation failure, and poor disease-free survival^[30]. A previous study of our group explored CSCs markers in tongue tumors and found that the overexpression of CD44 was related to worst overall survival, and Nanog and Oct4 were associated with regional metastasis and death^[31]. Ma et al. suggested that CD133⁺ cells could be responsible for aggressiveness and chemoresistance in oral tumors^[32]. A meta-analysis study by Fan et al. showed that the CSCs markers, CD133, Nanog, and Oct4, could have a prognosis value in HNSCC patients^[33]. In light of recent events in CSCs markers, there is now some discovery about non-coding RNAs (ncRNAs) used as biomarkers of cancer development and tumor stage determination^[34].

MicroRNAs are a type of sncRNA that regulate biological processes. Each miRNA can control target genes and accentuate their potential influence on almost every genetic pathway. Hsieh PL et al. demonstrated that ncRNA molecules associated with CSCs are responsible for acquiring and maintaining cancer stemness^[35]. Let-7 genes family act as a tumor suppressor. Lin28B-let-7 pathway positively regulates the expression of stemness factors Oct4 and Sox2; it causes a switch of non-CSCs to CSCs with tumor starting and self-renewal characteristics in oral CSC^[36].

MicroRNA-200 family is another group of genes related to CSC; expression levels of miR-200c were downregulated in ALDH1+/CD44+ HNSCC with BMI1 overexpression. Also, an expression of let-7c or let-7d in oral CSCs suppressed stemness and the radio/chemoresistance hallmarks through suppression of IL-8 or EMT markers, respectively^[37,38]. MicroRNA-494 acts as a tumor suppressor or oncogenic factor. An increase of miR-494 can inhibit ALDH1 activity, CD133 positivity, and other stemness signatures in ALDH1+CD44+ oral cancer cells. In the same way, activation of miR-494 inactivates Bmi-1 and ADAM10 expression in OSCC-CSCs^[39]; also, miR-494-3p may enhance the radiosensitivity and induce a senescence pathway in oral cancer cells^[40].

In this scenario, it is essential to highlight that CSCs are not easily eliminated by conventional therapies, meaning that after the effective depletion of the bulk of the tumor, residual CSCs populations may survive, drive and sustain cancer recurrence, invasiveness, and therapy resistance^[41]. Moreover, CSCs are considered intrinsically resistant to chemo and radiotherapy. It is also possible that the CSCs and their close descendants give rise to therapeutic-resistant malignant cells that accumulated mutations caused by genotoxic therapies^[42]. CSCs adopt different strategies to overcome the challenges imposed by treatment, including the acquisition of dormancy, which is influenced by the CSC niche and immune surveillance, increased drug efflux capacity, activation of DNA repair machinery and decreased activation of apoptosis^[43]. This review will focus on the mechanisms that lead to CSC resistance to radiotherapy and chemotherapy in HNSCC.

RADIORESISTANCE AND CSC

In HNSCC patients, radiotherapy (RDT) is a common choice of treatment to achieve cancer control after surgery and/or current chemotherapy^[6]. Usually, on weekdays patients receive a dose of 70 Gy that can be administered through standard fractionation (2 Gy, once a day) or via accelerated fractionation and hyperfractionation (twice a day)^[44]. Fractionation guarantees that cancer cells will eventually be exposed to radiation in all cell cycle phases, favoring DNA damage and cell fate. Nevertheless, this process also activates important protein regulators of DNA damage response, such as ataxia-telangiectasia mutated (ATM) and ataxia-telangiectasia and Rad3-related protein (ATR), which will be decisive in treatment response^[45].

Tumor response or failure to ionizing radiation is mainly associated with the classical 4 R’s of radiobiology: repair of sublethal DNA damage, reassortment of cells in the cell cycle, cell repopulation, and reoxygenation of hypoxic areas^[46]. Efficient cell death by RDT depends on producing unrepairable damage involving DNA double-strand breaks (DSBs); however, most radiation-induced DNA damage is sublethal. DNA repair systems include base excision repair (BER), nucleotide excision repair (NER), homologous recombination (HR), non-homologous end joining (NHEJ), and mismatch repair (MMR) pathways^[47]. In this context, CSCs seem to hold elevated levels of proteins responsible for NHEJ and HR and an increased DSB repair capacity^[23].

If tumor recurrences occur within six months following radiation, tumors are considered radioresistant^[48]. Mechanisms involved in radioresistance are not fully understood, but accumulated evidence indicates that cancer stem cells (CSCs) are decisive in this process^[46,49]. In general, therapeutic resistance refers to the ability of cancer cells to recover and repair DNA damage and regrow after tumor therapy^[50], being higher in CSCs than non-CSC^[51]. This ability is mainly related to the increased regulation of DNA repair genes, DNA-damage checkpoints, and anti-apoptotic proteins^[52,53].

Furthermore, it has been recognized that a CSC subpopulation exhibiting a mesenchymal profile (CD44^high/CD24^low) presents an even higher level of DNA repair following RDT^[23]. Besides, irradiation activates stemness pathways and induces CSC phenotypes in non-stem cancer cells. Up-regulation of CSCs genes, such as Sox2 and Oct3/4, may be observed after radiation, contributing to tumor radioresistance^[53]. The plasticity of CSCs dramatically interferes with identifying and eliminating CSCs during cancer therapies^[54].

Radiation promotes an arrest of CSCs in the G2/M phase, which allows active DNA repair. Moreover, after radiation, there is a noticeable discrepancy between the higher rates of self-renewal and proliferative abilities of CSCs compared to their lower apoptosis activation, favoring tumor growth^[52]. In oral cancer cell lines, changes in CSCs content (ALDH+) are associated with an increase in the rates of sub-lethal damage repair (SLDR), which enables efficient cell repair and reduces tumor control capabilities^[55]. Duration of the exposure to the fractionated dose-delivery of radiation seems to influence radioresistance mechanisms driven by SLDR, suggesting that reduced overall dose-delivery time on radiotherapy could favor CSCs control^[55].

Besides the DNA repair process, activation of checkpoint responses after radiation damage also participates in the radioresistance of several tumors, including HNSCC. Cell cycle progression is delayed to allow DNA repair through activation of signaling pathways such as ataxia telangiectasia mutated (ATM)-checkpoint kinase 2 (Chk2) and ATM-Rad3-related (ATR)- checkpoint kinase (Chk1)^[56]. CSCs appear to enhance response to DNA damage activating Chk2 in invasive oral cancer^[23]. Inhibition of Chk1 was suggested as a therapeutic target in HNSCC that contributes to the failure of DNA replication and intensification of DNA damage^[57].

Induction of apoptosis represents one of the primary mechanisms by which cancer cells are eliminated in cancer therapies^[58]. Reduced cleaved caspase proteins showed the apoptotic resistance of CSCs in oral cancer after irradiation^[23]. Resistance mechanisms evolving upregulation of anti-apoptotic proteins such as Bcl-2 and inhibitor of apoptosis (IAP) are commonly found in tumor cells, especially in CSCs^[59]. Radiation can activate X-linked IAP (XIAP), another IAP family member that inhibits apoptosis mediated by mitochondrial and caspase-3 pathways^[60]. Besides apoptosis regulation, Bcl-2 family members also participate in cell migration, invasion, and metastasis^[61]. In this focus, an inhibitor of Bcl-2 combined with Cetuximab and radiation showed excellent results in eliminating CSCs in HNSCC cell lines^[62].

Another widely studied mechanism of CSCs contributing to radioresistance and poor prognosis in HNSCC is related to hypoxia, i.e., low oxygen levels caused by insufficient blood supply to tumor tissues^[63,64]. A hypoxic tumor environment can interfere directly with the potential of radiation to damage DNA cells and indirectly regulate the expression of genes related to aggressiveness and response to treatment. Additionally, hypoxia is essential in protecting the CSCs niche from radiation effects and in acquiring and maintaining CSC-like phenotype^[65].

In HNSCC, hypoxia-inducible factor-1a (HIF-1α), a transcriptional regulator of oxygen homeostasis, is enhanced in CSCs subpopulations in response to radiation^[66]. Furthermore, hypoxia upregulates CSCs genes such as Sox2 and Nanog, consequently contributing to the survival of tumor cells after radiation^[67]. Linge et al. showed a correlation between high tumor recurrence after postoperative radiochemotherapy in locally advanced HNSCC patients, increased expression of CSCs markers, and high hypoxia-induced gene signature expression^[68]. Strategies for hypoxic modifications such as hyperbaric oxygenation or nitroimidazoles significantly reduced locoregional recurrence after radiation in HNSCC^[69].

In the same context, reactive oxygen species (ROS) and redox-regulatory mechanisms can regulate DNA damage and resistance to irradiation. Accumulation of ROS and DNA damage of cancer cells is associated with the effectiveness of radiotherapy^[70]. Unlike non-CSCs, CSCs present a high antioxidant capacity that coordinates the activity of free-radical scavengers and protects cells from induced-radiation death^[70,71]. This low ROS state presented by CSCs is also related to the quiescent state of HNSCC stem cells and enhanced tumorigenic potentials in vitro and in vivo^[72]. Interestingly, GDF15 (growth differentiation factor 15), a member of the TGF-β superfamily, participates in ROS suppression in HNSCC, contributing to radioresistance and acquisition of the CSC phenotype^[73]. Boivin et al. showed that redox-modulating by inhibiting GSH antioxidant system previous to radiation is an accurate strategy to eliminate highly tumourigenic CSCs^[74].

Considering the better prognosis of HNSCC HPV-positive patients, it seems that HPV may influence several molecular mechanisms involved in CSC’s radiosensitivity^[75,76]. Rieckmann et al. demonstrated a limited capacity of DSB repair in HPV/p16-positive cancer cells^[77]. HPV-positive tumors are believed to present less radioresistant CSCs subpopulations due to their reduced repopulation ability during radiation therapy^[78]. Reid et al. explored irradiation behavioral responses of CSCs with CD44⁺ ALDH⁺ phenotype in 6 HPV positive and negative HNSCC cell lines^[79]. Their principal findings showed that HPV status did not influence the inherent proportions of CSCs, which were changed in both groups in response to radiation. HPV-negative samples showed a significant increase in CSCs densities, probably reflecting their remarkable repopulating ability after treatment^[79]. Other studies demonstrated that HPV-negative cell lines seem more capable of dedifferentiating from non-CSCs to CSCs in response to radiation than HPV-positive cell lines^[80]. In addition, low levels of functional TP53 expressed by HPV-positive cells may contribute to inducing apoptosis following radiotherapy^[81].

In an attempt to address this issue, the literature has found that cisplatin-sensitization has helped overcome resistance to radiation in many patients. In a recent study, Routila et al. appointed Oct4 as a good marker for identifying radioresistance and cisplatin-sensitive tumors, which could help distinguish patients who should receive cisplatin-sensitization from those who would not benefit from this therapy^[82]. In summary, Oct4 positivity reduced cancer cell apoptosis, favoring cell viability after irradiation. At the same time, Oct4 can contribute to cisplatin mechanisms inhibiting DNA repair activation^[82]. In radioresistance, Oct4 driving activates the oncogene Cancerous Inhibitor of Protein Phosphatase 2A (CIP2A), which promotes malignant cell growth and proliferation^[50].

Despite technological developments, RDT still promotes long-term toxicities compromising the quality of life and is often associated with potential tumor resistance^[6,83]. CSCs act as key players in regulating different mechanisms of DNA damage repair and other regulators of cell death after irradiation, such as hypoxia, apoptosis, and ROS [Figure 1]. At this point, we believe that RDT is insufficient to eliminate CSCs in HNSCC, explaining the high recurrence rates of these tumors. Thus, further investigation is required to comprehend and overcome CSC’s radioresistance and improve treatment success and overall survival in cancer patients.

Figure 1. Mechanisms related to CSCs radioresistance in HNSCC. Radiation can activate stemness pathways such as Sox-2 and Oct3/4 and induce CSC phenotype in non-stem cancer cells. Radiation promotes an arrest of CSCs in the G2/M phase and activates Chk2 and Chk1, which delays cell cycle progression and allows DNA repair. Overexpression of CSC marker ALDH1 leads to increased rates of sub-lethal damage repair (SLDR), enabling efficient cell repair and reducing tumor control capabilities. CSCs upregulate anti-apoptotic proteins such as Bcl-2 and X-linked inhibitors of apoptosis (XIAP). Hypoxia upregulates CSCs genes (Sox2 and Nanog) and is essential in protecting the CSCs niche from radiation effects. GDF15 (growth differentiation factor 15) participates in ROS suppression in HNSCC, contributing to radioresistance and acquisition of the CSC phenotype.

CHEMOTHERAPY RESISTANCE AND CSC

HNSCC in stage I or II (early tumors) is curable with higher survival rates after surgery or radiotherapy alone. In contrast, over 60% of stage III or IV HNSCC (locoregionally advanced) require advanced therapeutic options such as surgery followed by radiotherapy with or without chemotherapy^[7]. Currently, the standard chemotherapy regimens for stage III or IV, as well as recurrent and metastatic HNSCC, are based on cisplatin, 5-fluorouracil (5-FU), and docetaxel/paclitaxel^[84-86].

Chemotherapeutic drugs exert different biological effects on tumor cells, relying on specific mechanisms of action. Cisplatin is a platinum-based alkylating agent that creates inter- or intra-strand cross-links or transfers alkyl groups to the guanine residues of DNA, generating mispairing formation in DNA bases and avoiding strand separation during DNA synthesis^[87]. On the other hand, 5-FU is a pyrimidine antagonist antimetabolite that interferes with essential biosynthetic pathways, disturbs the DNA/RNA synthesis, or causes the formation of DNA strand breaks through inhibition of particular enzymes or incorporation of false structural analogs of pyrimidine/purine into DNA^[88]. Docetaxel is a topoisomerase II inhibitor that impairs DNA replication and causes DNA strand breaks. Paclitaxel is a taxane that modifies the function/formation of spindle microtubules by inhibition of nuclear division (mitotic arrest in metaphase), leading to cell death^[87]. In this context, it is essential to highlight that most chemotherapeutics’ success relies on the drugs’ ability to decrease tumor size or induce short-term remission. This measure of success is intuitive, and many medications evaluated by these criteria are used in effective chemotherapeutic regimens^[89].

Although the chemotherapeutic scenario seems broad, mortality from HNSCC continues to rise worldwide^[90]. As reviewed by Bukowski et al., part of this problem may be a reflection of drug resistance, which leads to a reduction of the therapeutic efficacy and is related to over 90% mortality of cancer patients^[91]. Multi-drug resistance (MDR) of cancer cells during chemotherapy can be associated with a variety of mechanisms, including enhanced efflux of drugs, drug activation or inactivation, genetic factors (gene mutations, amplifications, and epigenetic alterations), growth factors, increased DNA repair capacity, inactivation of apoptosis machinery, increased autophagy, and elevated metabolism of xenobiotics, or even any combination of these mechanisms^[91-93]. In addition, establishing a tumor microenvironment (TME) promotes tumor progression and chemoresistance through a collection of soluble proteins and insoluble vesicles secreted by tumor cells. This cell-to-cell communication among various cell types required to form the TME, such as mesenchymal stromal cells, immune cells, and vascular endothelial cells, influences the function of cells in the TME, shapes the premetastatic niche, and is an essential contributor to the development of chemoresistance^[94].

Tumor heterogeneity is a significant complicating factor in cancer treatment and is also strictly associated with chemotherapy resistance, impacting poor prognosis for HNSCC patients^[95]. Specifically, the presence of the CSCs has been associated with resistance to chemotherapeutic agents such as cisplatin, bortezomib, etoposide, 5-FU, and doxorubicin^[92,96]. Most importantly, many studies have demonstrated that treatment with these drugs enhances the CSCs fraction in different solid tumors and favors EMT traits, leading to treatment resistance and cancer progression^[97,98]. In addition, the acquisition of resistance to a specific drug generally tends to multiply resistance to unrelated compounds in CSCs and malignant cells, which under treatment pressure, can acquire a stem-like phenotype and become therapeutic resistant^[18].

CSCs were identified as crucial players in the acquisition of drug resistance and unresponsiveness to current chemotherapies against cancer by activating different cellular signaling pathways and mechanisms [Figure 2]. The main reasons found in the literature rely on intrinsic properties of CSCs, such as the (1) inherent quiescent state that enables them to evade the actions of drugs that target rapidly proliferating cells; (2) high levels of drug efflux pumps and detoxifying enzymes; (3) increased DNA self-repair capacity; (4) specific expression of anti-apoptotic and prosurvival proteins; (5) acquisition of the EMT-phenotype; (6) oxidative modulation; (7) epigenetic modifications and (6) activation of the specific signaling pathways^[90,99-103]. In addition, the role of the TME in sustaining the CSCs niche is also gaining substantial importance in promoting resistance to chemotherapy as an extrinsic factor^[101]. The TME shapes the morphology and functional features of CSCs, mainly influencing (1) cellular plasticity; (2) hypoxia; (3) metabolic reprogramming; (4) activation of specific signaling pathways; and (5) cell-to-cell interactions^[100].

Figure 2. Mechanisms related to CSCs’ chemoresistance in HNSCC. Overexpression of Bmi1, Sox2, Sox8, Oct4, Slug, Snail, Nanog, and TSPAN1 genes leads to the acquisition of drug resistance and stemness, EMT, and metastasis. CSCs activate signaling pathways such as the NOTCH1, FGF2, and Wnt/β-catenin to promote chemoresistance and stemness. Increased expression of ABC transporters, mainly ABCG2, the activation of EMT, cell cycle deregulation, increased autophagy, and activation of epigenetic mechanisms, such as up-regulation of miR-10, are involved with CSC’s chemoresistance in HNSCC.

Several in vitro studies found that stemness-related genes are overexpressed in HNSCC cell lines resistant mainly to Cisplatin, 5-FU, doxorubicin, and docetaxel. Sox2, Oct4, CD44, Bmi1, ALDH1, and Nanog were the genes most frequently associated with the CSC phenotype. Also, EMT markers (Slug, ZEB1, Twist, Snail), as well as drug efflux transporters (ABCG2, ABCC1/ABCC2/ABCC3/ABCC4/ABCC5, ABCB1), epigenetic alterations (HDAC1/HDAC2, SIRT1, KAT6A,/KAT6B), and specific signaling pathways such as Wnt/β-catenin and NOTCH1. These mechanisms endow CSCs to survive against standard cancer therapies and promote tumorigenesis, recurrence, and metastasis after chemotherapy Table 1^[99,104-107].

Interestingly, when considering the CSC phenotype and plasticity in chemoresistant HNSCC tumor samples and cell lines, members of the regulator of embryonic stem cell Sox and Oct4 are highlighted over the classical CD44, Bmi1, and even ALDH1 CSCs biomarkers. Sox2 was associated with clinicopathological parameters of worse outcomes in HNSCC patients and a mediator of therapy resistance in vitro. Functionally, Sox2 induced the expression of the anti-apoptotic protein Bcl-2 and enhanced resistance to apoptosis-inducing agents, including cisplatin^[108]. Accordingly, Lee et al. found that Sox2 overexpression was correlated with tumor recurrence and poor prognosis in HNSCC, contributing significantly to the acquisition of stem cell traits in vitro^[109]. Ectopic expression of Sox2 in HNSCC cells induced stemness by positive regulation of Oct4 and Nanog and co-expression of CD44. In addition, endogenous levels of Sox2 were significantly higher in ALDH1 high cells. At the same time, the downregulation of Sox2 was followed by Oct4 and Nanog down-regulation, decrease in stemness, invasion, EMT mediators, in vivo tumorigenicity, and frequency of CD44⁺ cells. Moreover, Sox2 enhances the chemoresistance of CSCs to cisplatin, possibly by inhibiting ABCG2 expression and resistance to oxidative stress in CD44⁺ CD271⁺ CSCs in HNSCC^[109].

Xie et al. found that Sox8 expression was positively associated with chemotherapeutic resistance, higher lymph node metastasis, advanced tumor stage, and shorter overall survival in HNSCC patients^[110]. Also, the expression of Sox8 in cisplatin-resistant HNSCC cell lines is responsible for orchestrating the acquisition of the CSC phenotype via ABCG2, Sox2, Oct4, and Bmi1 expression but also resistance to therapy and activation of EMT and Wnt/β-catenin pathway, favoring tumor invasion and progression. These findings indicate that Sox8 could be used as a biomarker and a possible target to eradicate the CSCs and increase tumor response to standard therapies toward HNSCC^[110].

Several in vitro studies investigating the relevance of CSCs on chemoresistance initially characterize the CSCs subpopulation based on the expression levels of CD44 and ALDH1. Nör et al. showed that treatment with low doses of cisplatin promotes Bmi1 and Oct44 expression and increases the CSCs fraction identified as CD44^high ALDH^high, indicating that these cells are intrinsically resistant to treatment and can expand after therapy^[111]. The study by Chen et al. elegantly confirmed that Bmi1⁺ CSCs are enriched in vivo after treatment with cisplatin, being able to reconstitute the tumor heterogeneity and are the main responsible for recurrence^[112]. Kulsum et al. found that HNSCC cell lines resistant to cisplatin and 5-FU showed enrichment of CD44⁺ ALDH1⁺ subpopulation, stemness, expression of ABCG2, Sox2, Nanog, Oct4, and NOTCH1 genes, and G0/G1 or S phase arrest^[113]. One of the mechanisms by which the CD44^high/ALDH1^high cells become resistant may be the upregulation of the DOT1L and monomethyl l-H3K79 that lead to miR-10 activation, resulting in cytoskeleton remodeling via RhoC and upregulation of prosurvival molecules such as cIAP-2 and XIAP^[114]. Another mechanism associated with cisplatin resistance of CD44^high/ALDH1^high is the secretion of FGF2. Most importantly, cisplatin combined with FGFR2 inhibition decreased the percentage of CD44^high/ALDH1^high, and no CSCs enrichment was noticed after cisplatin exposure, indicating that blocking FGFR is an attractive target to eliminate the CSCs in HNSCC^[115].

CD44 is frequently associated with other potential markers of CSC aiming for efficient enrichment of this subpopulation within HNSCC cell lines and tissues. Galbiatti-Dias et al. identified the CSC profile of HNSCC cell lines as CD44^high CD133^high CD117^high profile^[116]. This CSCs subpopulation demonstrated higher migration capacity and more resistance to Paclitaxel chemotherapy, in addition to an up-regulation of CD44 and down-regulation of EGFR transcripts in the HN13 oral cancer cell line^[116]. Elkashty et al. combined the positivity of CD44 to CD271 (p75NTR), a described marker of CSC in many tumors^[117], to isolate an enriched subpopulation of CSCs, followed by their characterization in vitro, in vivo, and HNSCC tissue samples. The authors found that CD44⁺ CD271⁺ cells exhibited higher cell proliferation, sphere/colony formation, chemoresistance to cisplatin and 5-FU, and radioresistance, upregulation of CSCs-related genes (Sox2, Oct4, Bmi1, Smo, and GLI1), and in vivo tumorigenicity^[117]. These combined cell markers also showed increased expression in patients with advanced disease.

A study from Oh et al. demonstrated that CD44⁺ cells derived from primary HNSCC had increased expression of ABCG2 and enriched side population^[118]. Yu et al. found that side population cells characterized by the CD133⁺ phenotype show elevated chemoresistance and ABCG2 expression, which was abrogated by combining cisplatin with CD133-targeted therapy^[119]. Moreover, Snail is overexpressed in CD44⁺ CSCs and associated with cisplatin resistance and high expression of ABC transporters^[120]. Interestingly, the percentage of Oct4 positive cells increases significantly after treatment with 5-FU, cisplatin, and paclitaxel^[121], and increased expression of ABCC6 was associated with increased resistance to cisplatin in Oct4 overexpressing cells, indicating that this poor explored ABC transporter may be relevant to resistance acquisition in HNSCC^[122]. Thus, constitutive or acquisition of stem cell and EMT-associated genes are involved with the up-regulation of drug transporters pumps and multi-drug resistance.

The process of EMT is tightly linked with the CSC’s biology and chemoresistance in HNSCC. CSCs keep their EMT phenotype until depositing in the distant sites of metastasis (a migratory phenotype), where they change their phenotype toward attaining a MET morphology to proliferate rapidly, causing tumor outgrowth (a proliferative phenotype)^[123]. This rapid cellular proliferation leads to hypoxia in the nearby milieu, thereby exacerbating tumor resistance to therapy^[124]. Masui et al. observed that the CSC-like phenotype is induced after Snail-overexpression and is associated with increased CD44⁺/ALDH⁺ in HNSCC cell lines^[125]. The EMT and CSC phenotype acquisition in Snail overexpressing cells also decreased chemosensitivity. Similarly, Ota et al. demonstrated that Snail-induced EMT was associated with increased stemness, inducing in vivo cancer invasive progression and enhancement of chemoresistance^[126]. A recent study from Oliveira et al. demonstrated that the CSC subpopulation and activation of the EMT program, characterized by down-regulation of E-cadherin and up-regulation of vimentin, mainly via association of epigenetic regulators and ZEB1, is involved with resistance to cisplatin in HNSCC cell lines^[96].

It is worth mentioning that the ability of tumor cells to dynamically adapt to signals provided by the tumor microenvironment and/or induced in response to therapy is obtained by the property of cell plasticity at different stages of tumor progression. Cancer cell plasticity reflects genetic and epigenetic alterations in tumor cells, promoting phenotypical diversity and contributing to intra-tumor heterogeneity^[127]. EMT and CSCs states are the two most studied axes of tumor cell plasticity and are often tacitly assumed to be synonymous^[128]. This is because both cell plasticity axes appear to drive one another in silico, in vitro, and in vivo studies^[129]. Notably, both mathematical modeling studies and experimental observations have reported that EMT is also not a unidirectional process since there are one or more hybrid epithelial/ mesenchymal (E/M) states between the two extremes of pure epithelial or pure mesenchymal phenotypes^[130,131] during EMT. For this reason, the term Epithelial-Mesenchymal Plasticity (EMP) has been used as a more accurate description of the process.

The same is true for CSC since there may be subsets of CSCs defined as epithelial, mesenchymal, and hybrid E/M (E-CSCs, M-CSCs, H-CSCs)^[132,133]. According to Sahoo et al. 2022^[128], the emerging evidence points to EMT and stemness being semi-independent axes, i.e., not every cell undergoing EMT may acquire stemness and not every cell switching to be a CSC is mandated to show one or more features of EMT. These authors recently proposed a mathematical model to understand the interconnectivity between the EMP and stemness axes aiming to elucidate the critical cellular processes driving metastasis. This model allows many possible couplings between EMP and stemness, showing that all phenotypes - epithelial, mesenchymal, and hybrid E/M - have the potential to be stem-like; however, this potential is likely to be maximum for hybrid E/M cells^[128]. On the other hand, tumor cells exhibiting an amoeboid phenotype belong to the utterly mesenchymal end of the EMP spectrum but show high stemness and metastatic potential^[134]. So, many stem cell phenotypes exist across the EMP spectrum that would only be identified based on single-cell RNA sequencing approaches^[128].

Garcia-Mayea et al. showed that CSCs isolated by sphere formation in non-adherent conditions were more resistant to cisplatin and 5-FU, possibly due to the increased levels of LC3II/I, indicating that autophagy may be involved in within-drug resistance of CSCs^[135]. Recently, using the same CSC model, these authors identified by RNAseq the TSPAN1 (Tetraspanin 1) gene as an essential modulator of chemoresistance in HNSCC^[136]. Blocking TSPAN1 demonstrated encouraging in vivo results, leading to impaired tumor growth, EMT acquisition, and metastasis spreading. Another possible target to eliminate HNSCC CSCs and cisplatin resistance is the SDCBP (Syndecan-binding protein), a central contributor in different phases of the metastasis cascade^[137,138]. Upon fibronectin and extracellular molecule engagement, SDCBP, as an adaptor protein, interacts with Src and forms a stable complex with FAK in the cellular membrane leading to long-term Src activation. As a result, downstream target signaling pathways such as NF-kB and TGF-β are activated, promoting EMT, tumor migration, invasion, metastasis, and cisplatin resistance^[138]. Lee et al. showed that Wnt/β-catenin signaling is activated in CSCs, and β-catenin overexpression drives the acquisition of CSCs properties as self-renewal, stem cell marker expression, including Oct4, and chemoresistance^[139]. In hypoxic conditions, HIF-1α activates NOTCH1, which is responsible for stemness, EMT activation, and resistance to cisplatin in CD44⁺ cells^[140].

All these exposed findings reveal how broad and complex the process of resistance to the chemotherapeutics available today for treatment could be. It also guides us to seek new and innovative drugs focused on CSCs, such as targeted therapy and immunotherapy, for better treatment and prognosis of HNSCC patients. Notably, the plasticity of CSCs must also be considered since their dynamic phenotype switch may be responsible for different levels of resistance even in the same tumor type. As pointed out by Biddle & Marles^[141], an effective biomarker should be precise in correlating the presence of phenotypically plastic CSCs with tumor aggressiveness and therapeutic resistance. It would allow more accurate clinical decisions, such as neck dissection and chemotherapy regimens in HNSCC. More recent evidence highlights some meaningful advances, for example, monoclonal antibody therapy anti-CD44v6 and other markers related to EMT signaling pathways activation, such as the Notch, WNT, and ERK/ MAPK pathways. Although, in terms of clinical safety, targeting CSC-specific processes is not well established yet.

CONCLUDING REMARKS

The presence of CSCs in HNSCC and other solid tumors is associated with tumor heterogeneity and resistance to standard therapies. Target CSCs therapy is very challenging as these cells are a dynamic and plastic population capable of switching between different phenotypes and activation states according to the stimuli provided by the TME. As a result, the frequency of CSCs and their spatial localization in the primary tumor and metastatic foci may be variable, leading to different levels of tumor resistance after treatment. Many studies demonstrated that after radio and chemotherapy, CSCs are enriched and guide tumor recurrence and progression.

In this scenario, it is mandatory to characterize the CSCs and their mechanisms of interactions with the TME in HNSCC to better design targeted therapies that efficiently eliminate these cells in combination with standard treatment and/or immunotherapy. Disrupting the TME can lead to hypoxia inhibition and disturb the CSC niche, facilitating CSCs sensitization to chemotherapy. Moreover, CSCs interaction with different cell types in the TME may be impaired, facilitating its elimination and response to standard treatment. It is essential to highlight that CSCs have an efficient drug efflux machinery that should be considered as possible targets to improve drug accumulation within this subpopulation of tumor cells. Targeting signaling pathways involved with acquiring stemness, such as the Wnt/β-catenin, FGF, and NOTCH1 in HNSCC, may also be an attractive strategy to eliminate the CSCs and drug resistance. Taken together, CSCs are a relevant target to achieve control of disease and treatment response in HNSCC as they represent significant drivers of tumor resistance. Future studies, especially those using cutting-edge methodologies such as scRNAseq, will help to identify new CSCs targets and cellular interactions that can be used to develop new multi-faceted adjuvant therapies.