2.1 scRNA-seq workflow

The workflow of the critical steps in a typical scRNA-seq experiment is shown in Figure 1. Compared to traditional bulk RNA analyses, single cells must be isolated from the tumor tissues at the beginning of scRNA-seq. Second, specific protocols are needed to perform mRNA reverse transcription and cDNA amplification with high efficiency and low biases as the quantity of RNA in a single cell is lower than that in bulk RNA analyses. New strategies have also been developed to better analyze data from scRNA-seq because previously developed tools for bulk RNA-seq were not appropriate for these complex data analysis. A comparison of scRNA-seq and bulk RNA analysis workflows is shown in Figure 2.

Due to the unique workflow of scRNA-seq including single-cell isolation at the beginning and the subsequent mRNA reverse transcription from limited RNA in a single cell, multiple methods such as micromanipulation and microfluidics have been utilized and new approaches such as Smart-seq and CEL-seq are constantly being reported [24-26]. The following sections provide a brief introduction to the current strategies for single-isolation and specific protocols for scRNA-seq.

2.2 Single-cell isolation

Single-cell isolation is the first step in scRNA-seq. A brief overview of these experimental methods, along with their strengths and weaknesses, is presented in Table 1.

One simple approach is micromanipulation. Single cells are manually picked with a glass pipet under a microscope [5, 27], which enables sampling from a limited number of cells or fragile cells, such as early embryos [28]. Microscopic supervision of cell isolation ensures that each sample is indeed a single cell; however, this method is time-consuming and has low-throughput. Another weakness of this approach is the high technical requirements, which may cause cellular injury due to mechanical shearing [29]. Therefore, due to these obstacles, micromanipulation is now rarely being used.

Laser capture microdissection (LCM) is another approach to capture single cells from solid tissue. Under a microscope, a laser beam is focused on cells of interest and attaches these individual cells to a thin and transparent film [30]. The single cells on the film are then transferred to a microcentrifuge tube containing appropriate buffer solutions [31]. Similar to micromanipulation, LCM is also laborious and inconvenient. Another shortcoming is that this approach is technically challenging to capture the contents of a single cell cleanly, without contamination from flanking cells and without damaging the cell RNA [32]. However, LCM also has unique advantages. It enables single-cell isolation from solid samples and it can provide spatial information for the target cell, which makes it still usable in practice.

Generally, microfluidics refers to technologies using micro-scale structures to precisely control fluids at ultralow volumes, typically at the nanoliter-to-femtoliter scale [33-35]. Microfluidics platforms, such as the Fluidigm C1 microfluidic robotic platform or microdroplet-based microfluidics methods have been utilized to trap single cells [36]. On microfluidics platforms, automated single-cell isolation was followed by automated reverse transcription and preamplification-on-chip, which downscale reactions to nanoliter or picoliter volumes and further reduce costs [37]. Moreover, microdroplet-based methods have the potential to capture thousands of cells in a single experiment [38]. Although microfluidics methods have their own limitations that capture efficiency can be low for sticky cells [36], they have gained increasing popularity due to their high throughput and low analysis cost.

Fluorescence-activated cell sorting (FACS) is a specialized type of flow cytometry. First, a mixture of biological cells is combined with a specific fluorescently tagged antibody. By detecting the specific fluorescent signals from individual cells based on the tagged antibody, single cells of interest can be sorted from the heterogeneous cell populations. Due to its capability of sorting particular cells of interest and advantages of high throughput, low analysis cost, and flexibility of samples, FACS has become a widespread method for isolating thousands of single cells [29]. However, this method also has some drawbacks, including the risk of damaging cell viability during sorting and the difficulty in differentiating cell subpopulations expressing similar markers [39].

Collectively, different single-cell isolation methods have their own strengths. Micromanipulators ensure precise cell picking and minimize cell damage. LCM enables single-cell isolation from solid samples and reserve spatial information. Microfluidics methods downscale the volume of reagents which makes it cost-effective. FACS can capture particular cells of interest. Therefore, a comprehensive understanding of these single-cell separation technologies is necessary for the selection of proper single-cell isolation methods.

2.3 Specific protocols for scRNA-seq

The accuracy and sensitivity of scRNA-seq depend on its efficiency in reverse transcription and amplification processes to obtain sufficient cDNA from a limited amount of RNA in a single cell, which enables the transcriptomic analyses at a single-cell level. Thus, several specific protocols have been developed (Table 2).

Protocols in scRNA-seq are characterized by different strategies used in reverse transcription reaction and cDNA amplification. Poly(A) tailing and template-switching are two main strategies in the reverse transcription reaction. “Tang method” [5] and its improved version, Quartz-seq [40], are representative methods using poly(A) tailing while Smart-seq [41], Smart-seq2 [7], STRT-seq [42] and Drop-seq [10] use template-switching to complete second-strand synthesis. cDNA amplification after reverse transcription can be achieved by either polymerase chain reaction (PCR) or in vitro transcription (IVT). PCR can be carried out after the second-strand synthesis by poly(A) tailing or template-switching. As a nonlinear amplification process, its efficiency is sequence dependent [36]. IVT is a linear amplification method that incorporates the T7 promoter in the poly(T) primers [43] ScRNA-seq protocols that apply IVT instead of PCR include CEL-seq [44], MARS-Seq [45] and inDrops [9].

Among the multiple scRNA-seq protocols mentioned above, a key difference is that some provide full-length transcript data such as Smart-seq [41] and Smart-seq2 [7], whereas others specifically count only the 3’ or 5’ ends of the transcripts such as CEL-seq [44] and MARS-seq [45]. Therefore, the selection of a specific protocol depends on the nature of the research question. Compared to methods only capturing and sequencing the 3′ or 5′ ends of the cDNAs, protocols capable of full-length transcription are more suitable for alternative splicing pattern analyses, allelic expression detection, and RNA editing identification [39]. The strength of protocols sequencing the 3′ or 5′ end of the transcript is that they are able to combine unique molecular identifiers (UMIs) [46]. These tags allow for the identification and quantification of individual transcripts, which improve the gene-level quantification and throughput [24]. Therefore, in some cases that full-length transcript data was not required, these tag-based methods have become dominant for quantification purposes, especially for larger cell populations [2].

3.1 Tumor heterogeneity

Gene expression profiling has been widely used to characterize bulk tumors in individual cancer patients to discover potential target therapy [47-49] but cancers usually display intratumoral heterogeneity, which may influence their therapeutic response to specific targeted therapy as well as clinical outcomes. The emergence of scRNA sequencing allows the assessment of genetic heterogeneity in breast cancer at a single-cell resolution which can help researchers to uncover the immense biological complexity in tumors.

Triple-negative breast cancer (TNBC) is a representative breast cancer subtype characterized by extensive intratumoral diversity. Although TNBC exhibits significantly higher response rates to neoadjuvant chemotherapy, compared to estrogen receptor (ER)-positive tumors, patients with TNBC who failed to achieve a pathologic complete response (pCR) were usually associated with very poor outcomes [50]. This indicated the possible existence of a minor subpopulation of TNBC cells not sensitive to traditional chemotherapy that may induce subsequent metastasis. Thus, identifying and characterizing these distinct cells could guide targeted therapy development and survival improvement in TNBC. As scRNA-seq can provide complete gene expression patterns of individual cells obscured in bulk analysis, a research group from Harvard Medical School [12] applied single-cell profiling to discover the sub-clonal heterogeneity and aggressive disease states in TNBC. By performing scRNA-seq on untreated primary TNBC tumors, the researchers confirmed the cellular heterogeneity within primary TNBCs and identified five distinct clusters of cells by clustering analyses. Among the identified five clusters of epithelial cells named as cluster 1-cluster 5, cluster 2 had the highest proportion of high-cycling cells, indicating its high proliferation ability. Further investigation showed that cluster 2 was associated with a luminal progenitor (LP) signature that was considered as the cell of origin for breast cancers [51]. The study also validated that the high expression of the cluster 2 signature was related to worse survival outcomes. The researchers applied scRNA-seq to explore tumor heterogeneity and discovered a malignant cluster subpopulation that may drive tumor progression and, thereby, lead to poor survival outcomes. These findings not only expand our knowledge on tumor heterogeneity in TNBC but also provide potential predictors or treatment targets for TNBC with poor prognosis.

Besides TNBC, scRNA-seq has also been used to comprehensively characterize heterogeneous tumors of other breast cancer subtypes. Korean researchers [13] conducted transcriptome analysis of 515 single cells from 11 patients with different breast cancer subtypes, including luminal A, luminal B, human epidermal growth factor receptor 2 (HER2), and TNBC. The most versatile subtype composition was observed for cancer cells isolated from HER2-positive tumors, from predominant HER2 to mostly TNBC type. The researchers also found that cells from the ER/HER2 double-positive tumor tended to be the ER subtype due to the low expression of HER2 module genes and were related to the corresponding predominant ER downstream signaling pathway activation. These findings suggest that molecular profiling at the single-cell level may be more detailed and accurate compared to molecular subtypes defined by bulk analyses. Based on single-cell molecular profiling, clinicians may identify ER/HER2 double-positive tumors with prominent activation of the ER downstream signaling pathway that might benefit from intensive hormone therapy, or identify HER2-positive tumors with low levels of HER signaling pathway activation but higher basal gene expression that might be resistant to anti-HER2 target therapy. Guided by single-cell molecular profiling, clinicians could choose more effective therapeutic regimen and improve the patients’ prognosis.

Single-cell profiling of circulating tumor cells (CTCs) can also provide unique insights into tumor heterogeneity. Genotypic and phenotypic characterization of CTCs has the potential to provide a better understanding of tumor evolution and identify metastasis-initiating cells [52-55] to elucidate the onset of metastasis and identify potential therapeutic targets that could be used to prevent or treat metastasis. A recent single­cell study on CTCs isolated from blood samples of patients with primary and metastatic breast cancer [14] reported that individual CTCs had significant heterogeneity and identified two distinct clusters of CTCs named as Cluster I and Cluster II. Cluster I was characterized by the expression of genes associated with the epithelial-mesenchymal transition while Cluster II generally showed low to undetectable expression of these genes, which indicated that Cluster I may be related to cancer progression and poor prognosis. The researchers also compared the profiles of CTCs and breast cancer cell lines that are widely used in drug discovery. Significant differences between CTCs and cell lines were identified in the expression of certain genes, levels of growth factors and clinically informative phenotypes, which indicated that traditional cell lines may not perfectly simulate the seeding process of metastasis by CTCs and that conventional therapies designed and tested based on these cell lines may be inadequate for eliminating metastasis. Thus, appropriate experimental systems for drug discovery should be carefully selected with the help of cell-to-cell profiling of CTCs. Meanwhile, liquid biopsies may help to identify patients with specific CTC clusters and inform the selection of therapeutic interventions targeting overexpressed gene products and activated pathways in these CTC clusters.

3.2 Tumor microenvironment

The tumor microenvironment comprises of heterogeneous cellular populations including tumor cells and the surrounding nonmalignant cells, such as vascular, immune, and fibroblast cell types [56]. As previously described [57-59], constant cross-talk between tumor cells and the surrounding microenvironment was associated with treatment response and survival outcomes of malignancies. ScRNA-seq is a powerful tool to explore the complex tumor microenvironment and further facilitate individualized therapy and overcome drug resistance. Here, we briefly introduce a few key studies applying scRNA-seq methods to explore tumor environment heterogeneity.

Researchers from the Memorial Sloan Kettering Cancer Center conducted a single-cell analysis of the immune environment of eight primary breast carcinomas composed of one HER2-positive tumor, two TNBCs and five ER-positive tumors, and found a large degree of diversity in the immune composition of each tumor [60]. By assessing the immune cells captured in the normal and malignant breast tissues, lymph nodes, and peripheral blood, the researchers found that the immune phenotype was associated with the tissue of residence, suggesting that biomarkers based on blood samples may not reflect the immune phenotype states in tumors [15]. In-depth analyses of T cells, which are considered more clinically relevant, revealed that intratumoral T cells displayed continuous activation and differentiation transitions rather than a few discrete and stable cell states. Therefore, the traditional classification of T cells in tumors may underestimate the complexity of the T cell populations. The authors of this study suggested that the single-cell profiling of diverse immune cells facilitated the understanding of complex mechanisms in immune enhancement and suppression in the tumor environment, and the co-expression of checkpoint receptor genes identified in some regulatory T cells clusters may be potential immunotherapy targets.

Another single-cell profiling study of T cells in the breast cancer microenvironment conducted by a research group in Australia analyzed data from 6,311 T cells isolated from human breast cancer tissue. Their analyses revealed heterogeneity in the CD8⁺ population [16]. CD8⁺ T cells with features of tissue-resident memory T (TRM) cell differentiation were found to express both immune checkpoints and cytotoxic effector proteins. The expression of inhibitory receptor genes such as PDCD1 and CTLA4 was higher in these distinct TRM cells. Thus, CD8⁺ TRM cells may be associated with immunosurveillance and could be potential immunotherapy targets. As indicated in that study, scRNA-seq enabled the discovery of minor subgroups of immune cells that were related to immunosuppression or immunosurveillance. Biomarkers of these distinct immune cells may serve as prognostic factors or therapeutic targets, allowing individual risk assessment and stratification of patients for targeted therapies.

In summary, scRNA-seq helped to explore the complexity of cell populations in the breast cancer microenvironment and revealed distinct cell subpopulations with the potential to become individualized immunotherapy targets.

3.3 Diverse approaches beyond exploring heterogeneity

As mentioned above, many cancer studies have explored the cellular heterogeneity within a tumor. However, scRNA-seq can also be applied to many other types of breast cancer research beyond exploring heterogeneity. For instance, scRNA-seq has been used to study the cell-to-cell communications mediated by secreted factors and its role in tumor progression [18]. The different expression levels of receptors and their ligands between tumoral and normal tissues, and the correlation between the expressions of these specific ligand-receptor pairs could serve as indicators of cell-cell communication. By providing information on the expression level of ligands and receptors of individual cells, which determines “who talks to whom”, single-cell transcriptome data has enabled researchers to map a more-defined cell-cell communication network with distributions of receiving and signaling cells in distinct cell populations. Moreover, with increased knowledge of the heterogeneity of the tumor parenchyma, research on the communications between distinct cell populations have also gained attention. ScRNA-seq could integrate the ligand-receptor analysis and clonal analyses to construct a heterotypic cell-cell communication network, and further facilitate the study on cell-to-cell communications between cell populations of clinical significance, which might help to find novel targets of individualized therapy.

Besides genetic heterogeneity, tumors also display regulatory heterogeneity [61]. While the current scRNA-seq is a powerful tool to study heterogeneity in lineage, it is still difficult to reliably measure tumor-cell regulatory heterogeneities. Thus, complementary approaches are needed to profile single-cell regulatory states. One example is the in situ 10-cell RNA-seq (10cRNA-seq) [20], which combines the single-cell resolution of laser capture and improves the preamplification procedure; enabling RNA-seq of 10 micro-dissected cells and making it a reliable, unbiased, and sensitive method to measure cell-state heterogeneity in tumors.

Researchers have also used scRNA-seq to identify progenitor or stem-like cells. By applying a marker-free system biology approach called LandSCENT in single-cell transcriptome data from the human mammary epithelium, researchers have identified a novel bipotent stem-like state that was also correlated with poor prognosis in basal breast cancer [19]. The novel marker-free computational approach used in this study allowed the estimation of cell potency to enable the identification of rare cell populations representing progenitor or stem-like cells. Its future application in other tumor types to identify putative cancer stem cells warrants investigation.

With the increasing awareness of the importance of special cellular information in cancer research, novel computational approaches have been developed to reconstruct the spatial organization of cells lost in the process of scRNA-seq analysis. Researchers in Sweden applied spatial transcriptomics technology to identify immune cell distribution in HER2⁺ breast cancer [21]. They constructed and analyzed three-dimensional images of the transcriptional landscape to reveal immune cell infiltration patterns among different patients, which may provide an opportunity for the design of personalized treatments.

3.4 Therapeutic selection and monitoring

The continued discovery of genotypically and phenotypically distinct cell subpopulations has changed our view of cancer from a group of homogenous tumor cells to a constellation of heterogeneous, continuously evolving cancer subpopulations [14]. ScRNA-seq method can be an important tool to identify optimum combined therapies to effectively target these heterogeneous cell populations. In addition, scRNA-seq may identify alterations associated with therapeutic resistance in distinct cell clusters to support individualized cancer treatment.

To identify cell diversity by transcriptome profiling, scRNA-seq was utilized to explore possible predictors of clinical outcomes and drug targets in breast cancer by characterizing cell subsets with metastatic potential. For example, transcriptome profiling of migratory breast cancer cells using scRNA-seq by researchers at the University of Michigan [62] revealed that the over-expression of certain genes in migratory cells was associated with poor prognosis in breast cancer patients, suggesting that these regulators of cell migration may be potential prognostic markers or drug targets. Similarly, researchers at the University of California conducting scRNA-seq in matched primary breast cancer and micrometastases discovered a distinct transcriptome program related to poor prognosis in micrometastatic cells and identified mitochondrial oxidative phosphorylation (OXPHOS) as a top pathway upregulated during metastatic seeding [63]. Selective inhibition of OXPHOS may be a novel target therapy to prevent breast cancer metastasis. Ongoing research is expected to discover additional factors for more accurate prediction of prognosis and novel drug targets to facilitate pharmaceutical research, which may dramatically affect the current treatment of patients with breast cancer.

Currently, gene expression signatures are mainly generated by conventional molecular profiling methods such as micro-array or bulk RNA-seq, which measures the average values of gene expression. scRNA-seq enables profiling at the single-cell level and detection of gene expression differences within cell populations. Therefore, novel biomarkers related to cancer progression and response to target therapies might be revealed by scRNA-seq. By comparing single-cell transcriptome profiles between trastuzumab-treated and non-treated patients with HER2⁺ breast cancer, researchers found that trastuzumab-enhanced matrix Gla protein (MGP) gene expression was associated with a better prognosis [64]. Moreover, they constructed a 48-gene expression signature that was associated with cardiomyocyte death, which could serve as a predictor for trastuzumab-mediated cardiotoxicity. Although no other gene expression signature-based on scRNA-seq data has yet been reported, additional novel signatures will likely be discovered with increased application of scRNA-seq. These signatures may better guide therapeutic decision-making in clinical practice.

As a major obstacle in the treatment of breast cancer, drug resistance is related to tumor heterogeneity. Therefore, the detection of cell populations able to survive anticancer treatment is critical to the prediction and reversion of drug resistance. The discovery of minor groups of drug-resistant cells by genetic studies performed on bulk samples was difficult. However, the development of scRNA-seq enables extensive single-cell gene expression profiling to characterize these drug-resistant cells and identify the underlying mechanism of drug resistance acquisition. Researchers have performed scRNA-seq in docetaxel-resistant cells derived from the luminal-type breast cancer cell line MCF7 [65]. Epithelial-to-mesenchymal transition and stemness-related genes were found upregulated, while cell-cycle-related genes were found downregulated in these chemo-resistant cells. A similar gene-expression pattern was also identified in a subgroup of untreated cells, suggesting the existence of cell subpopulations with an inherent predisposition toward docetaxel resistance. Identification of distinctive gene expression patterns of cells not sensitive to specific drugs may inform the design of multi-gene panels to predict the efficacy of specific drugs and assist clinicians in choosing an individualized therapeutic regimen.

In metastatic breast cancer research, the significance of CTCs in selecting appropriate therapies, monitoring therapeutic response, and innovating new treatments has been widely recognized [66]. Multiple nucleic acid- and protein-based assays have been developed to assess the ER and HER2 status in CTCs, which help clinicians to identify candidates for endocrine therapy and anti-HER2 therapy [67, 68]. Researchers also found that other biomarkers beyond ER and HER2 in CTCs could guide treatment decisions and monitor treatment efficacy as well. For instance, CD133 expression in CTCs may serve as a potential marker for chemoresistance [69] while CTC-Endocrine Therapy Index might predict resistance to endocrine therapy in MBC patients [70]. The heterogeneity and rarity of CTCs warrant the use of single-cell technologies to provide us with a more comprehensive understanding of CTCs. In a recent study, the molecular analysis of single CTCs isolated from metastatic breast cancer patients has been achieved in a noninvasive way and researchers have identified the heterogeneity of PIK3CA mutational status at single-cell level in CTCs isolated from individual patients [71]. Thus, similar studies could be conducted to analyze other clinically relevant genetic mutations in CTCs. CTC-derived preclinical models could then be designed to develop new drugs targeting specific genetic mutations. Furthermore, corresponding clinical trials can be designed to investigate the prognostic value of gene mutations detected in CTCs and monitor the efficacy of newly developed drugs, which may help clinicians to more effectively tailor therapeutic regimens for individual patients.

ScRNA-seq is a powerful tool in discovering novel therapies associated with tumor mutation burden or immune checkpoint crosstalk. A Korean research group analyzed scRNA-seq data from breast cancer and immune cells to explore potential approaches for improving the therapeutic outcome of immune checkpoint inhibition. The study mainly analyzed the tumor mutation burden (TMB), immune checkpoint crosstalk, radiosensitivity, and their relationships to test the hypothesis that the radiosensitivity of tumor cells might be related to its programmed death-ligand 1 (PD-L1) expression or TMB [22]. According to their results, radioresistant tumor cells were associated with a higher rate of PD-L1 positivity and TMB. Additionally, increased immune checkpoint crosstalk between cancer cells and immune cells was observed in TNBC compared to those in the luminal and HER2-positive subtypes. These findings suggest that radiotherapy could be combined with immune checkpoint blockades to improve the outcome of tumors with radiosensitive cells characterized by high PD-L1 expression and higher TMB in TNBC. In another recent study that performed scRNA-seq analysis in TNBC murine models, a high mutation burden was also found associated with immune checkpoint therapy response [72]. In these models, B cell activation of T cells was identified as an important mechanism that mediated response to checkpoint inhibitors, which indicated that the combined novel drugs that could induce B cell-dependent T cell activation may enhance the anti-tumor response of immune checkpoint therapy [72]. Single-cell profiling also helped researchers to find a distinct immunosuppressive immature myeloid cell (IMC) population that infiltrated the fast-evolving cyclin-dependent kinases 4 and 6 (CDK4/6) inhibitor-resistant tumors [73]. According to this study, combinatorial immunotherapy of IMC-targeting tyrosine kinase inhibitor cabozantinib and immune checkpoint blockade could enhance anti-tumor immunity and overcome CDK4/6 inhibitor resistance [73]. Single-cell profiling showed its superiority in these studies to identify potential biomarkers and optimal combination strategies of immune therapy in the management of breast cancer.

ScRNA-seq was also performed to study the synergistic effect of the anti-tumor immune response induced by chemotherapy in breast cancer. A recent study conducted by Chinese researchers identified the phenotype switch of B cells induced by chemotherapy in breast cancer patients and further explored the underlying mechanism. scRNA-seq of tumor-infiltrating B cells was performed in paired clinical samples of pre- and post-neoadjuvant chemotherapy collected from breast cancer patients [23]. The results showed that a distinct B cell subset that expressed high levels of inducible T-cell co-stimulator ligand (ICOSL) significantly increased after neoadjuvant chemotherapy. Compared to patients with stable diseases or progression, the ICOSL⁺ B cell density in the residual tumor was markedly higher in patients with partial or complete remission (P < 0.001), indicating that the ICOSL⁺ B cell subset was also related to improved therapeutic efficacy. Meanwhile, survival analyses indicated that ICOSL⁺ B cell abundance was an independent positive prognostic factor for disease-free survival (hazard ratio [HR], 0.275; 95% confidential interval[CI], 0.134-0.564) and overall survival (HR, 0.232; 95% CI, 0.090-0.601). The underlying mechanism behind the B cell switch was studied in mouse models. The researchers found that the anti-tumor immune response was elicited by these distinct B cell subsets after chemotherapy, which enhanced therapeutic efficacy. Further analyses found the generation of ICOSL⁺B cells depended on complement receptor type 2 (CR2) signaling, whereas tumoral CD55 expression could inhibit this complement-dependent ICOSL⁺ B cell induction and thereby undermine anti-tumor immunity. Therefore, CD55 could be a potential target to enhance immune response and facilitate immunogenic cell death.

The combination of scRNA-seq with other technologies has further facilitated the development of personalized treatment. Combined with a scalable hydrodynamic scRNA-seq barcoding technique, a high-throughput contamination-free scRNA-seq, Hydro-Seq, enables the analysis of small numbers of CTCs in 10 mL of blood [74]. Hydro-Seq allowed the detection of ER, progesterone receptor (PR), HER2 expression, and molecular profiles in CTCs, which could be discordant compared to primary tumors. Identification of the clinical markers of CTC in lipid biopsy may help to identify drug targets for personalized therapy. Transcriptome analysis of CTCs also increases our understanding on the intra-tumor cellular heterogeneity and evolution of tumors following therapy, which could provide insights into monitoring target therapeutics and processes underlying tumor metastasis. Using functional cellular approaches combined with scRNA-seq, researchers have analyzed cells with different degrees of cancer stem cells (CSCs) properties in breast cancer cell lines and identified potential breast cancer biomarkers related to CSC properties [75]. Due to their self-renewal capacity and potential to differentiate into cancer cells, CSCs have been associated with tumor progression, metastasis, and treatment resistance, leading to poor prognosis [76]. Many useful markers for the identification of CSCs can also serve as therapeutic targets to eliminate CSCs, especially those involved in pathways associated with self-renewal and epithelial–mesenchymal transition (EMT), such as Notch, Hedgehog, and Wnt signaling pathways [77]. The inhibition of the Notch signaling can be achieved by γ-secretase inhibitors [78] and clinical trials have already been launched to evaluate the efficacy of γ-secretase inhibitor MK-0752 (Merck) in metastatic breast cancer [79]. Multiple CSC-based targets have been investigated to inhibit Hedgehog and Wnt signaling as well, such as vismodegib [80] and vantictumab [81]. Therefore, the identification of CSC markers and novel drugs targeting CSCs may provide new opportunities for individualized therapy. A newly developed method has enabled the detection of cell-level mutations from scRNA-seq data [82], which further broaden the application of scRNA-seq in future cancer research. Specific somatic alternations were found associated with drug responses in breast cancer. For example, GATA3 mutation may predict better response for aromatase inhibition [83] and PI3K and/or ERBB2 mutations may indicate the sensitivity of HER2⁺ tumors to neoadjuvant docetaxel, carboplatin, trastuzumab and lapatinib regimen [84]. Mutation profiling based on scRNA-seq in breast cancer research has not been conducted yet. However, with the development of new methods like the one mentioned above, it is believed that scRNA-seq will be used to identify cell-specific mutations in future studies and help to predict individual drug-response.

Individualized cancer management largely depends on the precise detection of specific molecular subtypes or therapeutic targets. Although heterogeneity-based personalized treatment has progressed substantially in breast cancer, with a growing list of target agents in adjuvant, neoadjuvant, and metastatic settings, there is an unmet need to enhance precise medicine to improve survival and eliminate late relapse. As mentioned above, scRNA-seq is a powerful tool in exploring tumor and tumor environment heterogeneity. The detection of distinct cell subsets not sensitive to conventional therapy may explain intrinsic drug resistance and targeted agents against these cell subsets could provide crucial clinical benefits. By using scRNA-seq, researchers have found that the endocrine-resistance of fulvestrant and tamoxifen was caused by a group of pre-existing genetically distinct cells not sensitive to endocrine therapy and these cells were highly selected during treatment [85]. Therefore, the innovation of novel drugs targeting these cells might reverse the endocrine resistance. ScRNA-seq could also be used to detect the emergence of resistant cell subsets after treatment, which could have significant clinical implications for second-line treatment decision-making on available or new target drugs. For instance, in a translational research mentioned above, researchers have used scRNA-seq to trace the tumor evolution during the combination treatment of anti-HER2/neu antibody and CDK4/6 inhibitor and discovered that the acquired resistance was related to infiltrating immunosuppressive immature myeloid cell (IMC) population [73]. They also found that combinatorial immunotherapy of IMC-targeting drugs and immune checkpoint blockade was an effective regimen to treat these fast-evolving CDK4/6 inhibitor-resistant tumors [73]. Collectively, integrating scRNA-seq into basic and translational research could promote personalized therapy by identifying potential treatment targets to develop novel drugs and reveal promising biomarkers to monitor treatment efficacy and guide therapeutic decision-making.

AUTHORSHIP

Conception and design: all authors. Manuscript writing: SND. Manuscript revision: XSC and KWS. All authors read and approved the final manuscript.

ETHICS APPROVAL AND CONSENT TO PARTICIPATE

Not applicable.

CONSENT FOR PUBLICATION

Not applicable.

CONFLICT OF INTEREST

The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

FUNDING

The authors received financial support from the National Natural Science Foundation of China (Grant Number: 81772797), Shanghai Municipal Education Commission—Gaofeng Clinical Medicine Grant Support (20172007); Ruijin Hospital, Shanghai Jiao Tong University School of Medicine “Guangci Excellent Youth Training Program” (GCQN-2017-A18). All these financial sponsors had no role in the study design, data collection, analysis, or interpretation.