Breast cancer is the most prevalent malignancy in women and its incidence is increasing globally, especially in developed countries (Bianchini et al., 2016; Siegel et al., 2020). Different treatment strategies, such as surgical resection, hormone therapy, targeted therapy, radiation therapy, and chemotherapy, have greatly improved survival of breast cancer patients (Bray et al., 2018; Burstein et al., 2014; Early Breast Cancer Trialists' Collaborative G, 2015; Khan et al., 2012; Saini et al., 2012). To date, clinical strategies for breast cancer treatment remain suboptimal. Although continuous treatment with tamoxifen for 10 years has reduced cancer recurrence and mortality of patients with luminal A breast cancer, a significant proportion of patients may be over-treated (Bianchini et al., 2016; O'Conor et al., 2018). Hence, it is urgently needed to discover novel biomarkers to predict treatment effectiveness and to improve treatment success and prognosis of breast cancer patients. Furthermore, the available therapies can also lead to a considerable number of patients at a risk to develop a delayed breast cancer metastasis, which occurs even 20–40 years after breast cancer diagnosis (Sharma, 2018). Notably, breast cancer metastasis occurring 5–8 years after initial surgical resection has become a significant cause of cancer relapse, progression, and poor survival in patients (Nishimura et al., 2013); thus, further researches on its molecular mechanisms and gene alterations may help identify novel biomarkers and targets for development of therapeutic strategies to effectively control breast cancer metastasis and progression.

Indeed, tumor metastasis is a multistep process, during which, cancer cells escape from the primary site, migration into a neighboring or distant tissues, extravasation, survival, and colonization, leading to the formation of new tumor nodules at a secondary site (Drabsch and ten Dijke, 2011; Klein, 2008; Scott et al., 2012; Syn et al., 2016). The rate-limiting step of cancer metastasis is cancer cell colonization and proliferation at the secondary site, because the initial metastatic cancer nodule usually lacks an efficient vasculature to provide sufficient nutrients to support cancer cell growth. Thus, the newly arrived tumor cells may grow to a certain size in the new and harsh microenvironment and undergo growth arrest in that organ. However, once they regain their proliferative ability, delayed metastasis will occur (Langley and Fidler, 2007). Molecularly, CD44⁺/CD24^- breast cancer cells from primary breast tumors are associated with distant metastasis (Abraham et al., 2005), and these cells display potent motility and invasiveness (Liu et al., 2010), similar to chemoresistance cancer stem cells (CSCs) (Velasco-Velázquez et al., 2011). Previous studies have shown that CD44⁺/CD24^- breast CSCs may be a dominant factor for the relapse of triple negative breast cancer (TNBC), due to their potent self-renewal and differentiation capacities (Geng et al., 2014; Wang et al., 2014). Indeed, injection with about 50 breast CSCs can induce a solid tumor mass in immunocompromised mice (Chaffer et al., 2011; Iliopoulos et al., 2011). Thus, the number of breast CSCs in the secondary site may affect the efficient formation of early metastatic nodules, and breast CSCs are commonly prone to be resistant to chemotherapy (De Angelis et al., 2019). Moreover, previous studies have reported that the differentiated cancer cells can spontaneously convert into CSCs to renew the CSC pool in breast cancer, pancreatic cancer and sarcomas, resulting in chemoresistance (Gruber et al., 2016; Kim et al., 2015; Ye et al., 2018). Thus, the dormant CD44^-/CD24^- breast cancer cells that have previously been colonized in the metastatic site may be able to spontaneously convert into CD44⁺/CD24^- breast CSCs to regain their potent proliferative ability and drug resistance, resulting in delayed breast cancer metastasis. Accordingly, it is reasonable to hypothesize that the frequency of CD44^-/CD24^- cells in human tumor specimens may be useful in the prediction of delayed breast cancer metastasis.

The current study aimed to explore the molecular mechanisms by which CD44^-/CD24^- cell conversion into CSC promotes delayed breast cancer metastasis. First, a retrospective analysis of CD44^-/CD24^- breast cancer cells in tissue specimens from patients enrolled from three academic medical centers was performed to investigate the potential associations between the frequency of CD44^-/CD24^- cells and postoperative tumor metastasis. Next, the spontaneous CD44^-/CD24^- cell conversion into CD44⁺/CD24^- CSCs was tested and the biological functions of the converted CSCs were analyzed in vitro and in vivo. The results may provide novel insights into the role of CD44^-/CD24^- tumor cells in delayed breast cancer metastasis and into the potential use of CD44^-/CD24^- cells as a biomarker to predict survival and metastasis in breast cancer patients. These findings also suggest that RHBDL2 may be a novel therapeutic target for the future studies.

The current study uncovered that CD44^-/CD24^- TNBC cells were able to spontaneously convert to CD44⁺/CD24^- CSCs, consistent with previous studies in vitro (Italiano and Shivdasani, 2003; Zoppino et al., 2010) both WT CSCs and CD44^-/CD24^- CSCs had similar biological properties in terms of stemness marker expression, self-renewal, differentiation, tumorigenicity and lung metastasis in vitro and in vivo a higher frequency (≥19.5%) of CD44^-/CD24^- cells in breast cancer tissues was associated with worse DFS and delayed postoperative distant metastasis a higher frequency of CD44^-/CD24^- and CD44⁺/CD24^- cells, higher tumor N stage, and molecular subtypes were predictors of DFS for breast cancer patients; and mechanistically, RHBDL2 silencing inhibited the YAP/USP31/NF-κB signaling and attenuated the spontaneous CD44^-/CD24^- cell conversion into CD44⁺/CD24^- CSCs and their mammosphere formation. Therefore, the findings may uncover new biomarkers for prognosis of delayed distant breast cancer metastasis and shed light on the molecular mechanisms underlying the distant metastasis of breast cancer, particularly for TNBC. It is necessary to further investigate whether RHBDL2 can be a therapeutic target in inhibiting the development and progression of breast cancer.

Previous studies have shown that non-CSC tumor cells can dedifferentiate (convert) into CSCs in human glioblastoma and intestinal stroma melanoma, contributing to intra- and inter-tumor heterogeneity (Stepanova et al., 2003; Tzimas et al., 2006; Wei et al., 2019). Furthermore, CSC plasticity and heterogeneity can promote tumor progression and resistance to therapy (Das et al., 2020; Kilmister et al., 2020; Martin-Castillo et al., 2013; Thankamony et al., 2020). For example, intra-tumoral heterogeneity is a major ongoing challenge for effective cancer therapy, while CSCs are responsible for intra-tumoral heterogeneity, therapeutic resistance, and metastasis, which may be because cancer cells exhibit a high level of plasticity and an ability to dynamically switch between CSC and non-CSC states or among different subsets of CSCs (Thankamony et al., 2020). Consistently, the differentiated non-CSCs can revert to be trastuzumab-refractory, CSS-like cells by enhancing their epithelial-to-mesenchymal transition process (Martin-Castillo et al., 2013). Similarly, the changes in tumor microenvironment and epigenetics can promote the conversion of non-CSCs into CSCs, leading to tumor progression and therapeutic resistance (Das et al., 2020). The current study further confirmed that CD44^-/CD24^- TNBC cells spontaneously converted into CD44⁺/CD24^- CSCs that possessed biological properties, similar to their CSCs isolated directly from parental TNBC cells. More importantly, a higher frequency of CD44^-/CD24^- cells in breast cancer tissues was associated with significantly with worse DFS and delayed distant metastasis. These, together with the dynamic process of spontaneous conversion of non-CSC CD44^-/CD24^- cells and CSC differentiation, indicate that the spontaneous conversion of CD44^-/CD24^- cells helps maintain the CSC pool size, contributing to the delayed distant metastasis and worse prognosis in breast cancer patients.

The data from the current study indicated that a higher frequency of CD44^-/CD24^- cancer cells, like higher frequency of CSCs, was one of the independent risk factors for delayed distant metastasis and worse DFS in this population. Particularly, the higher frequency of CD44^-/CD24^- cancer cells was a better predictor of delayed breast cancer metastasis (up to 12 years after initial breast cancer diagnosis). Evidently, while high frequency of CSCs was valuable for predicting distant metastasis at 5–7 years post-surgery a higher frequency of CD44^-/CD24^- cells effectively predicted delayed distant metastasis in cases with a low frequency of CSCs. These results suggest that CD44^-/CD24^- cells may spontaneously convert into CSCs and cause delayed distant metastasis when environmental and other factors cause a clonal evolution by accumulating successive mutations (Meacham and Morrison, 2013).

In addition, the RNA-seq analysis and subsequent RT-qPCR identified several DEGs during the dynamic process of CD44^-/CD24^- TNBC cell conversion into CSCs, such as RHBDL2, and its expression was up-regulated in the newly converted CSCs. The RHBDL2, also known as the rhomboid like 2, is one intramembrane serine protease of the secretase-A rhomboid family (Etheridge et al., 2013). Previous studies have identified that the RHBDL2 cleaves its substrates, including EGF, ephrin-β3, thrombomodulin, the C-type lectin family 14 member A (CLEC14A), cadherin, IL-6R, Spint-1 and the collagen receptor tyrosine kinase DDR1 (Adrain et al., 2011; Battistini et al., 2019; Etheridge et al., 2013). Although the functional consequence of cleaving these substrates has not been fully explored the available findings indicate that RHBDL2 functionally activates the EGF signaling, cadherin shedding, angiogenesis and promotes the wound healing and migration of different types of cells, including tumor cells (Adrain et al., 2011; Cheng et al., 2011). Furthermore, up-regulated RHBDL2 expression is associated with worse prognosis of several types of malignant tumors, such as breast cancer, pancreatic adenocarcinoma, clear cell kidney cancer, and low-grade glioma patient (Canzoneri et al., 2014; Johnson et al., 2017). However, little is known on how upregulated RHBDL2 expression promotes malignant behaviors of these types of tumors. Interestingly, the current study not only observed that the upregulated RHBDL2 transcription was associated with worse DFS of breast cancer patients in TCGA, but also found that RHBDL2 silencing inhibited the YAP1/NF-κB signaling by upregulating USP31 expression and attenuated the spontaneous conversion of CD44^-/CD24^- cells into CD44⁺/CD24^- CSCs and their mammosphere formation. Moreover, these data extended previous observation that YAP1-mediated suppression of USP31 enhances NF-κB activity to promote sarcomagenesis (Kemeny and Fisher, 2018; Mehta et al., 2018). YAP1 can enhance CSC stemness in several types of human cancers, and aberrant YAP1 activation is associated with a low level of TNBC differentiation and poor survival of breast cancer patients (Bora-Singhal et al., 2015; Hansen et al., 2015). Therefore, the findings from the current study indicated that RHBDL2 promoted the spontaneous conversion of CD44^-/CD24^- cells into CD44⁺/CD24^- CSCs by enhancing YAP1 stability through inhibiting its phosphorylation to suppress USP31 expression, enhancing the NF-κB signaling in breast cancer cells, whereas RHBDL2 silencing had the opposite effects in breast cancer cells (Figure 7G). Alternatively, RHBDL2 may cleave other substrates to activate the EGF receptor-mediated signaling, and modify cadherin and other molecules through their involved signal pathways to promote the spontaneous conversion of CD44^-/CD24^- cells into CD44⁺/CD24^- CSCs. Hence, these findings may provide new insights into the molecular mechanisms underlying the action of RHBDL2 in regulating the CSC pool to promote the metastasis of TNBC. Given that CSCs are crucial for breast cancer progression and metastasis, RHBDL2 may be a new therapeutic target for control of CD44^-/CD24^- cell conversion into CSCs to reduce CSC pool size and limit breast cancer progression and metastasis.

Definitely, this study had limitations. First, the sample size in some groups remained relatively smaller, which might affect statistical power. Second, this study only centered on the role of RHBDL2, but not other DEGs in regulating the spontaneous conversion of CD44^-/CD24^- TNBC cells into CD44⁺/CD24^- CSCs. The results might miss some important information on the molecular mechanisms underlying the spontaneous conversion. Furthermore, the method for identification of CD44^-/CD24^- TNBC cells did not include a positive tumor molecular marker, which might contaminate some other types of cells, although using histological tissue sections for identification of tumor cells. Thus, further studies are warranted with advanced cutting-edge technologies, such as combination of single-cell RNA-seq, multiple marker-based CyTOF mass cytometry or spatial proteomics to quantify the frequency or number of CD44^-/CD24^- breast cancer cells in fresh breast cancer tissues, together with prospectively following-up the patients, in a bigger population to investigate the prognostic value of CD44^-/CD24^- breast cancer cells and the molecular mechanisms underlying the spontaneous conversion of CD44^-/CD24^- TNBC into CD44⁺/CD24^- CSCs in the metastasis of breast cancer.

In summary, postoperative breast cancer metastasis, especially delayed metastasis (e.g. ≥5 years after diagnosis), is an important unresolved issue. The results from the current study indicated that patients with a higher frequency of CD44^-/CD24^- cells in their breast cancer tissues had a high risk to develop delayed distant metastasis 5–7 years after diagnosis. Mechanistically, CD44^-/CD24^- cells spontaneously converted into CD44⁺/CD24^- CSCs and both WT and newly converted CSCs had similar stemness properties in vitro and in vivo. RHBDL2 silencing significantly decreased YAP1 expression and increased USP31 expression to attenuate the NF-kB activation and CD44^-/CD24^- cell conversion into CSCs as well as their mammosphere formation. RHBDL2 may act as a positive regulator of the spontaneous conversion of CD44^-/CD24^- cells into CSCs by enhancing the YAP1/USP31/NF-kB signaling. Alternatively, RHBDL2 may cleave other substrates to activate the EGF and other signal pathways, enhancing the spontaneous conversion of CD44^-/CD24^- TNBC into CSCs and their metastasis. These novel findings may provide insights into the molecular process of non-CSCs conversion into CSCs, leading to breast cancer progression and delayed distant metastasis. Therefore, the current findings may uncover a novel biomarker for prognosis and therapeutic target for intervention of breast cancer, particularly for TNBC.

All data generated or analyzed during this study is included in the manuscript and supporting documents.

1. 1. Abraham BK
   2. Fritz P
   3. McClellan M
   4. Hauptvogel P
   5. Athelogou M
   6. Brauch H

   (2005)

   Prevalence of CD44+/CD24-/low cells in breast Cancer may not be associated with clinical outcome but may favor distant metastasis

   Clinical Cancer Research 11:1154–1159.

   • PubMed
   • Google Scholar
2. 1. Adrain C
   2. Strisovsky K
   3. Zettl M
   4. Hu L
   5. Lemberg MK
   6. Freeman M

   (2011) Mammalian EGF receptor activation by the rhomboid protease RHBDL2

   EMBO Reports 12:421–427.

   https://doi.org/10.1038/embor.2011.50

   • Google Scholar
3. 1. Battistini C
   2. Rehman M
   3. Avolio M
   4. Arduin A
   5. Valdembri D
   6. Serini G
   7. Tamagnone L

   (2019) Rhomboid-Like-2 Intramembrane Protease Mediates Metalloprotease-Independent Regulation of Cadherins

   International Journal of Molecular Sciences 20:5958.

   https://doi.org/10.3390/ijms20235958

   • Google Scholar
4. 1. Bianchini G
   2. Balko JM
   3. Mayer IA
   4. Sanders ME
   5. Gianni L

   (2016) Triple-negative breast cancer: challenges and opportunities of a heterogeneous disease

   Nature Reviews Clinical Oncology 13:674–690.

   https://doi.org/10.1038/nrclinonc.2016.66

   • Google Scholar
5. 1. Bora-Singhal N
   2. Nguyen J
   3. Schaal C
   4. Perumal D
   5. Singh S
   6. Coppola D
   7. Chellappan S

   (2015) YAP1 regulates OCT4 activity and SOX2 expression to facilitate Self-Renewal and vascular mimicry of Stem-Like cells

   Stem Cells 33:1705–1718.

   https://doi.org/10.1002/stem.1993

   • PubMed
   • Google Scholar
6. 1. Bray F
   2. Ferlay J
   3. Soerjomataram I
   4. Siegel RL
   5. Torre LA
   6. Jemal A

   (2018) Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries

   CA: A Cancer Journal for Clinicians 68:394–424.

   https://doi.org/10.3322/caac.21492

   • Google Scholar
7. 1. Burstein HJ
   2. Temin S
   3. Anderson H
   4. Buchholz TA
   5. Davidson NE
   6. Gelmon KE
   7. Giordano SH
   8. Hudis CA
   9. Rowden D
   10. Solky AJ
   11. Stearns V
   12. Winer EP
   13. Griggs JJ

   (2014) Adjuvant endocrine therapy for women with hormone receptor-positive breast Cancer: american society of clinical oncology clinical practice guideline focused update

   Journal of Clinical Oncology 32:2255–2269.

   https://doi.org/10.1200/JCO.2013.54.2258

   • PubMed
   • Google Scholar
8. 1. Canzoneri R
   2. Lacunza E
   3. Isla Larrain M
   4. Croce MV
   5. Abba MC

   (2014) Rhomboid family gene expression profiling in breast normal tissue and tumor samples

   Tumor Biology 35:1451–1458.

   https://doi.org/10.1007/s13277-013-1199-8

   • Google Scholar
9. 1. Chaffer CL
   2. Brueckmann I
   3. Scheel C
   4. Kaestli AJ
   5. Wiggins PA
   6. Rodrigues LO
   7. Brooks M
   8. Reinhardt F
   9. Su Y
   10. Polyak K
   11. Arendt LM
   12. Kuperwasser C
   13. Bierie B
   14. Weinberg RA

   (2011) Normal and neoplastic nonstem cells can spontaneously convert to a stem-like state

   PNAS 108:7950–7955.

   https://doi.org/10.1073/pnas.1102454108

   • Google Scholar
10. 1. Cheng T-L
    2. Wu Y-T
    3. Lin H-Y
    4. Hsu F-C
    5. Liu S-K
    6. Chang B-I
    7. Chen W-S
    8. Lai C-H
    9. Shi G-Y
    10. Wu H-L

    (2011) Functions of Rhomboid Family Protease RHBDL2 and Thrombomodulin in Wound Healing

    Journal of Investigative Dermatology 131:2486–2494.

    https://doi.org/10.1038/jid.2011.230

    • Google Scholar
11. 1. Das PK
    2. Pillai S
    3. Rakib MA
    4. Khanam JA
    5. Gopalan V
    6. Lam AKY
    7. Islam F

    (2020) Plasticity of Cancer stem cell: origin and role in disease progression and therapy resistance

    Stem Cell Reviews and Reports 16:397–412.

    https://doi.org/10.1007/s12015-019-09942-y

    • PubMed
    • Google Scholar
12. 1. De Angelis ML
    2. Francescangeli F
    3. Zeuner A

    (2019) Breast Cancer stem cells as drivers of tumor chemoresistance, dormancy and relapse: new challenges and therapeutic opportunities

    Cancers 11:1569.

    https://doi.org/10.3390/cancers11101569

    • Google Scholar
13. 1. Drabsch Y
    2. ten Dijke P

    (2011) TGF-β Signaling in Breast Cancer Cell Invasion and Bone Metastasis

    Journal of Mammary Gland Biology and Neoplasia 16:97–108.

    https://doi.org/10.1007/s10911-011-9217-1

    • Google Scholar
14. 1. Early Breast Cancer Trialists' Collaborative G

    (2015) Aromatase inhibitors versus tamoxifen in early breast cancer: patient-level meta-analysis of the randomised trials

    The Lancet 386:1341–1352.

    https://doi.org/10.1016/S0140-6736(15)61074-1

    • Google Scholar
15. 1. Edge SB
    2. Compton CC

    (2010) The american joint committee on Cancer: the 7th edition of the AJCC Cancer staging manual and the future of TNM

    Annals of Surgical Oncology 17:1471–1474.

    https://doi.org/10.1245/s10434-010-0985-4

    • PubMed
    • Google Scholar
16. 1. Etheridge SL
    2. Brooke MA
    3. Kelsell DP
    4. Blaydon DC

    (2013) Rhomboid proteins: a role in keratinocyte proliferation and cancer

    Cell and Tissue Research 351:301–307.

    https://doi.org/10.1007/s00441-012-1542-1

    • Google Scholar
17. 1. Geng S-Q
    2. Alexandrou AT
    3. Li JJ

    (2014) Breast cancer stem cells: Multiple capacities in tumor metastasis

    Cancer Letters 349:1–7.

    https://doi.org/10.1016/j.canlet.2014.03.036

    • Google Scholar
18. 1. Gruber R
    2. Panayiotou R
    3. Nye E
    4. Spencer-Dene B
    5. Stamp G
    6. Behrens A

    (2016) YAP1 and TAZ control pancreatic Cancer initiation in mice by direct Up-regulation of JAK-STAT3 signaling

    Gastroenterology 151:526–539.

    https://doi.org/10.1053/j.gastro.2016.05.006

    • PubMed
    • Google Scholar
19. 1. Hansen CG
    2. Moroishi T
    3. Guan K-L

    (2015) YAP and TAZ: a nexus for Hippo signaling and beyond

    Trends in Cell Biology 25:499–513.

    https://doi.org/10.1016/j.tcb.2015.05.002

    • Google Scholar
20. 1. Iliopoulos D
    2. Hirsch HA
    3. Wang G
    4. Struhl K

    (2011) Inducible formation of breast cancer stem cells and their dynamic equilibrium with non-stem cancer cells via IL6 secretion

    PNAS 108:1397–1402.

    https://doi.org/10.1073/pnas.1018898108

    • Google Scholar
21. 1. Italiano JE
    2. Shivdasani RA

    (2003) Megakaryocytes and beyond: the birth of platelets

    Journal of Thrombosis and Haemostasis 1:1174–1182.

    https://doi.org/10.1046/j.1538-7836.2003.00290.x

    • Google Scholar
22. 1. Johnson N
    2. Březinová J
    3. Stephens E
    4. Burbridge E
    5. Freeman M
    6. Adrain C
    7. Strisovsky K

    (2017) Quantitative proteomics screen identifies a substrate repertoire of rhomboid protease RHBDL2 in human cells and implicates it in epithelial homeostasis

    Scientific Reports 7:7283.

    https://doi.org/10.1038/s41598-017-07556-3

    • Google Scholar
23. 1. Kaverina N
    2. Borovjagin AV
    3. Kadagidze Z
    4. Baryshnikov A
    5. Baryshnikova M
    6. Malin D
    7. Ghosh D
    8. Shah N
    9. Welch DR
    10. Gabikian P
    11. Karseladze A
    12. Cobbs C
    13. Ulasov IV

    (2017) Astrocytes promote progression of breast cancer metastases to the brain via a KISS1-mediated autophagy

    Autophagy 13:1905–1923.

    https://doi.org/10.1080/15548627.2017.1360466

    • Google Scholar
24. 1. Kemeny LV
    2. Fisher DE

    (2018) Targeting the (Un)differentiated State of Cancer

    Cancer Cell 33:793–795.

    https://doi.org/10.1016/j.ccell.2018.04.007

    • Google Scholar
25. 1. Khan F
    2. Amatya B
    3. Ng L
    4. Demetrios M
    5. Zhang NY
    6. Turner-Stokes L

    (2012) Multidisciplinary rehabilitation for follow-up of women treated for breast Cancer

    The Cochrane Database of Systematic Reviews 12:CD009553.

    https://doi.org/10.1002/14651858.CD009553.pub2

    • PubMed
    • Google Scholar
26. 1. Kilmister EJ
    2. Patel J
    3. van Schaijik B
    4. Bockett N
    5. Brasch HD
    6. Paterson E
    7. Sim D
    8. Davis PF
    9. Roth IM
    10. Itinteang T
    11. Tan ST

    (2020) Cancer stem cell subpopulations are present within metastatic head and neck cutaneous squamous cell carcinoma

    Frontiers in Oncology 10:1091.

    https://doi.org/10.3389/fonc.2020.01091

    • PubMed
    • Google Scholar
27. 1. Kim T
    2. Yang S-J
    3. Hwang D
    4. Song J
    5. Kim M
    6. Kyum Kim S
    7. Kang K
    8. Ahn J
    9. Lee D
    10. Kim M
    11. Kim S
    12. Seung Koo J
    13. Seok Koh S
    14. Kim S-Y
    15. Lim D-S

    (2015) A basal-like breast cancer-specific role for SRF–IL6 in YAP-induced cancer stemness

    Nature Communications 6:10186.

    https://doi.org/10.1038/ncomms10186

    • Google Scholar
28. 1. Klein CA

    (2008) CANCER: The Metastasis Cascade

    Science 321:1785–1787.

    https://doi.org/10.1126/science.1164853

    • Google Scholar
29. 1. Langley RR
    2. Fidler IJ

    (2007) Tumor cell-organ microenvironment interactions in the pathogenesis of Cancer metastasis

    Endocrine Reviews 28:297–321.

    https://doi.org/10.1210/er.2006-0027

    • PubMed
    • Google Scholar
30. 1. Li X
    2. Wan L
    3. Shen H
    4. Geng J
    5. Nie J
    6. Wang G
    7. Jia N
    8. Dai M
    9. Bai X

    (2012) Thyroid transcription factor-1 amplification and expressions in lung adenocarcinoma tissues and pleural effusions predict patient survival and prognosis

    Journal of Thoracic Oncology 7:76–84.

    https://doi.org/10.1097/JTO.0b013e318232b98a

    • PubMed
    • Google Scholar
31. 1. Liu H
    2. Patel MR
    3. Prescher JA
    4. Patsialou A
    5. Qian D
    6. Lin J
    7. Wen S
    8. Chang Y-F
    9. Bachmann MH
    10. Shimono Y
    11. Dalerba P
    12. Adorno M
    13. Lobo N
    14. Bueno J
    15. Dirbas FM
    16. Goswami S
    17. Somlo G
    18. Condeelis J
    19. Contag CH
    20. Gambhir SS
    21. Clarke MF

    (2010) Cancer stem cells from human breast tumors are involved in spontaneous metastases in orthotopic mouse models

    PNAS 107:18115–18120.

    https://doi.org/10.1073/pnas.1006732107

    • Google Scholar
32. 1. Martin-Castillo B
    2. Oliveras-Ferraros C
    3. Vazquez-Martin A
    4. Cufí S
    5. Moreno JM
    6. Corominas-Faja B
    7. Urruticoechea A
    8. Martín ÁG
    9. López-Bonet E
    10. Menendez JA

    (2013) Basal/HER2 breast carcinomas: integrating molecular taxonomy with Cancer stem cell dynamics to predict primary resistance to trastuzumab (Herceptin)

    Cell Cycle 12:225–245.

    https://doi.org/10.4161/cc.23274

    • PubMed
    • Google Scholar
33. 1. Meacham CE
    2. Morrison SJ

    (2013) Tumour heterogeneity and cancer cell plasticity

    Nature 501:328–337.

    https://doi.org/10.1038/nature12624

    • Google Scholar
34. 1. Mehta A
    2. Kim YJ
    3. Robert L
    4. Tsoi J
    5. Comin-Anduix B
    6. Berent-Maoz B
    7. Cochran AJ
    8. Economou JS
    9. Tumeh PC
    10. Puig-Saus C
    11. Ribas A

    (2018) Immunotherapy resistance by Inflammation-Induced dedifferentiation

    Cancer Discovery 8:935–943.

    https://doi.org/10.1158/2159-8290.CD-17-1178

    • PubMed
    • Google Scholar
35. 1. Mylona E
    2. Giannopoulou I
    3. Fasomytakis E
    4. Nomikos A
    5. Magkou C
    6. Bakarakos P
    7. Nakopoulou L

    (2008) The clinicopathologic and prognostic significance of CD44+/CD24−/low and CD44−/CD24+ tumor cells in invasive breast carcinomas

    Human Pathology 39:1096–1102.

    https://doi.org/10.1016/j.humpath.2007.12.003

    • Google Scholar
36. 1. Najafi M
    2. Farhood B
    3. Mortezaee K

    (2019) Cancer stem cells (CSCs) in cancer progression and therapy

    Journal of Cellular Physiology 234:8381–8395.

    https://doi.org/10.1002/jcp.27740

    • Google Scholar
37. 1. Nishimura R
    2. Osako T
    3. Nishiyama Y
    4. Tashima R
    5. Nakano M
    6. Fujisue M
    7. Toyozumi Y
    8. Arima N

    (2013) Evaluation of factors related to late recurrence--later than 10 years after the initial treatment--in primary breast Cancer

    Oncology 85:100–110.

    https://doi.org/10.1159/000353099

    • PubMed
    • Google Scholar
38. 1. O'Conor CJ
    2. Chen T
    3. González I
    4. Cao D
    5. Peng Y

    (2018) Cancer stem cells in triple-negative breast Cancer: a potential target and prognostic marker

    Biomarkers in Medicine 12:813–820.

    https://doi.org/10.2217/bmm-2017-0398

    • PubMed
    • Google Scholar
39. 1. Saini KS
    2. Taylor C
    3. Ramirez AJ
    4. Palmieri C
    5. Gunnarsson U
    6. Schmoll HJ
    7. Dolci SM
    8. Ghenne C
    9. Metzger-Filho O
    10. Skrzypski M
    11. Paesmans M
    12. Ameye L
    13. Piccart-Gebhart MJ
    14. de Azambuja E

    (2012) Role of the multidisciplinary team in breast Cancer management: results from a large international survey involving 39 countries

    Annals of Oncology 23:853–859.

    https://doi.org/10.1093/annonc/mdr352

    • PubMed
    • Google Scholar
40. 1. Scott J
    2. Kuhn P
    3. Anderson ARA

    (2012) Unifying metastasis — integrating intravasation, circulation and end-organ colonization

    Nature Reviews Cancer 12:445–446.

    https://doi.org/10.1038/nrc3287

    • Google Scholar
41. 1. Sharma P

    (2018) Update on the Treatment of Early-Stage Triple-Negative Breast Cancer

    Current Treatment Options in Oncology 19:22.

    https://doi.org/10.1007/s11864-018-0539-8

    • Google Scholar
42. 1. Siegel RL
    2. Miller KD
    3. Jemal A

    (2020) Cancer statistics, 2020

    CA: A Cancer Journal for Clinicians 70:7–30.

    https://doi.org/10.3322/caac.21590

    • Google Scholar
43. 1. Stepanova T
    2. Slemmer J
    3. Hoogenraad CC
    4. Lansbergen G
    5. Dortland B
    6. De Zeeuw CI
    7. Grosveld F
    8. van Cappellen G
    9. Akhmanova A
    10. Galjart N

    (2003) Visualization of microtubule growth in cultured neurons via the use of EB3-GFP (end-binding protein 3-green fluorescent protein)

    The Journal of Neuroscience 23:2655–2664.

    https://doi.org/10.1523/JNEUROSCI.23-07-02655.2003

    • PubMed
    • Google Scholar
44. 1. Syn N
    2. Wang L
    3. Sethi G
    4. Thiery JP
    5. Goh BC

    (2016) Exosome-Mediated metastasis: from Epithelial-Mesenchymal transition to escape from immunosurveillance

    Trends in Pharmacological Sciences 37:606–617.

    https://doi.org/10.1016/j.tips.2016.04.006

    • PubMed
    • Google Scholar
45. 1. Tan PH
    2. Schnitt SJ
    3. van de Vijver MJ
    4. Ellis IO
    5. Lakhani SR

    (2015) Papillary and neuroendocrine breast lesions: the WHO stance

    Histopathology 66:761–770.

    https://doi.org/10.1111/his.12463

    • Google Scholar
46. 1. Thankamony AP
    2. Saxena K
    3. Murali R
    4. Jolly MK
    5. Nair R

    (2020) Cancer stem cell plasticity - A deadly deal

    Frontiers in Molecular Biosciences 7:79.

    https://doi.org/10.3389/fmolb.2020.00079

    • PubMed
    • Google Scholar
47. 1. Tzimas C
    2. Michailidou G
    3. Arsenakis M
    4. Kieff E
    5. Mosialos G
    6. Hatzivassiliou EG

    (2006) Human ubiquitin specific protease 31 is a deubiquitinating enzyme implicated in activation of nuclear factor-kappaB

    Cellular Signalling 18:83–92.

    https://doi.org/10.1016/j.cellsig.2005.03.017

    • PubMed
    • Google Scholar
48. 1. Velasco-Velázquez MA
    2. Popov VM
    3. Lisanti MP
    4. Pestell RG

    (2011) The role of breast Cancer stem cells in metastasis and therapeutic implications

    The American Journal of Pathology 179:2–11.

    https://doi.org/10.1016/j.ajpath.2011.03.005

    • Google Scholar
49. 1. Wang D
    2. Lu P
    3. Zhang H
    4. Luo M
    5. Zhang X
    6. Wei X
    7. Gao J
    8. Zhao Z
    9. Liu C

    (2014) Oct-4 and Nanog promote the epithelial-mesenchymal transition of breast cancer stem cells and are associated with poor prognosis in breast cancer patients

    Oncotarget 5:10803–10815.

    https://doi.org/10.18632/oncotarget.2506

    • Google Scholar
50. 1. Wei M
    2. Zhu Z
    3. Wu J
    4. Wang Y
    5. Geng J
    6. Qin Z-H

    (2019) DRAM1 deficiency affects the organization and function of the Golgi apparatus

    Cellular Signalling 63:109375.

    https://doi.org/10.1016/j.cellsig.2019.109375

    • Google Scholar
51. 1. Ye S
    2. Lawlor MA
    3. Rivera-Reyes A
    4. Egolf S
    5. Chor S
    6. Pak K
    7. Ciotti GE
    8. Lee AC
    9. Marino GE
    10. Shah J
    11. Niedzwicki D
    12. Weber K
    13. Park PMC
    14. Alam MZ
    15. Grazioli A
    16. Haldar M
    17. Xu M
    18. Perry JA
    19. Qi J
    20. Eisinger-Mathason TSK

    (2018) YAP1-Mediated suppression of USP31 enhances nfκb activity to promote sarcomagenesis

    Cancer Research 78:2705–2720.

    https://doi.org/10.1158/0008-5472.CAN-17-4052

    • PubMed
    • Google Scholar
52. 1. Zoppino FC
    2. Militello RD
    3. Slavin I
    4. Alvarez C
    5. Colombo MI

    (2010) Autophagosome formation depends on the small GTPase Rab1 and functional ER exit sites

    Traffic 11:1246–1261.

    https://doi.org/10.1111/j.1600-0854.2010.01086.x

    • PubMed
    • Google Scholar

Acceptance summary:

The manuscript describes a possible link between the presence of CD44-/CD24- cells and post-operative site distant metastasis. The functional significance of CSC, particularly of the phenotype CD44-/CD24- remains elusive and is worthy of careful evaluation. The strength of the study is the inclusion of a large number of patient samples, which were used in a retrospective analysis. The resulting data provide the rationale for the application of contemporary technologies in the evaluation of CDX and PDX models. As such, this work paves the way for further investigation at a mechanistic and translational level, which may lead to a better understanding of breast cancer metastasis.

Decision letter after peer review:

Thank you for submitting your article "Association of CD44 -/CD24 -Breast Cancer Cells with Late Stage Tumor Recurrence" for consideration by eLife. Your article has been reviewed by 2 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Mone Zaidi as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Robin Anderson (Reviewer #1).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission. Essential revisions:

1) Referees comments must be addressed in a convincing manner, with additional data.

2) Edits are required in all sections of the manuscript.

3) If extensively revised, this manuscript would represent an incremental contribution in the field of breast cancer biology.

Reviewer #1:

The authors set out to demonstrate the importance of a subset of cancer cells lacking CD44 and CD24, in promoting tumor progression, metastasis and resistance to therapy. They analyse archival breast tissues from patients where the clinical outcome – development of metastasis and therapy response – are known. They then take this information into cell culture experiments, seeking the mechanism that underlies the importance of these CD44-/CD24- cells. Their major premise is that these double negative cells can de-differentiate into cancer stem cells that subsequently promote metastasis and chemotherapy resistance.

The overall concept of this study is good and the cell-based results are mainly clear. However, the archival tissue analysis needs clarification, as does the RNAseq analysis, as detailed below.

The breast cancer tissues are scored by immunofluorescence for expression of CD44 and CD24, but there is no indication that these two markers are being scored only in tumor cells rather than in all lineages within the tumor microenvironment. This is an important point that must be clarified. Also, the RNAseq analysis is reported to be conducted on single cells, but the data presented do not indicate if this is actually the case and how many replicate samples were analysed and which sample yielded each heat map pattern. Again clarification is required before the data can be interpreted.

Finally, the conclusion from this study is that RHBDL2 induces spontaneous conversion of CD44-/CD24- cells into CD44+/CD24- cells via YAP1, USP31 and decreased NFkB signaling. This may well be the case, but has not been proven by the experiments presented here. Whilst DHBDL2 does alter YAP, USP31 and NFkB, it could also be working through an alternate pathway to induce CSCs.

1. Figure 1 and S1 and S2A score the numbers of samples with the varying CD44/CD24 protein levels. For Figure S1, this is shown by immunofluorescence, but there is no indication of whether the cells being analysed are tumor cells or host cells from the microenvironment. There should be co-staining for EpCAM or CK19 to confirm that tumor cells only are being scored. I am assuming that the data shown in Figure S2A are actually from the immunostaining and not from flow cytometry analysis as stated in the figure legend? If this was a flow analysis, there is no description of the origin or processing of the samples reported in Figure 1A for flow cytometry. The legend for Figure S2A states that the data are stratifying only metastatic tumors, but in Figure 1A, apparently many of the tumors showed no metastasis. All these points need to be clarified/corrected.

2. This first section of the manuscript is poorly presented and I am struggling to follow the arguments and data being provided. Many of the figures are mis-labelled in the text and they are not presented in consecutive order as discussed in the text. For some references to data in the text where the wrong figure has been cited, I cannot even find the correct figure.

3. Possibly the way the data are presented, with some figures referring to the training set and other figures to the test set is adding to the confusion when it is not clearly stated which set is under discussion at the time.

4. Late metastasis – better phrased as delayed metastasis.

5. The sentence in the Introduction: "Previous studies have shown that CD44+/CD24- breast CSCs might be a dominant factor in relapse of triple negative breast cancer (TNBC), due to their possession of potent self-renewal and differentiation capacities to differentiate into mature CD44-/CD24-, CD44+/CD24+, and CD44-/CD24+ cancer cells (Geng et al., 2014; Wang et al., 2014)". It is not at all clear to me that the references provided support this comment. Geng et al. do not mention CD44-/CD24- cells at all, and Wang et al. do not discuss differentiation of CD44+/CD24- into the other phenotypes. Only CD44- cells are mentioned, not CD24.

6. Sentence in Introduction: "Indeed, injection of up to 1000 breast CSCs was able to generate a solid tumor mass in immunocompromised mice (Chaffer et al., 2011; Iliopoulos et al., 2011)." Please clarify. It is not obvious to me that Chaffer et al. specified how many CSCs they injected and Iliopoulos et al. showed that as few as 50 CSCs could form tumors.

7. Sentence in Introduction: "Moreover, previous studies reported that non-CSCs, such as CD44-/CD24- TNBC cells, are able to spontaneously convert into CSCs to renew the CSC pool, resulting in chemoresistance (Gruber et al., 2016; Kim et al., 2015; Ye et al., 2018)." The statement being referenced is for non-CSCs in TNBC, but Gruber et al. and Ye et al. are not reporting on TNBC.

8. Figure 1B – please explain more fully how the cut-off of 19.5% CD44-/CD24- was determined.

9. Figure 1C – please explain how the ratios of different subtypes of breast cancer with high or low CD44-/CD24- were calculated. According to the figure, for luminal cancers, there was a metastasis rate of ~40% in the high group and less then 10% in the low group. How were the numbers of 63.1% and 32.6% derived from Figure 1C?

Then the same figures are provided for Figure 2A, when using the test set. Were the data in Figure 1C from the training set? Why are the final metastasis rates exactly the same? This needs a better explanation.

10. Page 6: reference to Figure S3C should be Figure S2C?

11. Page 7 – Should Figure S5A,B actually be Figure S3?

12. Page 7: The statement: "Moreover, a low CSC percentage led to different risks for developing tumor metastasis 5 years after diagnosis and adjuvant chemotherapy between the C1 and C0 patients (Figure S3D)." This does not appear to apply to Figure S3D and I cannot tell to which figure this statement is referring.

13. Page 7, last sentence: "In the present study, we first designed the experiments illustrated in Figure 3A to characterize the percentages of different cell subtypes among MDA-MB-231 cells using cell culture and flow cytometric cell sorting of parental MDA-MB-468 cells." If I understand correctly, this should re-worded as follows: "In the present study, we first designed the experiments illustrated in Figure 3A and Figure S5 to characterize the percentages of different cell subtypes among MDA-MB-231 and MDA-MB-468 cells using cell culture and flow cytometric cell sorting of parental cells.

14. Figure 3C raises questions about reproducibility of the data. The profile in Figure 3C after 7 days in culture looks very different from that shown in Figure 3B.

15. Figure 5C: please explain how you obtained a weight measurement for the lung metastases. You record lung metastasis weights of up to 2 grams. Lungs typically weight about 0.2 grams. Lungs of 2 grams seem improbable, let alone metastases in lungs weighing 2 grams. Also, tumor sizes of 700 mm2 do not seem possible in lungs.

16. The RNAseq analysis requires more explanation. In Figure 6B, please explain the three columns shown in the heat map. Without their designation, the rest of this figure is hard to interpret. It seems that only one single cell per timepoint has been analysed. Is this correct? How reliable is sequencing from a single cell? Or was the analysis from many cells grown up from a single cell? How many single cells (or derivative of single cells) were analysed at each time point? There is a discrepancy between the protocol described in the text and that provided in the Methods section, when describing the RNAseq analysis. It is difficult to relate the significance of the genes shown in Panel B to the Kaplan-Meier data when it is not clear what the three columns in panel B represent.

17. Figure 6D: I assume that the "*" and "#" are meant to convey some sort of significance, but they do not appear for any of the genes assessed by RT-PCR. Does that mean that none of the changes were significant?

18. Why does the legend for Figure S6 talk about SILAC for protein analysis? Was SILAC run on these samples? Impossible on single cells. No mention of SILAC in Methods.

19. Page 10: Explain why YAP was selected for analysis following knockdown of RHBDL2. Was there a previously known connection between these two genes? For the CD44-/CD24- cells in Figure 7B, was YAP phosphorylation also altered? Please show a western demonstrating that RHBDL2 is reduced after transfection with siRHBDL2.

20. Figure 7D: text says 0.7% CD44+/CD24- but figure says 0.5% and vice versa for the MDA-MB-468 cells.

21. The text on page 11 needs re-writing. The second paragraph, commencing with "We next selected…" is a repeat of the text above where you have already done this step and analysed the cells by flow cytometry.

Reviewer #2:

The manuscript describes a possible link between the presence of CD44-/CD24- cells and post-operative site distant metastasis. This is not a new concept, yet the actual biological function of CSC, particularly of the phenotype CD44-/CD24- remains inconclusive. The primary strength of the study is the large number of patient samples used in the retrospective analysis that provides the rationale for in vitro experiments using cultured cancer cells. Several technical and methodological limitations were identified.

The description of the patient cohort is very elusive. The bioinformatic analysis lacks in rigor as the methodology used to identify and quantitate CD44-/CD24- cells in patients is not properly described. For instance, Figure legends describe the use of flow cytometry for quantification of CD44-/CD24- negative cells from patient samples. Yet only an immunofluorescence staining is described in Results. No Methods description is given to FACS of patient samples. The manuscript seems focused on TNBC, yet the reader has a difficult time understanding how many samples are actually from TNBC patients. Two TNBC established cell lines are mentioned in the manuscript but no rationale is provided as to why some experiments are performed only with MDA-MB-468 and some with MDA-MB-231. The same is true for several other experiments. The extensive molecular description of CD44-/CD24- cell dedifferentiation into CSCs deserves distinction but will need to be confirmed in patient samples before conclusions can be drawn with regards to the biological significance of these findings to breast cancer patients and the medical community at large.

Clarify the methodology used to quantitate CD44-/CD24- in patient samples, as well as the bioinformatic analysis supporting the current hypothesis.

Provide a detailed analysis of clinical and pathologic information of the patient cohort, including histology.

Several references are 10 years old or older, new references supporting the authors finds would strength the hypothesis.

Reviewer #1:

The authors set out to demonstrate the importance of a subset of cancer cells lacking CD44 and CD24, in promoting tumor progression, metastasis and resistance to therapy. They analyse archival breast tissues from patients where the clinical outcome – development of metastasis and therapy response – are known. They then take this information into cell culture experiments, seeking the mechanism that underlies the importance of these CD44-/CD24- cells. Their major premise is that these double negative cells can de-differentiate into cancer stem cells that subsequently promote metastasis and chemotherapy resistance.

The overall concept of this study is good and the cell-based results are mainly clear. However, the archival tissue analysis needs clarification, as does the RNAseq analysis, as detailed below.

We thank the reviewer for his/her positive comments. We also recognize the limitations of our current study and have fully addressed the reviewers’ concerns by clarifying the analysis methods in the revision.

The breast cancer tissues are scored by immunofluorescence for expression of CD44 and CD24, but there is no indication that these two markers are being scored only in tumor cells rather than in all lineages within the tumor microenvironment. This is an important point that must be clarified. Also, the RNAseq analysis is reported to be conducted on single cells, but the data presented do not indicate if this is actually the case and how many replicate samples were analysed and which sample yielded each heat map pattern. Again clarification is required before the data can be interpreted.

We are sorry for the confusion. Actually, we analyzed the frequency of CD44-CD24- breast cancer cells in breast cancer tissue sections, but not stromal, infiltrates and other non-tumor cells, identified in the H&E stained corresponding series of breast cancer sections. Now, we have clarified it in the revision and provided the relevant H&E stained section in the supplementary Figure x. Please see our answers regarding the RNAseq method in question 12.

Finally, the conclusion from this study is that RHBDL2 induces spontaneous conversion of CD44-/CD24- cells into CD44+/CD24- cells via YAP1, USP31 and decreased NFkB signaling. This may well be the case, but has not been proven by the experiments presented here. Whilst DHBDL2 does alter YAP, USP31 and NFkB, it could also be working through an alternate pathway to induce CSCs.

We fully understand his/her concern. It is true that RHBDL2 acts as an intramembrane serine protease that can cleave and activate other substrates, including ephrin B and others, contributing to the development and progression of malignant tumors. Actually, ephrin B has been shown to function as a prognostic factor of uterine cervical (PMID 25205602 ). Unfortunately, we are unable to perform more experiments to test alternative mechanisms underlying the action of RHBDL2 in promoting distant metastasis of breast cancer due to the limited resources. In response to his/her concern, we have modified the conclusion of this study in a conservative manner and briefly discussed the potential alternative mechanisms underlying the action of RHBDL2.

1. Figure 1 and S1 and S2A score the numbers of samples with the varying CD44/CD24 protein levels. For Figure S1, this is shown by immunofluorescence, but there is no indication of whether the cells being analysed are tumor cells or host cells from the microenvironment. There should be co-staining for EpCAM or CK19 to confirm that tumor cells only are being scored. I am assuming that the data shown in Figure S2A are actually from the immunostaining and not from flow cytometry analysis as stated in the figure legend? If this was a flow analysis, there is no description of the origin or processing of the samples reported in Figure 1A for flow cytometry. The legend for Figure S2A states that the data are stratifying only metastatic tumors, but in Figure 1A, apparently many of the tumors showed no metastasis. All these points need to be clarified/corrected.

We fully understand his/her concern. In this study, we performed H&E staining and immunofluorescence simultaneously in a series of consecutive tissue sections. With H&E stained sections, we identified tumor cells and other types of non-tumor cells in the tumor regions. Subsequently, we analyzed the numbers of CD44-CD24- tumor cells in each section as described in the method section to determine the frequency of CD44-CD24- cells in total tumor cells. We have provided the relevant H&E stained sections in the revision and clarified the data in Figure S2A(now is Figure 1—figure supplement 2) by revising the figure legend in the revision. We are sorry for our carelessness to state wrong method in the Figure S2(now is Figure 1—figure supplement 2) legend

2. This first section of the manuscript is poorly presented and I am struggling to follow the arguments and data being provided. Many of the figures are mis-labelled in the text and they are not presented in consecutive order as discussed in the text. For some references to data in the text where the wrong figure has been cited, I cannot even find the correct figure.

We fully understand the reviewer’s concern and have corrected the figure citations accordingly.

3. Possibly the way the data are presented, with some figures referring to the training set and other figures to the test set is adding to the confusion when it is not clearly stated which set is under discussion at the time.

We appreciate his/her advice. Now, we have specified the figures and their descriptions in the revision, accordingly.

4. Late metastasis – better phrased as delayed metastasis.

We appreciate his/her advice. Now, we have changed the “late” into “delayed” metastasis in the revision.

5. The sentence in the Introduction: "Previous studies have shown that CD44+/CD24- breast CSCs might be a dominant factor in relapse of triple negative breast cancer (TNBC), due to their possession of potent self-renewal and differentiation capacities to differentiate into mature CD44-/CD24-, CD44+/CD24+, and CD44-/CD24+ cancer cells (Geng et al., 2014; Wang et al., 2014)". It is not at all clear to me that the references provided support this comment. Geng et al. do not mention CD44-/CD24- cells at all, and Wang et al. do not discuss differentiation of CD44+/CD24- into the other phenotypes. Only CD44- cells are mentioned, not CD24.

We thank the reviewer for his/her careful review. To avoid potential overstatement, we have modified the sentence into a general concept by deleting specific differential phenotypes in the revision.

6. Sentence in Introduction: "Indeed, injection of up to 1000 breast CSCs was able to generate a solid tumor mass in immunocompromised mice (Chaffer et al., 2011; Iliopoulos et al., 2011)." Please clarify. It is not obvious to me that Chaffer et al. specified how many CSCs they injected and Iliopoulos et al. showed that as few as 50 CSCs could form tumors.

We thank the reviewer for his/her careful review. We have re-checked the references and found that Chaffer et al. described that injection of up to 50 breast CSCs was able to generate a solid tumor mass in immunocompromised mice. Accordingly, we would like to keep the sentence as is.

7. Sentence in Introduction: "Moreover, previous studies reported that non-CSCs, such as CD44-/CD24- TNBC cells, are able to spontaneously convert into CSCs to renew the CSC pool, resulting in chemoresistance (Gruber et al., 2016; Kim et al., 2015; Ye et al., 2018)." The statement being referenced is for non-CSCs in TNBC, but Gruber et al. and Ye et al. are not reporting on TNBC.

We thank the reviewer for his/her careful review. Now, we have revised the text by changing “CD44-/CD24- TNBC cells” to “CD44-/CD24- breast cancer cells” to reflect the findings in the references.

8. Figure 1B – please explain more fully how the cut-off of 19.5% CD44-/CD24- was determined.

We thank the reviewer for his/her advice. the average frequency of CD44-/CD24- cancer cells in all samples was 19.7% and the median was 19.5%. The receiver-operating characteristic curve analysis also showed a decision threshold of 19.5% CD44^-/CD24^- cancer cells; thus, we used this cut-off point to perform a subgroup analysis.

9. Figure 1C – please explain how the ratios of different subtypes of breast cancer with high or low CD44-/CD24- were calculated. According to the figure, for luminal cancers, there was a metastasis rate of ~40% in the high group and less than 10% in the low group. How were the numbers of 63.1% and 32.6% derived from Figure 1C?

Figure 1C describes the effect of CD44-/CD24- levels on the metastasis rates among three molecular subtypes of breast cancer, i.e., 63.79% vs. 22.22% (high vs. low CD44-/CD24- tumor cells in TNBC), 40.85% vs. 6.82% (luminal), and 47.06% vs. 13.64% (HER-2).

Then the same figures are provided for Figure 2A, when using the test set. Were the data in Figure 1C from the training set? Why are the final metastasis rates exactly the same? This needs a better explanation.

Figure 2A contains an error in the description and has been corrected to luminal: 22.73% vs. 8.89%; HER-2: 50% vs. 18.42%; TNBC: 41.67% vs. 15% for high vs. low CD44-/CD24- cells.

10. Page 6: reference to Figure S3C should be Figure S2C?

We thank the reviewer for their careful review and have corrected this figure citation.

11. Page 7 – Should Figure S5A,B actually be Figure S3?

We thank the reviewer for their careful review and have corrected the figure citation of Figure S5A,B to Figure S4A,B(now is Figure 2—figure supplement 1).

12. Page 7: The statement: "Moreover, a low CSC percentage led to different risks for developing tumor metastasis 5 years after diagnosis and adjuvant chemotherapy between the C1 and C0 patients (Figure S3D)." This does not appear to apply to Figure S3D and I cannot tell to which figure this statement is referring.

We are sorry for the confusion. The correct description should be that we stratified patients with <2% CD44+/CD24- and >19.5% CD44-/CD24- tumor cells in the C1 group while those with <2% CD44+/CD24- and <19.5% CD44-/CD24- tumor cells in the C0 group. We have clarified it in the text and figure legend accordingly.

13. Page 7, last sentence: "In the present study, we first designed the experiments illustrated in Figure 3A to characterize the percentages of different cell subtypes among MDA-MB-231 cells using cell culture and flow cytometric cell sorting of parental MDA-MB-468 cells." If I understand correctly, this should re-worded as follows: "In the present study, we first designed the experiments illustrated in Figure 3A and Figure S5 to characterize the percentages of different cell subtypes among MDA-MB-231 and MDA-MB-468 cells using cell culture and flow cytometric cell sorting of parental cells.

We thank the reviewer for his/her advice. Now, we have revised it in the manuscript accordingly. The data in Figure 3A, B have been updated with data of primary cells from breast cancer patients.

14. Figure 3C raises questions about reproducibility of the data. The profile in Figure 3C after 7 days in culture looks very different from that shown in Figure 3B.

We are sorry for the confusion. Actually, the Figure 3C shows the flow cytometric data of primary tumor cell-derived tumor in nude mice, which is different from the flow cytometric chart of the cultured cell lines. We have specified it in the Figure 3C legend.

15. Figure 5C: please explain how you obtained a weight measurement for the lung metastases. You record lung metastasis weights of up to 2 grams. Lungs typically weight about 0.2 grams. Lungs of 2 grams seem improbable, let alone metastases in lungs weighing 2 grams. Also, tumor sizes of 700 mm2 do not seem possible in lungs.

We are sorry for our carelessness. Actually, we measured the lung weights, but not the tumor weights because we cannot dissect all metastatic tumor nodules in the lung. Similarly, the tumor size data were calculated wrongly, based on the magnified images. Now, we have corrected these mistakes in the revision. Overall, our data indicated the converted CSCs had similar tumorigenicity and differential capacity in mice.

16. The RNAseq analysis requires more explanation. In Figure 6B, please explain the three columns shown in the heat map. Without their designation, the rest of this figure is hard to interpret. It seems that only one single cell per timepoint has been analysed. Is this correct? How reliable is sequencing from a single cell? Or was the analysis from many cells grown up from a single cell? How many single cells (or derivative of single cells) were analysed at each time point? There is a discrepancy between the protocol described in the text and that provided in the Methods section, when describing the RNAseq analysis. It is difficult to relate the significance of the genes shown in Panel B to the Kaplan-Meier data when it is not clear what the three columns in panel B represent.

We are sorry that we did not describe the experimental protocol in detail and present the data in Figure 6B well. We cultured CD44-/CD24- MDA-MB-231 cells in a single cell manner for 24, 72 and 120 h and they were harvested for RNAseq analysis of three single cells at each time point. We have modified the experimental protocol to specify it in the revision. Although we did single cell RNAseq the data could not be explained by single cell RNAseq, rather than general RNAseq. Accordingly, we changed the term of single cell RNAseq into RNAseq to avoid potential misleading in the revision. From the data in the heatmap, the repeated single samples display similar levels of each gene, indicative of its reliable nature. We are sorry for our carelessness and now, we have modified the manuscript to ensure the consistence between the protocol and result sections. In addition, we have labeled the columns in Figure 6B to specific the samples and time points.

17. Figure 6D: I assume that the "*" and "#" are meant to convey some sort of significance, but they do not appear for any of the genes assessed by RT-PCR. Does that mean that none of the changes were significant?

We are sorry for the confusion. To clarify the significance among groups of cells, we have specified the significance as * or # vs. CD44-/CD24- cells in the Figure legend. There were significant difference in the levels of mRNA transcripts of some genes between these types of cells in our experimental system.

18. Why does the legend for Figure S6 talk about SILAC for protein analysis? Was SILAC run on these samples? Impossible on single cells. No mention of SILAC in Methods.

We are sorry for our mistake. We did not run SILAC in this study. We have deleted it in the revision. (now is Figure 6—figure supplement 1)

19. Page 10: Explain why YAP was selected for analysis following knockdown of RHBDL2. Was there a previously known connection between these two genes? For the CD44-/CD24- cells in Figure 7B, was YAP phosphorylation also altered? Please show a western demonstrating that RHBDL2 is reduced after transfection with siRHBDL2.

We thank the reviewer for his/her valuable comments. Indeed, the Hippo pathway is important for the stemness of tumor cells. Thus, we selected the YAP1 as a verification indicator. We did not find any available data on how RHBDL2 modulates YAP1 expression and phosphorylation in any type of tumors in the literature. In response to his/her concerns, we have performed additional experiments and found that transfection with RHDBL2-specific siRNAA effectively decreased the relative levels of its expression in CD44-/CD24- cells. More importantly, RHDBL2 silencing also decreased YAP1 expression, but increased YAP1 phosphorylation (Figure 7B), leading to increased USP31 expression and decreased NF-κB phosphorylation in CD44-/CD24- breast cancer cells. These data supported our conclusion.

20. Figure 7D: text says 0.7% CD44+/CD24- but figure says 0.5% and vice versa for the MDA-MB-468 cells.

We are sorry for the confusion. Actually, the Figure 7D describes the data from MDA-MB-231 cells, which were 0.5% of CD44+/CD24- cells while the Figure 7E displays the data (0.7% of CD44+/CD24- cells) from MDA-MB-468 cells following RHDBL2 silencing. Now,we have clarified the data in the revision.

21. The text on page 11 needs re-writing. The second paragraph, commencing with "We next selected…" is a repeat of the text above where you have already done this step and analysed the cells by flow cytometry.

We thank the reviewer for their careful review and have revised the paragraph accordingly.

Reviewer #2:

The manuscript describes a possible link between the presence of CD44-/CD24- cells and post-operative site distant metastasis. This is not a new concept, yet the actual biological function of CSC, particularly of the phenotype CD44-/CD24- remains inconclusive. The primary strength of the study is the large number of patient samples used in the retrospective analysis that provides the rationale for in vitro experiments using cultured cancer cells. Several technical and methodological limitations were identified.

The description of the patient cohort is very elusive. The bioinformatic analysis lacks in rigor as the methodology used to identify and quantitate CD44-/CD24- cells in patients is not properly described. For instance, Figure legends describe the use of flow cytometry for quantification of CD44-/CD24- negative cells from patient samples. Yet only an immunofluorescence staining is described in Results. No Methods description is given to FACS of patient samples. The manuscript seems focused on TNBC, yet the reader has a difficult time understanding how many samples are actually from TNBC patients. Two TNBC established cell lines are mentioned in the manuscript but no rationale is provided as to why some experiments are performed only with MDA-MB-468 and some with MDA-MB-231. The same is true for several other experiments. The extensive molecular description of CD44-/CD24- cell dedifferentiation into CSCs deserves distinction but will need to be confirmed in patient samples before conclusions can be drawn with regards to the biological significance of these findings to breast cancer patients and the medical community at large.

We fully understand the reviewer’s concerns and we are sorry for the confusion in the manuscript. As we addressed the similar concerns from the reviewer 1, we have clarified the experimental protocols and corrected couple mistakes in the revision. Actually, we used H&E staining to identify tumor cells in breast cancer tissue sections and employed immunofluorescence to quantify the number of CD44-/CD24- tumor cells in the consecutive sections to determine the frequency of CD44-/CD24- breast cancer cells in total tumor cells in individual tissue sections. Moreover, Supplementary file 1A (Table S1) shows the number of patients with three molecular classifications and their clinical information. In addition, our data have been verified with two cell lines MDA-MB-468 and MDA-MB-231, and these data have been added to the manuscript. In order to confirm our findings in patients’ samples, we utilized fresh specimens from breast cancer patients to isolate primary cells for data verification. The corresponding results have been added in Figure 3.

Clarify the methodology used to quantitate CD44-/CD24- in patient samples, as well as the bioinformatic analysis supporting the current hypothesis.

We appreciate his/her advice. As we stated above, we have clarified the methods for quantifying CD44-/CD24- breast cancer cells in tissue sections above and in the revision. In addition, we have modified the section of bioinformatics to clarify the strategies for anlaysis of data in the revision.

Provide a detailed analysis of clinical and pathologic information of the patient cohort, including histology.

We appreciate his/her advice. Now, we have provided the detailed information on demographic and clinical characteristics of all patients we studied in Supplementary file 1A (Table S1) of the revision.

Several references are 10 years old or older, new references supporting the authors finds would strength the hypothesis.

We appreciate his/her advice. Now, have updated the reference section with new references available in the literature.

Author details

1. Xinbo Qiao

   Department of Oncology, Shengjing Hospital, China Medical University, Shenyang, China

   Contribution

   Data curation, Formal analysis, Methodology, Writing - original draft, Writing - review and editing

   Contributed equally with

   Yixiao Zhang, Lisha Sun and Qingtian Ma

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-6759-921X

2. Yixiao Zhang

   1. Department of Oncology, Shengjing Hospital, China Medical University, Shenyang, China
   2. Dapartment of Urology, Shengjing Hospital, China Medical University, Shenyang, China

   Contribution

   Validation, Methodology

   Contributed equally with

   Xinbo Qiao, Lisha Sun and Qingtian Ma

   Competing interests

   No competing interests declared

3. Lisha Sun

   Department of Oncology, Shengjing Hospital, China Medical University, Shenyang, China

   Contribution

   Resources, Software, Writing - review and editing

   Contributed equally with

   Xinbo Qiao, Yixiao Zhang and Qingtian Ma

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-4095-5026

4. Qingtian Ma

   Department of Oncology, Shengjing Hospital, China Medical University, Shenyang, China

   Contribution

   Formal analysis, Writing - original draft, Project administration

   Contributed equally with

   Xinbo Qiao, Yixiao Zhang and Lisha Sun

   Competing interests

   No competing interests declared

5. Jie Yang

   Department of Oncology, Shengjing Hospital, China Medical University, Shenyang, China

   Contribution

   Software, Investigation

   Competing interests

   No competing interests declared

6. Liping Ai

   Department of Oncology, Shengjing Hospital, China Medical University, Shenyang, China

   Contribution

   Data curation, Formal analysis

   Competing interests

   No competing interests declared

7. Jinqi Xue

   Department of Oncology, Shengjing Hospital, China Medical University, Shenyang, China

   Contribution

   Formal analysis

   Competing interests

   No competing interests declared

8. Guanglei Chen

   Department of Oncology, Shengjing Hospital, China Medical University, Shenyang, China

   Contribution

   Formal analysis

   Competing interests

   No competing interests declared

9. Hao Zhang

   1. Department of Oncology, Shengjing Hospital, China Medical University, Shenyang, China
   2. Department of Breast Surgery, Liaoning Cancer Hospital and Institute, Cancer Hospital of China Medical University, Shenyang, China

   Contribution

   Data curation, Software

   Competing interests

   No competing interests declared

10. Ce Ji

    1. Department of Oncology, Shengjing Hospital, China Medical University, Shenyang, China
    2. Department of General Surgery, Shengjing Hospital, China Medical University, Shenyang, China

    Contribution

    Formal analysis, Funding acquisition

    Competing interests

    No competing interests declared

11. Xi Gu

    Department of Oncology, Shengjing Hospital, China Medical University, Shenyang, China

    Contribution

    Formal analysis

    Competing interests

    No competing interests declared

12. Haixin Lei

    Institute of Cancer Stem Cell, Cancer Center, Dalian Medical University, Dalian, China

    Contribution

    Writing - review and editing

    Competing interests

    No competing interests declared

13. Yongliang Yang

    Center for Molecular Medicine, School of Life Science and Biotechnology, Dalian University of Technology, Dalian, China

    Contribution

    Writing - review and editing

    Competing interests

    No competing interests declared

14. Caigang Liu

    Department of Oncology, Shengjing Hospital, China Medical University, Shenyang, China

    Contribution

    Conceptualization, Writing - review and editing

    For correspondence

    angel-s205@163.com

    Competing interests

    Reviewing Editor, eLife

    "This ORCID iD identifies the author of this article:" 0000-0003-2083-235X

• 1,879

  Page views

• 239

  Downloads

• 3

  Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.