The Role and Molecular Mechanism of SLC34A2 in Stemness Maintenance of CD44+CD166+ Lung Cancer Stem Cells (LCSCs) with AT-II cells’ characteristics


 Background

Evidence showed some non-small cell lung cancers (NSCLCs) had the type II alveolar epithelium cells’(AT-II cells) characteristics, and AT-II cells were a kind of original stem cells of NSCLCs. But how AT-II cells malignantly transformed into NSCLCs was unclear. Recent evidence indicated SLC34A2 was critical in the development of AT-II cells, and SLC34A2 might be a new gene in the initiation of NSCLCs. However, whether SLC34A2 participated in the malignant transformation of AT-II cells remained unknown. The exact role and mechanism of SLC34A2 in the initiation of NSCLCs needed to be further investigated.
Methods

The expression of Napi-IIb (encoded by SLC34A2) in the NSCLC cells was compared with that in AT-II cells using immunohistochemistry (IHC). Also coexpression of CD44 and CD166 was detected in these NSCLCs tissues by IHC. Then the CD44+CD166+ cells were sorted from lung tumor spheres by FACS. They were assessed by sphere, proliferation and tumorgenicity assay. Besides, their expression of surfactants C(SP-C) was stained by IHC. Next, the role and mechanism of SLC34A2 in CD44+CD166+ lung cancer stem cells were explored by siRNA-mediated SLC34A2 knockdown, related pathway pharmacological inhibition or activation. In vitro findings were furtherly validated in vivo and NSCLCs samples.
Results

The expression of SLC34A2 was downregulated in NSCLCs cells compared with AT-II cells in clinic samples. Then the CD44+CD166+ population was identified as CD44+CD166+ lung cancer stem cells (LCSCs). And LCSCs showed abundant expression of SP-C, the hallmark of AT-II cells. Higher expression of SLC34A2 was found in LCSCs compared to their origin NSCLC cells. Additionally, the expression of SLC34A2 was decreased after LCSCs were differentiated, and the morphology of the differentiated cells from LCSCs was similar to their origin NSCLC cells. Knockdown SLC34A2 made declined abilities of self-renewal, drug-resistance, migration and invasion in vitro as well as tumorigenicity in vivo in LCSCs. And SLC34A2 could maintain stemness of LCSCs via PI3K/AKT/STAT3/Sox2 axis. Besides, the connection between SLC34A2 maintaining stemness of lung cancer stem cells and PI3K/AKT/STAT3/Sox2 axis was also validated in vivo and in clinic samples.
Conclusions

For the first time, we illustrated the expression of SLC34A2 was downregulated in NSCLCs cells compared with AT-II cells. We discovered the downregulated expression of SLC34A2 performed a vital role in the malignant transformation of AT-II cells into NSCLCs. And SLC34A2 could maintain stemness of CD44+CD166+ lung cancer stem cells, which were with AT-II cell’s characteristic, via PI3K/AKT/STAT3/Sox2 axis. It had important significance in the revelation of a new mechanism for the initiation of NSCLCs.


Introduction
Non-small cell lung cancers (NSCLCs) accounted for 80% -85% of lung cancers [1]. As their non-ideal diagnosis at the early stage and treatment, the 5-year survival of NSCLCs was still less than 20% [2]. So, re ned investigation on the new mechanism under the origin of NSCLCs will strongly promote its early diagnosis and provide effective methods for its targeted therapy.
The cancer stem cell hypothesis stated that the origin of cancer was fuelled by cancer stem cells (CSCs), and the deregulated self-renewal pathways in CSCs contributed to the initiation of cancers [3,4]. The genetic alterations might make stem cells to malignantly transform into cancer stem cells, then cancer stem cells were differentiated into cancer cells [5][6][7]. Accumulated studies showed that lung cancer stem cells promoted the origin of NSCLCs [8,9]. And some genes and pathways involved in embryonic development were found to participate in retaining stemness of lung cancer stem cells [10]. In recent years, as the complex anatomical features of the lung, many types of stem cells were found in different anatomical sites of the lung [10]. Diverse lung stem cells or progenitor cell populations with the abilities of self-renewal and differentiation, were also found to have the ability to drive lung oncogenesis in different contexts [10]. For example, in the proximal lung, the tracheal basal cells had been proposed to be the cells of origin for NSCLCs; in the bronchoalveolar junction, the bronchioalveolar stem cells (BASCs) might be able to give rise to NSCLCs; at the end of the airway, the type II alveolar epithelial cells(AT-II cells) appeared to progress into NSCLCs [9]. However, how these stem cells malignantly transformed were rarely understood until now.
The alveolar was very important to maintain the physiological function of the lung. The alveolar was the main place for gas exchange, while it was highly susceptible to injury as it was directly exposed to the external environment [1]. And the alveolar was a major site of NSCLCs [11,12]. The alveolar consisted of the type I and type II alveolar epithelial cells (AT-I and AT-II) [13]. And the AT-II cells, which occupied about 4% of the alveolar surface, were cubic and located in the periphery of the alveoli [13]. The AT-II cells were critical to the physiological function of the alveolar, and they could prevent alveolar collapse for gas exchange by secreting surfactants [12]. Among the surfactants, the surfactant protein C(SP-C) was restricted exclusively to the AT-II cells and was identi ed as the hallmark of AT-II cells [12,14]. More noteworthy, the AT-II cells could be intermittently activated to self-renewal and differentiate into the AT-I cells for restoring alveolar in physiological condition and during lung injury, so the AT-II cells were identi ed as stem cells in alveolar [11,12]. Recently, some studies focused on the relationship between the malignant transformation of the AT-II cells and the origin of NSCLCs. The early histopathological studies uncovered some cells in NSCLCs tissues exhibited the features of AT-II cells, such as the presence of SP-C in the cell membrane [15,16]. Furthermore, in adult mouse lung, AT-II cells appeared to progress into NSCLCs by Kras mutation or by Kras mutation combined with Notch inhibition [17][18][19]. These results indicated that AT-II cells, the stem cells in alveolar, were a kind of original stem cells of NSCLCs. But how the AT-II cells malignantly transformed into NSCLCs was not clear until now. SLC34A2, encoding a type 2b sodium-dependent phosphate transporter(Napi-IIb), was responsible for transporting extracellular phosphorus ions into cells [20,21]. It was critical not only to the normal function of the lung but also to lung development. Among various organs, the highest expression of SLC34A2 was found in the lung and fetal lung [22]. And during the development of the human fetal lung, the expression of SLC34A2 was up-regulated [22]. More importantly, the latest study revealed that in terminal lung epithelium, SLC34A2 was speci cally expressed in mouse mature AT-II cells by single-cell RNA sequencing, and the expression of SLC34A2 was gradually upregulated during alveolar epithelial precursor cells differentiating into the AT-II cells [14]. Similarly, Jacob A et.al also found the expression of SLC34A2 was gradually increased, when pluripotent stem cells (iPS) differentiated into the mature AT-II cells [23]. These results indicated the key role of SLC34A2 in the development of the AT-II cells. Tumorigenesis resembled the reverse process of embryogenesis [3]. Kopantzev EP et.al found the expression of SLC34A2 in 12/12 NSCLCs tissues was signi cantly down-regulated [24]. Our previous studies further con rmed that expression of SLC34A2 was signi cantly downregulated in 59/90 NSCLCs tissues and 6 NSCLCs cell lines, and subsequently, the abilities of proliferation, invasion and migration were signi cantly inhibited in NSCLCs cells after SLC34A2 was overexpressed [25,26]. These results showed that the downregulated expression of SLC34A2 was involved in the initiation of NSCLCs. But the role and mechanism of SLC34A2 in the initiation of NSCLCs needed to be further investigated. Cancer stem cell hypothesis thought that abnormal expression of genes or pathways involved in embryonic development and self-renewal might lead to malignant transformation of stem cells [7]. Therefore, we further guessed the downregulated expression of SLC34A2 might be involved in the malignant transformation of AT-II cells into NSCLCs. However, until now little is known about it.
Therefore, in our present study, we aimed to investigate the relationship between downregulated expression of SLC34A2 and the malignant transformation of AT-II cells. And the role and mechanism of SLC34A2 in the initiation of NSCLCs would be further investigated. Firstly, we compared the expression of SLC34A2 in NSCLCs cells and AT-II cells in clinic samples. Then we identi ed CD44 + CD166 + lung cancer stem cells (LCSCs) from NSCLCs cell lines and tissue, and found LCSCs were with AT-II cell's characteristic. Furtherly we explored the role and molecular mechanism of SLC34A2 in stemness maintenance of CD44 + CD166 + lung cancer stem cells (LCSCs). For the rst time, our studies demonstrated SLC34A2 played an important role in the malignant transformation of AT-II cells into NSCLCs. And our results shed light on the revelation of the new mechanism for the initiation of NSCLCs.

NSCLCs Sample Collection
A total of 57 NSCLC tissues and their adjacent tissues were obtained from the Department of Thoracic Surgery, West China Hospital, Sichuan University. This study was performed with the approval of the Medical Ethical Committee of West China Hospital, Sichuan University. All patients were diagnosed with primary NSCLC and did not show other tumor occurrences. The detailed information of clinical parameters was shown in Additional les 1.

Quantitative Real-time RT-PCR Analysis
Differently treated cells were respectively washed with cold PBS and then extracted total RNA using Trizol (Invitrogen, USA) according to the manufacturer's protocol. Extracted total RNA was reverse transcribed into cDNA using the PrimrScript RT reagent Kit with gDNA Eraser (TaKaRa, Japan). As for the protocol of the quantitative real-time PCR analysis, it was described in Additional les 1.

Immunohistochemistry (IHC)
Immunohistochemistry was performed in NSCLC tissues, their adjacent tissues and mouse xenografts. All of the specimens were embedded in para n and cut as the sections of 4 µm thick, and then carried out the standard histological analysis. As for the protocol of immunohistochemistry, it was described in Additional les 1.

Animal and Cell Culture
A detailed description was made in Additional les 1.

Sphere Formation and Propagation
Single-cell suspensions were obtained from NSCLC tissue and NSCLC cell lines respectively. After mechanical and enzymatical disaggregation, cells were plated in serum-free medium, and the details were Additional les 1.

Colony Formation Assay
A detailed description was made in Additional les 1.

Xenotransplantation
A detailed description was made in Additional les 1.

Flow Cytometry Analysis and Cyto uorometric Cell Separation
A detailed description was made Additional les 1.

Immuno uorescence
A detailed description was made Additional les 1.

Hematoxylin-Eosin (H&E) Staining
Hematoxylin-eosin (H&E) staining was performed in mouse xenografts. All the specimens were embedded in para n and cut as sections of 4 µm thick, then they were stained with hematoxylin-eosin (H&E) according to standard histopathological techniques.

Differentiation of Stem Cell Progeny
A detailed description was made in Additional les 1.

Cell growth assay
A detailed description was made in Additional les 1.

Cell Viability Assay
A detailed description was made in Additional les 1.

Lentiviral Transduction
A549-LCSCs and H460-LCSCs were dissociated and spin-infected with 0.5 ml of NC-si-RNA (Shanghai Hanbio Biotechnology Co. Ltd, China) or SLC34A2-si-RNA lentiviral knockdown package (Shanghai Hanbio Biotechnology Co. Ltd, China). For both systems, cells were infected with lentiviral media at the MOI value of 40 overnight in a 37°C incubator. The transduction e ciency was evaluated by GFP expression. Stable clones with the GFP expression was > 99%, were then selected using puromycin. Note: LCSCs was short for CD44 + CD166 + lung cancer stem cells.

Western Blot
A detailed description was made in Additional les 1.
Transwell invasion /migration assay A detailed description was made in Additional les 1.

Chromatin immunoprecipitation analysis
A detailed description was made in Additional les 1.
The Cancer Genome Atlas (TCGA) analysis A detailed description was made in Additional les 1.

Down-regulated Expression of SLC34A2 Was Involved in Malignant Transformation of AT-II cells in Clinic Samples
Although it had been reported that the expression of SLC34A2 in NSCLCs was signi cantly lower than that in their adjacent tissues [24][25][26], yet there had no direct report about the relative expression of SLC34A2 between NSCLCs and AT-II cells. Therefore, we rst compared the expression of SLC34A2 in NSCLCs and AT-II cells in clinic samples. 57 poorly or moderately differentiated NSCLCs tissue samples were collected. And the downregulated expression of SLC34A2 was found in 45/57 collected poorly or moderately differentiated NSCLCs samples compared with their adjacent tissues (Fig. 1A). Namely, in most NSCLCs samples, the expression of SLC34A2 was found to be signi cantly downregulated compared with their adjacent tissues, which is consistent with our previous reports [25,26]. Then 45 NSCLCs samples with the down-regulated expression of SLC34A2 were chosen for further study. Classically, SP-C was identi ed as the hallmark of AT-II cells [12,14]. We detected the expression of SP-C by immunohistochemistry in the 45 NSCLCs tissues with down-regulated expression of SLC34A2. Cells with positive expression of SP-C were present in 18/45 NSCLCs samples (Additional File 2), indicating these 18 NSCLCs samples might be originated from the malignant transformation of AT-II cells. Secondly, the 18 NSCLCs samples with positive expression of SP-C and their adjacent tissues were chosen for further study. Referring to the previous literature [33], the corresponding adjacent tissues of the 18 NSCLCs were stained with both Napi-IIb and SP-C by double staining technology in serial sections to detect the expression of Napi-IIb in AT-II cells. The AT-II cells were cubic and located in the periphery of the alveoli [13]. In the sections, the cells with positive expression of SP-C were mostly found in the periphery of the alveoli and cuboidal. Similarly, the cells with positive expression of Napi-IIb were also mostly found in the periphery of the alveoli and their shape was cuboidal (Fig. 1B). Not only SP-C, but also SLC34A2 was identi ed as the hallmark of the AT-II cells recently [14]. Thus, we concluded the cells with positive expression of both Napi-IIb and SP-C were AT-II cells. Then we found that the staining intensity of Napi-IIb in the NSCLCs cells on the tumor tissue section was weaker than that in the AT-II cells on the adjacent tissue section (Fig. 1C). All these results showed the expression of SLC34A2 was downregulated in the NSCLCs cells compared with the AT-II cells. And it indicated down-regulated expression of SLC34A2 was involved in the malignant transformation of AT-II cells.
The Expressions of CD44 and CD166 Were Evaluated in NSCLCs Samples and Spheres from NSCLCs Cell Lines and NSCLCs Tissue The primary AT-II cells were di cult to culture and maintain in a well-differentiated state [12], so it was di cult to explore how AT-II cells malignantly transformed into NSCLCs directly. According to the cancer stem cell hypothesis stated that the genetic alterations might make stem cells to malignantly transform into cancer stem cells, then cancer stem cells were differentiated into cancer cells [5]. Therefore, we supposed AT-II cells might malignantly transform into lung cancer stem cells and then differentiate into NSCLCs cells. And we intended to explore the role of SLC34A2 in the malignant transformation of AT-II cells through the lung cancer stem cells which were with AT-II cells' characteristic.
The method for sorting lung cancer stem cells by single surface marker, such as CD133, CD326, CD44 and CD166 was controversial [28]. Researchers pointed out that it would be better to combine more than one marker to isolate CSCs from NSCLCs [29]. Evidence showed the expression of CD44 or CD166 was related to stem cell characteristics of AT-II cells [30,31]. Thus, we speculated the lung cancer stem cells with the surface markers of CD44 and CD166 might be related to the AT-II cells. In an attempt to preliminarily identify the markers of lung cancer stem cells, 18 primary NSCLC tissue samples with the down-regulated expression of SLC34A2 compared with their adjacent tissues and positive expression of SP-C were chosen to study. They were double-stained with the surface marker CD44 and CD166 by IHC in the serial section. Some cells were detected to show both positive expressions of CD44 and CD166 in 18 out of 18 NSCLC samples (include 12 lung adenocarcinoma and 6 lung squamous cell carcinoma) (Additional File 3, Fig. 2A). As a result, it indicated that the expression of CD44 + CD166 + was related to AT-II cells in NSCLCs specimens.
Furthermore, 3 NSCLC cell lines (A549, H460, and H1299) and 1 NSCLC tissue were cultured in the serumfree medium to enrich CSCs. After 2 weeks, some cells in 3 NSCLC cell lines and 1 NSCLC tissue could form spheres (Fig. 2B). And these spheres were found to possess signi cantly higher colony formation e ciency compared with their corresponding NSCLC cell lines and differentiated cells respectively ( Fig. 2C and 2D). Then the injection of as low as 10 5 sphere cells (except H1299 spheres) was uncovered to consistently produce tumor xenografts in 3-5 weeks ( Fig. 2E and Additional File 4). However, 10 6 NSCLCs cells (A549 and H460) and the differentiated cells from NSCLC tissue spheres could not generate tumor xenografts in 5 weeks (Additional File 4). All demonstrated that the spheres from NSCLC cell lines (A549 and H460) and NSCLC tissue had a higher capability of proliferation in vitro and tumorigenicity in vivo. And it indicated that these spheres contained lung cancer stem cells. Given that H1299 spheres failed to generate tumor xenografts in about 5 weeks, which pointed out the weaker CSCs characteristics of H1299, so we only took the spheres from A549, H460 and NSCLC tissue for the following study. After that, the higher proportion of CD44 + CD166 + cells was found in spheres than in their corresponding NSCLC cell lines (A549 and H460) or differentiated cells by FACS (Fig. 2F). It furtherly showed that CD44 + CD166 + could be the roust markers of lung cancer stem cells in spheres from NSCLCs cell lines and NSCLCs tissue.

CD44 + CD166 + Spheres from NSCLC Tissue and NSCLC Cell Lines Were Identi ed as Lung Cancer Stem Cells with AT-II cells' Characteristic
To identify whether CD44 + CD166 + could be a marker for lung cancer stem cells with AT-II cells' characteristic, rstly the fractions of CD44 + CD166 + , CD44 + CD166 − , CD144 − CD166 +, and CD44 − CD166 − were respectively sorted from the spheres by FACS. Subsequently, the strongest ability of sphere formation in vitro and tumor formation in vivo was detected in CD44 + CD166 + cells than CD44 + CD166 − and CD144 − CD166 + cells (Fig. 3A,3B and Additional File 5). Moreover, the highest potential of proliferation was found in transplants of CD44 + CD166 + cells by Ki-67 assay (Fig. 3C). And the mice xenografts derived by CD44 + CD166 + spheres were found to show positive expression of CD44 and CD166 (Fig. 3F). Besides, CD44 + CD166 + spheres had upregulated expression of the genes involved in stemness, including β-catenin, Sox2, Bmi-1, Oct-4, Nanog and CXCR4, compared with their corresponding NSCLC cell lines (A549 and H460) (Fig. 3E). And CD44 + CD166 + spheres also had increased expression of the genes related to drug resistance, such as ABCG1 and Notch3, as well as the enhanced viability against cisplatin and the higher value of IC 50 ( Fig. 3E and Additional File 6).
After being cultured in serum-medium, CD44 + CD166 + cells could differentiate into the cells which had similar morphological features with their original NSCLCs cells (Fig. 3D). At the same time, the expression of the genes involved in stemness, including β-catenin, Sox2, Bmi-1, Nanog and CXCR4, was notably reduced after CD44 + CD166 + spheres were differentiated (Fig. 3E). Similarly, the expression levels of the genes related to drug resistance, e.g. ABCG1 and Notch3, as well as the viability against cisplatin and value of IC 50, were also notably declined in differentiated CD44 + CD166 + spheres ( Fig. 3E and Additional File 6). Thus, we demonstrated a poorly differentiated state of CD44 + CD166 + spheres.
The common method used for identifying the AT-II cells was to detect the expression of the surfactant protein C(SP-C) at present [14]. To explore whether CD44 + CD166 + spheres were originated from AT-II cells, we detect the expression of SP-C in CD44 + CD166 + spheres. It was consistent with our expectations that CD44 + CD166 + spheres showed abundant expression of SP-C (Fig. 3G,3H). Also the mice xenografts derived by CD44 + CD166 + spheres showed abundant expression of SP-C (Fig. 3I). These showed that CD44 + CD166 + spheres were with AT-II cells' characteristic, and indicated that CD44 + CD166 + spheres might be originated from malignantly transformed AT-II cells.
Collectively, CD44 + CD166 + spheres were identi ed as CD44 + CD166 + lung cancer stem cells with the AT-II cells' characteristic. In the following study, we named CD44 + CD166 + lung cancer stem cells as LCSCs (Lung Cancer Stem Cells) for short.
SLC34A2 was Discovered to Maintain Stemness of CD44 + CD166 + Lung Cancer Stem Cells in vitro and in vivo Next, we tested the role of SLC34A2 played in LCSCs. The higher mRNA and protein expression levels of SLC34A2 were found in LCSCs than that in their corresponding NSCLC cell lines (Fig. 4A). Intriguingly, the differentiated cells that were derived from LCSCs had the lower expression of SLC34A2, compared with their LCSCs counterparts (Fig. 4A). The results above suggested that SLC34A2 might play a crucial role in maintaining stemness in LCSCs.
To verify this hypothesis, a lentiviral-based approach was used to knockdown SLC34A2 in LCSCs with si-RNAs, and the knockdown e ciency was con rmed by qRT-PCR and western blotting (Fig. 4B). Following, the weaker abilities of sphere formation, colony formation in vitro were detected in Si-SLC34A2-LCSCs than Si-NC-LCSCs and LCSCs ( Fig. 4C and 4D). Importantly, when implanted into the ank of nude mice, the tumorigenicity of si-SLC34A2-LCSCs was signi cantly lower than NC-LCSCs and LCSCs in vivo (Figs. 4E and Additional File 7). Moreover, compared with tumor xenografts of si-NC-LCSCs and LCSCs, tumor xenografts of si-SLC34A2-LCSCs also had the weaker ability of proliferation by Ki-67 staining (Figs. 4F). Then we tested the expression of genes involved in stemness, and found the mRNA expression level of the genes related to stemness -Sox2 was declined after SLC34A2 knockdown in A549-LCSCs and H460-LCSCs (Fig. 4G). Also the mRNA expression level of Oct-4, Bmi-1, CXCR4, Nanog and β-catenin was decreased after SLC34A2 knockdown in A549-LCSCs (Fig. 4G).
Collectively, our results demonstrated that SLC34A2 was necessary to maintain the stemness of LCSCs.
SLC34A2 was Discovered to Maintain Stemness of CD44 + CD166 + Lung Cancer Stem Cells by PI3K/AKT/STAT3/Sox2 Axis Following, we intended to search the molecular mechanism under SLC34A2 maintaining stemness of LCSCs. As the mRNA expression level of Sox2 was reduced after SLC34A2 knockdown both in A549-LCSCs and H460-LCSCs, therefore, we suggested that Sox2 might play an important role in the process that SLC34A2 maintaining stemness of LCSCs. Then the declined protein expression level of Sox2 was con rmed in si-SLC34A2-LCSCs than Si-NC-LCSCs and LCSCs (Fig. 4H). Moreover, the NaPi-IIb protein, encoded by SLC34A2, was located in the cell membrane [20], thus SLC34A2 might regulate the transcription of Sox2 in nuclear by its downstream factors. Later, the reduced expression of key proteins in PI3K/AKT (PI3K and p-AKT) and STAT3 pathway(p-STAT3) was discovered in Si-SLC34A2-LCSCs than Si-NC-LCSCs and LCSCs (Fig. 5A). Additionally, signi cantly lower expression of key proteins (p-NF-KB and p-STAT3) in STAT3 pathway was also found in LCSCs after inhibition of PI3K/AKT pathway (LY294002) or AKT(MK2206), while inhibition of STAT3 pathway (WP1066) did not affect the expression of the key proteins (PI3K and p-AKT) in PI3K/AKT pathway in LCSCs (Fig. 5B-5E and Additional File 8). These results showed that the PI3K/AKT/STAT3 axis was located downstream of SLC34A2 in LCSCs.
After that, we aimed to nd whether SLC34A2 regulated the transcription of Sox2 by the PI3K/AKT/STAT3 axis. Similar to the effect of SLC34A2 knockdown, reduced mRNA and protein expression levels of Sox2 were detected in LCSCs after inhibition of the PI3K/AKT/STAT3 axis (Fig. 5F). While signi cantly increased mRNA and protein expression levels of Sox2 were found in Si-SLC34A2-LCSCs after activation of the PI3K/AKT/STAT3 axis (Fig. 5G). The above showed that SLC34A2 could regulate the transcription level of Sox2 by the PI3K/AKT/STAT3 axis. Furthermore, binding sites of STAT3 to the promoter region of Sox2 were predicted via the bioinformatics prediction website (Genecards, JASPAR and Patch), and one site was con rmed by CHIP-PCR according to the reference [32] (Fig. 5H and 5I). Then, the CHIP-qPCR result uncovered the binding probability of STAT3 to the promoter region of Sox2 in LCSCs was decreased after SLC34A2 was interfered (Fig. 5J). Collectively, the above showed that SLC34A2 prompted STAT3 to bind to the promoter region of Sox2 via PI3K/AKT pathway in LCSCs.
Finally, we con rmed whether SLC34A2 maintained stemness of LCSCs via the PI3K/AKT/STAT3/Sox2 axis. Similar to the effect of SLC34A2 knockdown, the reduced stemness was found in LCSCs after inhibition of the PI3K/AKT/STAT3/Sox2 axis, such as the declined abilities of sphere formation, colony formation, growth, drug-resistance, migration, and invasion, as well as improved expression of differentiation markers CK8 and CK18 (Fig. 6A-6G and Additional File 9). While these declined abilities in si-SLC34A2-LCSCs, such as sphere formation, colony formation, growth, migration and invasion, as well as declined expression of differentiation markers CK8 and CK18, could be rescued by activation of the PI3K/AKT/STAT3/Sox2 axis (Fig. 6A-C, E-G). The ability of drug-resistance was also discovered to be largely strengthened in A549-si-SLC34A2-LCSCs after the PI3K/AKT/STAT3/Sox2 axis was activated (Fig. 6D). These results showed that SLC34A2 maintained stemness of LCSCs via the PI3K/AKT/STAT3/Sox2 axis in vitro.
The Connection between SLC34A2 Maintaining Stemness of Lung Cancer Stem Cells and PI3K/AKT/STAT3/Sox2 axis Was Validated in vivo and in clinic samples At last, we validated the connection between SLC34A2 maintaining stemness of lung cancer stem cells and PI3K/AKT/STAT3/Sox2 axis in vivo and in clinic samples. In vivo, the weaker positive expression of key proteins in the PI3K/AKT/STAT3/Sox2 axis (PI3K, p-AKT, p-STAT3 and Sox2) was found in xenografts derived from si-SLC34A2-LCSCs than si-NC-LCSCs (Fig. 6H). Combined with our in vivo nding that Si-SLC34A2-LCSCs had reduced ability of tumorigenicity than Si-NC-LCSCs, these results indicated the connection between SLC34A2 maintaining stemness of lung cancer stem cells and the activation of the PI3K/AKT/STAT3/Sox2 axis in vivo.
As we mentioned in the rst part, we found the downregulated expression of SLC34A2 in 45/57(most) poorly or moderately differentiated NSCLCs samples compared with their adjacent tissues (Fig. 1A). What's interesting was that we found the expression of SLC34A2 in LCSCs, which were enriched and originated from NSCLCs cells, was signi cantly higher than that in NSCLC cells in vitro (Fig. 4A). These results suggested that although the expression of SLC34A2 was signi cantly downregulated in NSCLC cells compared with normal cells, the lung cancer stem cells (a subset of NSCLCs) had relatively higher expression of SLC34A2 compared with other NSCLCs cells. To furtherly detect the connection between SLC34A2 maintaining stemness of lung cancer stem cells and PI3K/AKT/STAT3/Sox2 axis in clinic samples, the 45 NSCLCs samples, which had the down-regulated expression of SLC34A2 compared with their adjacent tissues, were chosen for further study. And later they were divided into two groups according to their IHC score of Napi-IIb: the group with relatively higher expression of SLC34A2(the SLC34A2 High group) (IHC score of Napi-IIb ≥ 6) and the group with relatively lower expression of SLC34A2(the SLC34A2 Low group) (IHC score of Napi-IIb < 6). Subsequently, data analysis displayed that the proportion of NSCLCs samples with high T stage had no signi cant difference between the two groups (Fig. 6I). It indicated that the downregulated expression of SLC34A2 might be associated with the early origin of NSCLCs. Besides, compared with the SLC34A2 Low group, the SLC34A2 High group had a higher proportion in NSCLCs samples with lymphatic metastasis and high expression of p-AKT ( Fig. 6J  and 6K). However, the proportion of NSCLCs samples with high expression of PI3K, p-STAT3 and Sox2 had no difference between the two groups ( Fig. 6J and 6K). Cancer stem cells had a stronger ability of invasion and metastasis than cancer cells [4]. Therefore, these results preliminary suggested the connection between SLC34A2 maintaining stemness of lung cancer stem cells and the activation of AKT in the collected NSCLCs samples.
Besides, the connection between SLC34A2 maintaining stemness of lung cancer stem cells and the activation of AKT and STAT3 was also preliminarily found in the datasets of the TCGA database (Additional File 10).

Discussion
The cancer stem cell hypothesis assumed cancer stem cells were originated from stem cells [4]. Some studies on histopathological sections and mouse models indicated that AT-II cells, the stem cells in alveolar, were the origin cells of NSCLCs [15][16][17][18][19]. But how AT-II cells malignantly transformed into NSCLCs is still unclear now. Extensive researches suggested that abnormal expression of genes or pathways involved in embryonic development and self-renewal might cause stem cells to malignantly transform into cancer stem cells and promote the initiation of cancers [3,6,7]. Recent research found SLC34A2 was speci cally expressed in mouse mature AT-II cells and the expression of SLC34A2 was signi cantly upregulated during alveolar precursor cells differentiating into the AT-II cells [14]. It showed upregulated expression of SLC34A2 played a key role in the development of AT-II cells. Tumorigenesis resembled the reverse process of embryogenesis [3]. Yet, our previous studies and Kopantzev EP et.al both found the expression of SLC34A2 in most NSCLCs tissues were signi cantly down-regulated compared with their adjacent tissues, while the abilities of proliferation, invasion and migration were signi cantly inhibited in NSCLCs cells after SLC34A2 was overexpressed [24][25][26]. These results showed downregulated expression of SLC34A2 was involved in the initiation of NSCLCs. Therefore, according to the cancer stem cell hypothesis, we furtherly supposed the downregulated expression of SLC34A2 might be related to the malignant transformation of AT-II cells.
Although it had been reported that the expression of SLC34A2 in NSCLCs was signi cantly lower than that in their adjacent tissues [24][25][26], there was no direct evidence about the relative expression of Similarly, in our study, we detected the expression of Napi-IIb and SP-C in AT-II cells referring to the previous methods [27]. The serial sections of 18 adjacent lung cancer tissues were stained Napi-IIb and SP-C by immunohistochemical staining. Some cells showed positive expression of SP-C in adjacent NSCLCs tissue sections and located mostly in the periphery of the alveoli, cubic. And the cells with positive expression of Napi-IIb had the similar position and morphology to the cells with positive expression of SP-C. According to the literature [27], we considered the cells with positive expression of Napi-IIb in the periphery of the alveoli and cubic as the AT-II cells. Recently, SLC34A2 was identi ed as the hallmark of the AT-II cells [14]. So, it furtherly con rmed our ndings that in the periphery of the alveoli, the cells with positive expression of Napi-IIb were indeed the AT-II cells. Then the weaker staining intensity of Napi-IIb was found in the NSCLCs cells in the 18 tumor tissue sections compared with the AT-II cells in their adjacent tissue sections in our present study. Namely, the expression of SLC34A2 in the NSCLCs cells was decreased compared with the AT-II cells. For the rst time, our results indicated that the downregulated expression of SLC34A2 might lead AT-II cells to malignantly transform into NSCLCs from the perspective of histology.
As it was di cult to culture and maintain primary AT-II cells in a well-differentiated state, as well as faithfully recapitulate the morphologic, functional, molecular, proteomic, and/or lipidomic markers of AT-II phenotype in stable cell lines [12], the research on how AT-II cells were malignantly transformed into NSCLCs was greatly hindered before. According to the cancer stem cell hypothesis, the genetic alterations might make stem cells to malignantly transform into cancer stem cells, then cancer stem cells were differentiated into cancer cells [5]. We supposed AT-II cells might be malignantly transformed into lung cancer stem cells and then differentiated into NSCLCs cells. Therefore, we intended to explore the role and mechanism of SLC34A2 in the malignant transformation of AT-II cells by the lung cancer stem cells with AT-II cells' characteristic. Evidence showed the expression of CD44 or CD166 was related to stem cell characteristics of AT-II cells [30,31]. Thus, we speculated the lung cancer stem cells with the surface markers of CD44 and CD166 might be related to AT-II cells. Then we preliminarily explored the connection between the expression of CD44 and CD166 and the AT-II cells in 18 NSCLCs samples. Histological analysis revealed that in the 18 NSCLCs tissues with downregulated expression of SLC34A2 and positive expression of SP-C, some cells showed coexpression of CD44 and CD166. These results suggested that the expression of CD44 + CD166 + was related to AT-II cells in NSCLCs specimens. However, neither CD44 nor CD166 was associated with TNM stage and differentiation in NSCLCs samples, which might most likely due to the limited number of NSCLC tissues. And no prognosis data was available for these NSCLCs samples, so we were unclear whether the expression of CD44 + CD166 + correlated with the prognosis of NSCLCs. In the future, we will add the number of NSCLCs samples and collected their prognosis data to re ne this part.
Futherly, we sorted the CD44 + CD166 + lung cancer stem cells from NSCLCs tissue and cell lines. Spheres, which possessed the higher abilities of colony formation and tumorigenic potential in vivo, were enriched from NSCLCs tissue and cell lines by cultivating in a serum-free medium. Then the higher proportion of CD44 + CD166 + was tested in spheres than that in their corresponding NSCLCs cell lines and differentiated cells. Subsequently, CD44 + CD166 + , CD44 + CD166 − , CD44 − CD166 + and CD44 − CD166 − cells were sorted from spheres by FACS. And the CD44 + CD166 + cells were found to have the strongest abilities of sphere formation and tumorigenesis than other fractions respectively. Besides, CD44 + CD166 + cells had a poor differentiation state and enhanced ability of drug resistance against cisplatin, as well as the increased expression of genes involved in stemness, compared with their adherent counterparts in vitro. These results identi ed the CD44 + CD166 + population as lung cancer stem cells. Next, we stained surfactant protein C(SP-C) in CD44 + CD166 + spheres to determine whether CD44 + CD166 + spheres had AT-II cells' characteristic. SP-C was one kind of surfactant proteins secreted by the AT-II cells [12]. SP-C was not only essential for the physical structure of the surfactant, but also the formation and stability of the surfaceactive monolayer at the air-water interface [13]. In the adult, only the expression of SP-C was restricted exclusively to the AT-II cells, and SP-C was identi ed as the hallmark of the AT-II cells [12,14]. Therefore, the common method used for identifying the AT-II cells was to detect the expression of SP-C at present [14]. In our studies, we found that both CD44 + CD166 + lung cancer stem cells and the mice xenografts derived from CD44 + CD166 + lung cancer stem cells showed abundant expression of SP-C. It showed that CD44 + CD166 + lung cancer stem cells (LCSCs) were with AT-II cells' characteristic, and indicated CD44 + CD166 + lung cancer stem cells might be originated from malignantly transformed AT-II cells.
Therefore, the identi cation of CD44 + CD166 + lung cancer stem cells provided an experimental basis for studying how SLC34A2 played in the malignant transformation of AT-II cells into NSCLCs.
Then, we tested the role of SLC34A2 in CD44 + CD166 + lung cancer stem cells (LCSCs). Firstly, the higher expression of SLC34A2 was found in LCSCs than in their corresponding NSCLC cell lines and differentiated cells, which preliminarily indicated SLC34A2 might be related to maintaining stemness of LCSCs. Furtherly, reduced abilities of sphere formation and colony formation in vitro, as well as tumorigenesis in vivo were found in LCSCs after SLC34A2 was interfered. Moreover, the mRNA expression levels of genes related to stemness, such as Bmi-1, CXCR4, Nanog, β-catenin, and Sox2 were signi cantly decreased in A549-LCSCs after SLC34A2 was interfered, and the mRNA expression level of Sox2 was reduced in H460-LCSCs after SLC34A2 was interfered. The evidence above demonstrated that SLC34A2 could maintain stemness of LCSCs.
Next, we studied how SLC34A2 maintained stemness of LCSCs. What caught our attention was that the reduced mRNA expression level of Sox2 was found both in A549-LCSCs and H460-LCSCs after SLC34A2 was interfered. Several reports identi ed that Sox2 played a critical role in maintaining the self-renewal of embryonic stem cells and the development of lung, as well as side population (SP) cells in NSCLCs [33][34][35]. To our knowledge, SLC34A2 played an important role in embryonic stem cells and the development of the lung [22,36]. And in this study, we found SLC34A2 had an important action in maintain stemness of LCSCs. Based on both Sox2 and SLC34A2 had alike functions in the development of embryonic stem cells, the lung and lung cancer stem cells, it suggested the key role of Sox2 in maintaining stemness of LCSCs by SLC34A2. Subsequently, the PI3K/AKT/STAT3 axis was detected to be downstream of SLC34A2 in LCSCs. Then we uncovered SLC34A2 could induce the transcription level of Sox2 via the PI3K/AKT/STAT3 axis. Then the enhancement effect of SLC34A2 on the binding of STAT3 to the promoter region of Sox2 was con rmed. Finally, SLC34A2 was discovered to maintain stemness of LCSCs through PI3K/AKT/STAT3/Sox2 axis in vitro. Besides, the connection between SLC34A2 maintaining stemness of lung cancer stem cells and the activation of PI3K/AKT/STAT3/Sox2 axis was veri ed in vivo. In the clinical specimens, we only found the SLC34A2 High group had a higher proportion of lymphatic metastasis and higher expression of p-AKT, compared with the SLC34A2 Low group. However, the proportion of NSCLCs samples with high expression of PI3K, p-STAT3, and Sox2 had no difference between the two groups. It was regrettable that we could only verify the connection between SLC34A2 maintaining stemness of lung cancer stem cells and activation of AKT. The imperfect results might be due to their small sample size. In the future, we would expand the sample size for identifying it. Above all, we illustrated that SLC34A2 could maintain stemness of LCSCs, which were with AT-II cell's characteristic. And the relative mechanism was that SLC34A2 promoted STAT3 to bind to the promoter region of Sox2 via PI3K/AKT pathway. Furthermore, the cancer stem cell hypothesis stated that the origin of cancer was fueled by cancer stem cells (CSCs), and deregulated self-renewal signals in CSCs contributed to the initiation of cancers [3,4]. Therefore, our results furtherly showed that the critical role and mechanism of SLC34A2 in the malignant transformation of AT-II cells into NSCLCs.
It was worth noting that in our in vivo studies, we injected CD44 + CD166 + lung cancer stem cells (LCSCs) into the back of nude mice. Then the xenografts derived from LCSCs were found to show positive coexpression of CD44 and CD166. It suggested that the NSCLCs cells, which were derived from CD44 + CD166 + lung cancer stem cells, had positive coexpression of CD44 and CD166. As we mentioned previously, 18 NSCLCs tissues with downregulated expression of SLC34A2 showed positive expression of SP-C. It indicated that the 18 NSCLCs tissues might be originated from AT-II cells and downregulated expression of SLC34A2 might lead to malignant transformation of AT-II cells. Moreover, the coexpression of CD44 and CD166 was also detected in these 18 NSCLCs tissues. Like our in vivo studies, the results in clinic samples suggested these 18 NSCLCs tissues might be originated from CD44 + CD166 + lung cancer stem cells. The results above furtherly prompted that downregulated expression of SLC34A2 might lead AT-II cells to malignantly transform into CD44 + CD166 + lung cancer stem cells (LCSCs).
Additionally, our in vitro ndings showed the connection between the downregulated expression of SLC34A2 and the differentiation of LCSCs. The genetic alterations might make stem cells to malignantly transform into cancer stem cells, then cancer stem cells were differentiated into cancer cells [5]. We found that the expression of SLC34A2 was decreased after LCSCs were differentiated. And the morphological type of the differentiated cells derived from LCSCs was similar to their origin NSCLC cells. Meanwhile, lentiviral-si-SLC34A2 reduced the protein expression of differentiation markers CK8 and CK18 in LCSCs.
These in vitro results preliminarily indicated that downregulated expression of SLC34A2 might induce LCSCs to differentiate into NSCLCs. However, until now we did not know why the expression of SLC34A2 was decreased during the differentiation of LCSCs and how SLC34A2 played in the differentiation of LCSCs. In future research, we will focus on this part.
However, what made us confused was that, there were signi cant differences in the expression and function of SLC34A2 between CD44 + CD166 + lung cancer stem cells and NSCLCs cells. In our previous study, we found that SLC34A2 was signi cantly down-regulated both in NSCLC tissues and cell lines compared to normal cells or adjacent tissues. And overexpressed SLC34A2 could diminish cell growth, motility and invasiveness of some NSCLC cell lines in vitro and signi cantly attenuate tumorigenicity in vivo [25,26]. And SLC34A2 has been indicated as a tumor-repressive role. However, in this work, we found the signi cant differential expression and function of SLC34A2 between CD44 + CD166 + lung cancer stem cells and NSCLCs cells. Paradoxically, we found both the mRNA and protein expression levels of SLC34A2 in LCSCs were relatively higher than that in NSCLCs cell lines. Furthermore, SLC34A2 could maintain stemness of CD44 + CD166 + lung cancer stem cells, such as maintain the ability of sphere formation, colony formation in vitro and tumorigenesis potential in vivo and so on. While knockdown SLC34A2 in LCSCs with si-RNA made the weaker abilities of sphere formation, colony formation in vitro and tumorigenesis potential in vivo. The contradictory expression and function of SLC34A2 between CD44 + CD166 + lung cancer stem cells and NSCLCs cells was similar to the pattern of Bone Morphogenetic Proteins (BMPs) in glioma stem cells and gliomas. BMPs had been demonstrated to drive astrocytic differentiation in CSCs, which indicated a tumor-repressive role and were proposed as anti-CSC therapies to induce differentiation [37]. But several groups have demonstrated elevated BMP expression, particularly BMP2, was detected in human gliomas. Later the researchers demonstrated the reason that CSCs could express Gremlin1, a BMP antagonist, to speci cally blocks differentiation effects of BMPs [37], which gave a reasonable explanation that contradictory expression of BMPs in glioma stem cells and gliomas. Although we rst found the signi cant differential expression and function of SLC34A2 between CD44 + CD166 + lung cancer stem cells and NSCLCs cells, however, we were not clear about the reason for the contradictory result now. We are interested in the seemingly contradictory result and would carry on exploration to this question in the future.
Above all, our studies demonstrated SLC34A2 played an important role in the malignant transformation of AT-II cells into NSCLCs. The downregulated expression of SLC34A2 was found in NSCLCs cells than the AT-II cells in clinic samples. Then SLC34A2 was detected to maintain the stemness of CD44 + CD166 + lung cancer stem cells (LCSCs), which were with AT-II cells' characteristic. Finally, the relative mechanism was uncovered that SLC34A2 induced transcription of Sox2 by promoting the binding of STAT3 to the promoter region of Sox2 via PI3K/AKT pathway. Additionally, we suggested the downregulated expression of SLC34A2 performed a vital role in the process that AT-II cells malignantly transformed into LCSCs, and then differentiated into NSCLCs (Additional File 11). It showed the critical role and mechanism of SLC34A2 in the malignant transformation of AT-II cells and initiation of NSCLCs. Also it provided a new insight for exploring how AT-II cells malignantly transform into NSCLCs.

Conclusions
Above all, we illustrated the expression of SLC34A2 was downregulated in NSCLCs cells compared with AT-II cells. We discovered the downregulated expression of SLC34A2 performed a vital role in the

Consent for publication
Written informed consent for publication was obtained from all participants.

Availability of data and material
All data generated or analyzed during this study were included in this published article and its supplementary information les. And all data supporting the ndings of this study are available from the corresponding author on reasonable request. TCGA database analysis was performed by querying the website: cBioPortal for Cancer Genomics (http://www.cbioportal.org/)

Competing interests
The authors declare that they have no competing interests.         The mechanism of SLC34A2 maintaining stemness of LCSCs was uncovered to be through with high expression of PI3K, p-AKT, p-STAT3 and Sox2 between the group with relative higher expression of SLC34A2 and relative lower expression of SLC34A2. Note: LCSCs was short for CD44+CD166+ lung cancer stem cells. si-NC was short for LCSCs transfected with si-NC and si-SLC34A2 was short for LCSCs transfected with si-SLC34A2.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.