Pathological Significance and Prognostic Value of Surfactant Protein D in Cancer

Surfactant protein D (SP-D) is a pattern recognition molecule belonging to the Collectin (collagen-containing C-type lectin) family that has pulmonary as well as extra-pulmonary existence. In the lungs, it is a well-established opsonin that can agglutinate a range of microbes, and enhance their clearance via phagocytosis and super-oxidative burst. It can interfere with allergen–IgE interaction and suppress basophil and mast cell activation. However, it is now becoming evident that SP-D is likely to be an innate immune surveillance molecule against tumor development. SP-D has been shown to induce apoptosis in sensitized eosinophils derived from allergic patients and a leukemic cell line via p53 pathway. Recently, SP-D has been shown to suppress lung cancer progression via interference with the epidermal growth factor signaling. In addition, a truncated form of recombinant human SP-D has been reported to induce apoptosis in pancreatic adenocarcinoma via Fas-mediated pathway in a p53-independent manner. To further establish a correlation between SP-D presence/levels and normal and cancer tissues, we performed a bioinformatics analysis, using Oncomine dataset and the survival analysis platforms Kaplan–Meier plotter, to assess if SP-D can serve as a potential prognostic marker for human lung cancer, in addition to human gastric, breast, and ovarian cancers. We also analyzed immunohistochemically the presence of SP-D in normal and tumor human tissues. We conclude that (1) in the lung, gastric, and breast cancers, there is a lower expression of SP-D than normal tissues; (2) in ovarian cancer, there is a higher expression of SP-D than normal tissue; and (3) in lung cancer, the presence of SP-D could be associated with a favorable prognosis. On the contrary, at non-pulmonary sites such as gastric, breast, and ovarian cancers, the presence of SP-D could be associated with unfavorable prognosis. Correlation between the levels of SP-D and overall survival requires further investigation. Our analysis involves a large number of dataset; therefore, any trend observed is reliable. Despite apparent complexity within the results, it is evident that cancer tissues that produce less levels of SP-D compared to their normal tissue counterparts are probably less susceptible to SP-D-mediated immune surveillance mechanisms via infiltrating immune cells.

Surfactant protein D (SP-D) is a pattern recognition molecule belonging to the Collectin (collagen-containing C-type lectin) family that has pulmonary as well as extra-pulmonary existence. In the lungs, it is a well-established opsonin that can agglutinate a range of microbes, and enhance their clearance via phagocytosis and super-oxidative burst. It can interfere with allergen-IgE interaction and suppress basophil and mast cell activation. However, it is now becoming evident that SP-D is likely to be an innate immune surveillance molecule against tumor development. SP-D has been shown to induce apoptosis in sensitized eosinophils derived from allergic patients and a leukemic cell line via p53 pathway. Recently, SP-D has been shown to suppress lung cancer progression via interference with the epidermal growth factor signaling. In addition, a truncated form of recombinant human SP-D has been reported to induce apoptosis in pancreatic adenocarcinoma via Fas-mediated pathway in a p53-independent manner. To further establish a correlation between SP-D presence/levels and normal and cancer tissues, we performed a bioinformatics analysis, using Oncomine dataset and the survival analysis platforms Kaplan-Meier plotter, to assess if SP-D can serve as a potential prognostic marker for human lung cancer, in addition to human gastric, breast, and ovarian cancers. We also analyzed immunohistochemically the presence of SP-D in normal and tumor human tissues. We conclude that (1) in the lung, gastric, and breast cancers, there is a lower expression of SP-D than normal tissues; (2) in ovarian cancer, there is a higher expression of SP-D than normal tissue; and (3) in lung cancer, the presence of SP-D could be associated with a favorable prognosis. On the contrary, at non-pulmonary sites such as gastric, breast, and ovarian cancers, the presence of SP-D could be associated with unfavorable prognosis. Correlation between the levels of SP-D and overall survival requires further investigation. Our analysis involves a large number of dataset; therefore, any trend observed is reliable. Despite apparent complexity within the results, it is evident that cancer tissues that produce less levels of SP-D compared to their normal tissue counterparts are probably less susceptible to SP-D-mediated immune surveillance mechanisms via infiltrating immune cells.
Keywords: innate immunity, surfactant protein D, immune surveillance, bioinformatics analysis, immuno histochemistry, cancers, tumor microenvironment inTrODUcTiOn Surfactant protein D (SP-D) is a collagenous glycoprotein encoded by SFTPD gene belonging to the collectins family (1). Like other members of the collectin family, SP-D has a primary subunit structure that comprises of an N-terminal cysteine-rich region, a triple-helical collagen-like domain, an α-helical coiled neck domain, and a C-terminal C-type lectin domain [also called carbohydrate recognition domain (CRD)] (2). Each subunit of human SP-D comprises three identical polypeptide chains of 43 kDa, which is assembled into a tetrameric structure with four of the homotrimeric subunits linked via their N-terminal regions, but trimers, dimers, and monomers also exist. Tetrameric structures can undergo further oligomerization to give SP-D multimers that could contain up to 96 individual chains. SP-D was originally described in association with pulmonary surfactant; in the lung, it is synthesized and secreted by type II alveolar cells and non-ciliated bronchiolar epithelial cells. It has a key role in the maintenance of surfactant homeostasis by reducing surface tension (3). Reduced SP-D expression or genetic variations (single-nucleotide polymorphism) have been associated with an increased risk of respiratory diseases (4,5).
Extra-pulmonary existence of SP-D has also been reported. SP-D is also expressed by epithelial cells lining various exocrine ducts, the mucosa of the gastrointestinal and genitourinary tracts, the nasal cavity, and in the brain (2). Furthermore, its presence has been demonstrated in healthy lacrimal gland, conjunctiva, cornea, and nasolacrimal duct samples (6). Other studies have shown the presence of SP-D in synovial fluid derived from patients with rheumatoid arthritis (7).
In addition to its role in surfactant homeostasis, SP-D has a critical function as a regulator of inflammation (3). It is involved in the recognition and neutralization of pathogens which promotes aggregation/agglutination and inhibition of microbial growth (8), SP-D has also been implicated in the clearance of necrotic and apoptotic cells (9). Thus, its function in the recognition of non-self and altered self makes it a potent and versatile humoral pattern recognition receptor (10)(11)(12). SP-D has also been described as a potent link between innate and adaptive immune mechanisms (13)(14)(15). Studies involving in vivo and ex vivo models of allergic inflammation revealed that SP-D can alleviate pulmonary hypersensitivity via suppression of IgE levels, promotion of Th2 to Th1 polarization (16), apoptosis induction in sensitized eosinophils via p53-mediated pathway (17), and inhibition of IgE synthesis by B cells (18). These studies highlighted a potential role of SP-D as an immune surveillance molecule. It has recently been shown that SP-D also plays a role in the control of lung cancer progression via epidermal growth factor (EGF) signaling (19). Very recently, Kaur et al. have shown that a recombinant fragment of human SP-D, composed of homotrimeric neck and C-type lectin domains, can induce apoptosis in pancreatic adenocarcinoma cell lines, such as Panc-1 (p53 mt ), MiaPaCa-2 (p53 mt ), and Capan-2 (p53 wt ), via Fas-mediated pathway (20).
In the current study, we performed a bioinformatics analysis in order to investigate whether SP-D can serve as a potential prognostic marker for human lung cancer. We extended our investigation to several non-pulmonary sites such as human gastric, breast, and ovarian cancer. We used the Oncomine dataset and the survival analysis platforms Kaplan-Meier plotter. Our results appear to suggest a likely pro-tumorigenic role of SP-D in gastric, breast, and ovarian cancers and an anti-tumor effect in lung cancer. Furthermore, we analyzed the presence of SP-D in normal and tumor human tissues via immunohistochemistry (IHC). Differential expression of SP-D was also investigated in human cells isolated from normal and tumor ovary tissues by real-time PCR. This in silico study, if validated via a retrospective study at the protein level, could be a step forward in ascertaining the importance of SP-D as a prognostic biomarker for different cancers.

Oncomine Database analysis
The expression level of SFTPD gene in various types of cancer was analyzed using Oncomine, 1 a cancer microarray database and web-based data mining platform from genome-wide expression analyses (21,22). We compared the differences in mRNA level between normal tissue and cancer. The mRNA expression level in neoplastic tissues compared to the healthy tissues was obtained as the parameters of p-value < 0.001, fold change > all, and gene ranking in the top 10%. Information about the datasets used in this study is summarized in Table 1.

Kaplan-Meier Plotter Database analysis
A Kaplan-Meier plotter database can assess the effect of 54,675 genes on survival using 10,461 cancer samples (5,143 breast, 1,816 ovarian, 2,437 lung, and 1,065 gastric cancer patients with a mean follow-up of 69/40/49/33 months) using probe sets on the HGU133 Plus 2.0 array. The prognostic significance of SP-D expression and survival in breast, ovarian, lung, and gastric cancer was analyzed by Kaplan-Meier plotter 2 (23). The hazard ratio with 95% confidence intervals and logrank p-value was also computed.

Patients and specimens
Eight fresh clinical specimens (four normal ovarian epithelial tissues and four malignant ovarian epithelial tumor tissues) were obtained from the Department of Gynaecology of IRCCS "Burlo Garofolo", in Trieste, Italy between 2016 and 2017. Cancer patients underwent laparoscopy for diagnosis of pelvic mass whereas control patients underwent laparoscopy for other indications. Tissue samples from patients were collected after informed consent following ethical approval by the Institutional Board of IRCCS "Burlo Garofolo", Trieste, Italy.

immunohistochemical analysis
For the immunohistochemical analysis, human normal and neoplastic tissues, including lung, breast, ovary, and stomach samples, were selected from the archives of the Department of rna isolation, cDna synthesis, and Quantitative realTime Pcr (qPcr) Total RNA was extracted from cells using EuroGOLD trifast (Euroclone), according to the manufacturer's instructions, and reverse-transcribed as previously described (24). qPCR was carried out using a Rotor-Gene 6000 (Corbett, Qiangen, Ancona, Italy) using iQ SYBR Green Supermix  resUlTs clinical significance of sPD expression in lung cancer We initially compared the differences in the mRNA level of SP-D between neoplastic and healthy tissues using the Oncomine platform. While analyzing Bhattacharjee's, Hou's, and Garber's datasets, we detected a significantly lower SP-D mRNA expression in lung adenocarcinoma, squamous cell carcinoma, large cell carcinoma, small cell carcinoma, and tumor carcinoid, compared to the normal lung tissue (Figure 1A, p < 0.05; Figure S1 in Supplementary Material, p < 0.05). We subsequently performed a bioinformatic analysis of SP-D mRNA expression using the Kaplan-Meier plotter dataset. As shown in Figure 1B, SP-D mRNA expression was positively related to an overall survival rate of the patients with lung cancer, stratified into lung adenocarcinoma and squamous cell carcinoma (p < 0.05). IHC staining for SP-D confirmed a differential expression in healthy and neoplastic pulmonary parenchyma. Moreover, in lung adenocarcinoma and squamous cell lung carcinoma tissues, we observed a lower expression of SP-D within its microenvironment compared to the healthy pulmonary parenchyma (Figure 2).
Pathological significance of sPD mrna expression in gastric, breast, and Ovarian cancers adenocarcinoma by Lauren's classification (Figure 3A, p < 0.05; Figure S2 in Supplementary Material, p < 0.05). According to the data from Kaplan-Meier plotter, SP-D mRNA expression was negatively related to an overall survival rate of the patients with gastric cancer (Figure 3B, p < 0.05). If stratified by Lauren's classification, SP-D mRNA expression had a statistically significant association with intestinal-type adenocarcinoma, whereas no association with diffuse-and mixed-type adenocarcinomas was found ( Figure 3C, p < 0.05). A higher expression of SP-D was negatively correlated with an overall survival rate in the patients without distant metastasis, HER2-negative and only intestinaltype adenocarcinoma (Figure 3D, p < 0.05).
The information regarding the SP-D mRNA expression in breast cancer was obtained from Zhao's, TCGA's, and Curtis's datasets, which showed that SFTPD was expressed at a lower level in invasive ductal breast carcinoma, male breast carcinoma, and breast phyllodes tumor, compared to normal breast tissues ( Figure 4A, p < 0.05; Figure S3A in Supplementary Material, p < 0.05). According to the data from Kaplan-Meir plotter, SFTPD expression was negatively linked to the high overall survival rate in breast cancer patients with Luminal-A grade-1 and grade-2 cancers (Figure 4B, p < 0.05; Figure S3B in Supplementary Material, p < 0.05). No correlation between SP-D mRNA expression and overall survival rate was observed in patients with the other characteristics (Luminal-B, HER2 + , Basal, grade-3, mutated p53, wild-type p53).
Using IHC, we observed a variable presence and distribution of SP-D in normal tissues with respect to their cancer counterpart. In fact, IHC performed on either healthy or neoplastic gastric mucosa highlighted a significantly reduced expression of SP-D in the intestinal-type adenocarcinoma compared to gastric control tissue (Figures 5A,C). Likewise, a higher expression of SP-D in the normal mammary parenchyma was detected compared to that observed within microenvironment within the invasive ductal breast carcinoma, Luminal-A (Figures 5B,D).
We collected the results from Yoshihara's and TCGA's datasets and analyzed SFTPD expression in ovarian cancer. We observed a lower expression of SFTPD mRNA expression in normal ovary than in serous cystadenocarcinoma ( Figure 6A, p < 0.05). The Kaplan-Meier plotter data, derived from stage-1 and -2 patients, showed a negative ratio between SFTPD expression and either overall or progression-free survival rates of patients with serous cystadenocarcinoma (Figure 6B, p < 0.05). However, no correlation was observed between SFTPD expression and these parameters (overall or progression-free survival rates) of patients with stage-3 and -4 ovarian cancer.

sPD expression in the Microenvironment of Ovarian cancer
The mRNA expression of SP-D was also evaluated by real-time PCR in primary cells isolated from four samples each of human ovarian serous cystadenocarcinoma and normal ovarian tissues. As shown in Figure 7A, the cells isolated from ovarian serous cystoadenocarcinoma tissues expressed more SP-D compared to the normal tissue, confirming the data obtained with the bioinformatics analysis. IHC analysis also revealed the presence and the distribution of SP-D in the normal ovary where it appeared to be localized in the ovarian epithelium lining and in the serous cystadenocarcinoma. In addition, we detected a differential expression in the normal as well as its malignant histotypes. Moreover, in the ovarian context, it showed an enrichment of SP-D expressing cells within the tumor microenvironment compared to the control tissue (Figures 7B,C).

DiscUssiOn
The importance of SP-D in the regulation of the inflammation and homeostasis and in the protection against infection and allergens in the lung and at a range of extra-pulmonary mucosal sites is well documented (26,27). However, there are recent evidences to implicate SP-D as an immune surveillance molecule against cancer (19,20). In this study, we examined the potential prognostic value of this protein in lung, gastric, breast, and ovarian cancers. We focused our attention on these tumor types because we performed a bioinformatics analysis using the Kaplan-Meier plotter dataset, a manually curated database containing the information of 54,675 genes on 5,143 breast, 1,816 ovarian, 2,437 lung, and 1,065 gastric cancer samples. This is the most updated and reliable dataset available that offers the possibility of stratifying the analysis based on different tumor settings. The bioinformatics analysis highlighted a favorable prognostic effect of SP-D mRNA expression in the lung cancer, both in adenocarcinoma and squamous cell carcinoma; on the contrary, an unfavorable prognostic effect in gastric, ovarian, and breast cancer was revealed. In particular, SP-D mRNA expression showed a negative correlation with the intestinal-type gastric adenocarcinomas, grade-1 and grade-2 breast cancers and with stage-1 and -2 ovarian cancers. No significant correlation was showed within stage-3 and -4 in breast and ovarian cancers.
Sin et al. (28) have suggested that low SP-D levels may be correlated with the development of lung cancer. They observed a reduction of the concentration of SP-D in the bronchoalveolar lavage fluid of heavy smokers that was linked to bronchial dysplasia. More recently, Hasegawa et al. (19) noted the presence of SP-D in lung cancer, and demonstrated that SP-D was able to interfere, via its CRD region, with the interaction between EGF and EGF receptor (EGFR), a tyrosine kinase receptor of the ErbB family, causing downregulation of the EGF induced signaling (19). EGFR is commonly altered in epithelial tumors and its dysregulation leads to cell proliferation, angiogenesis, invasion, and metastasis (29). Furthermore, it has been recently demonstrated that SP-D is also able to interact with the mutant form of EGFR, inhibiting its ligand-independent dimerization (29). In addition, Kaur et al. have reported the ability of a recombinant truncated form of human SP-D to induce apoptosis via TNF-α/Fas-mediated pathway in human pancreatic adenocarcinoma using Panc-1 (p53 mt ), MiaPaCa-2 (p53 mt ), and Capan-2 (p53 wt ) cell lines. Treatment of these cell lines with a recombinant form of truncated human SP-D (made up of homotrimeric neck and C-type lectin domains) for 24 h caused growth arrest in G1 cell cycle phase and triggered transcriptional upregulation of pro-apoptotic factors such as TNF-α and NF-κB. Translocation of NF-κB from the cytoplasm into the nucleus of pancreatic cancer cell lines was observed following treatment with SP-D. SP-D treatment caused upregulation of pro-apoptotic marker Fas, which then triggered cleavage of caspase 8 and 3. This study raises the possibility of using recombinant SP-D as a therapeutic molecule against pancreatic cancer irrespective of their p53 phenotype (20).
The EGFR is commonly overexpressed in non-small cell lung cancer (in 89% squamous cell carcinoma; 41% adenocarcinomas) (30), and therefore, it is considered a potential target for cancer therapy (30). The presence of SP-D in these cancers could exert a protective effect via downregulation of the EGFR pathway. It has also been shown that serum level of SP-D reflects its levels in the lung and that higher amount of SP-D in the serum correlated with better overall survival in patients with EGFR mutant adenocarcinoma undergoing treatment with gefitinib, a tyrosine kinase inhibitor (29).
Our study appears to highlight a more favorable prognosis for adenocarcinoma with respect to squamous cell carcinoma. A possible explanation of this observation may be that adenocarcinoma originates from peripheral airways progenitor cells that are able to produce SP-D. Moreover, more SP-D production may be indicative of a more differentiated cancer.
SFTPD, together with a number of genes selectively expressed in the respiratory epithelial cells, is under the control of the thyroid transcription factor 1 (TTF-1) (31,32). A recent meta-analysis showing that TTF-1 overexpression is related to a favorable prognosis for non-small cell lung carcinoma patients (33), appears to strengthen the results being reported here.
Although the overexpression of the EGFR gene has also been reported in a variety of other cancers including those of head and neck, ovary, cervix, bladder, esophagus, stomach, brain, breast, endometrium, and colon (24), the above-mentioned mechanisms cannot explain the opposite results obtained via the bioinformatics analysis of Kaplan-Mayer dataset for gastric, ovarian, and breast carcinomas, where SP-D showed an unfavorable prognostic effect. We think that the unfavorable prognostic effect of SP-D in other tumor settings can be due to its direct or indirect action on the immune population present in the tumor microenvironment (15). The following mechanisms can explain the role of SP-D in determining a tumor microenvironment favorable to tumor progression. For example, the protective effect of SP-D against breast cancer cells can be negated by the presence of hyaluronic acid, which is abundantly present in the microenvironment of a number of solid tumors (34) (Murugaiah, Bulla, and Kishore, unpublished data).
SP-D is able to reduce the expression of CD11c (15). CD11c is predominantly expressed on dendritic cells, but also on effector cells in the local tumor microenvironment, such as some macrophages, natural killer (NK), and activated T cells (25). It has been shown that low CD11c expression indicates unfavorable prognosis in patients with gastric cancer (35).
SP-D can promote production of TNF-α and IFN-γ (16,18,36). The anti-tumor effects of Th1 cells may reflect their known role in enhancing CD8 + T cell responses and activating macrophages, through the secretion of TNF-α and IFN-γ. IFN-γ can increase tumor cell class I MHC expression and sensitivity to lysis by NK cells and cytotoxic T lymphocytes (CTLs). Besides, antigen-presenting cells such as macrophages and dendritic cells can directly activate antigen-specific Th1 or CTLs, which can activate the anti-tumor immune response and are thus associated with favorable prognosis in a diverse range of cancers (37,38).
It has been demonstrated that SP-D binds to lymphocytes and suppress T cell proliferation (14) via apoptosis induction in activated PBMCs. SP-D has been shown to enhance expression of CTLA-4, a negative regulator of T cell activation and proliferation (39). In addition, monocytes expressed CTLA-4, but only the lymphocytes treated with SP-D show a significant overexpression of CTLA-4 (15). There are strong experimental and clinical evidence to suggest that T cell responses to some tumors are inhibited by the involvement of CTLA-4, one of the best-defined inhibitory pathways in T cells (40,41). In fact, tumor-infiltrating T cells often have a dysfunctional (exhausted) phenotype that is characterized by impaired effector functions and increased expression of CTLA-4 and other inhibitory molecules (40,41). Blockade of the CTLA-4 pathways is now being widely used in the clinic to reverse the dysfunctional phenotype of tumor-specific T cells and enhance their ability to kill tumor cells (41). Thus, SP-D, by increasing the expression of CTLA-4, may contribute to the inhibition of the anti-tumor immune responses.
SP-D is able to inhibit the IL-12p40 production by macrophages via the SIRPα/ROCK/ERK signaling pathway (12). IL-12p40 is a component of IL-12p70 and IL-23, and its regulation is important for both innate and adaptive immunity. IL-12p40 is a marker of M1-like macrophages and data indicate that IL-12p40 may be contributing to inducing Th 1 polarization (42,43). Macrophages derived from IL-12p40-deficient mice have a bias toward M2-like polarization (42). The production of IL-12p40 by macrophages and dendritic cells is associated with the ability to migrate to the lymph node and initiate T cell responses (44). We think that SP-D repressing the expression of IL-12p40 in macrophages may maintain the steady M2-like polarization and inhibit Th1 polarization.
SP-D has also been shown to interact with the leukocyteassociated Ig-like receptor-1 (LAIR-1) (45), known as CD305. This molecule is a transmembrane glycoprotein and is expressed on almost all immune cells as well as CD34 + hematopoietic progenitor cells. SP-D acts as a ligand for the inhibitory receptor LAIR-1, which inhibits the function of multiple types of immune cells (45), indicating that SP-D present in the tumor microenvironment may exert its immunomodulatory effect and inhibit the anti-tumor immune responses through LAIR-1 activation. Thus, the context of immune infiltration and composition of tumor microenvironment are likely to dictate the consequent effects of SP-D, and hence, tumor progression or resistance.
In summary, our in silico analysis, if confirmed with a retrospective study at the protein level, could highlight a possible role of SP-D as a novel marker for tumor prognosis in a range of cancers. The presence of SP-D could be associated with a favorable prognosis in lung cancer where it has been demonstrated to downregulate the EGF signaling, and unfavorable prognosis in non-pulmonary sites such as gastric, breast, and ovarian cancers.

eThics sTaTeMenT
This study was carried out in accordance with the recommendations of governmental guidelines, and approved by the CEUR (Comitato Etico Unico Regionale, FVG, Italy) with written informed consent from all subjects, who gave written informed consent in accordance with the Declaration of Helsinki.

aUThOr cOnTribUTiOns
Conception and design: AM and RB. Development of methodology: AG, IF, CA, and FR. Acquisition of data: BB and CA. Analysis and interpretation of data (e.g., statistical analysis, biostatistics, and computational analysis): CT, FZ, AM, BB, and CA. Writing, review, and/or revision of the manuscript: RB, UK, CA, and GR. Study supervision: RB.

FUnDing
This work was supported by grants from the Institute for Maternal and Child Health, IRCCS "Burlo Garofolo", Trieste, Italy (RC20/16); AIRC to CT; Fondazione Cassa di Risparmio Trieste to RB.