Glycopro ﬁ ling of early non-small cell lung cancer using lectin microarray technology

Objectives: Non-small cell lung cancer (NSCLC) is one of the most common malignancies in the world with a high incidence and it lacks e ﬀ ective biomarkers for early-stage detection. In this investigation, we aimed to investigate the alterations in plasma glycans related to NSCLC and assess the possibility of plasma glycopatterns as potential biomarkers for the diagnosis of NSCLC. Methods: First, plasma samples from 16 patients with early-stage lung adenocarcinoma (LUAD), 16 patients with early-stage Lung squamous cell carcinoma (LUSC), and 16 healthy volunteers, were selected for inclusion in this study to probe the di ﬀ erence in plasma glycopatterns using lectin micro-arrays. Then, the diagnostic e ﬀ ectiveness of the candidate lectins was evaluated using ROC. Results: In contrast to the NL group, seven candidate lectins o ﬀ ered potential diagnostic utility in the NSCLC (LUAD and LUSC) group. F17AG was signi ﬁ cantly altered in LUSC with an AUC of 0.818 (adj.P.Val<0.05) compared to NL samples. There were 20 di ﬀ erentially expressed lectins in the LUAD group compared to the NL group. Based on the AUC values (AUC>0.800) and the normalized ﬂ uorescence intensities of the lectins, we selected eight lectins, GAL2, PTL-1, GNA, SSA, LENTIL, CA, PHA-E, and MAA to perform logistic regression analysis, and found that the combination of these eight candidate lectins had high diagnostic potential. Conclusions: The results


Introduction
Lung cancer is one of the most prevalent cancers in the world, and its five-year survival is very low in the advanced stage [1,2].The mortality from lung cancer in China is predicted to rise by approximately 40 % between 2015 and 2030 [3].
The most common types of lung cancer are non-small cell lung cancer (NSCLC) and small cell lung carcinoma (SCLC), and NSCLC is further subdivided into lung adenocarcinoma (LUAD), large cell lung cancer, and lung squamous cell carcinoma (LUSC) [4,5].Lung cancer is frequently discovered at an advanced stage, which means that the ideal therapy window is lost and the patient's survival is poor.As lung cancer remains a disease with a high mortality rate, some researchers have used natural myricetin and its derivatives to develop new bioactive molecules to treat lung cancer [6].Another study analyzed scRNA-seq data from the tumor tissue of peripheral blood of NSCLC patients to bioinformatically explore cellular heterogeneity, DEGs, associated biological pathways, PPI networks, key genes, miRNAs and TFs, and other genetic biomarkers intended for the diagnosis and prognosis of NSCLC [7].Early-stage detection has emerged as a key strategy for reducing NSCLC incidence and mortality.Early detection methods for lung cancer include sputum cytology, breath testing, computed tomography (CT) scans, low-dose computed tomography screening, and CXR [8][9][10][11].There are some limitations to these diagnostic methods.For example, the sensitivity of CXR is low, and CT scan images contain much information about nodules, which need to be accurately evaluated [11].Therefore, a more efficient early-stage NSCLC diagnostic approach is still desperately needed in clinical practice.
At present, glycobiology is a hot topic in cancer research [12,13].Almost all secreted proteins and cell membranes carry glycans.Glycans are involved in biological processes such as cell communication and molecular signaling [14].Many tumor-associated glycans, such as sialic acid Tn (sTn), polysialic acid (PSA), and sialic acid Lewis X (sLex), are abnormally produced in cancer and are essential for biological processes such as growth and migration [15][16][17][18].Therefore, it is of great clinical significance to explore the association between glycans and diseases.
Studies of glycoprotein-bound glycans, however, are not simple.The structure of glycans has diversity, which is not the diversity of constituent glycans but more the diversity of isomers.High-performance liquid chromatography (HPLC) and mass spectrometry (MS) can make it difficult to distinguish glycans from isomers of different structures, and these methods require glycans to be dissociated from proteins [19].Lectins are glycan-binding proteins that are purified from trees, legumes, etc., and they are highly specific for specific glycans.The lectin microarray composed of lectins can provide differential glycan analysis in a high-throughput manner, similar to an antibody microarray, which does not require prior dissociation of glycans from proteins.Therefore, the lectin microarray is one of the most effective molecular tools for the analysis of glycans [20].As a result, lectins are frequently used for glycosylation research, particularly in the identification of glycans that change during cancer [20,21].For example, a study explored the changes in salivary protein glycosylation associated with gastric cancer through 37 different lectins and constructed a diagnostic model of gastric cancer and atrophic gastritis through 15 candidate lectins [22].Another study mined potential biomarkers by lectin analysis of the entire glycan profile of serum samples from prostate cancer patients, benign prostatic hyperplasia, and healthy individuals [23].
Currently, due to the lack of effective biomarkers for the early-stage diagnosis of NSCLC, plasma is an easily accessible body fluid that is rich in protein, amino acids, and lipids in the human body [24].Therefore, in this study, we analyzed a total of 48 human plasma samples, including 16 lung squamous cell carcinoma (LUAD), 16 lung squamous cell carcinoma (LUSC), and 16 healthy volunteers (NL), using a lectin microarray containing 95 lectins, aiming to explore the relevance of changes in plasma glycopatterns associated with NSCLC.The alterations in glycopatterns were also compared between the LUAD, LUSC, and NL groups, and ROC analysis was performed based on candidate lectins to assess the potential of altered plasma glycopatterns as a biomarker for early-stage NSCLC diagnosis.

Plasma sample collection
This study was approved by the Research Ethics Committee of West China Hospital, Sichuan University.And it involved 48 human plasma samples, including 16 LUAD, 16 LUSC, and 16 NL samples.All patients in the study were in the early-stage.All recruited patients had diagnostic pathological information and clinical characteristics (Table 1).All plasma samples were collected, processed, and stored at −80 °C.

Lectin microarray
The reagents used in this experiment are contained in the GA-Lectin-95 Kit, purchased from RayBiotech, located in Peachtree Corners, Georgia.This sample was first dialyzed and then biotin-labeled samples were obtained.The slides containing the lectin chips were removed and equilibrated at room temperature for 20-30 min.Subsequently, 100 µL of 1 × closure solution was added to each chip well and closed at room temperature.The closure solution was withdrawn, and 100 µL of the sample was added to each well.Dilute the biotin-labeled plasma 50-fold according to the pre-experimental closure solution.Draw the samples and wash the slides.Afterward, the Cy3 equivalent was added to the chip by adding 1.4 mL of the closure solution and incubating for 1 h protected from light.The slide was washed again.An InnoScan 300 laser scanner was used to scan the signal with the following parameters: Instrument model: InnoScan 300 Microarray Scanner; Manufacturer: Innopsys; Place of origin: Parc d'Activités Activestre, 31,390 Carbonne -France; WaveLengh: 532 nm; Resolution: 10 µm.

ROC analysis
ROC analysis was performed on the target lectins using the pROC package via R4.0.2 software to assess their sensitivity and specificity.
The AUC was used to estimate the diagnostic value.A logistic regression analysis was constructed for the lectin combination.

Results
This

Alterations in plasma glycopatterns among the NSCLC and NL groups
The NFIs of each lectin showed different expression levels of plasma glycopatterns.The lectin signal patterns were divided into three categories to evaluate whether the changes in plasma glycopatterns were significant, (1) the results showed significant upregulation (fold change>1.2,adj.P.Val<0.05),(2) significant downregulation (fold change<0.83,adj.P.Val<0.05),and (3) the results showed no significant difference (fold change was 0.83-1.2,adj.P.Val>0.05).Glycopatterns were altered in NSCLC (LUAD and LUSC) compared to NL samples, with seven NFIs for lectins significantly altered (Figure 2A) (Table 2).Among them, the GlcNAc binder F17AG, and the NeuAcα2-6Gal/GalNAc binder SAMB exhibited significantly increased NFIs in patients with NSCLC compared with the NL group (all adj.P.Val<0.05).Gal and GalNAc binder PTL-1, N-acetylgalactosamine binder SSA and SHA, αMan and αGlc binder LcH A, branched LacNAc and Gal binder GAL1 exhibited significantly decreased NFIs in patients with NSCLC compared with the NL group (all adj.P.Val<0.05).In this study, ROC curves were drawn to judge the role of each differentially expressed lectin in the diagnosis of different comparison groups.If AUC>0.700, the lectin can be considered an independent diagnostic potential factor.In the NSCLC group and the NL group, the AUC values of F17AG (AUC=0.809)and PTL-1 (AUC=0.867)were greater than 0.800, which had better diagnostic potential for NSCLC.The AUC values of the other five lectins were all between 0.700 and 0.800 (Figure 2B), which offered only moderate performance.As shown in Figure 2C, these seven differential lectins clustered 48 samples well.Plasma glycopatterns in the NSCLC group and the NL group were well differentiated.Therefore, plasma glycopatterns may be used as biomarkers to provide diagnostic information.

Alterations in plasma glycopatterns among the LUSC and NL groups
In our study, changes in NFIs for lectins in the LUSC group were not as significant as those in the LUAD group compared to the NL group.According to lectin screening of differential NFIs according to the corrected p value (adj.P.Val<0.05).p value correction was utilized to correct for false positives and to reduce the number of false positives.It was found that F17AG (LogFC=5.40,adj.P.Val=0.0008)conformed to screening conditions, and F17AG was significantly upregulated in LUSC, indicating that GlcNAc in LUSC was changed compared with that in the NL group (Figure 4A and B).The AUC of F17AG was 0.818, the specificity was 0.688, and the sensitivity was 0.938; thus, F17AG can be used as a potential diagnostic biomarker for LUSC and healthy people and provides a basis for distinguishing LUSC from healthy people (Figure 4C).

Correlation analysis and combination analysis of candidate lectins
We analyzed the correlation of each group of candidate lectins, blue represents a positive correlation, red represents a negative correlation, the darker the color, the stronger the correlation, and X represents that there is no significant difference in the correlation.As shown in Figure 5A.Our study found that the correlation of PTL-1 with SHA and SSA was greater than 0.7500 in the comparison of the NSCLC and NL groups.In Figure 5B, we found that among the 20 candidate lectins in the LUAD and NL groups, the correlation coefficient between MOA and GAL1 was 0.9479, with a significant correlation.The correlation coefficients between SSA and PTL-1, GAL3, SHA, and VVA-M were all greater than 0.8000.The correlation coefficients between GAL3 and GNA    and GAL2 were 0.8041 and 0.8275, respectively, indicating that they were significantly correlated.In addition to correlation analysis, we conducted a combined analysis of the candidate lectin factors.In the NSCLC group compared with the NL group, seven candidate differential lectins provided excellent diagnostic performance (AUC=0.990)(Figure 5C).In the LUAD group compared with the NL group, there were 20 candidate lectins.According to the AUC value and the intensity of NFIs of the lectin, we selected eight lectins, GAL2, PTL-1, GNA, SSA, LENTIL, CA, PHA-E, and MAA (AUC values are greater than 0.800), and carried out the logistic analysis.The combination of these eight candidate lectins provided excellent diagnostic value (AUC=1) (Figure 5D).

Discussion
Protein glycosylation plays an important role in the study of cancer and alterations in glycoproteins as well as proteoglycans are a frequent feature of cancer cells [13].
Glycans are rich in biological information and are associated with important biological roles, including intercellular adhesion, angiogenesis, and cell migration [13].Changes in the sialylation and fucosylation of N-and O-glycans play a role in the development of brain cancer [24].In another study of altered serum glycans in NSCLC, the precursor of O-linked glycans, Tn antigen, was found to be significantly increased in lung adenocarcinoma serum compared to sera from healthy volunteers [25].One study analyzed 633 colorectal cancer (CRC) patients and 478 age-and sex-matched controls to explore changes in plasma N-glycosylation between colorectal cancer patients and controls, and the study found statistically significant differences observed in plasma N-glycosomes at all stages of CRC [26].Plasma indicators for α2-6 sialylation, β1-4 branching, β1-6 branching, and outer-arm fucosylation can differentiate between bladder cancer patients and healthy controls [27].Consequently, glycosylation alterations in glycoconjugates present on surfaces or secreted by cancer cells are commonly used in the field of cancer biomarkers [28].New glycan analysis tool lectin microarrays make it possible to quickly and sensitively analyze complex glycans without releasing them from their protein-bound state.High-throughput screening of glycans in intricate biological samples is now possible because of recent advancements in lectin microarray technology [29].This study's objective was to compare healthy samples with those from patients with early-stage NSCLC (LUAD and LUSC).Differences in NFIs of various lectins, which are detected by lectin chip technology, imply distinct glucose patterns in the plasma of NSCLC patients.A total of 48 samples were involved in our study, including 16 LUAD, 16 LUSC, and 16 NL samples.Both LUAD and LUSC were early-stage samples.The lectin chip we used contains 95 lectin factors, so it covers most of the glycan structures, which can more comprehensively detect the changes in plasma glycopatterns in early-stage NSCLC patients compared with NL patients and evaluate whether it can be used as a biomarker for early diagnosis.Through lectin signal detection and ROC analysis, it was found that PTL-1 and F17AG had significant changes in early-stage NSCLC patients, and the AUC of PTL-1=0.867 and the AUC of F17AG=0.809, which had good clinical diagnostic potential.The specific glycan structures that bind to PTL-1 are D-galactose and N-acetylglactosamine, and the one that binds to F17AG is GlcNAc, which means that these glycan structures in the healthy group exhibit signified changes in early NSCLC (LUAD + LUSC).It has been shown that the treatment of cancer cells with D-Gal transcriptionally upregulated genes is associated with necrophagocytosis and autophagy (Atg5, TIGAR).D-Gal does not activate caspase-3 but activates indicators of cellular autophagy [30].Therefore, the downregulation of D-Gal in our study may be associated with the regulation of inhibition of tumor autophagy and promotion of tumor growth.Later, we discussed the LUAD and NL groups and the LUSC and NL groups.The study found that in LUAD, GalNAcα1-3Gal (Blood Group A) binder GAL2 was found to have good clinical diagnostic potential by ROC analysis, and its AUC=0.922,specificity was 1.000, and sensitivity was 0.750.GalNAcα1-3Gal was identified as a prognostic marker in cervical cancer [31].It has been proposed that a macrophage galactose-type lectin (MGL) expressed by macrophages and dendritic cells specifically binds GalNAcα1-3Gal, and the loss of this antigen on tumor cells reduces or prevents recognition of MGL by expressing macrophages and dendritic cells, leading to evasion of host defenses [32].Human serum contains GalNAcα1-3Gal binding antibodies, and tumor cells lacking GalNAcα1-3Gal avoid recognition by these antibodies; thus, tumor immune escape occurs [33].Therefore, in our study, downregulation of GalNAcα1-3Gal in early LUAD plasma may be associated with immune escape from tumors and promote tumor development.
The analysis of the experimental results of our study was screened with a more stringent corrected p value.It was found in our study that F17AG was significantly increased in LUSC.The AUC of F17AG was 0.818.Five of our LUSC samples had COPD, which may also have been a factor in the results.Chronic obstructive pulmonary disease (COPD) is a common comorbidity of NSCLC, and patients with COPD have a higher risk of developing NSCLC than those without airway obstruction [34].Therefore, further studies with larger samples are needed to explore the alterations in plasma glucose profiles in the healthy population vs. COPD patients, in COPD patients vs. early NSCLC patients, in COPD patients vs. early NSCLC patients with concurrent COPD, and in the healthy population vs. early NSCLC patients with concurrent COPD, to further validate plasma glycan biomarkers for the diagnosis of early LUSC.
Overall, the NFIs of F17AG were significantly enhanced in the three groups comparing the early-stage NSCLC vs. NL group.LUAD vs. NL group, and LUSC vs. NL group.The glycan that binds to F17AG is N-acetylglucosamine.The results indicated that N-acetylglucosamine was significantly increased in the plasma of patients with early-stage NSCLC.N-acetylglucosamine (GlcNAc) is a monosaccharide derivative of glucose, and can also be found in glycoproteins, proteoglycans, GAGs, and other components of connective tissue [35].One study reported that GlcNAc was identified by metabolic methods as the most predictive metabolic marker for the severity of systemic mastocytosis, a hematopoietic tumor [36].Compared with the healthy group, the increase in GlcNAc in NSCLC patients may be related to the metastatic growth of cells and metabolism.Thus, N-acetylglucosamine is altered in the plasma of patients with early-stage NSCLC and it may serve as a potential biomarker to differentiate between healthy and early-stage NSCLC patients.
This study still has some limitations.First, the sample size is small, and more samples need to be included in future experiments.Second, the content of the components in plasma is complex and has some individual variation related to the age of the patient, their physical condition, and whether they have other comorbidities.In our study, the LUSC group was older than the healthy group, and 87.5 % of the LUSC patients had a history of smoking.Smoking increases oxidative stress and the resulting DNA damage, which inhibits DNA repair and may be one of the driving factors for NSCLC [37].Therefore, these could have been factors that influenced the results of our experiment.Therefore, the role of altered glycopatterns in plasma for the diagnosis of early-stage NSCLC needs to be further explored and validated to determine the diagnostic value of these diagnostic values of glycan biomarkers to differentiate NSCLC from benign lung disease and other types of cancer.Although there are limitations to this study, our study explored changes in plasma glycopatterns to help explore potential biological markers for early-stage NSCLC.We are also curious to see if mining potential biomarkers through lectins can be combined with other diagnostic techniques to improve the efficiency of early-stage lung cancer diagnosis.

Conclusions
This study investigated the altered plasma glycan patterns associated with early-stage NSCLC, systematically comparing the different plasma glycan patterns between NSCLC (LUAD and LUSC) and NL.The glycan pattern was altered in NSCLC compared to NL samples, with seven candidate lectins (F17AG, PTL-1, SSA, SAMB, GAL1, LcH A, SHA) having significantly altered NFIs.Combined analysis of these seven lectins revealed that these candidate lectins may have a discriminatory ability for NSCLC (AUC=0.990).F17AG was also significantly upregulated in LUSC with an AUC of 0.818 (adj.P.Val<0.05),suggesting that F17AG may have the potential to discriminate NL from LUSC.There were 20 candidate differential lectins in the LUAD group compared to the NL group, of which GAL2, PTL-1, GNA, SSA, LENTIL, CA, PHA-E, and MAA, the combination of eight candidate lectins, had a high potential to discriminate LUAD from NL (AUC=1.000).This study provides key information for differentiating early-stage NSCLC from NL based on altered plasma glycopatterns, which have great potential as biomarkers for the diagnosis of non-small cell lung cancer.

Figure 1 :
Figure 1: Detection of glycopatterns in LUAD samples, LUSC samples, and NL samples using the lectin chip.(A) Flow chart of the lectin chip.The samples were labeled with biotin, blocked and incubated, incubated with Cy3 equivalent dye-streptavidin, and then the signal was scanned with a laser scanner such as InnoScan 300, with scanning parameters WaveLengh: 532 nm; resolution: 10 µm.(B) 48 samples are columns and 95 lectins are rows.The color and intensity of each square represent the level of expression relative to the rest of the data in the row.Red, high, blue, low; white is medium.

Figure 2 :
Figure 2: Detection of glycopatterns in NSCLC (LUAD + LUSC) samples and NL samples using the lectin chip.(A) Difference volcano plot of NSCLC vs. NL group.Red indicates up-regulated lectins, blue indicates down-regulated lectins, and grey indicates insignificant differences.(B) Lectin factors with significantly different signals and ROC analysis of candidate lectins in the NSCLC group compared with the NL group, * p≤0.05.(C) The samples are rows, and the seven candidate lectins are columns.The color and intensity of each square represent the expression level relative to the rest of the data in the row.Red, high, blue, low; white is medium.

Figure 3 :
Figure 3: The altered plasma glycopatterns among the LUAD and NL groups.(A) Volcano plot of the difference between the LUAD group and the NL group.Red indicates up-regulated lectins, blue indicates down-regulated lectins, and grey indicates insignificant differences.(B) Lectin factors with significantly different signals and ROC analysis of candidate lectins in the LUAD group compared with the NL group, * p≤0.05.(C) The samples are rows, and the 20 candidate lectins are columns.The color and intensity of each square represent the expression level relative to the rest of the data in the row.Red, high, blue, low; white, medium.

Figure 4 :
Figure 4: Changes in plasma glycopattern in the LUSC and NL groups.(A) Volcanic map of differences between the LUSC and NL groups.Red indicates up-regulated lectins, blue indicates down-regulated lectins, and grey indicates insignificant differences.(B) Differential expression of F17AG, * p≤0.05.(C) ROC analysis of F17AG.

Figure 5 :
Figure 5: Correlation analysis and combination analysis of candidate lectins.(A) Correlation analysis of the NSCLC group and the NL group.(B) Correlation analysis of the LUAD group and NL group.(C) ROC analysis of candidate lectins between the NSCLC group and NL group.(D) ROC analysis of candidate lectins between the LUAD group and NL group.

Table  :
Clinical information of samples.
study was divided into three groups: the LUAD group, the LUSC group, and the NL group, with each group, having 16 cases.All plasma samples were detected by the lectin microarray (Figure1A.Figure created with BioRender.com).The lectin microarray layout contains 95 lectins and is currently the largest lectin array, which can cover different glycans to the greatest extent, and details are shown in TableS1.The original data obtained by chip scanning were processed by Raybiotech's Excel-based analysis software for chip background removal and normalization.The specific binding of lectins can recognize different carbohydrates, so the different normalized fluorescence intensities (NFIs) of lectins can reflect the different glycopatterns in plasma samples.As shown in Figure1B, plasma samples were analyzed in the heatmap with 48 samples as columns and 95 lectins as rows, showing different glycopatterns in 48 different samples.The glycopatterns of the LUAD group were quite different from those of the NL group and LUSC group.In our study, changes in NFIs for lectins in the LUSC group were not as significant as those in the LUAD group compared to the NL group.

Table  :
The differentially expressed lectins in the NSCLC group and the NL group.

Table  :
The differentially expressed lectins in the LUAD group and the NL group.