Comprehensive Integrative Analysis Reveals the Association of KLF4 with Macrophage Infiltration and Polarization in Lung Cancer Microenvironment

Macrophage polarization and infiltration to the tumor microenvironment (TME) is a critical determining factor for tumor progression. Macrophages are polarized into two states—M1 (pro-inflammatory, anti-tumorigenic and stimulated by LPS or IFN-γ) and M2 (anti-inflammatory pro-tumorigenic and stimulated by IL-4) phenotypes. Specifically, M2 macrophages enhance tumor cell growth and survival. Recent evidences suggest the pivotal role of microRNAs in macrophage polarization during the development of Non-small cell lung cancer (NSCLC), thus proposing a new therapeutic option to target lung cancer. In silico analysis determined cogent upregulation of KLF4, downregulation of IL-1β and miR-34a-5p in NSCLC tissues, consequently worsening the overall survival of NSCLC patients. We observed a significant association of KLF4 with macrophage infiltration and polarization in NSCLC. We found that KLF4 is critically implicated in M2 polarization of macrophages, which, in turn, promotes tumorigenesis. KLF4 expression correlated with miR-34a-5p and IL-1β in a feed-forward loop (FFL), both of which are implicated in immune regulation. Mechanistic overexpression of miR-34a-5p in macrophages (IL-4 stimulated) inhibits KLF4, along with downregulation of ARG1, REL-1MB (M2 macrophage specific markers), and upregulation of IL-1β, IL-6, (M1 macrophage specific markers), demonstrating macrophage polarization switch from M2 to M1 phenotype. Moreover, co-culture of these macrophages with NSCLC cells reduces their proliferation, wound healing, clonogenic capacity and enhanced NO-mediated apoptosis. Further, transfection of miR-34a-5p in NSCLC cells, also degrades KLF4, but enhances the expression of KLF4 regulated genes—IL-1β, IL-6 (pro-inflammatory mediators), which is further enhanced upon co-culture with IL-4 stimulated macrophages. Additionally, we observed a significant increase in i-NOS/NO content upon co-culture, suggesting polarization reversion of macrophages from M2 to M1, and eventually leading to anti-tumor effects. Our findings thus show a significant role of KLF4 in tumorigenesis and TAM polarization of NSCLC. However, miR-34a-5p mediated targeting of these molecular networks will provide a better therapeutic intervention for NSCLC.


Introduction
Lung cancer (LC) is one of the prime causes of cancer-associated mortality worldwide. It is divided into two types: (a) more fatal but less recurrent Small cell lung cancer KLF4 is a highly conserved nuclear TF, containing a zinc-finger like domain, bearing N-terminal transcriptional activation domain, C-terminal transcription inhibitory region along with nuclear localization sequences for the regulation of protein interactions [19]. It is critically intricated into several aspects of cellular processes, including proliferation, differentiation, somatic reprogramming, tissue homeostasis and apoptosis [20]. Reports have identified its role in induction of fibroblasts into pluripotent stem cells, and designated it as the first gene regulating the transcription of C-myc, Oct4, and Sox2 resulting in transformation of non-neoplastic epithelial cells to cancer stem cells [21][22][23]. It is found to express in both nucleus and cytoplasm; however, the subcellular localization affects the formation of cancer stem cells, thereby affecting drug resistance too [24]. The diversity of biological roles of KLF4 can be attributed to multilevel regulation at transcription, miRNA, alternative splicing, post-translational modification (PTM), protein stability, protein interactions and subcellular localization levels. It is a bifunctional TF that may either activate or repress transcription, depending upon the target gene. Thus, cell type and context may determine the tumor suppressive and oncogenic role of KLF4. It is implicated in several roles in cancer; however, nearly all of the studies have suggested its tumor suppressive action, e.g., in gastric, colon and colorectal cancers, esophageal, lung and bladder cancer, T-cell leukemia and medulloblastoma [20,[25][26][27][28][29][30][31]. Contrarily, it is overexpressed in head and neck, breast and skin cancer and acts as a transforming oncogene [32,33]. Moreover, tumor subtypes and microenvironment also play a convincing role. Further, its overexpression is associated with poor prognosis in several cancers [34]. Studies have identified its correlation with the clinicopathological features of patients with LC and is considered to be a probable biomarker of LC [20,35]. A recent study has identified a differential expression of KLF4 in LC subtypes, with a significant downregulation in NSCLC and upregulation in SCLC. In addition, a compelling overexpression was also observed in advanced stages of NSCLC [20]. Furthermore, it is a demanding regulator of macrophage polarization. It has been found to be highly induced in STAT6 regulated M2 macrophages and reduced in M1 macrophages via reclusion of coactivators needed for NF-κβ activation. Moreover, KLF4 deficient macrophages display increased expression of pro-inflammatory genes, bactericidal effects and altered metabolism [36].
In this study, we have comprehensively analyzed the messenger RNA (mRNA) and microRNA (miRNA) transcriptome of NSCLC patients obtained from National Center for Biotechnology Information (NCBI)-Gene Expression Omnibus (GEO) datasets and their correlation with the prognosis of cancer patients using the Kaplan-Meier (KM) plotter database. We performed the pathway enrichment analysis to identify the immune related pathways in which differentially expressed genes (DEGs) were functionally enriched. Further, we analyzed the interactions between DEGs and differentially expressed miRNAs (DEMs) to identify the significant interacting Feed-Forward Loop (FFL) among them. We validated the expression levels using UALCAN and cBioportal databases. Further, we explored the infiltration of immune cells using Tumor Immune Estimation Resource (TIMER) database. We also performed in vitro co-culture of macrophages with NSCLC cells to validate the significant associations with components of FFL and macrophage polarization. The findings of this study have shed light on the considerable role of KLF4 in infiltration and polarization of TAMs. Further, the inhibition of KLF4 by miR-34a-5p leads to consequent inhibition of M2 markers and promotion of M1 markers. This has enabled us to identify the probable strategy of reversing the polarization of TAMs. More strikingly, reversal of TAM polarization promotes antitumor activities in NSCLC.

Identification of NSCLC-Specific DEGs, DEMs and Pathway Enrichment Analysis
The miRNA and mRNA expression profiles of NSCLC patients were downloaded from the NCBI-GEO (https://www.ncbi.nlm.nih.gov/geo/, accessed on 15 March 2021) database [37]. GEO was queried utilizing 'Non-small cell lung cancer' and 'NSCLC' as appropriate keywords. The search results were further trimmed down by applying following inclusion criteria: (1) the datasets must be obtained from non-tumor and adjacent tumor tissues of NSCLC patients; (2) the datasets must be standardized or raw; (3) the datasets must be miRNA/mRNA transcriptome data of the whole genome; (4) the datasets should be "expression profiling by array" type and its samples must belong to "Homo Sapiens"; and (5) the datasets must have more than 100 samples. GEO2R ( https://www.ncbi.nlm.nih.gov/geo/geo2r/, accessed on 15 March 2021) was utilized to detect the DEGs and DEMs between NSCLC and normal controls expression profiles. The probe IDs with their corresponding Hugo Gene Nomenclature Committee (HGNC) gene symbols as NA were discarded. Duplicate gene symbols corresponding to multiple probe IDs were removed by averaging their relative expression values. The adjusted p-values (adj.p) were computed using the Benjamini-Hochberg (BH) False Discovery Rate (FDR) method to correct for the likelihood of false-positive results [38]. DEGs were screened using an adj.p < 0.001 and |log 2 ( f old change)| > 2 as the preferred threshold. Additionally, DEMs were screened using an adj.p < 0.0001. The up and downregulated DEGs were filtered considering adj.p < 0.0001 with log 2 ( f old change) > 2 and adj.p < 0.0001 with log 2 ( f old change) < −2. Additionally, up, and downregulated DEMs were filtered considering adj.p < 0.0001 with log 2 ( f old change) > 0 and adj.p < 0.0001 with log 2 ( f old change) < 0. Reactome library available within the Enrichr database [39] was utilized for pathway enrichment analysis of DEGs where top 10 significant pathways corresponding to p-value < 0.05 were selected.

Extraction of Significant NSCLC-Specific miRNAs and TFs
TF-gene regulation: TF-gene interaction pairs were extracted from the ChIPBase v2.3 database [40] and Enrichr platform TF libraries like ChEA, TRANSFAC and JASPAR PWMs. TFs with binding sites located within 1kb upstream region were retrieved from ChIPBase database, whereas the TFs with a significant p-value < 0.05 were retrieved from Enrichr TF libraries. The list of TFs were scrutinized via literature studies and only NSCLC-specific TFs were retained miRNA-gene/TF repression: miRNA-gene/TF pairs were extracted from miRWalk v3.0 [41], miRSearch v3.0 [42] and Starbase v2.0 [43] databases, respectively. Parameters like binding gap = 1, 3 UTR region, and score > 0.95 were considered as the cut-off for significant miRNAs retrieval from miRWalk. miRNAs having a significantly high score and denoted by the green-colored band were extracted from miRSearch.
The list of miRNAs obtained from miRNA-gene/TF repression pairs overlapping with NSCLC-specific DEMs and literature-validated NSCLC-specific miRNAs were retained. Finally, all the three types of regulatory interaction pairs were altered with respect to these finalized NSCLC-specific TFs and miRNAs. All these molecular interactions were then merged to construct a NSCLC-specific 3-node miRNA FFL [44] and subsequently visualized using Cytoscape v3.8.2 [45].

Overall Survival (OS) Analysis
The KM plotter database (https://kmplot.com/analysis/, accessed on 31 March 2021) [46] was used to evaluate the prognostic value of KLF4, IL-1β and miR-34a-5p in The Cancer Genome Atlas (TCGA)-LUAD cohort. Clinical data from 513 LUAD patients (mRNA-Seq and miRNA-Seq) including smoking history, pathological grades, histology, clinical staging, and gender were retrieved from the KM plotter. The cohorts of patients were divided by median expression values via auto-select best cut-off. In accordance with the median expression level of these genes, the LUAD patient samples were bifurcated into two groups, i.e., high, and low expression groups. Then, the log-rank p-values, number at risk, hazard ratio (HR), and 95% confidence interval (CI) were computed. p-value < 0.01 was considered as the statistically significant threshold.

Validation of KLF4 and IL-1β Using UALCAN and cBioPortal
UALCAN (http://ualcan.path.uab.edu/, accessed on 3 April 2021) [47] and cBioPortal (https://www.cbioportal.org/, accessed on 3 April 2021) [48,49] web-based tools were queried in order to validate the roles of KLF4 and IL-1β in TCGA-LUSC and Pan-lung cancer cohorts, respectively. Validation of expression levels of KLF4 and IL-1β was done using UALCAN with respect to clinical characteristics such as nodal metastasis and TP53 mutation status in LUSC. Graphical summarization of genomic alterations including mutations and CNA (amplifications and homozygous deletions) within non-small cell lung cancer dataset (TCGA, Pan Cancer Atlas) was presented using cBioPortal.

Tumor Immune Infiltration Analysis
TIMER web-based tool (http://timer.cistrome.org/, accessed on 3 April 2021) [50] was queried to explore the interconnection between tumor-infiltrating immune cells in TCGA-LUAD and LUSC patients and expression levels of KLF4 and IL-1β. Spearman's test was used and p-value < 0.05 was regarded as the statistically significant threshold.

Co-Culture
Transfected THP-1 cells were directly co-cultured with A549 cells in a density of 1: 10 for 24 h. On the other hand, H1299 cells (transiently transfected with miR-mimic and scrambled) were directly co-cultured with THP-1 cells (stimulated with 20 ng/mL IL-4) in a density of 1:10 for 24 h. Co-culture conditioned medium (CM) was collected, briefly centrifuged, and stored at −80 • C for further use.

Quantitative Real-Time PCR (qRT-PCR)
Total RNA was isolated from transfected THP-1 and H1299 cells alone as well as when co-cultured using TRIzol reagent (Ambion, Austin, TX, USA) according to the manufacturer's protocol. cDNA was then reverse transcribed using the iScript cDNA synthesis kit (Bio-Rad, Hercules, CA, USA). qRT-PCR analyses for mRNA of KLF4, REL-1MB, IL-6, IL-1β, and GAPDH were performed by using iTaq Universal SYBR Green Supermix (Bio-Rad, Hercules, CA, USA). PCR was performed using 7900HT Fast Real-time PCR System (Applied Biosystems, Waltham, MA, USA) using GAPDH as an endogenous control. Relative quantification from real-time data is presented, based on the calculation of 2 −∆∆Ct . The primer sequences are available in Supplementary Material (Oligonucleotide sequences and list of antibodies used).

One-Step qRT-PCR
To confirm the transfection of miR-34a-5p, cDNA was synthesized in 5X miScript Hi-Spec Buffer using mi-Script (II) RT kit (Qiagen, Hilden, Germany) using manufacturers' protocol. miR-34a-5p expression was measured by the miScript SYBR Green PCR Kit (Qia-gen, Hilden, Germany) using manufacturers' protocol. Relative expression was quantified using ∆∆Ct method and RNU6 as a normalizing control.

Flow Cytometry
Measurement of apoptosis was done in A549 and H1299 cells after treatment with CM for 48 h using Dead Cell Apoptosis Kit with AnnexinV-FITC and PI, for flow cytometry (Invitrogen, Waltham, MA, USA) using the manufacturers' protocol. Fluorescence emission was measured by BD FACS Aria™ III (BD Biosciences, Franklin Lakes, NJ, USA) using BD FACSDiva software.

Cell Viability and Cell Proliferation
A549 and H1299 cells were treated with CM for 48 h, and the cell viability was determined using 0.5% Trypan blue staining. Cell proliferation was determined using MTT assay (Cat #634844, Merck, Bangalore, India). Briefly, 7500 cells/well were seeded in a 96-well plate and treated with CM for 48 h, after which 0.5 mg/mL MTT was added for 3 h. DMSO was added, and the cells were incubated at room temperature for 30 min. Finally, absorbance was measured at 540 nm.

Clonogenic Assay
A549 and H1299 cells were treated with CM for 48 h and then allowed to form colonies in a serum-deprived medium for 12 days. Colonies were stained with 0.5% crystal violet in methanol, counted, and solubilized in a destain solution of 10% acetic acid. OD of the extracts were spectrophotometrically determined at 570 nm.

In Vitro Scratch Assay
A549 and H1299 cells were treated with CM for 48 h and were allowed to grow as a monolayer after which a scratch was created using a sterile pipette tip. Scratched cells were removed by washing with PBS and allowed to migrate through the scratch for 48 h in a serum-deprived medium. The cells were photographed at different intervals of time and the area covered by scratch was measured using ImageJ software.

Nitrite Assay
The concentration of nitrites in CM was determined using Griess reagent (Chromous Biotech, Bangalore, India) using the manufacturers' protocol. NaNO2 was used to prepare respective standards.

Dual-Luciferase Assay
p-MIR reporter plasmid containing 3 UTR of KLF4 and plasmid containing the mutated sequence of KLF4 were purchased from Addgene. Dual-luciferase assay was performed using the DLR assay kit (Cat # E1910, Promega, Madison, WI, USA) as per the manufacturer's protocol using plasmid expressing the Renilla luciferase gene (pRL, Promega, Madison, USA, 20 ng) as a control. Luciferase activity was assayed on an InfiniteM200 Pro Multimode Reader (TECAN, Seestrasse, Männedorf, Switzerland). Relative f old change in luciferase activity was estimated, following normalization to Renilla luciferase activity.

Statistical Analysis
All experiments were performed thrice independently, and the data is presented as mean ± SEM. Student's t-test, One-Way Analysis of Variance (ANOVA) and Two-Way ANOVA were used to analyze the statistical significance of data. p-value < 0.05 was considered to be statistically significant. Statistical calculations were performed using Graph Pad Prism 7.0.

Identification of NSCLC-Specific DEGs, DEMs and Pathway Enrichment Analysis
Based on the specified searching and inclusion criteria, we chose NSCLC-specific mRNA and miRNA expression profiles possessing accession numbers GSE75037 and GSE53882. In consideration with the abovementioned threshold, a total of 612 DEGs and 467 DEMs were identified from datasets GSE75037 and GSE53882, respectively. Amongst these, a total of 199 and 413 DEGs along with 247 and 220 DEMs were categorized as up and downregulated (Supplementary Tables S1 and S2). Chord plot representing the association of top 10 significant pathways with 4 participating DEGs (i.e., ARG1, TNF, IL-6, IL-1β) is shown in Figure 1A. Clearly, the interaction edges in the plot show that IL-6 was present in maximum number of pathways (i.e., 5). All four pathway enriched DEGs were used for further analysis.

Validation of KLF4 and IL-1β Using UALCAN and cBioPortal
UALCAN database was used to validate the KLF4 and IL-1β expression in TCGA-LUSC cohort based on various clinicopathological features (i.e., nodal metastasis and TP53 mutation status). As shown in Figure 2A, the expression levels of KLF4 significantly correlated with nodal metastasis [Normal vs. N0, Normal vs. N1, Normal vs. N2, Normal vs. N3] and TP53 mutation status in LUSC cohort ( Figure 2B). Additionally, the expression levels of IL-1β significantly correlated with nodal metastasis [N0 vs. N2] ( Figure 2C) and non-significantly with TP53 mutation status in LUSC cohort ( Figure 2D). cBioPortal was used to investigate the specific genetic alterations of KLF4 and IL-1β in NSCLC dataset (TCGA, Pan Cancer Atlas) with 1144 patient samples (660 LUAD + 484 LUSC). The Lollipop plot as shown in Figure 2E displayed the frequency and location of all possible mutations in Pfam protein domains. KLF4 had a somatic mutation frequency of 1.0% (i.e., 12 missense mutations) and IL-1β had a somatic mutation frequency of 0.7% (i.e., 7 missense mutations + 1 truncating mutation), suggesting the significance of somatic mutations in their functional alterations ( Figure 2F). Figure 2G showed the alteration frequency barplot of KLF4 and IL-1β with missense mutation frequency of 2.12% (14 samples), deep deletion frequency of 0.15% (1 sample), amplification frequency of 0.15% (1 sample), and multiple alterations/truncating mutation frequency of 0.15% (1 sample) in case of LUAD. Whereas in LUSC, a missense mutation frequency of 1.03% (5 samples) and amplification frequency of 0.62% (3 samples) was observed.   3.5. KLF4, IL-1β and miR-34a-5p Is Associated with Macrophage Polarization KLF4 is known to be associated with M2 phenotype of macrophages [51]. Besides, IL-1β and IL-6, are also implicated in macrophage polarization [52], but their association with miR-34a-5p is not yet explored. For this, we analyzed the expression levels of KLF4 and IL-1β in different phenotypic subtypes of macrophages-M0 (unstimulated) and M2 (stimulated by IL-4). Macrophages were actually divided in 3-subsets, M0, M2 as stated above along with M1 (LPS-stimulated). Macrophage stimulation was checked by semi-quantitative RT-PCR (Supplementary Figure S4). Further, we have used only M0 and M2 subsets in the study, as M2 macrophages are highly predominant in tumor microenvironment). qRT-PCR analysis revealed higher expression of KLF4 ( Figure 4A) and reduced expression of IL-1β ( Figure 4B) in M2 subset as compared to M0 subset. To scrutinize the role of miR-34a-5p in macrophage polarization, the M0 and M2 subsets were transfected with scrambled and miR-34a-5p mimic. The relative expressions of M1 and M2 specific markers in macrophage subsets, were checked by semi quantitative RT-PCR ( Supplementary Figures S5 and S6). We observed an increase in the expression of IL-1β ( Figure 4C) and IL-6 ( Figure 4D) in M0 subset transfected with miR mimic. However, there was a decrease in KLF4 ( Figure 4E) and an increase in the expression of IL-1β ( Figure 4F) and IL-6 ( Figure 4G), upon transfection of M2 subset with miR-mimic. These results demonstrated that KLF4 is highly expressed in M2 subset and miR-34a-5p favors an increase in M1 pro-inflammatory markers (IL-1β and IL-6). Furthermore, we found that KLF4 is a direct target of miR-34a-5p, identified by TargetScan ( Figure 4H). To confirm this, we have performed luciferase assay and found an eloquent decrease (>50%) in luciferase activity upon co-transfection of pMIR-KLF4-wild type 3 UTR (pMIR-KLF4-3 UTR), pRL vector, and miR-34a-5p mimic as compared to the mutated (co-transfection of pMIR-KLF4-3 UTR mutated, pRL vector, and scrambled) in A549 cells. In contrast, changes in luciferase activity upon co-transfection of miR-34a-5p mimic with the pMIR-KLF4 mutant 3 UTR (pMIR-KLF4-m3 UTR) were not so noteworthy, connotating the direct binding between miR-34a-5p and KLF4 3 UTR ( Figure 4I). To assess whether or not the expression of KLF4 protein is affected by miR-34a-5p transfection, we transfected pMIR-KLF4-3 UTR, and pMIR-KLF4-3 UTR in combination with miR-34a-5p mimic in unstimulated THP-1 cells. We analyzed the protein expression of transfected cells and found that miR-34a-5p reduced KLF4 and ARG1 (M2 specific marker) expression at the protein levels too ( Figure 4J). These findings highlighted the significance of KLF4 in polarization of TAMs towards M2 phenotype; however, miR-34a-5p directly targets KLF4 and inhibits M2 phenotype, but promotes M1 phenotype, demonstrating their relevance in macrophage reprogramming.  (I) dual-luciferase reporter assay-plasmid containing 3 UTR of KLF4 K(N) and mutated 3 UTR of KLF4 K(M) were co-transfected with scrambled (Scr), miR-34a-5p mimic and pRL-vector containing Renilla luciferase gene, respectively, in A549 cells, and luminescence was measured. Measurements were expressed as relative luminometer units and normalized by the luminescence of Renilla luciferase; (J) THP-1 cells were transfected with KLF4 overexpression plasmid and miR-mimic. Total proteins were extracted and evaluated for degradation of KLF4 and ARG1 by Western blotting. β-actin was used as an endogenous control. * p < 0.05, ** p < 0.01, **** p < 0.0001.

KLF4-Acting as an Oncogene in NSCLC
Bioinformatics analysis has shown that KLF4 is upregulated (Supplementary Table S1) in NSCLC cells. Thus, to explore the role of KLF4 in NSCLC, we transfected A549 and H1299 cells, respectively, with empty pcDNA3.1, pcDNA3.1-KLF4-full length (FL), and pcDNA3.1-KLF4 (FL) along with miR-34a-5p mimic. There was a significant increase in cellular proliferation, migration, and clonogenic capacity of NSCLC cells upon transfection of pKLF4 (FL) ( Figure 5A-F). Meanwhile, the percentage of apoptotic cells (A549) was significantly decreased with pKLF4 (FL) transfection. However, all the above hallmarks were significantly decreased, and apoptosis increased upon co-transfection of pcDNA3.1-KLF4 (FL) and miR-34a-5p mimic ( Figure 5G,H). These results thus show that KLF4 acts as an oncogene in the progression of NSCLC and its inhibition by miR-34a-5p counter these effects. were co-transfected with scrambled (Scr), miR-34a-5p mimic and pRL-vector containing Renilla luciferase gene, respectively, in A549 cells, and luminescence was measured. Measurements were expressed as relative luminometer units and normalized by the luminescence of Renilla luciferase; (J) THP-1 cells were transfected with KLF4 overexpression plasmid and miR-mimic. Total proteins were extracted and evaluated for degradation of KLF4 and ARG1 by Western blotting. βactin was used as an endogenous control. * p < 0.05, ** p < 0.01, **** p < 0.0001.

KLF4-Acting as an Oncogene in NSCLC
Bioinformatics analysis has shown that KLF4 is upregulated (Supplementary Table S1) in NSCLC cells. Thus, to explore the role of KLF4 in NSCLC, we transfected A549 and H1299 cells, respectively, with empty pcDNA3.1, pcDNA3.1-KLF4-full length (FL), and pcDNA3.1-KLF4 (FL) along with miR-34a-5p mimic. There was a significant increase in cellular proliferation, migration, and clonogenic capacity of NSCLC cells upon transfection of pKLF4 (FL) ( Figure 5A-F). Meanwhile, the percentage of apoptotic cells (A549) was significantly decreased with pKLF4 (FL) transfection. However, all the above hallmarks were significantly decreased, and apoptosis increased upon co-transfection of pcDNA3.1-KLF4 (FL) and miR-34a-5p mimic ( Figure 5G,H). These results thus show that KLF4 acts as an oncogene in the progression of NSCLC and its inhibition by miR-34a-5p counter these effects.  . Oncogenic role of KLF4 in NSCLC. pcDNA, pcDNA-KLF4 and pcDNA-KLF4 along with miR-34a-5p mimic were transfected, respectively, in A549 and H1299 cells and analyzed for cellular proliferation, wound healing, and clonogenic capacity after 48 h of treatment; (A). Cellular proliferation was checked via MTT assay; (B). In vitro scratch assay was performed to evaluate the migrative capacity of cells. % wound closure was recorded for all the groups in both A549 and H1299 cells till 24 h; (C). Pictures showing wound closure after 24 h in pcDNA, pcDNA+KLF4, and pcDNA+KLF4+miR transfected groups; (D). Clonogenic assay-number of colonies counted by ImageJ software for determination of clonogenic capacity of both A549 and H1299 cells upon CM treatment; (E). Pictures showing colonies appeared after treatment in H1299 cells; (F). OD at 570 nm was measured after destaining the colonies appeared after transfection; (G). Apoptotic analysis using AnnexinV-FITC staining of transfected A549 cells; (H). Fluorescence emission by transfected cells measured by flow cytometry. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. All the experiments were done thrice (N = 3).

miR-34a-5p/KLF4 Mediated Macrophage Polarization Reduce Tumorigenesis
To elucidate macrophage polarization effects on NSCLC cells, miR-34a-5p mimic transfected macrophages (M0, M2) were co-cultured with A549 cells. The co-culture CM was used to treat A549 and H1299 cells and analyzed for cancer hallmarks. We observed a significant decrease in percent proliferation, wound closure, and number of colonies, upon treatment with CM (from M2 co-culture) in both A549 ( Figure 6A-F) and H1299 cells ( Figure 6G-L). Additionally, there was a concomitant increase in the concentration of nitrites in CM of both M0 and M2 (miR-transfected) co-culture ( Figure 6N). CM treatment from M2 co-culture also increased the percentage of apoptosis in A549 cells ( Figure 6O,P). Besides this, an increase in expression of pro-caspase-3 was also observed by Western blotting in both A549 and H1299 cells ( Figure 6Q). The relative expression of Caspase-3 was checked by semi-quantitative qRT-PCR (Supplementary Figure S13). These results demonstrate that miR-34a-5p mediated modulation of macrophage polarization reverses the processes of tumorigenesis in NSCLC cells.

miR-34a-5p/KLF4 Mediated Macrophage Polarization Reduce Tumorigenesis
To elucidate macrophage polarization effects on NSCLC cells, miR-34a-5p mimic transfected macrophages (M0, M2) were co-cultured with A549 cells. The co-culture CM was used to treat A549 and H1299 cells and analyzed for cancer hallmarks. We observed a significant decrease in percent proliferation, wound closure, and number of colonies, upon treatment with CM (from M2 co-culture) in both A549 ( Figure 6A-F) and H1299 cells ( Figure 6G-L). Additionally, there was a concomitant increase in the concentration of nitrites in CM of both M0 and M2 (miR-transfected) co-culture ( Figure 6N). CM treatment from M2 co-culture also increased the percentage of apoptosis in A549 cells ( Figure 6O,P). Besides this, an increase in expression of pro-caspase-3 was also observed by Western blotting in both A549 and H1299 cells ( Figure 6Q). The relative expression of Caspase-3 was checked by semi-quantitative qRT-PCR (Supplementary Figure S13). These results demonstrate that miR-34a-5p mediated modulation of macrophage polarization reverses the processes of tumorigenesis in NSCLC cells. . Nitrite content when M0 and M2 macrophages were transfected with scrambled and miR-mimics, respectively; (N). Nitrite content upon co-culture of transfected macrophages with A549 cells, as deter. * p < 0.05, ** p < 0.01, *** p < 0.001,**** p < 0.0001; (O,P). Flow cytometric measurement of apoptosis in A549 cells using Annexin V-FITC. CM from co-culture of M2 transfected with scrambled and miR-mimics, respectively, with A549 cells was used to treat A549 and H1299 cells for 48 h. Cells were stained with Annexin-V/FITC and PI, fluorescence of A549 cells was measured using FACS and; (Q). Western blotting. CM treated A549 and H1299 cells were lysed, total protein was extracted in RIPA extraction buffer and expression of Procaspase-3 was detected using Western blotting. Protein expression was normalized by expression of β-actin, an endogenous control. * p < 0.05, ** p < 0.01. Three independent experiments were performed (N = 3). M0-scrambled-unstimulated THP-1 cells transfected with scrambled, M2-scrambled-IL-4 stimulated THP-1 cells transfected with scrambled, M0-miR-34a-5p-unstimulated THP-1 cells transfected with miR mimic, M2-miR-34a-5p-IL4 stimulated THP-1 cells transfected with miR mimic.

KLF4 and Tumor Inversion in NSCLC Microenvironment
To decipher the functional significance of KLF4 in progression of NSCLC, we overexpressed H1299 cells with control (scrambled) and miR-34a-5p. The effect of transfection on expression of KLF4 and its regulated genes-IL-6 and IL-1β was checked semi quantitatively (Supplementary Figure S7). The transfected cells were co-cultured with IL-4 stimulated THP-1 cells. A significant increase in the relative expression of miR-34a-5p was observed upon transfection ( Figure 7A). We observed a significant reduction in expression of M2 specific markers-KLF4 ( Figure 7B) and REL-1MB ( Figure 7C), with a compelling increase in expression of M1 specific markers-IL-1β ( Figure 7D) and IL-6 ( Figure 7E) upon co-culture. Further, co-culture CM was used to treat A549 cells and analyze the cancer hallmarks. Cellular proliferation, wound healing potential and clonogenic capacity of NSCLC cells was markedly decreased in cells treated with co-culture CM from miR-transfected cells ( Figure 7F-K). Moreover, the concentration of nitrites in H1299cells was reduced upon miR-34a-5p transfection ( Figure 7L), but was significantly increased upon co-culture ( Figure 7M). An increase in apoptosis of A549 cells ( Figure 7N,O) and protein levels of pro-caspase3 upon treatment with co-culture CM was also observed ( Figure 7P).  H1299) were lysed, total protein was extracted in RIPA extraction buffer and expression of Pro-caspase-3 was detected using Western blotting. Protein expression was normalized by expression of β-Actin, an endogenous control. * p < 0.05, ** p < 0.01. All the experiments were independently done thrice (N = 3).

Discussion
The tumor microenvironment has brought great interest in last few years, owing to its impact on initiation, development, progression, and prognosis of cancers. Macrophages are the key components of the immune system that exert tumor-promoting and tumor-inhibiting activities in TME. Clinical data has demonstrated that around 70% of TAMs are M2 type and the remaining ones are M1 which further acquires M2 phenotype during cancer progression [53][54][55]. The exclusive M2-like TAMs are correlated with poor prognosis, evade host immune responses, and promote cell migration and angiogenesis [56,57]. Besides, small (~22 nts long) non-coding RNAs, called miRNAs have also been implicated in almost all cellular processes including cellular, cell cycle, and apoptosis along with immune mechanisms. Thus, it is logical to propose the involvement of miR-NAs in modulating components of TME.
The in silico analysis of NSCLC data available on public databases has displayed the differential expression of more commonly immune system related genes in our study. Pathway enrichment analysis of IL-1β, IL-6, TNFα, and ARG1 (macrophage polarization

Discussion
The tumor microenvironment has brought great interest in last few years, owing to its impact on initiation, development, progression, and prognosis of cancers. Macrophages are the key components of the immune system that exert tumor-promoting and tumorinhibiting activities in TME. Clinical data has demonstrated that around 70% of TAMs are M2 type and the remaining ones are M1 which further acquires M2 phenotype during cancer progression [53][54][55]. The exclusive M2-like TAMs are correlated with poor prognosis, evade host immune responses, and promote cell migration and angiogenesis [56,57]. Besides, small (~22 nts long) non-coding RNAs, called miRNAs have also been implicated in almost all cellular processes including cellular, cell cycle, and apoptosis along with immune mechanisms. Thus, it is logical to propose the involvement of miRNAs in modulating components of TME.
The in silico analysis of NSCLC data available on public databases has displayed the differential expression of more commonly immune system related genes in our study.
Pathway enrichment analysis of IL-1β, IL-6, TNFα, and ARG1 (macrophage polarization related genes) demonstrated their denoting enrichment in cytokine signaling in the immune system ( Figure 1A). Further, KLF4, IL-1β, and miR-34a-5p are associated in the form of an FFL ( Figure 1B). Noticeably, KLF4 is highly upregulated (~3-fold upregulation), while IL-1β (2.67-fold downregulation) (Supplementary Table S1) and miR-34a-5p are downregulated (Supplementary Table S2) in NSCLC. Higher expression of KLF4 and lower expression of miR-34a-5p are correlated with the poor OS of NSCLC patients, indicative of their roles as prognostic biomarkers (Table 2). Later, UALCAN-based analysis in NSCLC advocated that higher KLF4 and lower IL-1β expressions were closely associated with advanced stages of the disease and hence could be potent diagnostic biomarkers. The results presented that KLF4 expression exhibited a progressive increase in NSCLC from stage 1 to stage 4 and the maximal expression can be found in stage 4 (Figure 2A). Similarly, the expression of IL-1β was found to be gradually reduced from stage 1 to stage 4, with the minimum expression in stage 4 ( Figure 2C). cBioportal-based somatic mutation analysis of both KLF4 and IL-1β has displayed a combined alteration frequency of 2.8% and 1.8% in LUAD and LUSC, respectively, indicating high mutational burden in KLF4 ( Figure 2G). Furthermore, immune infiltration investigation pinpoint towards KLF4 mediated infiltration of TAMs in both LUAD and LUSC, particularly, favoring infiltration of M2 macrophages, and negative correlation with NK and B cells suggesting the involvement of KLF4 in the evasion of immunity. Congruently, IL-1β, is critically involved in the infiltration of M1 macrophages ( Figure 3). In addition, KLF4 also displayed negative correlation with tumor purity in both LUAD and LUSC. Tumor purity is defined as the percentage of cancer cells in solid tumors, which may act as a prognostic factor indicator or predictor of chemotherapy benefit. It is an important tool in analyzing the patient's condition in clinical practices as contamination of normal cells in tumor tissues may subsequently hamper genomic analysis [58,59]. Moreover, cancers with high genomic instability will possess more genomic diversity leading to the formation of more neoantigens and greater infiltration of immune cells [60,61]. This is in concordance with our results where high mutational burden in KLF4 leads to low tumor purity and hence greater tumor infiltration of immune cells, specifically macrophages.
Further, we analyzed the association of KLF4, IL-1β, and miR-34a-5p in macrophage polarization. We found that KLF4 was upregulated in IL-4 stimulated M2 subset of macrophages, but IL-1β was upregulated in M0 macrophages ( Figure 4A,B), however, miR-34a-5p inhibits M2 and promotes M1 subtype of macrophages via directly targeting KLF4 ( Figure 4C-G). Additionally, KLF4 overexpression increased cell proliferation, migration, and clonogenic capacity, but decreased apoptosis of NSCLC cells, suggesting its oncogenic role within the regulatory network of NSCLC. Moreover, all the above effects were rescued back, when pKLF4 (FL) was co-transfected with miR-34a-5p ( Figure 5), suggesting the implication of KLF4 in the tumor-suppressive functions of miR-34a-5p and its oncogenic transformation. The oncogenic nature of KLF4 in NSCLC is quite contradictory, as most of the studies have depicted the tumor suppressive role of KLF4 in NSCLC. However, tumor suppressive and oncogenic functions of KLF4 are cell type, context, and subcellular localization dependent, but how KLF4 exert these differential functions and the associated molecular mechanisms remains unraveled. Contemporary studies have demonstrated the role of alternative splicing of KLF4. Multiple isoforms of KLF4 have been identified in several cancer types, with KLF4α, being most significant in tumorigenesis [62]. KLF4 transcript is about 3kb long and multiple shorter isoforms (KLF4α, β, γ and δ) have been found. KLF4 (FL) (isoform 2) shows a band at around 1440bp; however, KLF4α displays a band at 440bp. This isoform lacks exon 3, causing a frameshift in exon 4 to a premature stop codon in exon 5. Moreover, it lacks all the three zinc finger domains of KLF4 (FL) and its nuclear localization signal (NLS) sequence, due to which it is mainly located in the cytoplasm, while KLF4 (FL) mainly exists in nuclei. Moreover, KLF4α interacts with KLF4 (FL) and prevents its nuclear translocation, thereby modulating the transcriptional landscape and exerting antagonistic effects on KLF4 (FL) [62,63]. Several miRNAs, in-cluding miR-34, have been found to regulate KLF4 post transcriptionally and alter the splicing patterns [64]. Further, it has been found that KLF4α is highly expressed in tumors and although KLF4 (FL) exerts tumor inhibitory effects, but excessive KLF4α opposes its effect, thereby promoting tumorigenesis. This effect of KLF4α is seen in Figure 5, when pc-DNA-KLF4 (FL) is transfected in A549 cells, it acts as an oncogene. This is because KLF4α is highly expressed in A549 cells, which may have masked the KLF4 (FL) from nuclear translocation and exert its tumor suppressing effects. However, miR-34a-5p modulates the alternative splicing of KLF4 and reduces the KLF4α isoform, thereby rendering the KLF4 (FL) free to exert its tumorigenic effects (Supplementary Figure S7).
Subsequently, to retrospect the significance of macrophage polarization in development and progression of NSCLC, we co-cultured miR-34a-5p transfected macrophages with NSCLC cells and analyzed the effects on cancer hallmarks. There was a concomitant decrease in cellular proliferation, migration and clonogenic capacity of both A549 and H1299, accompanied by an increase in nitrite content and hence apoptosis upon treatment with co-culture CM from macrophages transfected with miR-mimic, suggesting the tumoricidal potential of reprogrammed macrophages ( Figure 6).
Furthermore, to persuade the role of KLF4 in polarization of TAMs in NSCLC microenvironment, we transfected miR-34a-5p mimic in H1299 cells ( Figure 7A). The relative expression of KLF4 was reduced, and its regulated genes-IL-1β, IL-6 were increased (Supplementary Figure S7). We co-cultured them with IL-4 stimulated macrophages. We found out a convincing decrease in expression of M2 signature genes-KLF4, REL-1MB ( Figure 7B,C) and ARG1 (Supplementary Figure S8, nonsignificant), along with an increase in expression of M1 signature genes-IL-1β, IL-6 ( Figure 7D,E) and iNOS (Supplementary Figure S8). These results thus demonstrate that miR-34a-5p when overexpressed in NSCLC cells, inhibits KLF4 by degradation of oncogenic KLF4α isoform of NSCLC cells, accompanied with an increase in expression of KLF4 regulated genes-IL-1β and IL-6 (Supplementary Figure S7). Increased secretion of pro-inflammatory mediators (Il-1β, IL-6) provide a stimulus for polarization of M1 macrophages in co-culture. M1 macrophages, then exert positive feedback to enhance the release of more IL-1β, IL-6 ( Figure 7D,E) and iNOS (Supplementary Figure S7), upon co-culture. These pro-inflammatory cytokines, however, exert anti-tumor effects, via an autocrine mechanism, as demonstrated by reduced proliferation, wound healing, clonogenic capacity, and enhanced apoptosis of NSCLC cells.
We also observed a significant decrease in nitrite content of H1299 cells upon transfection of miR-34a-5p, while the content increased upon co-culture of miR-transfected H1299 cells with macrophages ( Figure 7L,M). Higher nitrite content upon co-culture leads to increased apoptosis of NSCLC cells. It has been reported that patients with LC exhale high levels of NO and its metabolites-nitrite, and nitrotyrosine, which are associated with advanced stages and poor survival. NO has also been shown to possess both pro and antiapoptotic activities, based on the cellular context, dosage, and oxidative state of the cells. A high concentration of NO is known to induce apoptosis resistance by S-nitrosylation of FLIP and Bcl2 (KLF4 regulated anti-apoptotic protein) which makes them resistant to proteasomal degradation and, hence preventing apoptosis. NSCLC cells release high amounts of NO, which can thus be correlated with their aggressive behavior and resistance to apoptosis [53]. However, miR-34a-5p overexpression in NSCLC reduces NO content which would reverse the above effects on Bcl2, thereby presuming apoptosis. However, in case of co-culture with THP-1 cells, we have observed an increase in nitrites along with increased apoptosis. This can be explained as due to miR-34a mediated polarization of M2 TAMs to M1, the expression of iNOS also increases. iNOS secretes high levels of NO which creates an oxidative state via its activity on mitochondrial membrane permeability. This causes a release of cytochrome c into the cytoplasm, which initiates apoptosis via binding to Apaf and thereby activating caspase 9, caspase 7 and 3, ultimately leading to downstream events of apoptosis [65]. These observations are clear indications of both pro and anti-apoptotic activities of NO based upon dosage and cellular oxidation states. Further, excessive secretion of NO and cytokines like TNF-α by M1 macrophages ( Supplementary  Figures S4 and S5) also activate the immune system, promoting inflammatory responses and generating anti-tumor immunity.
Overall, our study has suggested two important aspects of TAM polarization in NSCLC. The first aspect is the association of KLF4 with tumorigenesis in NSCLC. Our results have suggested that IL-4 mediated induction of KLF4 not only induces M2 phenotype, but also inhibits the M1 pathway. This was consistent with the observation that KLF4 deficiency debilitates the ability of IL-4 to inhibit M1 targets, as demonstrated by an increase in IL-1β (M1 gene, when KLF4 was degraded in IL-4 stimulated M2 THP-1 cells upon miR-34a-5p transfection ( Figure 4F, Supplementary Figure S5)). This is suggestive of the role of IL-4 mediated cross-talk between TAMs and NSCLC cells, and promotion of tumorigenesis.
The second aspect is related to KLF4 mediated regulation of NO content. Tumor cells secrete a large amount of NO, which promotes tumor progression by induction of tumor-cell invasion, proliferation, and the expression of angiogenic factors [55]. However, immune cells present in TME, like macrophages, when polarized to M1 state upregulates iNOS (Supplementary Figure S5), leading to an accumulation of more NO. This macrophage secreted NO exerts tumoricidal activity due to which host iNOS might suppress tumor growth and metastasis. Our study has also demonstrated the potential of miR-34a-5p to polarize TAMs from M2 to M1 phenotype by directly targeting KLF4 in the LC microenvironment. Consequently, these reprogrammed M1 macrophages exhibit cytotoxic effects on NSCLC cells via NO associated processes (Figure 7). Although NO, mediated apoptosis is the mechanism by which M1 macrophages have exerted their tumoricidal actions here, but there are several other factors such as MMP induction and decreased secretion of growth factors, that are also accountable and need to be further explored. Additionally, a better understanding of the circuits associated with inflammatory cytokines, other growth factors, and miRNAs will aid in unravelling the molecular mechanisms associated with macrophage polarization in TME, which would pave a way towards the development of novel and more efficient therapeutic strategies for NSCLC.  Figure S1: NSCLC-specific 3-node miRNA-FFL regulatory network comprising 49 nodes and 242 edges, Figure S2: Topological Graphs, Figure S3: Scatter plots exhibiting correlations of KLF4 with infiltrating levels of immune cells, Figure S4: Macrophage stimulation, Figure S5: Relative expression of M1 macrophage specific markers, Figure S6: Relative expression of M2 macrophage specific markers, Figure S7: Relative expression of KLF4, IL-1β and IL-6 upon transfection of NSCLC cells with Scrambled and miR-34a-5p, Figure S8: Differential expression of macrophage specific markers, upon Co-culture of M2 macrophages with transfected H1299 cells, Figure S9: Western Blotting, Figure S10: SSC/FSC plots of Flow cytometry based apoptotic analysis using AnnexinV /FITC-PI staining of A549 cells transfected with pcDNA 3.1, pcDNA 3.1 + KLF4 and pcDNA 3.1+ KLF4 + miR, Figure S11: SSC/FSC plots of Flow cytometry based apoptotic analysis using AnnexinV /FITC-PI staining of A549 cells treated with Co-culture Conditioned medium of M2 macrophages (THP-1) cells transfected with Scrambled and miR-34a-5p mimic, Figure S12: SSC/FSC plots of Flow cytometry based apoptotic analysis using AnnexinV /FITC-PI staining of A549 cells treated with Co-culture Conditioned medium of H1299 cells transfected with Scrambled and miR-34a-5p mimic, Figure S13: Relative expression of Caspase-3; Supplementary Tables, Table S1: differentially expressed genes (DEGs) in NSCLC, Table S2: Differentially expressed miRNAs (DEMs) in NSCLC; Oligonucleotide sequences of primers and list of antibodies used.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.