Introduction

The platinum(II) complex, cis-diamminedichloroplatinum(II) (cisplatin; Fig. 1), is a bifunctional chemotherapy drug that upon aquation of its two chloride leaving ligands and entry into cancer cell nuclei, typically forms intrastrand crosslinks with DNA causing the nucleotide strands to bend and alteration of gene transcription1,2. As cisplatin chemotherapy is frequently associated with severe side-effects3,4,5,6, there has been considerable interest in identifying alternative platinum complexes that are toxic to cancer cells without causing severe side-effects. The monofunctional platinum(II) complex, cis-[Pt(NH3)2Cl (phenanthridine)]+ (phenanthriplatin; Fig. 1) has similar or superior anticancer efficacy as cisplatin in lung, breast, prostate, testicular, colorectal, glioblastoma and cervical cell lines7,8. Unlike cisplatin, monofunctional compounds have only one chloride leaving ligand and bind to a single strand of DNA without causing DNA to bend7,9. Studies have shown that phenanthriplatin can prevent RNA transcription and has a distinct DNA residue binding profile compared to cisplatin7,10.

Figure 1
figure 1

Chemical structures of cisplatin and phenanthriplatin. Cisplatin is a bifunctional platinum(II) complex with two chloride leaving ligands. The monofunctional platinum(II) complex, phenanthriplatin, has one chloride leaving ligand, and steric bulk in the form of a heterocyclic phenanthridine ligand placed cis to the platinum coordinating site.

Full size image

Microarray studies suggest that cisplatin can regulate genes associated with drug resistance, metabolism, cell proliferation, apoptosis, cell adhesion, stress response, cell cycle control and DNA repair11,12,13,14. Long non-coding RNAs (lncRNAs) are RNAs more than 200 nucleotides in length and may interact with protein, chromatin and RNA targets where they can modulate epigenetic, transcriptional, and post-transcriptional function15,16. LncRNA expression can become dysregulated in lung cancer tissue, where their aberrant expression can promote tumor cell growth, apoptosis, invasion, and metastasis15. MicroRNAs (miRNAs) are non-coding RNAs that function post-transcriptionally by binding to messenger RNA (mRNA) to prevent transcription or promote mechanisms that increase mRNA degradation and can regulate tumorigenesis and metastasis16. Further, microarray analysis of lung cancer cells has shown that cisplatin can alter the expression of a lncRNA which acts as a regulatory sponge for a miRNA and modulates chemosensitivity17. Other studies conducted with microarrays in cisplatin-treated cells have shown that lncRNAs are involved in the regulation of cisplatin sensitivity in non-small cell lung cancer (NSCLC)18,19,20. However, the effect of monofunctional platinum(II) complexes on lncRNA expression and their potential regulatory interaction with miRNAs has not yet been characterized.

This study used microarray analysis to investigate whether phenanthriplatin utilizes different lncRNA signaling than cisplatin. The A549 non-small cell lung cancer (NSCLC) cell line is routinely used to evaluate platinum-based gene modulation, and the non-cancer lung fibroblast cell line, IMR90, has been used to assess cisplatin’s effect on gene regulation21,22. Here, we treated A549 and IMR90 cells with the monofunctional complex, cisplatin or a negative media only control. Then, RNA samples were extracted from the cell lines and subjected to microarray analysis. Next, lncRNAs found in the microarray results were bioinformatically analyzed to identify miRNAs with high targeting homology with the lncRNA sequences. qRT-PCR analysis was also used to validate the microarray results and identify miRNAs subject to distinct modulation by the platinum compounds. We found that phenanthriplatin treatment caused up- and downregulation of lncRNAs that were not modulated by cisplatin in either A549 or IMR90 cells. Also, in A549 cells, lncRNAs downregulated by the monofunctional complex have several potential miRNA binding partners associated with Wnt/β-catenin and transforming growth factor-β (TGF-β) signaling that were predicted to also target lncRNAs that were upregulated by cisplatin treatment. This result suggests that cisplatin and phenanthriplatin may utilize distinct lncRNA and miRNA networks to promote anticancer function. Further, we found that phenanthriplatin modulated different lncRNAs in IMR90 cells and that some of the miRNAs predicted to cause cell toxicity during phenanthriplatin treatment in A549 NSCLC cells might promote an opposite effect in IMR90 fibroblast cells.

Results

Microarray and lncRNA qRT-PCR results

Affymetrix microarray analysis was performed on samples taken from cisplatin and phenanthriplatin treated A549 and IMR90 cells to identify regulated lncRNAs. We used a fold change threshold of ≥ 1.5 to identify the most up- and downregulated lncRNAs in A549 cells treated with cisplatin or phenanthriplatin compared to control (Table 1).

Table 1 Top up- and downregulated lncRNAs in cisplatin or phenanthriplatin treated A549 cells versus control (N = 4; p ≤ 0.05; fold change ≥ 1.5).
Full size table

Microarray analysis was also used to identify regulated lncRNAs in IMR90 cells treated with cisplatin or phenanthriplatin compared to control samples (Table 2).

Table 2 Top up- and downregulated lncRNAs in cisplatin or phenanthriplatin treated IMR90 cells versus IMR90 controls (N = 4; p ≤ 0.05; fold change ≥ 1.5).
Full size table

To confirm the microarray data, we also performed qRT-PCR on samples taken from both the A549 and IMR90 cell lines and found that both methods detected similar expression profiles for many of the lncRNAs (Fig. 2). Cisplatin treated A549 samples had the following relative fold expression values (microarray compared to qRT-PCR respectively): PTBP1-2, 1.71–0.63; IMMP2L-1, 1.90–0.99; RIPPLY3-5, 2.20–2.71, which were within approximately one fold change value of one another or less (Fig. 2a). Phenanthriplatin treated A549 samples had very similar relative fold change values for two lncRNAs, TRIM55-1 (2.39–1.82) and RARB-1 (1.82–1.75), while RIPPLY3-5 (2.04–4.83) and PTBP1-2 (2.75–0.33) were within three relative fold change values of each other (Fig. 2a). For cisplatin treated IMR90 samples, relative fold expression values (microarray compared to qRT-PCR respectively) were within approximately one fold change value of one another: MRPS5-1 (1.87–0.71), HYPM-1 (1.30–0.59), NLRP12-1 (1.93–0.75), while several phenanthriplatin treated samples, HYPM-1 (2.90–0.57), FAM237B-2 (3.27–8.06), TMEM243-1 (2.36–0.59) differed by more than a fold change value, with the exception of TRIM55-1 (4.87–5.79), where relative expression was within a fold change value (Fig. 2b).

Figure 2
figure 2

Validation of long non-coding RNA microarray expression data using qRT-PCR. (a) A549 cell line results for PTBP1-2 (PT), IMMP2L-1 (IM), RIPPLY3-5 (RI), TRIM55-1 (TR), and RARB-1 (RA). (b) IMR90 cell line results for MRPS5-1 (MR), HYPM-1 (HY), NLRP12-1 (NL), FAM237B-2 (FA), TMEM243-1 (TM), and TRIM55-1 (TR). Key: “-M” = microarray (light gray); “-P” = qRT-PCR (dark gray). Error bars represent mean ± standard deviation, N = 3.

Full size image

Bioinformatics and miRNA qRT-PCR results

RNA22 v2 microRNA software was then used to identify microRNAs predicted to target the three most upregulated lncRNAs in A549 cells after cisplatin treatment. We found that cisplatin upregulated the lncRNAs, ATAD2B-5, CBR3-1, and IMMP2L-1, which may be targeted by several miRNAs with sequence homologies to lncRNAs downregulated by phenanthriplatin, including: AGO2-1, COX7A1-2 and SLC26A3-1 (Table 3; Supplementary Table S1). Also, we found that cisplatin downregulated the lncRNA, lnc-PGPEP1L-3, which is predicted to sponge miR-25-5p, a miRNA with high sequence homology to lnc-MRPL39-10, a lncRNA upregulated by phenanthriplatin (Table 3; Supplementary Table S1). These miRNAs are associated with Wnt/β-catenin and TGF-β signaling, e.g., miR-25-5p, -30a-3p, -138-5p, -149-3p, -185-5p, -378j, -608, -650, -708-5p, -1253, -1254 and -4516 (Table 3)23,24,25,26,27,28,29,30,31,32,33,34,35.

Table 3 MiRNAs projected to target regulated lncRNAs in A549 cisplatin and phenanthriplatin treated cells and predicted pathway signaling (p ≤ 0.05).
Full size table

As signaling integrating lncRNAs and miRNAs in IMR90 cells might modulate distinct targets than those in A549 cells, we also identified microRNA targets of the three most upregulated lncRNAs after cisplatin treatment. We found that cisplatin upregulated the lncRNAs, ATAD2B-5, GNG11-3, and MRPS5-1, which may be targeted by miRNAs with sequence homologies to lncRNAs downregulated by phenanthriplatin, including: AGO2-1, ARCN1-1, HYPM-1 (Table 4; Supplementary Table S2). Cisplatin and phenanthriplatin treatment also downregulated the lncRNA, HYPM-1, which may sponge miR-608, and the monofunctional complex upregulated the lncRNA, TMEM243-1, with high sequence homology to miR-608 (Table 4; Supplementary Table S2). As in the A549 results, these miRNAs can potentially act through Wnt/β-catenin and TGF-β signaling (Table 4)23,26,27,29,32,33,34,36.

Table 4 MiRNAs projected to target regulated lncRNAs in IMR90 cisplatin and phenanthriplatin treated cells and predicted pathway signaling (p ≤ 0.05, except GNG11-3, p ≤ 0.0581).
Full size table

To determine the expression of miRNA targets of lncRNAs identified in the bioinformatics analysis, we also performed qRT-PCR using locked nucleic acid primers on select miRNAs and found that cisplatin and phenanthriplatin treatment caused different regulation of three miRNAs, miR-25-5p, -378j, and -4516, in the A549 cell line and one miRNA, miR-4458, in the IMR90 cell line (Fig. 3).

Figure 3
figure 3

Measurement of miRNA expression using qRT-PCR. (a) A549 cell line results for miR-25-5p, -378j and -4516. (b) IMR90 cell line results for miR-149-3p, -608 and -4458. Key: “cis” = cisplatin; “phen” = phenanthriplatin. Error bars represent mean ± standard deviation, N = 3, “*” = p ≤ 0.05; “**” = p ≤ 0.01; “****” = p ≤ 0.0001.

Full size image

As a final step, we also analyzed how cisplatin and phenanthriplatin regulated lncRNAs might target miRNAs and potentially regulate Wnt/β-catenin and TGFβ signaling in both A549 (Fig. 4) and IMR90 cells (Fig. 5).

Figure 4
figure 4

Potential modulation of Wnt/β-catenin and TGF-β signaling via cisplatin and phenanthriplatin regulation of lncRNAs and predicted miRNA binding partners in A549 cells. Key: Grey miR text = miRNA associated with downregulated Wnt/β-catenin and TGF-β signaling, Black miR text = miRNA associated with upregulated Wnt/β-catenin and TGF-β signaling, grey open down arrow = downregulated, black open up arrow = upregulated.

Full size image
Figure 5
figure 5

Potential modulation of Wnt/β-catenin and TGF-β signaling via cisplatin and phenanthriplatin regulation of lncRNAs and predicted miRNA binding partners in IMR90 cells. Key: Grey miR text = miRNA associated with downregulated Wnt/β-catenin and TGF-β signaling, Black miR text = miRNA associated with upregulated Wnt/β-catenin and TGF-β signaling, grey open down arrow = downregulated, black open up arrow = upregulated.

Full size image

Discussion

Cisplatin can act through lncRNAs that regulate disease progression in lung cancer by sponging miRNAs, whose interaction with lncRNAs modulates a variety of genes in cancer apoptosis, migration, proliferation and resistance37,38,39,40,41,42. As the role of lncRNAs in the anticancer effect of monofunctional platinum(II) complexes had not yet been assessed, we used microarray analysis to identify regulated lncRNAs in A549 and IMR90 cells after treatment with cisplatin or phenanthriplatin (Tables 1, 2). To validate our microarray results, we performed qRT-PCR on total RNA extracts from cisplatin and phenanthriplatin treated cell preparations and found that the relative fold change expression of regulated lncRNAs calculated via PCR amplification generally corroborated the microarray data (Fig. 2). Next, we applied bioinformatics methods to the microarray results and discovered that a set of miRNAs predicted to target the three most upregulated lncRNAs by cisplatin treatment also had high sequence homology to the three most downregulated lncRNAs, AGO2-1, COX7A1-2 and SLC26A3-1, after phenanthriplatin treatment (Tables 1, 3). Then, we used LNA qRT-PCR to comparatively measure the effect of cisplatin and phenanthriplatin treatment on the expression of several miRNAs that are predicted to be targeted by these lncRNAs (Fig. 3). These results suggest that cisplatin might upregulate a set of lncRNAs that sponge and nullify the gene regulatory effect of these miRNAs; whereas, phenanthriplatin, by reducing the expression of lncRNAs that could prevent the action of the same miRNAs, might promote miRNA gene silencing. The results also suggest that cisplatin and phenanthriplatin treatment might cause distinct regulation of miRNAs, particularly in A549 cells.

Several of the miRNAs detected in our analysis are associated with Wnt/β-catenin signaling. Expression of miR-25-5p may be targeted by two lncRNAs regulated by phenanthriplatin, e.g., lnc-COX7A1-2, that the monofunctional complex reduces, and lnc-MRPL39-10, which phenanthriplatin increases (Table 3). The sponging of miR-25-5p via increased lnc-MRPL39-10 expression could promote anticancer effect through the Wnt pathway as the reduction of this miRNA in A549 cells is associated with decreased cell viability, invasion and migration by decreasing Wnt signaling (Fig. 4)25. Our miRNA qRT-PCR results may corroborate lnc-MRPL39-10 suppression of miR-25-5p as phenanthriplatin treated A549 cells had decreased miR-25-5p expression compared to cisplatin treated samples (Fig. 3a).

As phenanthriplatin reduced the expression of the lncRNAs, AGO2-1, COX7A1-2, and SLC26A3-1 (Table 3), this could promote the action of miRNAs that might be sponged by these lncRNAs. Increased miR-149-3p and miR-185-5p expression downregulates Wnt/β-catenin signaling reducing cancer proliferation in A549 and lung adenocarcinoma cells33,34. Overexpression of miR-608 induced cell death in A549 cells via signaling that integrates the Wnt pathway23. MiR-708-5p expression also suppressed DNA methylation decreasing Wnt/β-catenin signaling and impairing the stemness characteristics of NSCLC cells26. Therefore, phenanthriplatin could act to suppress NSCLC by repressing lncRNAs that regulate miR-25-5p, -185-5p, -608, and -708-5p, through the Wnt/β-catenin pathway (Fig. 4).

However, increasing the expression of some of the miRNAs may also promote cancer through the Wnt/β-catenin pathway. Suppression of miR-378 in A549 cells inactivates Wnt/β-catenin signaling leading to reduced cell proliferation28, and as phenanthriplatin treatment caused reduced miR-378j compared to cisplatin (Fig. 3a), the monofunctional complex may act through this pathway to prevent cancer proliferation. MiR-650 can be upregulated in A549 cells where it promoted proliferation and invasion through the Wnt-1/β‑catenin pathway by acting on inhibitor of growth 429. Increased expression of miR-1253, which directly targets Wnt5A, is associated with the proliferation, migration, and invasion of NSCLC cells27. Similarly, miR-1254 targets the Wnt/β-catenin pathway antagonist, secreted frizzled related protein 1 (SFRP1), and was upregulated in lung cancer tissues and cells promoting proliferation24. Our results suggest that the monofunctional complex might act against NSCLC via Wnt/β-catenin signaling if the miRNAs, miR-149-3p, -185-5p, -608, and -708-5p were not sponged by the phenanthriplatin downregulated lncRNAs, AGO2-1, COX7A1-2 and SLC26A3-1. Nonetheless, increased expression of the miRs, -25-5p, -650, -1253, and -1254, when these lncRNAs are suppressed by the monofunctional complex, could promote cancer by modulating the Wnt/β-catenin pathway possibly in conjunction with other pathway signaling (Fig. 4).

LncRNAs are also able to act in NSCLC through the TGF-β pathway and integrated MAPK/ERK and PI3K/AKT signaling43,44. We identified several miRNAs that might integrate TGF-β signaling components through lncRNAs regulated by phenanthriplatin (Fig. 4). Overexpression of miR-608 induced cell death in A549 cells through the MAPK/ERK pathway23. As the three lncRNAs, AGO2-1, COX7A1-2, and SLC26A3-1, which are predicted to sponge miR-608, were downregulated by the monofunctional complex (Table 3), we would expect that expression of this miRNA would lead to NSCLC cell death via ERK signaling. However, suppression of AGO2-1, COX7A1-2, and SLC26A3-1, could also have an opposite effect in NSCLC cells by upregulating the miRs, -25-5p, -30a-3p and -378, which act via MAPK/ERK, and are projected targets of these lncRNAs (Table 3), and have reduced anti-cancer efficacy when their expression is increased25,28,30. We found that phenanthriplatin treatment downregulated miR-378j more than cisplatin (Fig. 3a), which suggests that the monofunctional complex may signal via the TGF-β/MAPK/ERK pathway to promote anti-cancer effect. As miR-25-5p may be sponged via phenanthriplatin-mediated upregulation of lnc-MRPL39-10 (Table 3), this miRNA could be downregulated in A549 cells, which is associated with anti-cancer effect via PI3K signaling25. Reduction of AGO2-1, COX7A1-2, and SLC26A3-1 by phenanthriplatin would also be expected to lead to increased miR-138-5p and -608, whose expression modulates PTEN/PI3K/AKT signaling and is associated in A549 cells with anticancer effect23,31,45,46,47. In retinoblastoma tumors, increased miR-4516 expression promotes tumor growth through the PTEN/AKT pathway32. As phenanthriplatin treatment caused increased miR-4516 expression in A549 cells compared to cisplatin (Fig. 3a), the monofunctional complex may be able to promote PTEN/AKT signaling more than its bifunctional counterpart.

We also examined whether cisplatin and phenanthriplatin might regulate Wnt/β-catenin and TGF-β signaling differently in IMR90 compared to A549 cells through lncRNA regulation of miRNAs. Interestingly, a majority of the miRNAs, e.g., miR-149-3p, -185-5p, -608, -650, -708-5p, and -4516, identified as potential targets of lncRNAs in A549 cells, were predicted to be regulated identically in IMR90 cells albeit by different lncRNAs (Table 4, Fig. 5). Specifically, cisplatin upregulates lncRNAs, e.g., ATAD2B-5, GNG11-3, and MRPS5-1, that collectively should sponge all of these miRNAs, and phenanthriplatin downregulates lncRNAs, e.g., AGO2-1, ARCN1-1, and HYPM-1, that should allow expression of these miRNAs. Cisplatin also downregulates HYPM-1, which should allow expression of miR-608; whereas, the monofunctional complex upregulates TMEM243-1, a lncRNA with high sequence homology to this miRNA. Therefore, regulation of Wnt/β-catenin and TGF-β signaling through miR-149-3p, -185-5p, -608, -650, -708-5p, and -4516 would be predicted to be similar in both IMR90 and A549 cells. However, it is possible that simultaneous upregulation of multiple lncRNAs, e.g., ATAD2B-5 and MRPS5-1 in cisplatin treated IMR90 cells, and downregulation of multiple lncRNAs, e.g., AGO2-1, ARCN1-1, and HYPM-1 by phenanthriplatin, could synergistically promote or prevent respectively the action of other lncRNAs and might alter Wnt/β-catenin and TGF-β signaling differently than would occur from the effect of these lncRNAs and miRNAs alone. When we measured miR-149-3p and -608 expression in the IMR90 cell line, we found that cisplatin and phenanthriplatin did not exhibit distinct regulation of either miRNA (Fig. 3b). Thus, it is possible that the action of specific lncRNAs to modulate miRNA expression, and Wnt/β-catenin and TGF-β signaling, could be masked by the effect of larger combinations of up- and downregulated lncRNAs acting together.

Further, we also identified a potential miRNA target, miR-4458, unique to IMR90 cells. Cisplatin treatment upregulated lncRNA, ATAD2B-5, which might sponge miR-4458, while phenanthriplatin downregulated both AGO2-1 and HYPM-1, lncRNAs with targeting homology to this miRNA (Table 4). Interestingly, cisplatin treatment in IMR90 cells caused increased miR-4458 expression compared to phenanthriplatin (Fig. 3b), which could mean that increased ATAD2B-5 is less effective as a suppressor of this miRNA than the effect of decreasing either AGO2-1 or HYPM-1. As increased expression of miR-4458 prevents proliferation and migration in NSCLC cell lines via the PI3K/AKT pathway36, cisplatin treatment might suppress miR-4458 with ATAD2B-5 allowing cancer progression, while phenanthriplatin, by decreasing the expression of AGO2-1 and HYPM-1, could allow miR-4458 expression through modulating the PI3K/AKT pathway to promote anti-cancer activity.

As in A549 cells, phenanthriplatin regulated a distinct set of lncRNAs compared to cisplatin in IMR90 cells (Tables 1, 2) and might modulate Wnt/β-catenin and TGF-β signaling differently in normal lung fibroblast cells than in lung cancer cells (Figs. 4, 5). In A549 cells, miR-25-5p expression may be sponged by upregulated lnc-MRPL39-10, but we found in IMR90 cells that this miRNA would be potentially decoyed by lnc-HYPM-1, which is downregulated by phenanthriplatin (Tables 2, 3, 4). The absence of suppression of miR-25-5p by lnc-HYPM-1 in IMR90 cells could promote cell proliferation and viability via Wnt/β-catenin and/or TGF-β signaling integrating MAPK/ERK as found in A549 cells25. Similarly, miR-138-5p and miR-608 are potentially not subject to sponging in A549 cells treated with the monofunctional complex as their lncRNA targets are downregulated, but in IMR90 cells, these miRNA could be sponged by upregulated lnc-TMEM243-1 (Tables 2, 3, 4). Therefore, reduced expression of miR-138-5p and -608 due to TMEM243-1 decoying in IMR90 cells could lead to reduced cytotoxicity via TGF-β signaling (miR-138-5p) or Wnt/β-catenin and TGF-β signaling (miR-608) as suggested by the effects of suppressing these miRNAs in A549 cells23,46,47,48,49.

Materials and methods

Cell culture and treatment

The A549 and IMR90 cell lines were obtained from ATCC (Manassas, VA). A549 cells were cultured in F12K media with 10% FBS and 1% penicillin/streptomycin supplementation. IMR90 cells were cultured in Eagle's Minimum Essential Medium (EMEM) with 10% FBS and 1% penicillin/streptomycin supplementation. We seeded 6-well dishes with 3 × 105 cells per well and placed them in an incubator (37 °C, 5% CO2) for 24 h. Sets of three wells each were treated with either a negative control (media only), 5 µM cisplatin (Sigma-Aldrich, St. Louis, MO), or 5 µM phenanthriplatin (synthesized by Dr. Kevin M. Williams, Western Kentucky University Department of Chemistry, Bowling Green, KY according to a published protocol8), a platinum concentration that efficiently targets the nucleus at 24 h8. Then, the dishes were placed back into the incubator for 24 h after which they were prepared for RNA isolation.

RNA isolation

The Qiagen RNeasy kit (Hilden, Germany) was used to isolate RNA samples from control and platinum complex-treated dishes per kit instructions. Media was aspirated out of the wells followed by washing (1×) with phosphate buffered saline (PBS). The PBS was removed, and Buffer RLT Plus with β-mercaptoethanol was introduced into each well and homogenate was then transferred to a microcentrifuge tube. Each tube was vortexed and the lysate was transferred to a Gdna Eliminator spin column and centrifuged. Flow-through was collected and 70% ethanol was added. The samples were then transferred to an RNeasy spin column placed in a collection tube. Spin columns were centrifuged and the flow-through was discarded. Buffer RW1 was added to each spin column, followed by centrifugation. Flow-through was discarded and Buffer RPE with ethanol was added followed by centrifugation. The flow-through was again discarded and Buffer RPE with ethanol was then added followed by centrifugation. The spin column was then placed in a collection tube, centrifuged, and then placed in a new collection tube. Two elutions using RNase-free water were performed with centrifugation. All RNA samples were stored at − 80 °C for subsequent analysis.

Microarray assay

RNA transcription profiles from experimental and control samples were created using a microarray assay. RNA samples were analyzed using a NanoDrop 2000 (Thermo Fisher, Waltham, MA) and quality was assessed with an Agilent 2100 Bioanalyzer (Agilent, Wilmington, DE). RNA integrity was expressed as RNA integrity number (RIN) values and only samples with values greater than 10 were used. The Low RNA Input Linear Amplification kit with one-color (Agilent, Wilmington, DE) was used to produce fluorescent cRNA. Labeled cRNA was then purified using the RNeasy Mini Elute kit (Qiagen, Valencia, CA). Fluorescent cRNA samples were then hybridized onto Whole Human Genome 4 × 44 K microarrays GeneChips (Affymetrix, Santa Clara, CA). Four replicates were performed with array hybridization conducted for 17 h at 65 °C. Slides were then scanned with an Agilent microarray scanner (G2565BA) using Feature Extraction software (v 9.5.1, Agilent). GeneSpring GX 10.0.2 software (Agilent Technologies) was used for statistical analysis, background correction, normalization and summary of expression measure.

Quantitative real-time and locked nucleic acid polymerase chain reaction analysis

Microarray results were additionally validated using quantitative real time-polymerase chain reaction (qRT-PCR) on RNA samples also used for the microarrays. Custom probes were designed and specificity checked using the National Institutes of Health Primer-BLAST tool and obtained from Integrated DNA Technologies (Coralville, IA). Samples were placed in an Agilent Technologies Stratagene Mx3000P real-time PCR System (Santa Clara, CA) with MxPro software with thermal cycling at 95 °C for 10 min, followed by 40 cycles of denaturing at 95 °C for 30 s, annealing at 55 °C for 30 s, and extension at 72 °C for 30 s. Additionally, RNA samples were analyzed using locked nucleic acid PCR (LNA) miRNA to measure the expression of miRNA targets (A549: miR-25-5p, -378j, -4516; IMR90: miR-149-3p, -608, -4458). Custom LNA probes were obtained from Qiagen (Hilden, Germany). Samples were analyzed with the Mx3000P and MxPro software as per the qRT-PCR experiments except that thermal cycling was performed at 95 °C for 2 min (initial activation), followed by 40 cycles of denaturing at 95 °C for 10 s, and annealing/extension at 56 °C for 1 min. Three independent reactions were performed for each qRT-PCR and LNA PCR experiment with three technical replicates prepared in each. Quantitative data for qRT-PCR and LNA PCR experiments were normalized to the rRNA 5S gene.

Data and statistical analysis

Long noncoding RNA transcript IDs identified by the microarray assay were sequence characterized using the LNCipedia database (https://lncipedia.org/50. Identification of miRNA targeting of lncRNAs used the RNA22 v2 microRNA target detection software (https://cm.jefferson.edu/rna22/Interactive/51 with p ≤ 0.05 as the standard (except that for GNG11-3, p ≤ 0.0581). Microarray data was statistically analyzed using GeneSpring GX 10.0.2 software (Agilent Technologies) with p ≤ 0.05 as the standard. qRT-PCR and LNA PCR data were analyzed using the Livak equation to determine relative fold values and standard deviation fold change was calculated using the formula: s = ((s1)2 + (s2)2)1/2. GraphPad PRISM (version 9.0.0; La Jolla, CA) was used for all other statistical analysis with p ≤ 0.05 as the standard.