Dynamical Analysis of a Boolean Network Model of the Oncogene Role of lncRNA ANRIL and lncRNA UFC1 in Non-Small Cell Lung Cancer

Abstract

Long non-coding RNA (lncRNA) such as ANRIL and UFC1 have been verified as oncogenic genes in non-small cell lung cancer (NSCLC). It is well known that the tumor suppressor microRNA-34a (miR-34a) is downregulated in NSCLC. Furthermore, miR-34a induces senescence and apoptosis in breast, glioma, cervical cancer including NSCLC by targeting Myc. Recent evidence suggests that these two lncRNAs act as a miR-34a sponge in corresponding cancers. However, the biological functions between these two non-coding RNAs (ncRNAs) have not yet been studied in NSCLC. Therefore, we present a Boolean model to analyze the gene regulation between these two ncRNAs in NSCLC. We compared our model to several experimental studies involving gain- or loss-of-function genes in NSCLC cells and achieved an excellent agreement. Additionally, we predict three positive circuits involving miR-34a/E2F1/ANRIL, miR-34a/E2F1/UFC1, and miR-34a/Myc/ANRIL. Our circuit- perturbation analysis shows that these circuits are important for regulating cell-fate decisions such as senescence and apoptosis. Thus, our Boolean network permits an explicit cell-fate mechanism associated with NSCLC. Therefore, our results support that ANRIL and/or UFC1 is an attractive target for drug development in tumor growth and aggressive proliferation of NSCLC, and that a valuable outcome can be achieved through the miRNA-34a/Myc pathway.

Keywords: ANRIL; Boolean model; Long non-coding RNA; Myc; NSCLC; UFC1; apoptosis; feedback loops; miRNA-34a; senescence.

Figure 1

The gene regulatory network for the G1/S checkpoint in response to DNA damage. Arrows denote activations in green and hammer-head arcs represent inhibitions in black, respectively. Dashed hammer-head arcs in red represent targets of miR-34a. The yellow elliptic node represents miR-34a, whereas the blue elliptic nodes represent lncRNAs: ANRIL and UFC1, respectively. The input rectangular node in red denotes DNA Damage. The model outputs in orange in the rectangular nodes are Proliferation, Senescence, and Apoptosis, respectively.

Figure 2

Binding sites of miR-34a and wild-type case attractor of the Boolean networks. (A) On the right, predicted binding sites of miR-34a within ANRIL, whereas on the left within UFC1 (B) The stable states or fixed points identified for different scenarios are: WT, UFC1 KO + miR-34a KO, UFC1 KO, UFC1 E1, ANRIL KO + miR-34a KO, ANRIL KO, ANRIL E1, miR-34a KO, miR-34a E1 and Myc KO. Gain-of-function (GoF) and loss-of-function (LoF) perturbations correspond to its referential experiments [2,3]. The leftmost column shows the DNA damage level (highlighted in the black box) and the rightmost column presents the model outputs: proliferation, senescence, and apoptosis. Each line represents a fixed point corresponding to the input. Each cell/dot or circle represents a molecule. Light red cells indicate a zero value, whereas red ones indicate activation (value 1). Molecules that are involved in the regulation of miR-34a according to its referential studies (highlighted in the orange box). Orange arrows indicate up-regulation of the molecule, whereas the inverted black arrow represents the downregulation of the molecule, respectively.

Figure 3

Probabilities of perturbed cases of Boolean networks using Monte Carlo simulations (100,000 runs) and comparison with experimental studies by Zang et al. [3] and Nie et al. [2]. Each bar represents a specific case of the perturbation. As such, the first case is UFC1 KO in A549 cells, the second case is UFC1 KO along with miR-34a E1 in A549 cells. The third case is ANRIL KO, while the fourth case is ANRIL KO together miR-34a E1 in H1299 cells. The fifth case is ANRIL KO in SPC-A1 cells, whereas the sixth is ANRIL KO along with miR-34a E1 in SPC-A1 cells. E1 represents gain-of-function (GoF) and knockdown (KO) represents loss-of-function (LoF), respectively. See Section 3.3 for more information.

Figure 4

Duplex structure of miR-34a and model perturbations. (A) Binding site between miR-34a and Myc. (B) Perturbations of miR-34a & Myc in response to DNA damage, miR-34 E1, Myc KO and Myc E1. Gain-of-function (GoF) and loss-of-function (LoF) perturbations corresponding to knockdown (KO) and overexpression (E1) of their referential experiments (He et al. [6]). The leftmost column shows the DNA damage level (highlighted in the black box) and the rightmost column presents the model outputs, including proliferation, senescence, and apoptosis. Each line represents a fixed point corresponding to the input. Each cell/dot or circle represents a molecule. Light red cells indicate a value of zero, whereas red indicates activation (value 1). Activity of miR-34a on the Myc expression as suggested by [6] (highlighted in the orange box). Orange- arrows indicate up-regulation of the molecule, whereas the inverted black arrow represents the downregulation of the molecule, respectively.

Figure 5

Workflow of the model construction and validation by the cycle of 1 to 4-steps. Step 1, The model was simulated using GINsim 3.0.0b, based on the biochemical information provided by PubMed works of literature and databases. In Step 2, we verified the coherence between the outcome produced by the perturbed nodes and experimental observations using the same perturbations. Step 3, Once the model was collaborated with its referential studies we decided to perform “what if?” scenarios i.e., the hypotheses derived from the “what if” simulations can be used to design-focused experiments. In Step 4, these experimental designs lead to new in-vivo/in-vitro experiments in NSCLC as we see in Step 1. Worth noting is that this analysis was performed using the data used to build the network (see Table S1). To validate the Boolean Network (see Table 1), we used experimental studies that were not utilized in its construction.

Figure 6

Functional Versus Disrupted positive circuits in NSCLC. The left-hand side: disrupted positive circuits due to the lack of DNA damage in NSCLC cells, i.e., inactivation of miR-34a expression, consequently promoting uncontrolled proliferation. On the Right-hand side: In response to DNA damage, the positive circuits between (miR-34a/Myc/lncRNA ANRIL, miR-34a/E2F1/lncRNA ANRIL, and miR-34a/E2F1/lncRNA UFC1) are stimulated. Activated miR-34a inhibits Myc and E2F1 i.e., inducers of lncRNA ANRIL and lncRNA UFC1. Targeting Myc and E2F1 by the miR-34a prevented tumor growth through the induction of senescence and apoptosis in NSCLC cells. In this way, these positive circuits playing an important role to inhibit proliferation in DNA damage response. Green Arrows represent activation, whereas red hammer-head arrows represent inhibition. Black hammer-head arrows represent non-functional/inactivation of miR-34a, respectively.