Fig 1.
An example of the network model in the study.
miRNA X was known to target the pathways P1 and P2; miRNA Y, on the other hand, was related to the pathways P2, P5 and P8. The pathway based network model helped us to connect miRNA X and miRNA Y through their common pathway P2. The weight of the edge between miRNA X and miRNA Y was set to be one, since they shared one single pathway. The corresponding significance value was calculated using hypergeometric distribution for each miRNA pair. The less significant miRNA pairs with p-value larger than 0.05 were filtered out of the network.
Table 1.
20 responsive miRNAs with the greatest difference in expression levels in trastuzumab treated SKBR3 and BT474 cell lines.
Table 2.
Detailed information of the pathway based miRNA-miRNA network dataset in the trastuzumab treated BT474 and SKBR3 cell lines.
Table 3.
Top 30 genes targeted by responsive miRNAs in trastuzumab treated BT474 and SKBR3 cells that were used for the functional validation.
Table 4.
Lengths of miRNA target sites for the most targeted genes in the trastuzumab miRNA-miRNA networks.
Fig 2.
The expression values of top 30 targeted genes in the network in TCGA breast cancer data.
(A-B) TCGA breast invasive carcinoma (BRCA) gene expression (AgilentG4502A_07_3 array) was obtained by using 1247 samples in total. The expression values of the genes were given according to molecular subtypes of breast cancer (Luminal A, Luminal B, HER2+, Basal Like and normal like). The expression levels were indicated as in log2 lowess normalized ratio of sample signal to reference signal (cy5/cy3) collapsed for each gene. In order to view the differential expression between samples more easily, the default view was set to center each gene or exon to zero by independently subtracting the mean of each gene or exon on the fly. The data sets were visualized by using Xena Browser.
Table 5.
15 central nodes with the highest degree scores in trastuzumab treated SKBR3 and BT474 miRNA networks.
All the nodes values were statistically significant with P-value <0.05.
Fig 3.
The miRNA-miRNA network in SKBR3 cell.
In SKBR3 cells, 73 trastuzumab responsive miRNAs are found to be functionally relevant with each other. Each node represents a responsive miRNA and the nodes are sized by their degree centrality scores. hsa-miR-3976, hsa-miR-548b-5p and hsa-miR-3194-5p are identified to be the most central nodes with the degree scores of 9 and 8 (P<0.05 for each miRNA pair, red nodes:upregulated miRNAs, green nodes: downregulated miRNAs).
Fig 4.
The miRNA-miRNA network in BT474 cell.
In BT474 cells, 150 trastuzumab responsive miRNAs are defined as functionally relevant with each other. Each node represents a responsive miRNA and the nodes are sized by their degree centrality scores. hsa-miR-3671 is the most central node in the network with the degree centrality score of 19 (P<0.05 for each miRNA pair, red nodes:upregulated miRNAs, green nodes: downregulated miRNAs).
Fig 5.
The largest cluster in the BT474 miRNA-miRNA network.
The most powerful interaction consisted of the thick edges presented as a triangle (red) between hsa-miR-3064-3p, hsa-miR-32-3p and hsa-miR-216b. The edges are comprised of the metabolic pathways that also dominate the interactions between the other nodes in the complete cluster. The aforementioned pathways were shown in red boxes in left side. (The edge weight minimum value = 2.9, P<0.05 for each miRNA pair, red nodes:upregulated miRNAs, green nodes: downregulated miRNAs).
Fig 6.
The second and third largest clusters in the BT474 miRNA-miRNA network.
(A) The most powerful interaction consisted of one thick edge (red) presented between hsa-miR-5692a and hsa-miR-3121-3p. The edges are made of the cancer related pathways that control majority of the interactions between the other nodes in the complete cluster. The aforementioned pathways are shown in red boxes in left side. (B) In the last cluster, hsa-miR-29b-2-5p (shown in yellow) has important role as a hub node to unite two different groups of miRNAs that enriched in path:hsa05220 (Chronic myeloid leukemia), path:hsa04010 (MAPK signaling pathway) and path:hsa04060 (Cytokine-cytokine receptor interaction), path:hsa04630 (Jak-STAT signaling pathway) pathways (The edge weight minumum value = 2.9, P<0.05 for each miRNA pair, red nodes:upregulated miRNAs, green nodes: downregulated miRNAs).
Fig 7.
The largest cluster in the SKBR3 miRNA-miRNA network.
The most powerful interaction is consisted of the thick edges presented as a triangle (red) between hsa-miR-200a-3p, hsa-miR-513a-3p and hsa-miR-216b. The edges are once again made of the metabolic pathways that also dominate the rest of interactions in the cluster. The afermentioned pathways were shown in red boxes in left side. (The edge weight minumum value = 2.9, P<0.05 for each miRNA pair, red nodes:upregulated miRNAs, green nodes: downregulated miRNAs).
Fig 8.
The second and third largest clusters in the SKBR3 miRNA-miRNA network.
(A) The most powerful interaction is consisted of one thick edge (red) presented between hsa-miR-3942-5p and hsa-miR-298. The edges are made of the cancer related pathways that control majority of the interactions between the other nodes in the cluster. The afermentioned pathways are shown in red boxes in left side. (B) In the last cluster, the interactions are determined by the upregulated miRNAs mostly and they are related to each other through path:hsa04810 (Regulation of actin cytoskeleton) and path:hsa05200 (pathways in cancer) (The edge weight minumum value = 2.9, P<0.05 for each miRNA pair, red nodes:upregulated miRNAs, green nodes: downregulated miRNAs).