We identified candidate drugs to attenuate pulmonary fibrosis.
To explore the novel therapeutic candidate of pulmonary fibrosis, we applied a practical in silico screening approach using multi-sources data information and integrative molecular network bioinformatics. First, we obtained the microarray data of IPF from the GEO database and compared microarray data from IPF patients (n = 26) with those from normal patients (n = 9) registered with the GEO as GSE5774 in order to detect DEGs (defined as an adjusted P-value of < 0.05 and |log2FC| > 0.585). As a result, we detected 833 DEGs.
We next searched for therapeutic candidates for IPF by inputting DEGs into the L1000CDS2 (Fig. 1A). When the condition was |log2FC| > 0.585, nintedanib, which is currently approved for IPF treatment, was listed as the second-best candidate (Fig. 1B). We selected BI2536, a polo-like kinase 1/2 inhibitor, as a therapeutic candidate drug. PLK is serine/threonine kinase that regulates the cell cycle [19]. We selected BI2536 because it had no previous reports concerning any anti-fibrotic effects, was ready-made and available, and was ranked higher based on the strict criterion of |log2FC| > 1 (Fig. 1C).
The PLK1/2 inhibitor accelerated the body weight loss and increased the mortality of bleomycin-treated mice.
To investigate whether or not BI2536 had anti-fibrotic effects, we induced pulmonary fibrosis in mouse lungs with bleomycin. C57BL/6 mice received a single intra-tracheal instillation of bleomycin (3 mg/kg) on day 0. BI2536 (10 mg/kg or 20 mg/kg) or vehicle was then administered via intraperitoneal instillation twice a week until day 21. Lung tissues were harvested and analyzed on day 21 (Fig. 2A). The weight of mice in the bleomycin-alone group gradually started to decrease on day 2 (Fig. 2B). Surprisingly, in the bleomycin + BI2536 group, the body weight started to decrease soon after the intratracheal instillation of bleomycin. The rate of reduction was higher than in the bleomycin-alone group, and a significant difference was found on days 4, 7, 9, 11, 14, 16, 18, and 21 (Fig. 2B). Along with the decrease in the body weight, the survival rate in the bleomycin + BI2536 group also began to decline from an early period, and as a result, the total survival rate was lower than that in the bleomycin-alone group (Fig. 2C).
However, by the continuous administration of BI2536, the number of fibrotic lesions in the lungs of bleomycin-treated mice was reduced (Figure E1A). A hydroxyproline colorimetric assay showed that there was a decreasing trend in the collagen content in the lungs of mice treated with BI2536 (Figure E1B). A quantitative histological analysis showed that the Ashcroft fibrotic score was significantly lower in mice treated with bleomycin + BI2536 than in those treated with bleomycin alone (Figure E1C). These results suggest that the administration of a PLK1/2 inhibitor improves pulmonary fibrosis but worsens the rate of weight loss and survival.
PLK1 expression was dominant in myofibroblasts, whereas PLK2 expression was dominant in lung epithelial cells.
To examine why the PLK1/2 inhibitor BI2536 increased the mortality of bleomycin-treated mice, we next focused on the localization of PLKs. Although PLKs consists of five families (PLK1-5), the localization of each PLK in the lung has not been fully studied [19]. Among the vertebrate PLK family members, PLK1 has been most extensively studied, and it along with PLK2 have been reported to be involved in cell proliferation in the G2-M phase and G1 phase, respectively [19].
Immunofluorescence staining revealed that the number of PLK1+ cells was increased in fibrotic regions (Fig. 3A, E2A). In particular, PLK1+ cells were co-localized with alpha smooth muscle actin (α-SMA)+ cells (Fig. 3A) but not with pro-surfactant Protein C (pro-SPC)+ cells (Figure E2A). Conversely, the number of PLK2+ cells was decreased in fibrotic regions (Fig. 3B, E2B). Immunofluorescence staining revealed that PLK2+ cells were co-localized with pro-SPC+ cells (Fig. 3B) but not with α-SMA+ cells (Figure E2B).
We extracted mRNA from CD45− epithelial-cell adhesion molecule (Ep-CAM)− cells, a fibroblast-enriched population, and CD45−Ep-CAM+ cells and performed quantitative polymerase chain reaction (PCR) for Plk1 and Plk2. The mRNA expression of Plk1 was significantly higher in CD45−Ep-CAM− cells than in CD45−Ep-CAM+ (Figure E2C), but the expression of Plk2 was higher in CD45−Ep-CAM+ than in CD45−Ep-CAM− cells (Figure E2D).
These results suggest that PLK1 is mainly expressed in myofibroblasts, while PLK2 is mainly expressed in lung epithelial cells.
A selective PLK1 inhibitor attenuated pulmonary fibrosis with acceptable mortality and weight loss.
In developing new anti-fibrotic drugs for pulmonary fibrosis, especially when targeting factors related to proliferation, such as PLKs, it is desirable for an agent to be effective on fibroblasts but have less effect on lung epithelial cells. Therefore, we hypothesized that the inhibition of PLK1 alone would have an anti-fibrotic effect without suppressing the recovery of lung epithelial cells from injury and examined the anti-fibrotic effect of GSK461364, a selective PLK1 inhibitor, in a mouse model of bleomycin-induced pulmonary fibrosis.
After mice received intratracheal instillation of bleomycin, GSK461364 (5 mg/kg) was administered by intraperitoneal instillation twice a week until day 21. Lung tissues were then harvested and analyzed on day 21 (Fig. 4A). In contrast to the experiments with BI2536, there was no marked difference in the rate of weight loss or survival between the bleomycin-alone and bleomycin + GSK461364 groups (Fig. 4B, 4C). Furthermore, GSK461364 attenuated bleomycin-induced lung fibrosis (Fig. 4D) and reduced the collagen content in the lungs (Fig. 4E). A quantitative histological analysis also showed that the Ashcroft fibrotic score was significantly lower in mice treated with bleomycin + GSK461364 than in those treated with bleomycin alone (Fig. 4F). These results suggest that a selective PLK1 inhibitor attenuated pulmonary fibrosis without worsening the rate of weight loss or survival.
Selective PLK1 inhibition does not affect the infiltration of inflammatory cells into the lungs.
To investigate why selective PLK1 inhibition ameliorated lung fibrosis in bleomycin-treated mice, we next examined whether or not GSK461364 treatment affected the number of inflammatory cells and lymphocyte fractions in BALF. The BALF was harvested from mice in each group on day 21. A BALF analysis showed no marked differences in the cell count, macrophage percentage, lymphocyte percentage, or neutrophil percentage between the bleomycin-alone group and the bleomycin + GSK461364 group (Figure E3A-E3D).
Selective PLK1 inhibitor does not inhibit the proliferation of lung epithelial cells in vivo.
To examine the effects of BI2536 on lung fibroblasts or epithelial cells, we harvested lung tissue from mice administered bleomycin or bleomycin and BI2536 20 mg/kg on day 21. Paraffin-embedded lung sections were stained for the fibroblast marker α-SMA or the epithelial cell marker pro-SPC and the proliferation marker Ki-67. Immunohistochemical staining revealed that the number of a-SMA+Ki-67+ cells in the bleomycin + BI2536 group was significantly lower than that in the bleomycin-alone group (Figs. 5A). Furthermore, the number of pro-SPC+Ki-67+ cells in the bleomycin + BI2536 group was also significantly lower than in the bleomycin-alone group (Figs. 5B).
In contrast, although the number of a-SMA+Ki-67+ cells in the bleomycin + GSK461364 group was significantly lower than that in the bleomycin alone-group (Fig. 5C), the number of pro-SPC+Ki-67+ cells was not decreased by GSK461364 (Fig. 5D).
Taken together, these data suggest that a PLK1/2 inhibitor inhibited the proliferation of both lung fibroblasts and epithelial cells, while a selective PLK1 inhibitor inhibited the proliferation of lung fibroblasts only.
Growth factors upregulates PLK1 in lung fibroblasts and PLK2 in lung epithelial cells.
The fact that growth factors play a pivotal role for the fibrotic process by inducing fibroblast proliferation is now widely accepted. Indeed, numerous reports suggest that FGF and PDGF are involved in the pathogenesis of pulmonary fibrosis [20–25][17]. Therefore, we next examined whether or not these growth factors affect the expression of PLK, a regulator of the cell cycle.
Because PDGFR is expressed on lung fibroblasts but not on lung epithelial cells, the effect of PDGF was examined only on lung fibroblasts [26]. At the mRNA level, the stimulation with PDGF-BB or FGF-2 significantly upregulated the expression of Plk1 in murine primary lung fibroblasts (Fig. 6A, 6D). In contrast, the expression of Plk2 was not changed by PDGF-BB or FGF-2 stimulation (Fig. 6B). Similarly, the protein level of PLK1 in murine primary lung fibroblasts was also significantly upregulated by PDGF-BB or FGF-2 stimulation (Fig. 6C).
Conversely, murine lung epithelial cells stimulated with FGF-2 significantly upregulated the mRNA expression of Plk2 but not Plk1 (Fig. 6E). The PLK2 protein level was also significantly upregulated by FGF-2 stimulation (Fig. 6F).
These data suggest that the stimulation with growth factors induce PLK1 expression in lung fibroblasts and PLK2 in lung epithelial cells.
PLK1/2 inhibition suppresses both lung fibroblast and epithelial cell proliferation, whereas PLK1 inhibition inhibits only lung fibroblast proliferation.
We next examined whether or not PLK inhibition affect the biological responses of cells to growth factors using a 3H-TdR incorporation assay. FGF-2 or PDGF-BB significantly upregulated the proliferation of murine primary lung fibroblasts and murine lung epithelial cells (Fig. 7A-7E). Proliferation of murine primary lung fibroblasts induced by FGF-2 or PDGF was significantly suppressed by BI2536 or GSK461364 administration at ≥ 30 nM (Fig. 7A-7E). Conversely, murine lung epithelial cell proliferation induced by FGF-2 was significantly suppressed by BI2536 administration at ≥ 30 nM (Fig. 7C) but not by GSK461364 (Fig. 7F).
Taken together, these data suggest that PLK1/2 inhibitors suppressed the proliferation of both lung fibroblasts and epithelial cells, while a selective PLK1 inhibitor suppressed lung fibroblast proliferation but not lung epithelial cell proliferation.
BI2536 downregulates the mRNA expression of Col1a1.
Based on the result so far, both BI2536 and GSK461364 seemed to have anti-fibrotic effects. However, only BI2536 (not GSK461364) was selected by our in silico analysis. To determine the reason for this discrepancy, we re-examined genes that were suppressed or upregulated by BI2536 in the data that had been initially analyzed by the drug discovery tool L1000CDS2 to select BI2536 as a therapeutic candidate of IPF (Figure E4A). As a result, BI2536 was found to suppress collagen-related genes, such as COL1A1, COL1A2, COL3A1, COL5A1, COL5A2, and COL6A1 (Figure E4A), which was assumed to be the main reason that BI2536 was selected in the in silico analysis. Therefore, we next tried to confirm these inhibitory effects for collage-related genes using quantitative PCR.
The mRNA expression of Col1a1 upregulated by TGFβ-1 was significantly inhibited by BI2536 (Figure E4B, C). In contrast, it was not inhibited by GSK461364 (Figure E4D, E). The mRNA expression of Acta2 was not affected by BI2536 or GSK461364.
Therefore, the reason why GSK461364 was not selected by the in silico analysis was deemed to be due to its lack of an inhibitory effect on pro-fibrotic gene expression.