Inflammatory molecules upregulated by ECPsp are central in the functional linkage network

To understand the gene expression profile triggered by ECPsp, we used DNA microarray technology to analyze the transcriptome. All microarray analyses were conducted by Phalanx Biotech (Taiwan). Each total RNA, derived from the cells transfected with control plasmids or plasmids containing ECPsp, was duplicated on the array. We performed two technical replicates of all experiments to confirm the data and account for experimental variations. According to the gene expression data, 93 genes were upregulated by more than twofold with a statistics p value less than 0.05 in ECPsp-expressing cells relative to controls (Table1). The 93 genes were submitted to the functional linkage network obtained from the STRING 9.0 database, and a network was generated by choosing the default setting (Figure1).

Gene Symbol	Fold	p-value	Gene symbol	Fold	p-value
IFI44L	5.627	6.13E-04	FZD4	2.470	4.53E-06
CMPK2	5.477	5.24E-07	SULF1	2.455	8.71E-03
RSAD2	5.401	2.36E-06	TNFRSF8	2.452	4.53E-04
CXCL11	5.269	1.02E-04	IFIT5	2.439	3.46E-04
CCL5	4.823	6.01E-05	IFI30	2.431	3.64E-03
ANGPT1	4.542	1.11E-05	EGR1	2.423	6.42E-05
TNFSF10	4.531	2.63E-04	IFI35	2.413	1.26E-03
SAMD9L	4.208	6.56E-08	VCAM1	2.412	8.27E-04
IFIH1	4.199	3.38E-08	GBP1	2.397	4.11E-10
ISG20	4.169	1.14E-04	FGF2	2.385	2.62E-15
IDO1	3.956	1.50E-04	IFI16	2.360	4.28E-06
OAS1	3.919	2.71E-03	STAT1	2.350	2.85E-06
TLR3	3.908	2.72E-07	PLSCR1	2.331	4.79E-05
MX2	3.860	5.31E-04	IRF7	2.305	3.98E-03
CXCL10	3.827	6.56E-10	CD7	2.304	2.87E-04
HERC5	3.783	2.17E-07	IFI27	2.285	3.08E-02
RASGRP3	3.745	2.65E-05	DHX58	2.276	5.73E-04
OAS2	3.695	1.56E-04	PLCG2	2.251	6.55E-07
DDX58	3.596	5.24E-09	UBE2L6	2.248	8.97E-04
XAF1	3.522	1.65E-04	IFI6	2.243	3.62E-02
IFIT2	3.496	2.09E-09	BST2	2.222	2.91E-02
GPR109B	3.435	6.36E-05	NCF2	2.216	1.03E-13
LAMP3	3.295	3.38E-04	ZC3HAV1	2.210	4.61E-06
GBP4	3.194	1.68E-07	PLCB4	2.204	3.35E-10
OASL	3.074	1.79E-05	EDN1	2.204	4.94E-02
HERC6	3.027	8.91E-06	C1R	2.203	4.43E-02
TRIM22	2.961	6.67E-08	COL1A2	2.193	3.44E-03
USP18	2.952	8.08E-04	SP100	2.185	3.74E-08
SP110	2.941	1.82E-06	PARP12	2.185	3.27E-05
MX1	2.891	9.80E-04	GRB10	2.177	5.06E-10
IFI44	2.863	8.60E-08	TLR2	2.163	1.86E-03
PARP9	2.859	8.82E-03	ISG15	2.148	4.58E-04
IFIT1	2.843	1.36E-04	DTX3L	2.134	3.69E-06
IFIT3	2.819	1.91E-04	IFNB1	2.122	4.83E-05
OLR1	2.811	8.00E-05	UBA7	2.116	1.62E-02
TMEM140	2.802	2.26E-04	TYMP	2.105	2.76E-02
HSH2D	2.786	1.43E-02	NT5C3	2.097	2.00E-05
TRANK1	2.759	5.27E-04	GMPR	2.080	1.90E-04
PARP14	2.741	3.27E-07	OAS3	2.069	6.65E-04
STAT2	2.687	2.46E-05	AIM2	2.060	2.46E-07
CASP1	2.678	2.02E-03	CXCL16	2.046	9.54E-04
IFITM1	2.660	4.49E-03	PRSS23	2.028	6.38E-12
CEACAM1	2.612	4.38E-05	TRIM5	2.010	1.12E-03
NEDD9	2.529	1.75E-03	STON2	2.009	1.13E-02
RARRES3	2.520	4.00E-08	C4A	2.006	3.18E-02
EPSTI1	2.484	3.17E-03	C4B	2.006	3.18E-02
CFB	2.470	2.39E-02

Within this sub-network, a clearly evident area of dense interconnections contains nodes that are related to inflammatory responses. These include chemokines, interferon-induced/related molecules, and inflammatory receptors, such as Toll-like receptors. The focused area thus reveals that ECPsp may induce the expression of genes that encode proinflammatory molecules for triggering inflammatory pathways. This discovery is crucial to understanding the biological function of ECPsp. The function of ECPsp as a conventional signal peptide for the secretion of mature ECP is well documented[23]. We used differential display technique and real-time RT-PCR to demonstrate that ECPsp induces TGF-α and EGFR overexpression[22]. Another study indicated that ECPsp is cytotoxic to bacteria and yeast, but not to mammalian cells[21]. These findings demonstrated that ECPsp regulates gene expression in addition to executing its conventional function, protein secretion.

Patients with asthma usually experience severe inflammation in the respiratory tract. The activated eosinophils that gather in the airways release mature ECP, which in turn damages bronchial epithelial cells. At the same time, it is possible that ECPsp might induce and release cytokines and chemokines to attract immune cells, such as T cells, NK cells, eosinophils and macrophages.

To explain the ontology of the functional network, Network Ontology Analysis (NOA) which applied the categories of Gene Ontology to network analysis[24] would be employed. In the NOA results, the category ‘biological process’ is related to inflammation that is involved in the immune response, including viral infection and biotic stimulation. The ‘cellular component’ is consistent with the extracellular release of cytokines, chemokines, and interferons. Finally, the ‘molecular function’ is related to the production of chemokines and cytokines and activation of the JAK/STAT (signal transducers and activators of transcription) signaling pathway, which is involved in interferon signaling and chemokine production (Table2)[25]. To confirm the upregulation of molecules related to chemokine and JAK/STAT signaling, the mRNA levels of CCL5, CXCL10, CXCL11, CXCL16, STAT1, and STAT2 were measured by semi-quantitative RT-PCR. Indeed, levels of mRNAs encoding CCL5, CXCL10, CXCL11, CXCL16, STAT1, and STAT2 were upregulated 2.07-, 4.21-, 7.52-, 2.6-, 3.58-, and 1.67-fold, respectively, in cells that expressed ECPsp-eGFP as compared with cells that expressed eGFP only (Figure2).

Gene ontology	GO: Term	p-value	Term name
Biological process	GO:0009615	2.1E-24	response to virus
	GO:0009607	1.3E-20	response to biotic stimulus
	GO:0051707	9.7E-20	response to other organism
	GO:0002376	4.3E-19	immune system process
Cellular component	GO:0005615	1.8E-6	extracellular space
	GO:0005737	9.7E-6	cytoplasm
	GO:0044421	1.0E-4	extracellular region part
	GO:0005576	0.0023	extracellular region
Molecular function	GO:0005062	3.0E-6	hematopoietin/interferon-class (D200-domain) cytokine receptor signal transducer activity
	GO:0016763	4.9E-6	transferase activity, transferring pentosyl groups
	GO:0003950	1.1E-5	NAD+ ADP-ribosyltransferase activity
	GO:0008009	1.2E-4	chemokine activity

Integrating insights into ECPsp-mediated gene induction with knowledge of TGF-α/EGFR signaling

We previously demonstrated that ECPsp induced TGF-α and EGFR expression at the mRNA and protein levels in A431 cells[22]. The growth factor TGF-α is one of the ligands that stimulates EGFR phosphorylation and triggers the signaling pathways downstream of this event. Since the analysis of the KEGG database revealed that the STAT pathway is triggered by TGF-α via EGFR, resulting in cell proliferation, cell survival, migration and invasion, and the STATs could be activated by chemokines and chemokine receptors and then drive the expression of cytokine-encoding genes, we hypothesized that STATs may serve as key transcriptional regulators in the inflammatory response triggered by ECPsp. Thus, incorporation of TGF-α, EGFR, and the ECPsp-induced genes may enable the discovery of new regulatory networks initiated by ECPsp. To reduce the complexity of this analysis, the area of dense interconnections in Figure1 was extracted and combined with TGF-α and EGFR signaling pathway components.

The area of dense interconnections was determined manually, which was enlarged and extracted from a seed of IFNB1 to the borders of a proinflammatory molecule and a non-proinflammatory molecule. There are 40 nodes in the area of dense interconnections. After their incorporation, the resulting network showed that the TGF-α/EGFR pathway connected with inflammatory molecules via STAT1 and STAT2, important transcriptional factors that regulate cytokine expression and release (Figure3)[26]. Therefore, ECPsp may initiate alternative inflammatory responses in cells when ECP is expressed simultaneously at asthma/inflammatiory condition. In addition, within these 41 nodes, 9 (CEACAM1, PLCG2, EGR1, GRB10, CASP1, FGF2, ANGPT1, and TNFSF10) were not related with inflammatory responses. However within the 9 nodes, 7 (CEACAM1, PLCG2, EGR1, GRB10, CASP1, FGF2, ANGPT1) were located within the TGF-α/EGFR pathway network and these nodes functioned as cell proliferation, angiogenesis, and apoptosis, corresponding to the proposed functions of this subnetwork. The other two nodes, TNFSF10 and DDX58, were not proinflammatory molecule within the proinflammation network composed with 34 genes. Thus we concluded that within the proinflammation network, 94.12% (32/34) of nodes were inflammation-related ones.

Since it is hypothesized that STAT1 and STAT2 might serve as key regulators between the subnetworks in BEAS-2B cells, we also investigated the mRNA and protein levels of STAT1 and STAT2 in A431 cells in which the TGF-α and EGFR are upregulated by ECPsp. If the STAT1 and STAT2 also upregulated by ECPsp, it is confident to merge the TGF-α/EGFR signaling network and the proinflammatory network. As shown in Additional file1: Figure S1, the levels of mRNAs of STAT1 and STAT2 were increased in the presence of ECPsp by 1.67- and 1.57-fold, respectively, as well as the protein level by 1.54- and 2.72- fold, respectively. These results indicated that STATs might serve as real hubs to connect two subnetworks and regulate the gene expression induced by ECPsp.

STAT1 and STAT2 serve as hubs that connects TGF-α pathways and cytokine/chemokine production

To obtain a high-confidence network, we further studied the molecules that were induced by ECPsp by more than fourfold. The underlying functional linkage network is shown in Figure4. The functional linkage network was combined with the current knowledge of TGF-α, EGFR, and STAT1/2 signaling. As shown in Figure4, STAT1 and STAT2 still serve as hubs that connect two functional clusters. One cluster contains interferon-inducing factors and chemokines, such as IFI44L, IFIH1, CCL5, and CXCL11. Another cluster comprises molecules related to cell growth and proliferation, such as TGF-α and EGFR. This network provides potentially valuable avenues for future research concerning the novel functions of ECPsp.

Medium conditioned by ECPsp-expressing cells causes migration of RAW 264.7 cells

Our observation that the expression of genes encoding the chemokines CCL5, CXCL10, CXCL11, and CXCL16 is induced by ECPsp suggests that cells expressing ECPsp may induce chemoattraction. Among these four chemokines, some may be responsible for novel biological functions of ECPsp. Several reports show that CXCL10[27], CXCL11[28], and CXCL16 can initiate the homing of Th1 cells, NK cells, and eosinophils[29] and that CCL5/RANTES attracts macrophages and eosinophils[30–33]. We assayed macrophage migration using the human macrophage-like cell line RAW 264.7 to investigate whether the cytokines produced by ECPsp regulate the recruitment of immune cells. Compared with medium conditioned by culturing ECPsp/BEAS-2B and eGFP/BEAS-2B cells, the respective rates of chemoattractant-mediated macrophage migration were 2.93- and 3.11-fold higher, respectively, for medium conditioned by cells expressing ECPsp-eGFP (Figure5). Given that ECPsp is expressed by activated eosinophils at sites of inflammation or infection, secretory chemokines might attract macrophages to eliminate damaged cells or pathogens. This is the first time that ECPsp has been proposed to function not only in secretion but also in assisting immune-cell migration. In order to demonstrate whether STAT1 and STAT2 serve as key transcriptional regulators or not experimentally, the individual siRNA against STAT1 or STAT2 was transfected into BEAS-2B cells which express eGFP or ECPsp-eGFP. Figure6 reveals that while the mRNA and protein levels of STAT1 and STAT2 were knocked down, the migration of macrophages, which was induced by ECPsp, was downregulated by 2.38- and 2.50-fold, respectively. It indicated that the migration of macrophages enhanced by ECPsp was indeed regulated via STAT1 and STAT2. It might reflect our hypothesis that the STAT1 and STAT2 serve as hubs to regulate the gene expression of chemokines and cytokines in the ECPsp/BEAS-2B cells.

Cells and cell culture

We cultured the human bronchial epithelial cell line BEAS-2B (ATCC CRL-9609) and the mouse macrophage RAW 264.7 cell line in RPMI-1640 (Invitrogen, USA) supplemented with 10% FBS. Human epidermoid carcinoma cell line A431 was cultured in DMEM (GIBCO) supplemented with 10% heat-inactivated fetal bovine serum. All cells were cultured at 37°C in an incubator under 5% CO₂ and 95% air.

Transfection

The plasmids pEGFPC1 and pEGFPN1-ECPsp were individually transfected into BEAS-2B cells using TurboFect^TM (Thermo Fisher Scientific Inc., USA). For the transfection, 2 × 10⁶ BEAS-2B or A431 cells were seeded onto a 100-mm culture plate. After 24 h in culture (~80-90% confluency), 3 μg of plasmid DNA and 6 μl of TurboFect^TM reagent were mixed in 1 ml of serum-free RPMI-1640. The transfection mixture was gently vortexed and incubated for 15 min at room temperature to allow the formation of transfection complexes. The 1-ml transfection mixture was then added drop-wise to the cells and incubated at 37°C for another 48 h, at which point the cells were harvested for RNA isolation and Western blotting.

For plasmid and siRNA co-transfection, 3 × 10⁵ BEAS-2B cells were seeded onto a 60-mm culture plate. After 24 h, the pEGFPN1-ECPsp and siRNA (against STAT1 or STAT2, final concentration: 50 nM) were added into 200 μl jetPRIME® buffer (Polyplus-transfection) and mixed by pipetting. Then, 8 μl jetPRIME® reagent was added. The transfection mixture was gently vortexed, and incubated at room temperature. After 15 min, the transfection mixture was added drop-wise to the cells and incubated at 37°C. The original medium was replaced by fresh one at 24 h post-transfection. After additional 24 h, the cell lysates were harvested for Western blotting.

RNA extraction and cDNA preparation

Total RNA from BEAS-2B cells was isolated using TRIzol reagent (Invitrogen, USA). Cells that had been cultured in a 100-mm dish were washed with cold PBS. TRIzol reagent (1 ml) was added, and the cells were lysed by pipetting the mixture up and down several times. The cell lysate was then transferred to a 1.5-ml microcentrifuge tube. After the addition of 300 μl of cold chloroform (Sigma, USA), the lysate was gently mixed for 15 seconds, incubated on ice for 15 min, and then centrifuged at 10,000 ×g at 4°C for 15 min. The RNA, which remained exclusively in the aqueous phase, was transferred to a new tube. An equal volume of 100% isopropanol was then added and mixed gently by inverting the microcentrifuge tube repeatedly. The precipitated RNA was recovered by centrifugation at 12,000 rpm for 15 min at 4°C. After removal of the supernatant, the RNA pellet was washed twice with 75% ethanol dissolved in diethylpyrocarbonate (DEPC)-treated water and then air-dried. The RNA was dissolved in sterile DEPC-treated water. RT-PCR was used to determine gene expression based on the level of RNA abundance. Total RNA (2 μg) and 1 μl of 10 μM oligo-dT primer (MDBio, Taiwan) were mixed in DEPC-treated water and heated at 70°C for 5 min. The reverse transcription mixture containing 5 μl of Moloney Murine Leukemia Virus (M-MLV) 5× reaction buffer (Promega, USA), 5 μl of 2.5 mM dNTP (2.5 mM each of dATP, dTTP, dCTP, and dGTP; Protech, Taiwan), 1 μl of M-MLV reverse transcriptase (Promega, USA), and DEPC-treated water added to a final volume of 20 μl. The cDNA was generated at 42°C.

Semi-quantitative RT-PCR

Coding regions of the genes CCL5, CXCL10, CXCL11, CXCL16, STAT1, STAT2, and GAPDH were amplified using the following primer sets: CCL5-F’ (ccctcgctgtcatcctcattg)/CCL5-R’ (gtgacaaagacgactgctgg), CXCL10-F’ (ggcattcaaggagtacctct)/CXCL10-R’ (attcagacatctcttctcac), CXCL11-F’ (agttgttcaaggcttccccatg)/CXCL11-R’ (gggatttaggcatcgttgtcc), CXCL16-F’ (actcagccaggcaatggcaa)/CXCL16-R’ (tccaggaaaggagctggaac), STAT1-F’ (tgcgcgcagaaaagtttcat)/STAT1-R’ (ggattcaaccaaaggagcag), STAT2-F’ (ccagaactggcaggaagctg)/STAT2-R’ (atgtcccggcagaatttccg), and GAPDH-F’ (accacagtccatgccatcac)/GAPDH-R’ (tccaccaccctgttgctgta), respectively. cDNA (1 μl) was used as the template and was mixed with 0.5 μl of each primer (10 μM), 2 μl of dNTP (Protech, Taiwan), 0.25 μl of GoTaq Flexi DNA polymerase (5 U/μl; Promega, USA), 4 μl of 5× Colorless GoTaq reaction buffer (Promega, USA), 1.2 μl of 25 mM MgCl₂, and sterile deionized water to a final volume of 20 μl. The sample was heated at 95°C for 5 min, followed by 30 cycles, each comprising 95°C for 30 sec, 55°C for 30 sec, and 72°C for 30 sec, and one cycle of 72°C for 10 min in a thermocycler (Multigene, USA).

DNA microarray analysis

where It is one probe intensity of the treatment sample, Ic is one probe intensity of control sample. If a probe P’s R value is bigger than the threshold (usually set as 1), P is regarded as different expression on treatment and control. Furthermore we can calculate the p-value p (differential expression) to test the statistical significance of R, which means the probability of null hypothesis that treatment sample and control sample are not differential expression for one probe. The detailed procedure to estimate p is formally described in[43]. In our implementation, the spots with p <0.05 and R ≥1 were identified as differentially expressed genes for further pathway analysis. The DNA microarray data has been submitted to and approved by the Gene Expression Omnibus Database (GEO) with an accession number of GSE39122.

String

According to the gene expression profiles derived from DNA microarray, 93 genes upregulated more than 2-fold (log₂ ≥1) and with the statistic p less than 0.05 were selected and assembled as a gene expression subdataset (Table1). The functional linkage network was generated using STRING 9.0 database[44]. The gene symbols listed in the gene expression subdataset were input into the frame of “multiple names” in STRING and the organism was set Homo sapiens. After executing the search, STRING would show the possible genes we input and we had to select correct ones. Then, the functional linkage could be generated using the default setting. Within this completely functional network, the evident area of dense interconnections could be identified. In this network, 93 nodes are connected by 676 edges, and in the dense area, there are 618 edges to connect at least one of the 40 nodes. Although only 43.1% (40/93) of nodes are located in the dense area, there are 91.42% (618/676) edges to connect the nodes in dense area with other nodes.

Furthermore, to analyze the relationship of the functional network, and the TGF-α/EGFR pathway which we discovered previously[22], the genes within the evident area, TGF-α and EGFR were input into STRING again. The setting of STRING database is also default. Then, the functional network linked evident area and TGF-α/EGFR pathway could be generated. The figures presented here are redrawn using Cytoscape software which could calculate the numbers of nodes and edges efficiently.

NOA

The NOA (http://app.aporc.org/NOA) database was designed for identifying the enrichment of gene ontology based on biological networks as classified by systems biology[24]. According to microarray data (Table1), the 93 genes of interest were input into and analyzed by the NOA server to determine the relationship of network ontology.

Western blotting

After SDS-PAGE analysis, the proteins were transferred onto a polyvinylidene fluoride (PVDF) membrane (Pall, Pensacola). The membrane was incubated in 5% fetal bovine serum (FBS) in PBST (1X PBS and 0.1% Tween 20) at room temperature for 1 h and then the diluted antibody (anti-STAT1 mAb, 1:2,500 or anti-STAT2 mAb, 1:2,500, Santa Cruz, USA) in 5% FBS/PBST was added to react with the target proteins with shaking at 4°C for 16 h. The membrane was washed with PBST three times for 15 min each. The secondary antibody (anti-rabbit IgG conjugated with HRP, 1:5,000, Jackson, USA) was diluted in 3% Milk/PBST with shaking at 25°C. After 1.5 h, the membrane was washed with PBST for three times and the target protein was visualized by addition of the mixture of reagents A and B (Pierce) to the PVDF membrane, and exposed by the ImageQuant^TM LAS 4000 (GE Healthcare).

Macrophage migration assay

Migration assays were performed in Transwell plates (Corning Costar, USA) with a 6.5-mm diameter and an 8-μm pore size for the membrane. Conditioned medium was prepared before adding RAW 264.7 cells into the top well. The BEAS-2B cells were seeded in 60-mm dishes. After 24 h, both plasmids (pEGFP-C1 and pEGFP-N1-ECPsp) were transfected into BEAS-2B cells (as described above) for 6 h. The culture medium containing the transfection mixture was removed, and fresh RPMI-1640 medium supplemented with 10% FBS was added for a further 18-h incubation at 37°C in a humidified 5% CO₂ incubator. The culture medium was removed, and fresh serum-free RPMI-1640 was added prior to incubation at 37°C for 24 h. Three hundred microliters of the serum-free medium was collected, centrifuged to remove the debris, and then transferred into the bottom wells. At the same time, 1.5 × 10⁵ RAW 264.7 cells were transferred to each top well. After incubation for 8 h at 37°C, cells in the top wells were fixed using 3.7% formaldehyde for 5 min and then stained with 0.05% crystal violet for 30 min, with three further rinses with PBS. The number of cells migrating to the lower membrane surface was counted, and these data were used for further two-tailed student’s t test analysis.