Differential Expression of Stress and Immune Response Pathway Transcripts and miRNAs in Normal Human Endothelial Cells Subjected to Fractionated or Single-Dose Radiation

Radiation oncology remains a mainstay of cancer therapy as both curative and palliative therapy used alone or as a component of combined modality therapy. Routinely, in clinical practice, radiation therapy is administered as multiple fractions of 2 to 2.5 Gy per day for 5 days per week for 1 to 7 weeks to allow repair, repopulation, and recovery of the collateral damage to the normal tissue (1, 2). Individualized radiation therapy with development of modern techniques such as intensity-modulated radiation therapy and image-guided radiation therapy can deliver more controlled single or fewer fractions of high-dose radiation (hypofractionation) to tumor focusing on areas deemed at highest risk (3–5). The newer technology can reduce high doses to normal tissues but can increase the amount of tissue receiving daily dose (6). The incidental radiation exposure of normal tissues is a topic of concern in radiotherapy (7).

Recent work from our laboratory showed that prostate carcinoma cells that survive multifractionated (MF) radiation exposure have a different genomic signature compared with the cells exposed to single-dose radiation (SD; refs. 8–10). Exposure to 10 Gy radiation delivered as fractionated irradiation (1 Gy × 10 or 2 Gy × 5) resulted in more robust differential gene expression changes in PC3 and DU145 cells, whereas in LNCaP cells, 10 Gy radiation delivered as a single dose was more effective (9, 10). These studies also revealed that the mRNA expression profiles following fractionated irradiation were influenced by p53 status. In LNCaP cells, harboring wild-type p53 DNA replication/recombination/repair and cell cycle were the top gene ontology categories affected by radiation, whereas in p53-mutated PC3 cells, genes from IFN, immune, and stress response categories were altered significantly. miRNAs play an important role in regulation of gene expression at the posttranscriptional level by base pairing with the complementary sequences within 3′-untranslated regions of target mRNAs, resulting in translational repression or mRNA degradation (11). As observed for the mRNA expression profiles, in the prostate carcinoma cells, treatment with fractionated irradiation significantly altered more miRNAs as compared with the cells exposed to SD radiation (12).

Although normal tissue exposure remains a major concern in radiation therapy, few studies have investigated the molecular effects of various radiation treatment regimens in normal cells. The purpose of the present study was to investigate global gene and miRNA alterations in normal cells exposed to radiation protocols simulating hypofractionated and conventionally fractionated radiation regimens typically used for radiotherapy in clinic. For this study, we treated normal human coronary artery endothelial cells (HCAEC) with 10 Gy radiation delivered as a SD radiation or as 5 fractions of 2 Gy radiation (MF). The differentially expressed mRNAs and miRNAs were identified by microarray analysis at 6 and 24 hours after an SD and 6 and 24 hours after the final dose of fractionated irradiation. These data showed that in HCAEC more mRNAs and miRNAs were differentially expressed by exposure to MF compared with SD, and the magnitude of changes was higher in MF-irradiated cells. Gene ontology classification showed that in addition to cell cycle, genes regulating DNA replication, DNA damage stimulus, and DNA repair, and genes related to immune response were significantly altered following exposure to MF. Using ingenuity target filter program, we identified miRNAs associated with the target genes from different cellular pathways that were differentially expressed in response to SD and MF. The present study suggests that endothelial cells may play an important role in the outcome of radiotherapy in the clinical settings.

Cryopreserved HCAEC and the media were purchased from Lonza Walkersville Inc. Cells were thawed and maintained in EBM-2 basal medium supplemented with FBS and growth factors (EGM-2 MV BulletKit CC-3202) according to the supplier's instructions. Cells from passages P1 to P3 were used.

Cells were irradiated in a PANTAK high-frequency X-ray generator (Precision X-ray Inc.), operated at 300 kV and 10 mA. The dose rate was 1.6 Gy per minute. Cells were plated into T75 cm² flasks (1–1.5 × 10⁶ for SD radiation and 0.6–0.8 × 10⁶ for fractionated radiation). After 24 hours, cells were exposed to a total of 10 Gy radiation administered either as a SD radiation or as MF radiation of 2 Gy × 5. These nonisoeffective doses were selected to simulate clinical hypofractionated and conventionally fractionated radiotherapy regimens. For the MF protocol, cells were exposed to 2 Gy radiation twice a day, at 6-hour interval. The cells were approximately 90% confluent at the time of harvesting. For both protocols, radiation-induced changes were analyzed at 6 and 24 hours after an SD and 6 and 24 hours after the final dose of fractionated irradiation. Separate controls were maintained for SD and MF radiation protocols.

Cells were pelleted at 6 and 24 hours after an SD and 6 and 24 hours after the final dose of MF irradiation and stored in liquid nitrogen. Total RNA including small RNAs was isolated using phenol/chloroform extraction followed by purification over spin columns (Ambion Cat. No. AM9738). The concentration and purity of total RNA were measured by spectrophotometry at OD260/280, and the quality of the total RNA sample was assessed using an Agilent Bioanalyzer with the RNA6000 Nano Lab Chip (Agilent Technologies).

The mRNA microarray analysis was performed using Agilent Technologies Human Gene Expression 4 × 44 K V2 microarrays (product number G4845A, design ID 026652) designed to target 27,958 Entrez Gene RNAs.

The miRNA microarray analysis was performed using Agilent Technologies Human miRNA 8 × 15 K V2 microarrays (product number G4470B, design ID 019118) with probes for 723 human and 76 human viral miRNAs sourced from Sanger miRBase (release 10.1).

The mRNA and miRNA microarray data were analyzed using Gene Spring Software (Agilent Technologies) as described previously (12). To ensure that mRNAs and miRNAs were reliably measured, ANOVA was used to compare the means of each condition (n = 3). For mRNA analysis, cutoff ratios of gene expression greater than 2.0 and less than 0.5 and a P value of <0.05 relative to the respective control group were selected. For miRNA analysis, cutoff ratios greater than 1.5 and less than 0.66 with a P value of <0.05 relative to the respective control were selected.

The mRNA and miRNA microarray data discussed in this publication have been deposited in NCBI's Gene Expression Omnibus and are accessible through GEO Series Accession No. GSE57059 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE57059), and Accession No. GSE56824 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE56824), respectively.

Separate experiments were set up to extract RNA at 6 and 24 hours after an SD and 6 and 24 hours after the final dose of fractionated irradiation for real-time RT-PCR analysis. RNA was isolated using the RNAeasy mini Kit (Cat. No 74104; Qiagen) as described previously (12). Purified RNA was reverse transcribed to cDNA and RT-PCR was carried out as described previously (13). Alterations in selected differentially expressed genes were confirmed using Taq-Man Custom Express Plate (Part # 4391524; Applied Biosystem) and ABI PRISM 7500 Sequence Detection System instrument equipped with the SDS version 1.4 software. Each plate was designed to contain 18S and PES1 endogenous controls and 22 individual Taqman Gene Expression Assays in quadruplets in specified well locations (see Supplementary Data for assay IDs and expanded method).

The functional significance of differentially expressed mRNAs (2-fold change and P < 0.05) following SD and MF irradiation was evaluated using Ingenuity Pathway Analysis (IPA) software (Ingenuity Systems Version 8.7-3203) as described previously (12, 13). Datasets were uploaded into the IPA, which were next mapped to the functional networks available in the Ingenuity Pathway Knowledge Base and ranked by score as described previously (12).

To identify the target mRNAs associated with differentially expressed miRNAs, datasets of differentially expressed miRNAs (1.5-fold change and P < 0.05) and differentially expressed mRNAs (2-fold change and P < 0.05) were uploaded into an IPA “MicroRNA Target Filter” program. For the data analysis, only the experimentally verified and highly predicted targets from IPA database were selected.

Cells were fixed in 70% ethanol 6 and 24 hours after an SD and 6 and 24 hours after the final dose of MF. Data were collected and analyzed as described previously (13).

Cell extracts were prepared 6 and 24 hours after an SD and 6 and 24 hours after the final dose of MF. Proteins were separated and protein bands were captured by digital CCD camera (Fuji, LAS 3000) as described previously (13). The membranes were stripped and reprobed for actin. Signal intensities were quantified using Image J 1.44p software (NIH), normalized to the loading control actin, and expressed as fold change compared with the unirradiated control.

p53 (Sc-6243), p21 (Sc-756), MDM2 (Sc-5304), RAD51 (Sc-8349) and STAT-1 (Sc-346; Santa Cruz Biotechnology, Inc.), cyclin D2 (3741) and caspase 1 (3866; Cell Signaling Technology), and actin (MAB1501R; Millipore).

Each data point represents average ± SEM of 3 experiments. Differences between the groups were statistically evaluated by two-tailed paired t test. A P value of <0.05 was considered statistically significant.

The surviving fraction (SF) of HCAEC exposed to 2 Gy × 5 MF irradiation was 0.003 ± 0.0002 and for those exposed to 10 Gy SD irradiation was 0.00008 ± 0.00002. This is in keeping with the selected SD regimen having a higher biologically effective dose than the MF regimen.

Of the total 27,958 genes represented in the Agilent microarrays, treatment with 10 Gy SD and 2 Gy × 5 MF resulted in differential expression (>2 fold, P < 0.05) of combined 2,255 genes in HCAEC cells. The Venn diagrams and the heat map in Fig. 1 show that the MF exposure resulted in more robust gene expression changes compared with the SD radiation (Fig. 1). Of the total 2,255 genes altered, 89 genes were differentially expressed in response to SD, 1,873 genes were differentially expressed in response to MF, and 293 genes were commonly differentially expressed following SD and MF treatment (Fig. 1A and B). In cells irradiated with SD, more genes were differentially expressed at 24-hour compared with 6-hour time point (Fig. 1C). Significant gene expression changes were evident at 6 hours in cells irradiated with MF, and the changes persisted up to 24 hours (Fig. 1C).

The 2,255 differentially expressed genes were classified in to functional categories by gene ontology classification (Supplementary Table S1). The enrichment factor of the cell-cycle regulatory genes was the highest. The other significantly altered categories were stress response, DNA replication, response to DNA damage, DNA repair, immune response, apoptosis, p53, and inflammatory response. The number of genes altered and the magnitude of change in these categories were much higher after exposure to MF than SD. Genes from cytokines, inflammatory response, and growth factor activity categories that could influence other cells and tissues were significantly altered only by MF.

Functions associated with top 10 networks (score > 10) of genes differentially expressed by SD and MF are shown in Table 1. At 6 hours following SD, only 4 networks with score more than 10 were generated, with cell cycle and cell death as the top functions. At 24-hour time point after SD irradiation, there were 14 networks with scores more than 10. In addition to cell cycle, the other top functions included DNA replication, recombination, and repair. At 6- and 24-hour time point after the final dose of MF irradiation, there were >20 networks with score >10. The main functional category in cells treated with MF at 6- and 24-hour time point included DNA replication, recombination, and repair. RNA posttranscriptional modification was another significant category observed at both time points following MF (Table 1).

Because cell cycle was the topmost category affected by radiation, cell-cycle distribution in the HCAEC treated with single and fractionated radiation was examined. Figure 2 shows the distribution of cells in G₁, S, and G₂ compartments after treatment with 10 Gy SD (Fig. 2A) and 2 Gy × 5 MF (Fig. 2B). Exposure to SD resulted in reduction in cells in G₁ at 6 hours compared with the untreated cells, which persisted at 24 hours. There was an increase in the percentage of cells in G₂ at 24 hours. Exposure to MF resulted in reduction in the percentage of cells in S phase at 24 hours.

Radiation-induced changes in individual genes from selected functional categories were color coded to demonstrate the expression patterns of individual genes within a category for each radiation treatment regimen. Figure 3A shows heat map of genes from stress response category, which includes DNA damage stimulus and DNA repair gene subsets and other stress response genes. The majority of genes from DNA damage stimulus and DNA repair subsets were downregulated in response to SD and MF. However, the magnitude of downregulation was much higher with MF. Many of the downregulated genes following SD did not pass the cutoff (<2 fold, P > 0.05) because the ratio of fold change, although statistically significant, was less than 2-fold. A complete list of genes from stress response category with fold changes is given in Supplementary Table S2. Some of the downregulated genes involved in homologous recombination from DNA repair category included H2AFX (H2AX), BRCA1, BRCA2, BARD1, RPA1, RAD51, RAD51AP, RAD54B, RAD54L, and BLM. Other downregulated genes from DNA repair category were DNA polymerases POLA1, POLD2, POLD3, POLE2, and POLQ, DNA primase PRIM1, replication factor RFC5, ribonucleotide reductases RRM1, RRM2, and DCK. The downregulated genes related to p53 included RFWD3, GTSE1, BLM, BRCA1, BRCA2, HSPD1, and MTBP; the upregulated genes related to p53 were PML, MDM2, C16orf5, CDKN1A (p21), ATM, TP53INP1, and TP53INP2.

Figure 3B shows heat map of genes from immune response category, which includes inflammatory genes subset. In the immune response category, the majority of genes were upregulated following fractionated irradiation. These included adhesion molecules ICAM1 and VCAM1, chemokines CXCL10, CXCL11, CXCL12, CXCL16, CCL2, CCL5, CCL20, and CCL23, cytokines IFNE, IFNA4, IL1A, IL1B, IL15, TGFB1, and TGFB2, receptors for chemokines CXCR4 and CXCR7 and cytokines FAS, IFN-induced proteins and transcription factors, and molecules in integrin signaling pathways ITGA4, ITGB3, and ITGAV. Genes regulating HLA-A, B, C, F, G, and J MHC class I molecules were upregulated following fractionated irradiation. From HLA class II, HLA DPA1 and DPB1 were downregulated and HLA DQB1 was upregulated. A complete list of genes from immune response category with fold changes is given in Supplementary Table S3.

The miRNA microarray analysis revealed that 123 miRNAs were differentially expressed with high confidence (>1.5 fold, P < 0.05) in the irradiated cells, and the majority of them were upregulated (Fig. 4A–C). Exposure to SD resulted in the differential expression of 17 miRNAs, whereas exposure to MF altered 101 miRNAs. Five miRNAs were commonly expressed after SD and MF irradiation. At 6-hour time point, more miRNAs were differentially expressed in cells exposed to MF compared with the 24-hour time point, and the majority of them were upregulated (Fig. 4C). These included the members of tumor suppressor let-7 family (let-7a, let-7e, and let-7f). Tumor suppressor miR34a was common for SD and MF, and was upregulated at 24 hours after SD, and 6 and 24 hours after MF. The members of the oncomir miR17-92 cluster (miR17, miR18a, miR18b, miR19a, miR19b, miR20a, and miR92a) were all downregulated after MF at 24 hours. The miRNAs associated with cardiovascular functions (miR195, miR21, miR221, miR222, miR27b, miR29b, all upregulated), hypoxia response (miR210, miR424, upregulated), and senescence (downregulated: miR15a, miR20a; upregulated: miR410 and miR431) were also differentially expressed in cells exposed to MF.

IPA miRNA–mRNA target filter program was used to identify the target mRNAs associated with the differentially expressed miRNAs in HCAEC treated with SD and MF radiation. Only the highly predicted or experimentally verified targets were included for the target analysis. The differentially expressed miRNAs and their target mRNAs showed inverse correlation (altered in opposite direction) as well as direct correlation (both altered in same direction).

Table 2A shows the differentially expressed miRNAs after an SD at 6 and 24 hours and the number of mRNA targets of these miRNAs observed in the present data. At 6-hour time point, 3 miRNAs were upregulated and 5 mRNAs were downregulated showing inverse correlation between the miRNAs and mRNAs. At the same time point, 5 upregulated miRNAs showed direct correlation with 8 mRNAs that were also upregulated. At 24-hour time point, 5 differentially expressed miRNAs showed inverse correlation with 13 differentially expressed mRNAs. At the same time point, a total of 8 differentially expressed miRNAs showed direct correlation with 16 differentially expressed mRNAs. Exposure to MF altered more number of miRNAs and mRNAs compared with SD, especially at 6-hour time point (Table 2B). At 6 hours after the final dose of MF, a total of 48 miRNAs were differentially expressed (47 up and 1 down) and showed inverse correlation with 617 target mRNAs (596 down and 21 up). Upregulated 44 miRNAs also showed direct correlation with 997 upregulated mRNAs at this time point. At 24 hours after the final dose of MF treatment, 10 miRNAs were upregulated and 10 were downregulated. They showed inverse and direct correlation with 264 and 270 mRNAs, respectively.

Table 3 shows the differentially expressed mRNAs from ATM, p53 signaling and cell cycle checkpoint pathways, and the inversely regulatory miRNAs associated with these mRNA targets, identified by IPA target filter analysis. Most of the genes from ATM signaling pathway were downregulated in cells treated with MF (Table 3). Activation of H2AFX (H2AX) immediately after DNA double-strand break results in recruitment of specific DNA repair proteins in ATM signaling pathway. H2AX was downregulated at 6 hours after the final dose of MF (MF 6 h), and its regulatory miRNA miR24 was upregulated. Several other genes, including CDC25A, FANCD2, SMC1A, SMC2, BRCA1, CHEK1, CDK1, and CCNB1, were downregulated after MF exposure. The miRNAs showing inverse correlation with these target genes are shown in the table (Table 3). CDC25A was downregulated after both SD and MF, and its regulatory miRNA miR34a was upregulated in SD 24 h, MF 6 h, and MF 24 h. However, at MF 6 h in addition to miR34a several other miRNAs regulating CDC25A were also upregulated. At 6 hours after MF, BRCA1 showed inverse correlation with miR146a as well as miR24, but only with miR146a at 24-hour time point. In addition to BRCA1, miR24 inversely correlated also with E2F2 and CDK1. miR17-92 cluster was downregulated at 24 hours after MF irradiation. At this time point, p53-regulated targets CCND2, CDKN1A (p21), and SERPINE2 were upregulated showing inverse correlation with members of the miR17-92 cluster and miR16. Other upregulated gene in p53 pathway was FAS, and it showed inverse correlation with miR1275. miR23a/b was upregulated and showed inverse correlation with TOPBP1.

The mRNA targets associated with cell-cycle checkpoints and miRNAs inversely correlated with these targets are shown in Table 3. E2F2 was commonly downregulated in all radiation treatment groups and showed inverse correlation with miR326 after SD 6 h, SD 24 h, and MF 6 h. In addition to miR326, several other miRNAs (let-7a/f, miR24, miR222/221, miR365, and miR495) also showed inverse correlation with E2F2 in MF at 6 hours. However, 24 hours after fractionated irradiation these miRNAs were no longer differentially expressed, and E2F2 showed inverse correlation with miR31. CCND2 was upregulated at 6 and 24 hours in cells treated with fractionated radiation. However, at 6-hour time point, all the miRNAs associated with CCND2 were also upregulated (not shown). At 24-hour point, several miRNAs from miR17-92 cluster and miR16 showed inverse correlation with CCND2. CCNE2 was downregulated and showed inverse correlation with miR34. Several other miRNAs (miR30b, miR374a/b, and miR495) also showed inverse correlation with CCNE2 at 6 hour after MF. miR195 inversely correlated with targets CDC25B, CHEK1, and PLK1. Other miRNAs and their inverse targets associated with cell-cycle checkpoints were PKMYT1/miR27b/a, TOP2A/miR410, CKS1B/miR361-5p, and CCNB1/miR379.

Table 3 also demonstrates that at some time points, although target genes were differentially expressed, no regulatory miRNA were identified. For example, PLK1 was downregulated in SD 6, SD 24, MF 6, and MF 24. However, miR195, which showed inverse correlation with PLK1, was differentially expressed only in MF at 6 hours.

Table 4 shows the differentially expressed genes from immune response category that showed inverse correlations with the differentially expressed miRNAs in the present microarray data by target filter analysis. At 24 hours following SD, miR7 showed inverse correlation with RELB. Some of the immune response genes and the inversely correlated miRNAs in cells exposed to fractionated radiation were chemokines CXCL10/miR16, CXCL12/miR19b/a, cytokine TNFSF9/miR16 and cytokine receptors TNFRSF9/miR1275, TNFRSF1B/miR338-3p, and genes associated with integrin signaling ITGB3/miR19b/a, ITGA4/miR20a/miR17-5p, and ITGAV/miR92a. Expression of IFN regulatory transcription factor IRF9 was inversely correlated with miR20. Other prominent upregulated immune response genes and the miRNAs inversely correlating with them were FAS/miR1275, STAT2/miR19b/a, and PTGS2/miR16. IL1RAP, associated with synthesis of acute phase and proinflammatory proteins, was downregulated and showed inverse correlation with upregulated miR31 and miR146a.

Radiation-induced differentially expressed miRNAs associated with cardiovascular diseases and angiogenesis in HCAEC are given in the Supplementary Data. MiRNAs and genes associated with cardiac hypertrophy, hypoxia signaling, atherosclerosis signaling, and β adrenergic signaling in cardiovascular pathway are shown in Supplementary Table S4. Many of the miRNAs differentially expressed in HCAEC treated with fractionated irradiation have been implicated in cardiovascular events and angiogenesis and are shown in Supplementary Table S5.

Selected differentially expressed stress and immune genes from the microarray data were analyzed by real-time RT-PCR. The RT-PCR data substantially confirmed the microarray data (Supplementary Table S6).

The expression of selected differentially expressed genes, P53, CDKN1A, MDM2, RAD51, Cyclin D2, CASP1, and STAT1 at protein level was confirmed by Western blot analysis. P53 and P53-regulated P21, MDM2 proteins were upregulated in response to SD and MF, whereas RAD51 was downregulated only after MF. CYCLIN D2, CASPASE 1, and STAT-1 were upregulated after MF (Supplementary Fig. S1).

Current radiation therapy techniques expose both normal tissue and tumors to a wide range of dose size and fractionation, with a substantial amount of normal tissue potentially being irradiated (6). This study was undertaken to understand the effect of SD and MF radiation on endothelial cells using clinical-relevant schedules to complement our recently reported tumor data (6, 8–10, 12). The radiation protocols for the present study were selected to simulate hypofractionated (10 Gy SD) and conventionally fractionated (2 Gy × 5 MF) regimens typically used for radiotherapy in the clinic. The microarray data showed that exposure to MF resulted in more robust changes in gene and miRNA expressions in terms of number and magnitude, compared with the SD. The MF radiation induced a cohort of mRNAs and miRNAs associated with stress response, immune response, cell cycle, apoptosis, fibrosis, cardiovascular events, and angiogenesis. Using the ingenuity pathway target filter program, we identified the miRNAs that showed inverse and direct correlations with the differentially expressed target genes in response to single and fractionated irradiation.

Cell cycle and DNA replication/DNA damage stimulus/repair were the top gene ontology categories altered in the irradiated cells, and the effect was more pronounced in cells exposed to MF. Ionizing radiation-induced DNA damage results in activation of various DNA repair pathways and cell-cycle checkpoints resulting in a temporary arrest in cell-cycle progression to allow cells to repair damaged DNA. If the damage is too severe, cells are eliminated by apoptosis (14). BRCA1, ATM, and P53 play key roles in DNA damage response (15–17). The two major pathways of DNA damage repair are homologous recombination (HR) and nonhomologous end joining (NHEJ; ref. 18). BRCA1 regulates DNA repair by promoting HR in concert with BRCA2 and RAD51, and also inhibits NHEJ to restrict the extent of deletion at the break site (17). The present gene expression analysis revealed that in addition to BRCA1, both BRCA2 and RAD51 from HR pathway were downregulated in cells exposed to MF. Treatment with single and fractionated radiation resulted in the upregulation of ATM, and several p53-regulated genes including MDM2, which in turn controls p53 activity, and CDKN1A, which regulates cell-cycle checkpoints (19, 20). Cdkn1a, cyclin-dependent kinase (cdk) inhibitor (p21), inhibits cyclin E-cdk2, cyclin D-cdk4, and cyclin A-cdk2 complexes (21). The p53-independent check points following ionizing radiation operating at the G₂–M transition are mediated by the ATM-Chk1-cdc25C-cyclin B/cdc2 pathway (21). In the present study, many cell-cycle regulatory genes including those encoding cyclins CCNB1, CCNE2; kinases CDK1 (CDC2), CDK2; and phosphatases CDC25A, CDC25B, and CDC25C were downregulated. Although p53-regulated DNA repair genes DDB2, GADD45a, DDIT4, and TRIM22 were upregulated, the majority of DNA repair genes including those encoding DNA polymerases, primases, and replication factors were downregulated in the irradiated HCAEC.

MiRNAs play an important role in controlling the regulation of DNA damage response (14). The present study identified several miRNAs associated with target genes from ATM and P53 signaling pathways and cell-cycle checkpoints in the irradiated HCAEC. Although a few microRNAs were differentially regulated in HCAEC by SD such as miR34a, miR136, miR140-3p, miR326, miR338-3p, and miR874, these miRNAs have been implicated to coordinate the induction of cell death by apoptosis under various stresses (22–26). miR874 has been shown to induce G₂–M arrest and cell apoptosis by targeting HDAC1 (25). In agreement with these findings, exposure to SD resulted in a reduction in the percentage of HCAEC in G₁ and an increase in cells in G₂. The surviving fraction of cells exposed to the higher biologically effective dose regimen SD was much lower than the surviving fraction of cells treated with MF. These data suggest that the majority of cells accumulated in G₂ block following SD exposure did not recover and were eliminated. On the contrary, exposure to MF resulted in differential expression of proapoptotic as well as antiapoptotic miRNAs indicating that the final outcome would depend on the cumulative effect of these opposing miRNAs. For instance, although miR17-92 cluster and miR7, which are shown to be inhibitors of apoptosis (27), were reduced in HCAEC by MF, miR15 and miR16a, which induce apoptosis by targeting BCL-2 (28), were downregulated, and miR21, which is implicated in suppression of apoptosis (29), was upregulated suggesting that miRNAs coordinate apoptotic pathway in the irradiated HCAEC. Importantly, the upregulation of FAS by SD and MF may also contribute to apoptosis in the irradiated cells.

The irradiated endothelial normal cells showed significant downregulation of DNA repair genes. Similar results were observed in the irradiated LNCaP prostate cancer cells harboring wild-type p53 (10). This is in contrast to the response of p53-mutated PC3 prostate cancer cells to radiation treatment observed in our earlier study (9). Also, no significant change in DNA repair genes was observed in the irradiated MCF-7, SF539, and DU145 tumor cells, although MCF7 and SF539 cells express wild-type p53 (8). These findings indicate that the expression of DNA repair genes in response to radiation exposure in tumor cells is not strictly dependent on the p53 status. In fact, p53-independent pathways for repair such as P21-PCNA have been reported (30).

Radiation-induced vascular damage is considered to be related to the inflammatory changes in the microvasculature (31). Preclinical in vitro and in vivo studies have demonstrated that ionizing radiation triggers proimmunogenic and inflammatory changes in the tumor cells/tumor microenvironment, making tumors more susceptible to immunotherapy (32–36). The ability of radiation to promote the antitumor immunity has been a subject of great interest, and preclinical studies have reported that the outcome depends on the radiation dose and fractionation protocols used (1, 5). Moreover, in 2 mouse tumor models, the combination of fractionated radiotherapy and anti–CTLA-4 antibody to one tumor site induced systemic tumor control as observed by a complete regression in a second palpable tumor outside the radiation field (abscopal effect; ref. 37). The molecular changes in the irradiated tumor cells that contribute to immunogenic cell death include degradation of proteins, release of “danger signals” calreticulin and high mobility group protein B1 (HMGB1), and ATP which promote priming of antitumor T cells by dendritic cells (33, 36). Radiation-induced upregulation of chemokines enhances immune cell trafficking to attract activated T cells to the irradiated tumor site (35). The cancer cells that survive the radiation insult display enhanced expression of adhesion molecules ICAM-1, death receptor Fas, and MHC-1 antigen-presenting molecules, resulting in an improved recognition and killing by antitumor T cells (38). Interestingly, the microarray analysis revealed that several genes from immune response category were differentially expressed in the irradiated HCAEC. The majority of genes in this category were upregulated and the gene expression was more robust in cells exposed to MF. The immune response genes differentially expressed in the irradiated HCAEC included genes regulating adhesion molecules, chemokines and cytokines, receptors for chemokines, and HLA MHC class I and II antigens. There is increasing evidence that miRNAs function as an effective system to regulate the magnitude of inflammatory responses (39). Accordingly, many miRNAs that activate and dampen the immune response are altered in HCAEC exposed to SD or MF, indicating that this process is very tightly regulated. For instance, at 24 hours after fractionated radiation exposure, the miRNAs from miR17-92 cluster were downregulated and showed inverse correlation with several of the immune response genes upregulated at this time point. miR146 is considered to be a key regulator in innate as well as adaptive immune responses. Although miR146a was upregulated by SD and MF in HCAEC, the target filter analysis revealed inverse correlation between miR146a and only 2 differentially expressed immune response genes. However, CCL5, CXCR4, DDX58, IL1F10, IRAK2, LTB, MR1, and STAT1 showed direct correlation with miR146a. The target filter analysis identified several other miRNAs showing inverse correlations with immune response genes emphasizing the role of miRNAs in immune response and inflammation. These data suggest that the irradiated endothelial cells may contribute to radiation-induced immune response during radiation therapy.

The activation of growth factor, cytokine, and chemokine cascades in response to the radiation-induced vascular injury also contributes to the radiation-induced fibrosis of normal tissue (40). As mentioned above, several genes regulating cytokines and chemokines were upregulated in the irradiated endothelial cells. Among all radiation-induced cytokines, TGFβ activation is of particular relevance, as it elicits strong and long-lasting microenvironmental changes (7, 41). TGFβ plays a central role in fibrosis by stimulating production of new matrix proteins such as fibronectin, collagens, and proteoglycanes (7, 42). The present gene expression analysis showed an increase in TGFB as well as COL1A2 and FBN1 at 6 and 24 hours after MF. The upregulation of COL1A2 inversely correlated with the downregulation of miR7 and miR92a at 24-hour time point. FBN1 showed inverse correlation with miR1275 and miR92a. Previous studies showed that upregulation of collagens and fibrillin 1 in the regions adjacent to infarct during remodeling after myocardial infarction is regulated by downregulation of miR29 (43). Postinfarct cardiac fibrosis on the other hand was inhibited by forced expression of miR101 (44). Interestingly, miR29 blocks fibrosis by inhibiting the expression of extracellular matrix components, whereas miR21 promotes fibrosis in SMCs after vascular injury by stimulating MAPK signaling (45). In the irradiated HCAEC, miR29, miR21, and miR101 were upregulated, whereas miR7 and 92a were downregulated. These observations indicate that induction of fibrosis is also a balance between the actions of these miRNAs, in agreement with the mechanism of action of miRNAs acting as rheostats to fine-tune and modulate the outcome based on the intensity of damage (11).

As seen with the HCAEC, the miRNA-based gene-regulatory system provides a flexible and conditional option that would be particularly useful when mRNA expression must be fine-tuned to different levels in different cell types. The posttranscriptional dampening of gene expression by miRNAs not only offers both a mechanism for more uniform gene expression for cells of a particular type and a simple means to customize this expression level for each distinct cell type, but also offers a mechanism to rapidly respond to stress situations. The cohort of miRNAs influenced by different regimens of radiation indicates this. Although miRNAs expressed in response to SD coordinate the induction of immune response factors and apoptosis, the miRNAs in MF fine-tune several processes such as DNA repair, fibrosis, angiogenesis in addition to immune response and apoptosis.

Our previous studies have shown that the tumor cells that survive MF have substantially different phenotype than the untreated cells or the cells treated with SD and present a unique opportunity to exploit the radiation-induced changes to improve cancer therapy (9, 10), with the underlying theme of radiation as “focused biology” (46). The gene expression profile of endothelial cells in response to fractionated radiation resembles to that seen in p53 wild-type (cell cycle, DNA replication/repair) as well as p53-mutated (immune response) tumor cells observed in our previous studies (9, 10). Several studies have demonstrated a potential role for radiation as an immunologic adjuvant (1, 5, 32, 33, 36, 38). The present study suggests that endothelial cells may contribute to systemic changes during radiotherapy recognizing that modern radiation therapy techniques can be used to target tumors and reduce normal tissue hot spots, but at the same time more normal tissue, and thus endothelial cells, receive some dose.

The importance of the tumor microenvironment is greatly emphasized in cancer therapy. Tumors are complex structures with the stroma and infiltrating cells impacting tumor survival and progression, and the acquired ability for epithelial–mesenchymal transition (47). Radiotherapy significantly alters tumor microenvironment. Certainly, normal tissue effects of radiation depend on changes to parenchyma cells that are organ-specific and also to endothelial cells that are ubiquitous. Recent studies indicate that tumor-derived endothelial cells differ from normal endothelial cells at both functional and molecular levels, and endothelial cells derived from different tumors are shown to be divergent dependent on the origin of the tumor (48). Although these differences remain to be better understood and exploited, dissecting out the contributions of the various tissue components to radiation response is necessary to best understand the aggregate picture. The present data warrant further investigations on radiation response in both normal as well as tumor-derived endothelial cells.

Improving the therapeutic ratio is critical to effective clinical radiotherapy and treatment in general. In our current focus on understanding and targeting the cells that survive MF (6), the changes in endothelial cells would be of strategic importance. Furthermore, it may be possible to use changes induced by the endothelial cells including factors found in the blood to understand how the tumor is responding and use this information for immunotherapy or other molecularly targeted treatments. Although much remains to be done, having normal tissue data facilitate developing better and improved therapeutic approaches.

M.T. Falduto is chief technology officer and has ownership interest in GenUs BioSystems, Inc. S.R. Magnuson is president and has ownership interest in GenUs BioSystems, Inc. No potential conflicts of interest were disclosed by the other authors.

Conception and design: S.T. Palayoor, M. John-Aryankalayil, C.N. Coleman

Development of methodology: S.T. Palayoor, M. John-Aryankalayil, M.T. Falduto, S.R. Magnuson, C.N. Coleman

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. John-Aryankalayil, A.Y. Makinde, M.T. Falduto, S.R. Magnuson

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S.T. Palayoor, M. John-Aryankalayil, M.T. Falduto, C.N. Coleman

Writing, review, and/or revision of the manuscript: S.T. Palayoor, M. John-Aryankalayil, A.Y. Makinde, M.T. Falduto, C.N. Coleman

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S.T. Palayoor, M. John-Aryankalayil, A.Y. Makinde, M.T. Falduto, S.R. Magnuson

Study supervision: S.T. Palayoor, C.N. Coleman

The authors thank Dr. T. Adilakshmi for critical reading of the article and editorial help, Dr. Charles B. Simone II (Department of Radiation Oncology, Hospital of the University of Pennsylvania, Philadelphia, PA) for his expert advice for application of clinical radiotherapy regimens in the laboratory settings.

This work was supported by the Intramural Research Program of the Center for Cancer Research, NCI, NIH.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.