- Email: [email protected]
Gene expression changes during repopulation in a head and neck cancer xenograft
Radiotherapy and Oncology xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Radiotherapy and Oncology journal homepage: www.thegreenjournal.com
Original article
Gene expression changes during repopulation in a head and neck cancer xenograft George D. Wilson a,b,⇑, Bryan J. Thibodeau b, Laura E. Fortier b, Barbara L. Pruetz b, Sandra Galoforo a, Jan Akervall b,c, Brian Marples a, Jiayi Huang a,d a Department of Radiation Oncology; b Beaumont BioBank; c Department of Otolaryngology, William Beaumont Hospital, Royal Oak; and d Department of Radiation Oncology, Washington University School of Medicine, St. Louis, USA
a r t i c l e
i n f o
Article history: Received 23 April 2014 Received in revised form 24 June 2014 Accepted 17 August 2014 Available online xxxx Keywords: Head and neck Gene expression Repopulation Interferon
a b s t r a c t Background/purpose: To investigate temporal changes in global gene expression and pathways involved in the response to irradiation during phases of growth inhibition, recovery and repopulation in a human head and neck squamous cell cancer (HNSCC) xenograft. Methods and materials: Low passage head and neck squamous cancer cells (UT-14-SCC) were injected into the flanks of female nu/nu mice to generate xenografts. After tumors reached a size of 500 mm3, they were treated with either sham RT or 15 Gy in one fraction. At different time points, days 0, 3, and 10 for controls and days 4, 7, 12, and 21 after irradiation, the tumors were harvested for global gene expression analysis and pathway analysis. Results: The tumors showed growth inhibition through days 4–7 and began the transition to regrowth around the day 12 time point. When comparing the pooled controls to each day of treatment, there were 22, 119, 125, and 25 differentially expressed genes on days 4, 7, 12, and 21 respectively using a p 6 0.01 and a 2-fold cut-off. Gene Ontology (GO), gene set enrichment analysis (GSEA) and sub-network enrichment analysis (SNEA) identified different biological processes, cell process pathways and expression targets to be active on each time point after irradiation. An important observation was that the molecular events on day 12 which represented the transition from growth inhibition to regrowth identified interferon and cytokine related genes and signaling pathways as the most prominent. Conclusion: The findings in this study compliment research which has identified components of interferon-related signaling pathways to be involved in radioresistance. Further work will be required to understand the significance of these genes in both radioresistance and treatment response leading to new therapeutic strategies and prognostic tools. Ó 2014 Elsevier Ireland Ltd. All rights reserved. Radiotherapy and Oncology xxx (2014) xxx–xxx
The main reasons for treatment-associated failure in tumors such as lung, head and neck and brain are loco-regional tumor progression and distant metastases. The clinical consequences of local or distant relapse differ between the different tumor types. For advanced inoperable head and neck cancer (HNSCC) treated with radiotherapy or chemoradiation, loco-regional progression is the principal cause of treatment failure and cancer-related death, whereas distant metastases are less common. As a result, strategies to improve local control rates in HNSCC should result in improved overall survival rates [1]. However, current treatment modalities still fall short of delivering major improvements despite altered fractionation scheduling and addition of biologically targeted agents [2]. ⇑ Corresponding author at: Department of Radiation Oncology, William Beaumont Hospital, 3811 W. Thirteen Mile Rd., Royal Oak, MI 48073, USA. E-mail address: [email protected] (G.D. Wilson).
As there is considerable variation between patients in the chance of recurrence and survival even after accounting for stage and tumor volume, the large remaining variation can only be explained by biological factors that differ markedly between tumors. There are several classic radiobiological aspects that have been advocated to determine the response of HNSCC to fractionated radiotherapy, specifically intrinsic radiosensitivity of the tumor cells [3], the magnitude of tumor hypoxia [4], the influence of the microenvironment [5], and the capacity of surviving tumor cells to repopulate during and after treatment [6]. More recently, other emerging characteristics of HNSCC have also entered the discussion of local treatment failure in the form of cancer stem cells (CSC) [7] and immune effects [8]. Clearly, local failures following radiation therapy are multifactorial and no single biological characteristic can predict the outcome of treatment [9]. There have been several studies that have defined a gene signature that correlates with local failure both in
http://dx.doi.org/10.1016/j.radonc.2014.08.022 0167-8140/Ó 2014 Elsevier Ireland Ltd. All rights reserved.
Please cite this article in press as: Wilson GD et al. Gene expression changes during repopulation in a head and neck cancer xenograft. Radiother Oncol (2014), http://dx.doi.org/10.1016/j.radonc.2014.08.022
2
Gene expression changes during repopulation
experimental tumors [10] and in human tumor studies [11,12], but there have been no studies that investigate the temporal changes in global gene expression during treatment, recovery and repopulation of HNSCC. In this study, we designed an experimental protocol of sub-curative radiation treatment which would inflict significant cell death but allow subsequent recovery and repopulation. Global gene expression was studied at key time points during the repopulation response to assess both genes and pathways that were significantly implicated in the response of the tumor to radiation.
ND-8000 spectrophotometer (NanoDrop Technologies, Inc., Wilmington, DE) and quality assessed on a Model 2100 Bioanalyzer (Agilent technologies, Santa Clara, CA). High-quality RNA (i.e., RIN >9.5) was used for the experiments. Differential mRNA expression analysis between cell populations was performed according to the GeneChipÒ Whole Transcript (WT) Sense Target Labeling Assay protocol (Affymetrix Inc., Santa Clara, CA). The labeled fragments were hybridized overnight with the Human Exon 1.0 ST Array and then scanned with a GeneChip Scanner 3000.
Methods and materials
Data analysis
Experimental animal model and radiation treatment (RT) The UT-SCC-14 cell line was obtained from Dr. Reidar Grénman, University of Turku, Finland and has been maintained at low passage number such that it maintains phenotypic and morphological characteristics similar to the primary tumor which was a T3N1M0, moderately differentiated squamous cell carcinoma of the tongue. The experimental protocol was approved by the William Beaumont Hospital Animal Care Committee. Four- to 6-week old female nude NIH III mice were used in these studies. Mice were caged in sterile housing, in groups of five, and were fed a diet of animal chow and water ad libitum. Xenografts were established by harvesting UTSCC-14 cells in mid-log phase growth and injecting them subcutaneously into the flank of the mice, at a concentration of 2 x 106 cells per 100 lL of Matrigel (BD, Franklin Lakes, NJ). Tumor volume was measured by digital calipers and calculated using the standard formula (pab2)/6 where a is the largest and b is the smallest diameter. When the tumor volume reached a volume of 300–400 mm3, animals were randomly assigned to the experiment groups. Tumors were measured three times each week. The endpoint of the experiment was when the tumors grew to a volume of 3000 mm3. Animals were irradiated with a Faxitron Cabinet X-ray System, Model 43855F (Faxitron X-ray, Wheeling, IL) at a dose rate of 0.69 Gy/ min, tube voltage of 160 KVp and current of 4 mA. Animals were immobilized (without anesthetic) in a custom-designed jig that only exposed the hind flank to the radiation beam.
The .CEL files containing the raw intensity data from the Affymetrix GeneChip arrays were imported into Partek Genomics Suite (version 6.6 beta, build 6.11.1115) and normalized using the robust multichip average with a guanine-cytosine content background correction, quantile normalization, log2-transformation, and median polish probeset summarization. Exons were then summarized to genes using the average of the probesets. Differentially expressed genes were identified using a 1-way ANOVA comparing the samples from a given irradiated time point to the controls. Gene set enrichment analysis (GSEA) and pathway analyses were performed using Pathway Studio 10.3 (Ariadne Genomics, Rockville, MD, USA). GSEA identifies highly regulated categories by considering all genes without any pre-filtering based upon pvalue or fold change [14]. The expression microarray data were also analyzed using Ariadne Pathway Studio’s Sub-network Enrichment Analysis (SNEA) tool [15,16]. Pathway Studio utilizes MedScan, the literature mining program that searches publicly available literature such as PubMed for relationships between entities [17]. A sub-network consists of a single seed (i.e., disease or cell process) and genes associated to this seed by regulation of/ by the seed [18]. The expression microarray dataset is interrogated with no prior significance filtering, and enrichment of the sub-network is determined by both the level of regulation in the network and the size of the network. The visualized sub-network was limited to include only those genes that met p 6 0.01 and 1.5-fold change.
Experimental design
Results
Nine xenografts were randomized to receive sham treatment (control group) and twelve were randomized to receive 15 Gy (RT group). Groups of 3 mice from each treatment cohort were sacrificed at different time points after treatment. At each time point the tumor was rapidly excised, snap-frozen and stored at 80 °C. For the control group, tumors were harvested at days 0, 3, and 10 after reaching the starting volume of 400–500 mm3. For the RT group, tumors were harvested at days 4, 7, 12, and 21 after treatment. Isolation of RNA and gene expression Laser capture microdissection was used to isolate cells from the peripheral regions of the tumor based on our previous observation of central necrosis after radiation treatment [13]. Frozen tissue samples were embedded in OCT (Tissue-Tek; Sakura Finetek, USA) and 8 lm sections were cut and mounted onto PEN (polyethylene naphthalate) membrane glass slides (two sections per slide). Regions of periphery were identified on corresponding H&E slides of the tissue sections. The stained slides were microdissected within 2 h of sectioning using an ArcturusXT™ Microdissection System (Molecular Devices) onto CapSureÒ Macro LCM Caps (Molecular Devices). RNA isolation was carried out using RNeasy Plus Micro Kit (Qiagen, Valencia, CA). RNA concentration was determined on a
Tumor characteristics, growth rate and response to radiation In this study the untreated UT-SCC-14 xenografts had a volume doubling time of 4.8 ± 0.7 days (Fig. 1). In a different aspect of this present study, we showed that early radiation necrosis (days 4–12) was characterized by central coagulative necrosis with pyknotic nuclei while late radiation necrosis (day 12 onwards) was characterized additionally by extensive necrosis with fragmentation and dystrophic calcifications [13]. These changes coincided with a period of static tumor growth over the first 12 days (Fig. 1) followed by a transition to regrowth which occurred from day 12 onward. The repopulation of the tumor occurred from the peripheral region and this was the reason why gene expression analysis was restricted to only this region in the control and treated animals by the use of laser capture microdissection. Initial analysis of the radiation time course Comparing all controls to treated samples, there were 507 genes differentially expressed (p 6 0.01); these genes separately clustered the control and treated samples. The gene expression data from the controls was then pooled and each post-RT time point was compared with the pooled data. Using a criteria of p < 0.01 and at least a 1.5-fold change in gene expression, there were 110, 423, 381, and 137 genes significantly altered at day 4,
Please cite this article in press as: Wilson GD et al. Gene expression changes during repopulation in a head and neck cancer xenograft. Radiother Oncol (2014), http://dx.doi.org/10.1016/j.radonc.2014.08.022
G.D. Wilson et al. / Radiotherapy and Oncology xxx (2014) xxx–xxx
3
SNEA: Expression target sub-networks The results of SNEA focusing on expression target sub-networks are presented in Table 3 where the top 15 most highly regulated gene-related sub-networks are listed. Some interesting subnetworks of genes emerge from SNEA analysis. There is a striking upregulation of several sub-networks involved in interferon signaling particularly at day 12 during the transition from growth inhibition into renewed tumor growth. At day 12, STAT2, ISGF3, IF1H1, IRF9, IFNA2 and interferon are all among the top 15 sub-networks of genes with regulated expression. At day 7, the major changes are seen in the E2F transcription factor family and in APC/C (the anaphase promoting complex). Day 7 and day 12 have 2 subnetworks in common with TNFSD10 and TP53. PDGF was highly regulated at the later time points (days, 7, 12 and 21). SNEA: Regulating cell processes Fig. 1. Tumor growth in response to irradiation. The graph shows the growth curves for control (solid line) and tumors irradiated with 15 Gy (dotted lines).
7, 12, and 21 respectively compared to the pooled controls. It is clear that most differences are seen at day 7 and day 12 after irradiation which coincides with the transition from growth inhibition to repopulation. Supplemental Tables 1–4 list the most significant gene changes on each day using a p < 0.01 and 1.5-fold cut-off.
The results of the SNEA analysis focusing on the top 15 most highly regulated cell process sub-networks are shown in Table 4. Again, interestingly, there are several cell processes that are deemed highly regulated at multiple time points. The cell processes of ‘‘cell invasion’’, ‘‘epidermal cell differentiation’’, ‘‘epithelial to mesenchymal transition’’, ‘‘keratinocyte differentiation’’, ‘‘skin barrier’’, and ‘‘viral reproduction’’ are all regulated at days 4, 12, and 21. Interestingly there is no ‘‘3 time point’’ cell process that is regulated on day 7 where the majority of process sub-networks are cell cycle or DNA repair related.
GSEA: Gene Ontology biological processes The results of the GSEA analysis focusing on Gene Ontology biological process are presented in Table 1. The table presents the top 15 most highly regulated biological process at each time point after radiation. Virus–host interaction is regulated at all 4 time points. Apoptotic process and cellular protein metabolic process are regulated at each time point except day 7. Cell cycle and gene expression are regulated at days 7, 12, and 21 but not day 4. As mentioned previously, the most abundant changes in gene expression were seen at day 7 and day 12. This is reflected in the changing pattern of gene expression in Table 2. At day 4 the GO categories include some expected biological processes in response to a damaging dose of radiation such ‘‘negative regulation of cell proliferation’’, ‘‘apoptotic process’’, ‘‘response to oxidative stress’’ and ‘‘antigen processing and presentation of peptide antigen via MHC class I’’. At day 7, the tumors are still experiencing a growth inhibition (Fig. 1), but there is a clear alteration in cell cycle-related gene expression with the GO categories of ‘‘cell cycle’’, ‘‘mitotic cell cycle’’, and ‘‘cell division’’ being the top three categories showing highly significant changes in regulation. By day 12, during the transition into renewed physical growth of the tumors, a different set of biological processes become dominant in the tumor which seem to involve a general increase in various aspects of metabolic activity, gene expression and protein synthesis within the tumors. By day 21, during the regrowth phase, the most regulated biological processes tended to become more varied representing many different cellular activities without an obvious pattern.
GSEA: Cell process pathways The results of the GSEA analysis focusing on cell process pathways are presented in Table 2 which presents the top 15 most highly regulated processes at each time point. The cell process pathway ‘‘co-translational ER protein import’’ was highly ranked at all times. There was also a high representation of histone and chromatin remodeling-related pathways. The histone and chromatin-related cell process pathways were mainly associated with day 4 and day 7.
Discussion The response of tumors to radiation has typically been described in terms of the 4 ‘‘R’’s of radiation biology namely repair, redistribution, reoxygenation and repopulation with a fifth R being attributed to radiosensitivity. While the 4(5) Rs have contributed enormously to our understanding of why fractionated therapy works (or does not work), they shed little light on the molecular mechanisms involved in the tumor response. Recently, the 5 ‘‘R’’s were shown to apply to the larger single doses [19] used in this investigation providing more substantiation for the model used in this study. A significant body of work has described the association of genome-wide microarray analysis with radiosensitivity in cultures of both normal cells [20,21] and tumor cells [21,22]. However, to our knowledge, no studies have investigated the temporal response to irradiation in a solid tumor model designed to interrogate the genes, pathways and processes involved in growth inhibition, transition to regrowth and repopulation. As expected, the cellular response to, and recovery from, radiation damage were complex and multidimensional although patterns did emerge. The early response, during the period of profound growth arrest (day 4), showed a wide array of GO biological processes including epidermis development and keratinization which might be expected in a radiation-damaged squamous cell tumor. Negative regulation of cell proliferation, response to oxidative stress and apoptosis were also represented as significantly regulated processes at this time point. In terms of GSEA, the early highly regulated cellular process pathways were dominated by histone modifications and chromatin remodeling. The changes in histones and chromatin are also expected in response to significant damage, repair and modifications initiated by the extensive damage cause by 15 Gy. In terms of individual genes, Supplemental Table 1 lists the most significantly altered genes on day 4 compared to the controls using a p < 0.01 from ANOVA. Complement component 3 (C3) was the most significantly altered gene with a 6.8-fold increase over controls. C3 is a multifunctional glycoprotein
Please cite this article in press as: Wilson GD et al. Gene expression changes during repopulation in a head and neck cancer xenograft. Radiother Oncol (2014), http://dx.doi.org/10.1016/j.radonc.2014.08.022
4
Gene expression changes during repopulation
Table 1 List of the top 15 most highly regulated categories of biological process (Gene Ontology) as identified by GSEA. Categories in bold with Day 4 vs Controls
⁄
are found at >1 time point.
Day 7 vs Controls
GO biological process
p-Value
GO biological process
p-Value
⁄
1.01E 1.08E 1.24E 2.35E 4.07E 8.57E 1.97E 2.56E 3.10E 6.08E 6.22E 7.04E 4.46E 5.33E 7.74E
⁄
1.60E 1.52E 1.11E 1.48E 1.50E 8.33E 2.99E 4.12E 5.41E 6.80E 9.55E 1.56E 4.24E 5.17E 6.94E
Epidermis development Sterol biosynthetic process ⁄ Keratinization Negative regulation of viral genome replication Cholesterol biosynthetic process Negative regulation of cell proliferation ⁄ Apoptotic process Response to oxidative stress Antigen processing and presentation of peptide antigen via MHC class I Type I interferon-mediated signaling pathway ⁄ Cellular protein metabolic process ⁄ Virus–host interaction Respiratory electron transport chain ⁄ Small molecule metabolic process Oxidation–reduction process
07 07 07 07 07 07 06 06 06 06 06 06 05 05 05
Day 12 vs Controls
Cell cycle Mitotic cell cycle Cell division ⁄ Gene expression DNA replication M phase of mitotic cell cycle Mitosis Chromosome segregation DNA repair Mitotic prometaphase Response to DNA damage stimulus G1-S transition of mitotic cell cycle ⁄ Virus–host interaction ⁄ Small molecule metabolic process Regulation of cell cycle
23 20 15 12 12 12 11 11 10 10 10 09 09 09 09
Day 21 vs Controls
GO biological process
p-Value
GO biological process
p-Value
⁄
1.01E 1.67E 6.33E 1.98E 9.66E 2.37E 4.36E 6.16E 7.20E 8.24E 1.81E 1.93E 3.10E 8.14E 1.90E
⁄
1.23E 9.22E 1.62E 1.70E 2.71E 1.67E 2.27E 2.93E 3.51E 9.05E 9.55E 2.37E 2.41E 4.75E 3.21E
Cellular protein metabolic process ⁄ Gene expression ⁄ Epidermis development ⁄ Virus–host interaction ⁄ Apoptotic process Response to virus ⁄ Translation ⁄ RNA splicing ⁄ RNA metabolic process Endoplasmic reticulum unfolded protein response ⁄ Keratinocyte differentiation Negative regulation of apoptotic process ⁄ Cell cycle Type I interferon-Mediated signaling pathway Defense response to virus
that plays a central role in the activation of the complement system and is involved in multiple biological processes including immune response, innate immunity and inflammatory response [23]. Several other genes linked to inflammatory response were also upregulated on day 4 including ICAM1, ABCA1, AKR1B1, XDH, SOD2, CTNNB1 and CREB1. At day 7 after irradiation the UT-SCC-14 tumors were still profoundly inhibited but there was a major shift toward cell cycle-related GO categories and DNA repair categories; in particular mitosis-associated processes featured prominently. This was also evident in the GSEA cell process pathways where cell cycle, sister chromatid adhesion, kinetochore assembly and chromatin condensation were highlighted as significantly regulated processes in addition to persistent chromatin remodeling and histone modifications. Again, SNEA emphasized cell-cycle related expression targets as the most highly regulated including several members of the E2F family, APC/C and CDC20. However, somewhat surprisingly there were very few cell cycle-related genes that showed significant alteration (Supplemental Table 2). Perhaps, two key genes that are significantly downregulated, PLK1 and CDC45, and one gene that is upregulated, GADD45A, in conjunction with a closer look at the most significantly regulated sub networks (Table 4) where DNA replication initiation and response to DNA damage are highly regulated would suggest there is a continuing inhibition of the cell cycle at day 7. PLK1 expression is directly associated with the progression through each of the four main cell cycle phases where it peaks at the G2?M transition, plateaus throughout mitosis, and has a sharp reduction upon mitotic exit [24]. CDC45 has an essential role in the regulation of the initiation and elongation stages of eukaryotic chromosomal DNA replication and is the main target for a CHK1-regulated DNA damage checkpoint [25]. GADD45A is a p53-
12 12 12 11 11 10 10 10 10 10 09 09 09 09 08
Epidermis development ⁄ Gene expression ⁄ Cell cycle Transforming growth factor beta receptor signaling pathway ⁄ Virus–host interaction ⁄ Translation Keratinization mRNA metabolic process ⁄ Apoptotic process ⁄ RNA splicing Wound healing ⁄ RNA metabolic process ⁄ Cellular protein metabolic process ⁄ Keratinocyte differentiation Transcription from RNA polymerase II promoter
12 12 11 11 10 09 09 09 09 09 09 08 08 08 07
and BRCA1-regulated stress-inducible gene that plays a central role in cellular response to a variety of DNA damage agents including radiation [26]. Other noteworthy changes that began to emerge at day 7 included the upregulation of stem cell–cell related cellular processes and genes including ALDH1A3, C-MET, WNT7A, PLAUR and ABCA1. TIMP2 and TIMP3 showed significant upregulation as did TGFA. Other genes that showed highly significant changes without obvious connection with radiation included RNASE7, SERPINA12, SERPINA3 and TSPAN1. Day 12 represents the transition from growth arrest to regrowth in this model and was characterized by a different set of highly regulated GO categories that included cellular protein metabolic process, gene expression, epidermis development, virus–host interaction and apoptotic process in the top five. GSEA identified the cellular process pathways of co-translational ER protein transport, translation, cell cycle, histone phosphorylation and protein folding to be the most significant. SNEA highlighted the related sub-networks of STAT2, IRF9, ISGF3, interferon and IFNA2. The prominence of the virus-response sub-networks at this important transition in tumor growth after radiation treatment was also supported by a plethora of interferon and cytokine/chemokinerelated genes showing significant upregulation (Supplemental Table 3). These included CXCL11, IL8, IFIT3, IFIT2, CXCL1, IL32, CXCL16, IL1RL1, IL6ST, IL28A, TGFBR2, TNFAIP2, IL31RA1, IRF1, TGFA, IL6 and IL1R2. Further examination of Table 1 also reveals three other GO categories, in addition to virus host interaction, that were highly regulated at day 12 consisting of response to virus, type 1 interferon-mediated signaling and defense response to virus. The viral response mechanisms involving the interferon/STAT1/STAT2/IRF9 axis and the transcription factor ISGF3 are known to provide resistance to DNA damage in addition to their anti-viral response [27].
Please cite this article in press as: Wilson GD et al. Gene expression changes during repopulation in a head and neck cancer xenograft. Radiother Oncol (2014), http://dx.doi.org/10.1016/j.radonc.2014.08.022
5
G.D. Wilson et al. / Radiotherapy and Oncology xxx (2014) xxx–xxx Table 2 List of the top 15 most highly regulated cellular process pathways as identified by GSEA. Categories in bold are found at >1 time point. Day 4 vs Controls
Day 7 vs Controls
Cell process pathways
p-Value
Cell process pathways
p-Value
Co-translational ER Protein Import Histone phosphorylation Histone ubiquitination Cell cycle Histone and DNA methylation TRRAP/TIP60 chromatin remodeling Protein folding NURF chromatin remodeling Secretory pathway: Golgi transport NURD chromatin remodeling Histone sumoylation ER-associated degradation CHRAC chromatin remodeling Translation rRNA transcription and processing
4.72E 7.34E 3.77E 3.99E 7.72E 1.17E 1.22E 2.15E 4.23E 4.47E 7.45E 1.00E 1.37E 1.49E 1.50E
Cell cycle Sister chromatid cohesion Histone and DNA methylation Histone phosphorylation Histone ubiquitination Co-translational ER protein import Kinetochore assembly INO80 chromatin remodeling Chromosome condensation TRRAP/TIP60 chromatin remodeling NURD chromatin remodeling Protein folding NURF chromatin remodeling Histone sumoylation Histone acetylation
1.30E 7.75E 2.13E 3.16E 3.56E 7.66E 9.97E 7.15E 4.53E 5.16E 8.95E 8.22E 1.21E 2.86E 3.31E
08 07 06 06 06 05 05 05 05 05 05 04 04 04 04
Day 12 vs Controls
22 16 13 12 12 12 12 11 10 10 10 09 08 08 08
Day 21 vs Controls
Cell process pathways
p-Value
Cell process pathways
p-Value
Co-translational ER protein import Translation Cell cycle Histone phosphorylation Protein folding ER-associated degradation Histone ubiquitination Apoptosis Desmosome assembly Nuclear envelope Presentation of endogenous peptide antigen TRRAP/TIP60 chromatin remodeling INO80 chromatin remodeling Chromosome condensation Histone acetylation
1.73E 09 2.7E 09 3.79E 08 2.38E 07 4.25E 07 4.36E 07 5.8E 06 7.51E 06 8.82E 06 9.27E 06 1.23E 05 1.91E 05 2.3E 05 2.31E 05 3.43E 05
Co-translational ER protein import Translation Cell cycle Protein folding Desmosome assembly mRNA transcription and processing Histone ubiquitination Histone acetylation Tight junction assembly (occludin) Histone sumoylation NURD chromatin remodeling Nuclear envelope rRNA transcription and processing NURF chromatin remodeling Histone and DNA methylation
5.84E 3.35E 1.03E 1.51E 7.18E 8.16E 1.29E 1.92E 6.45E 1.51E 4.87E 4.97E 5.86E 9.92E 1.07E
13 10 09 08 08 08 07 07 07 06 06 06 06 06 05
Table 3 The top 15 sub-networks of genes with expression regulated by a given entity. Categories in bold are regulated at >1 time point. Day 4
Median change
Day 7
Median change
Day 12
Median change
Day 21
Median change
Interferon MIR29C DDX58 MAP3K5 STAT2 MKNK1 PML IRF9 CYR61 miR-30 STAT1 Tg(KRT5-cre)1Tak BGN KRT10 ITG
(1.01) 1.13 1.28 (1.03) 1.21 1.19 1.12 1.28 1.09 1.01 (1.02) 1.44 1.21 (1.32) (1.01)
E2F APC/C S100A4 E2F2 CDC20 E2F3 Caspase MIR155 ITGAV Fzr1 TNFSF10 E2F1 ERBB2 PEBP1 TP53
(1.06) (1.13) 1.15 (1.07) (1.18) (1.14) 1.03 1.04 1.29 (1.34) (1.02) (1.04) 1.01 1.06 (1.02)
STAT2 DDX58 ISGF3 CYR61 HRAS IFIH1 PDGF IRF9 TP53 TNFSF10 MAP3K5 Interferon IFNA2 BSG S1PR2
1.38 1.67 1.86 1.14 1.03 1.70 1.03 1.67 1.00 1.04 1.03 1.04 1.11 1.10 1.28
PDGF IL17RA PROK1 S1PR2 HRAS LRP1 KLF5 CA12 CTGF TNC PRKCD MAPK6 PRDX6 NOV MIR29C
1.03 1.15 1.15 1.23 1.02 1.09 (1.03) 1.46 1.05 1.11 1.03 (1.56) 1.27 1.09 1.22
Indeed, Weichselbaum and colleagues [28] described an IFNrelated DNA damage resistance signature (IRDS) which was associated with resistance to chemotherapy and/or radiation resistance across a panel of different cancer cell lines. A similar finding was also found in a pair of nasopharyngeal cell lines one of which was induced to become radioresistant [29]; the most significant GO category was type I interferon-mediated signaling. Fig. 2 shows the sub-network of genes at day 12 that are expression targets of interferon emphasizing how these pathways are highly regulated compared to controls. Day 21 had fewer differentially regulated genes and a heterogeneous assortment of highly regulated GO categories; the most significant biological processes were epidermis development, gene
expression, cell cycle, TGFB receptor signaling pathway, virus host interaction and translation. The cellular processes identified by GSEA again included co-translational ER protein transport, translation, cell cycle, protein folding and several associated with histone modifications. SNEA highlighted a disparate group of sub-network of genes including PDGF, IL17RA, PROK1, SIPR2 and HRAS. Similarly, the function of the differentially expressed individual genes (Supplemental Table 4) was equally assorted with C3 and SERPINA3 prominent again as on day 12 as were the stem cell-associated genes HAS2 and ALDH1A3. Interferon signaling was not prominent during the regrowth period. This is essentially a descriptive study utilizing one model of HNSCC so the study has limitations. However, the observation that
Please cite this article in press as: Wilson GD et al. Gene expression changes during repopulation in a head and neck cancer xenograft. Radiother Oncol (2014), http://dx.doi.org/10.1016/j.radonc.2014.08.022
6
Gene expression changes during repopulation
Table 4 The top 15 sub-networks of genes regulating cell processes. Categories in bold are regulated at >1 time point. Day 4
Day 7
Day 12
Day 21
Keratinocyte differentiation Cell invasion Epidermal cell differentiation Response to drug Senescence Epithelial to mesenchymal transition Trophoblast migration Viral reproduction Skin barrier Biomineral formation Cell aging Adherens junction assembly Skin development Sterol biosynthesis Tubulogenesis
Mitosis S phase Response to DNA damage Genome instability Kinetochore assembly G1 phase DNA Damage DNA repair G2/M transition Spindle assembly DNA replication initiation Cytokinesis Chromosome segregation G1/S transition Genome stability
Cell invasion Keratinocyte differentiation Epidermal cell differentiation Epithelial to mesenchymal transition Autophagy Viral reproduction Oncogenesis Cell motility Invasive growth Skin barrier Protein degradation Wound healing Response to viruses Cell aging Cytoskeleton organization and biogenesis
Keratinocyte differentiation Epidermal cell differentiation Cell invasion Cell aging Epithelial to mesenchymal transition Desmosome assembly G1/S transition Oncogenesis Wound healing Skin barrier Keratinocyte proliferation Senescence Autophagy Cytokinesis Viral reproduction
Fig. 2. Sub-network of genes that are expression targets of interferon. Category identified by SNEA as highly regulated at day 12 versus controls (p-value <0.01, 1.5-fold cutoff). Genes in red are upregulated in the irradiated sample; blue are downregulated. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
interferon and cytokine related genes and signaling pathways were prominent during the transition from growth inhibition to regrowth compliment other studies that have implicated these pathways in radioresistance [28]. Further work will be required to understand the significance of the interferon-related signaling pathways in treatment response but they are likely to lead to new therapeutic targets, predictors of response and prognostic tools.
Conflict of interest There is no conflicts of interest for any of the authors. Acknowledgment We thank Dr. Reidar Grénman for generously supplying the UTSCC-14 cell line.
Please cite this article in press as: Wilson GD et al. Gene expression changes during repopulation in a head and neck cancer xenograft. Radiother Oncol (2014), http://dx.doi.org/10.1016/j.radonc.2014.08.022
G.D. Wilson et al. / Radiotherapy and Oncology xxx (2014) xxx–xxx
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.radonc.2014.08. 022. References [1] Bourhis J, Overgaard J, Audry H, et al. Hyperfractionated or accelerated radiotherapy in head and neck cancer: a meta-analysis. Lancet 2006;368:843–54. [2] Numico G, Franco P, Cristofano A, et al. Is the combination of Cetuximab with chemo-radiotherapy regimens worthwhile in the treatment of locally advanced head and neck cancer? A review of current evidence. Crit Rev Oncol Hematol 2013;85:112–20. [3] Bjork-Eriksson T, West C, Karlsson E, Mercke C. Tumor radiosensitivity (SF2) is a prognostic factor for local control in head and neck cancers. Int J Radiat Oncol Biol Phys 2000;46:13–9. [4] Nordsmark M, Bentzen SM, Rudat V, et al. Prognostic value of tumor oxygenation in 397 head and neck tumors after primary radiation therapy. An international multi-center study.. Radiother oncol 2005;77:18–24. [5] Dewhirst MW, Chi JT. Understanding the tumor microenvironment and radioresistance by combining functional imaging with global gene expression. Semin Radiat Oncol 2013;23:296–305. [6] Withers HR, Taylor JM, Maciejewski B. The hazard of accelerated tumor clonogen repopulation during radiotherapy. Acta Oncol 1988;27:131–46. [7] Moncharmont C, Levy A, Gilormini M, et al. Targeting a cornerstone of radiation resistance: cancer stem cell. Cancer Lett 2012;322:139–47. [8] Liang H, Deng L, Chmura S, et al. Radiation-induced equilibrium is a balance between tumor cell proliferation and T cell-mediated killing. J Immunol 2013;190:5874–81. [9] Begg AC. Predicting recurrence after radiotherapy in head and neck cancer. Semin Radiat Oncol 2012;22:108–18. [10] Starmans MH, Zips D, Wouters BG, Baumann M, Lambin P. The use of a comprehensive tumour xenograft dataset to validate gene signatures relevant for radiation response. Radiother Oncol 2009;92:417–22. [11] Akervall J, Nandalur S, Zhang J, et al. A novel panel of biomarkers predicts radioresistance in patients with squamous cell carcinoma of the head and neck. Eur J Cancer 2014;50:570–81. [12] Pramana J, Van den Brekel MW, van Velthuysen ML, et al. Gene expression profiling to predict outcome after chemoradiation in head and neck cancer. Int J Radiat Oncol Biol Phys 2007;69:1544–52.
7
[13] Huang J, Chunta JL, Amin M, et al. Detailed characterization of the early response of head-neck cancer xenografts to irradiation using (18)F-FDG-PET imaging. Int J Radiat Oncol Biol Phys 2012;84:485–91. [14] Subramanian A, Tamayo P, Mootha VK, et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A 2005;102:15545–50. [15] Nikitin A, Egorov S, Daraselia N, Mazo I. Pathway studio – the analysis and navigation of molecular networks. Bioinformatics 2003;19:2155–7. [16] Kotelnikova E, Ivanikova N, Kalinin A, Yuryev A, Daraselia N. Atlas of signaling for interpretation of microarray experiments. PLoS ONE 2010;5:e9256. [17] Novichkova S, Egorov S, Daraselia N. MedScan, a natural language processing engine for MEDLINE abstracts. Bioinformatics 2003;19:1699–706. [18] Yuryev A, Mulyukov Z, Kotelnikova E, et al. Automatic pathway building in biological association networks. BMC Bioinformatics 2006;7:171. [19] Brown JM, Carlson DJ, Brenner DJ. The tumor radiobiology of SRS and SBRT: are more than the 5 Rs involved? Int J Radiat Oncol Biol Phys 2014;88:254–62. [20] Kis E, Szatmari T, Keszei M, et al. Microarray analysis of radiation response genes in primary human fibroblasts. Int J Radiat Oncol Biol Phys 2006;66:1506–14. [21] Ding LH, Shingyoji M, Chen F, et al. Gene expression profiles of normal human fibroblasts after exposure to ionizing radiation: a comparative study of low and high doses. Radiat Res 2005;164:17–26. [22] Jeong J, Hong SJ, Ju YJ, et al. Temporal cDNA microarray analysis of gene expression in human hepatocellular carcinoma upon radiation exposure. Oncol Rep 2006;15:33–48. [23] Barratt-Due A, Pischke SE, Brekke OL, et al. Bride and groom in systemic inflammation – the bells ring for complement and Toll in cooperation. Immunobiology 2012;217:1047–56. [24] Cholewa BD, Liu X, Ahmad N. The role of polo-like kinase 1 in carcinogenesis: cause or consequence? Cancer Res 2013;73:6848–55. [25] Broderick R, Nasheuer HP. Regulation of Cdc45 in the cell cycle and after DNA damage. Biochem Soc Trans 2009;37:926–30. [26] Zhan Q. Gadd45a, a p53- and BRCA1-regulated stress protein, in cellular response to DNA damage. Mutat Res 2005;569:133–43. [27] Cheon H, Holvey-Bates EG, Schoggins JW, et al. IFNbeta-dependent increases in STAT1, STAT2, and IRF9 mediate resistance to viruses and DNA damage. EMBO J 2013;32:2751–63. [28] Weichselbaum RR, Ishwaran H, Yoon T, et al. An interferon-related gene signature for DNA damage resistance is a predictive marker for chemotherapy and radiation for breast cancer. Proc Natl Acad Sci U S A 2008;105:18490–5. [29] Guo Y, Zhu XD, Qu S, et al. Identification of genes involved in radioresistance of nasopharyngeal carcinoma by integrating gene ontology and protein-protein interaction networks. Int J Oncol 2012;40:85–92.
Please cite this article in press as: Wilson GD et al. Gene expression changes during repopulation in a head and neck cancer xenograft. Radiother Oncol (2014), http://dx.doi.org/10.1016/j.radonc.2014.08.022
Contents lists available at ScienceDirect
Radiotherapy and Oncology journal homepage: www.thegreenjournal.com
Original article
Gene expression changes during repopulation in a head and neck cancer xenograft George D. Wilson a,b,⇑, Bryan J. Thibodeau b, Laura E. Fortier b, Barbara L. Pruetz b, Sandra Galoforo a, Jan Akervall b,c, Brian Marples a, Jiayi Huang a,d a Department of Radiation Oncology; b Beaumont BioBank; c Department of Otolaryngology, William Beaumont Hospital, Royal Oak; and d Department of Radiation Oncology, Washington University School of Medicine, St. Louis, USA
a r t i c l e
i n f o
Article history: Received 23 April 2014 Received in revised form 24 June 2014 Accepted 17 August 2014 Available online xxxx Keywords: Head and neck Gene expression Repopulation Interferon
a b s t r a c t Background/purpose: To investigate temporal changes in global gene expression and pathways involved in the response to irradiation during phases of growth inhibition, recovery and repopulation in a human head and neck squamous cell cancer (HNSCC) xenograft. Methods and materials: Low passage head and neck squamous cancer cells (UT-14-SCC) were injected into the flanks of female nu/nu mice to generate xenografts. After tumors reached a size of 500 mm3, they were treated with either sham RT or 15 Gy in one fraction. At different time points, days 0, 3, and 10 for controls and days 4, 7, 12, and 21 after irradiation, the tumors were harvested for global gene expression analysis and pathway analysis. Results: The tumors showed growth inhibition through days 4–7 and began the transition to regrowth around the day 12 time point. When comparing the pooled controls to each day of treatment, there were 22, 119, 125, and 25 differentially expressed genes on days 4, 7, 12, and 21 respectively using a p 6 0.01 and a 2-fold cut-off. Gene Ontology (GO), gene set enrichment analysis (GSEA) and sub-network enrichment analysis (SNEA) identified different biological processes, cell process pathways and expression targets to be active on each time point after irradiation. An important observation was that the molecular events on day 12 which represented the transition from growth inhibition to regrowth identified interferon and cytokine related genes and signaling pathways as the most prominent. Conclusion: The findings in this study compliment research which has identified components of interferon-related signaling pathways to be involved in radioresistance. Further work will be required to understand the significance of these genes in both radioresistance and treatment response leading to new therapeutic strategies and prognostic tools. Ó 2014 Elsevier Ireland Ltd. All rights reserved. Radiotherapy and Oncology xxx (2014) xxx–xxx
The main reasons for treatment-associated failure in tumors such as lung, head and neck and brain are loco-regional tumor progression and distant metastases. The clinical consequences of local or distant relapse differ between the different tumor types. For advanced inoperable head and neck cancer (HNSCC) treated with radiotherapy or chemoradiation, loco-regional progression is the principal cause of treatment failure and cancer-related death, whereas distant metastases are less common. As a result, strategies to improve local control rates in HNSCC should result in improved overall survival rates [1]. However, current treatment modalities still fall short of delivering major improvements despite altered fractionation scheduling and addition of biologically targeted agents [2]. ⇑ Corresponding author at: Department of Radiation Oncology, William Beaumont Hospital, 3811 W. Thirteen Mile Rd., Royal Oak, MI 48073, USA. E-mail address: [email protected] (G.D. Wilson).
As there is considerable variation between patients in the chance of recurrence and survival even after accounting for stage and tumor volume, the large remaining variation can only be explained by biological factors that differ markedly between tumors. There are several classic radiobiological aspects that have been advocated to determine the response of HNSCC to fractionated radiotherapy, specifically intrinsic radiosensitivity of the tumor cells [3], the magnitude of tumor hypoxia [4], the influence of the microenvironment [5], and the capacity of surviving tumor cells to repopulate during and after treatment [6]. More recently, other emerging characteristics of HNSCC have also entered the discussion of local treatment failure in the form of cancer stem cells (CSC) [7] and immune effects [8]. Clearly, local failures following radiation therapy are multifactorial and no single biological characteristic can predict the outcome of treatment [9]. There have been several studies that have defined a gene signature that correlates with local failure both in
http://dx.doi.org/10.1016/j.radonc.2014.08.022 0167-8140/Ó 2014 Elsevier Ireland Ltd. All rights reserved.
Please cite this article in press as: Wilson GD et al. Gene expression changes during repopulation in a head and neck cancer xenograft. Radiother Oncol (2014), http://dx.doi.org/10.1016/j.radonc.2014.08.022
2
Gene expression changes during repopulation
experimental tumors [10] and in human tumor studies [11,12], but there have been no studies that investigate the temporal changes in global gene expression during treatment, recovery and repopulation of HNSCC. In this study, we designed an experimental protocol of sub-curative radiation treatment which would inflict significant cell death but allow subsequent recovery and repopulation. Global gene expression was studied at key time points during the repopulation response to assess both genes and pathways that were significantly implicated in the response of the tumor to radiation.
ND-8000 spectrophotometer (NanoDrop Technologies, Inc., Wilmington, DE) and quality assessed on a Model 2100 Bioanalyzer (Agilent technologies, Santa Clara, CA). High-quality RNA (i.e., RIN >9.5) was used for the experiments. Differential mRNA expression analysis between cell populations was performed according to the GeneChipÒ Whole Transcript (WT) Sense Target Labeling Assay protocol (Affymetrix Inc., Santa Clara, CA). The labeled fragments were hybridized overnight with the Human Exon 1.0 ST Array and then scanned with a GeneChip Scanner 3000.
Methods and materials
Data analysis
Experimental animal model and radiation treatment (RT) The UT-SCC-14 cell line was obtained from Dr. Reidar Grénman, University of Turku, Finland and has been maintained at low passage number such that it maintains phenotypic and morphological characteristics similar to the primary tumor which was a T3N1M0, moderately differentiated squamous cell carcinoma of the tongue. The experimental protocol was approved by the William Beaumont Hospital Animal Care Committee. Four- to 6-week old female nude NIH III mice were used in these studies. Mice were caged in sterile housing, in groups of five, and were fed a diet of animal chow and water ad libitum. Xenografts were established by harvesting UTSCC-14 cells in mid-log phase growth and injecting them subcutaneously into the flank of the mice, at a concentration of 2 x 106 cells per 100 lL of Matrigel (BD, Franklin Lakes, NJ). Tumor volume was measured by digital calipers and calculated using the standard formula (pab2)/6 where a is the largest and b is the smallest diameter. When the tumor volume reached a volume of 300–400 mm3, animals were randomly assigned to the experiment groups. Tumors were measured three times each week. The endpoint of the experiment was when the tumors grew to a volume of 3000 mm3. Animals were irradiated with a Faxitron Cabinet X-ray System, Model 43855F (Faxitron X-ray, Wheeling, IL) at a dose rate of 0.69 Gy/ min, tube voltage of 160 KVp and current of 4 mA. Animals were immobilized (without anesthetic) in a custom-designed jig that only exposed the hind flank to the radiation beam.
The .CEL files containing the raw intensity data from the Affymetrix GeneChip arrays were imported into Partek Genomics Suite (version 6.6 beta, build 6.11.1115) and normalized using the robust multichip average with a guanine-cytosine content background correction, quantile normalization, log2-transformation, and median polish probeset summarization. Exons were then summarized to genes using the average of the probesets. Differentially expressed genes were identified using a 1-way ANOVA comparing the samples from a given irradiated time point to the controls. Gene set enrichment analysis (GSEA) and pathway analyses were performed using Pathway Studio 10.3 (Ariadne Genomics, Rockville, MD, USA). GSEA identifies highly regulated categories by considering all genes without any pre-filtering based upon pvalue or fold change [14]. The expression microarray data were also analyzed using Ariadne Pathway Studio’s Sub-network Enrichment Analysis (SNEA) tool [15,16]. Pathway Studio utilizes MedScan, the literature mining program that searches publicly available literature such as PubMed for relationships between entities [17]. A sub-network consists of a single seed (i.e., disease or cell process) and genes associated to this seed by regulation of/ by the seed [18]. The expression microarray dataset is interrogated with no prior significance filtering, and enrichment of the sub-network is determined by both the level of regulation in the network and the size of the network. The visualized sub-network was limited to include only those genes that met p 6 0.01 and 1.5-fold change.
Experimental design
Results
Nine xenografts were randomized to receive sham treatment (control group) and twelve were randomized to receive 15 Gy (RT group). Groups of 3 mice from each treatment cohort were sacrificed at different time points after treatment. At each time point the tumor was rapidly excised, snap-frozen and stored at 80 °C. For the control group, tumors were harvested at days 0, 3, and 10 after reaching the starting volume of 400–500 mm3. For the RT group, tumors were harvested at days 4, 7, 12, and 21 after treatment. Isolation of RNA and gene expression Laser capture microdissection was used to isolate cells from the peripheral regions of the tumor based on our previous observation of central necrosis after radiation treatment [13]. Frozen tissue samples were embedded in OCT (Tissue-Tek; Sakura Finetek, USA) and 8 lm sections were cut and mounted onto PEN (polyethylene naphthalate) membrane glass slides (two sections per slide). Regions of periphery were identified on corresponding H&E slides of the tissue sections. The stained slides were microdissected within 2 h of sectioning using an ArcturusXT™ Microdissection System (Molecular Devices) onto CapSureÒ Macro LCM Caps (Molecular Devices). RNA isolation was carried out using RNeasy Plus Micro Kit (Qiagen, Valencia, CA). RNA concentration was determined on a
Tumor characteristics, growth rate and response to radiation In this study the untreated UT-SCC-14 xenografts had a volume doubling time of 4.8 ± 0.7 days (Fig. 1). In a different aspect of this present study, we showed that early radiation necrosis (days 4–12) was characterized by central coagulative necrosis with pyknotic nuclei while late radiation necrosis (day 12 onwards) was characterized additionally by extensive necrosis with fragmentation and dystrophic calcifications [13]. These changes coincided with a period of static tumor growth over the first 12 days (Fig. 1) followed by a transition to regrowth which occurred from day 12 onward. The repopulation of the tumor occurred from the peripheral region and this was the reason why gene expression analysis was restricted to only this region in the control and treated animals by the use of laser capture microdissection. Initial analysis of the radiation time course Comparing all controls to treated samples, there were 507 genes differentially expressed (p 6 0.01); these genes separately clustered the control and treated samples. The gene expression data from the controls was then pooled and each post-RT time point was compared with the pooled data. Using a criteria of p < 0.01 and at least a 1.5-fold change in gene expression, there were 110, 423, 381, and 137 genes significantly altered at day 4,
Please cite this article in press as: Wilson GD et al. Gene expression changes during repopulation in a head and neck cancer xenograft. Radiother Oncol (2014), http://dx.doi.org/10.1016/j.radonc.2014.08.022
G.D. Wilson et al. / Radiotherapy and Oncology xxx (2014) xxx–xxx
3
SNEA: Expression target sub-networks The results of SNEA focusing on expression target sub-networks are presented in Table 3 where the top 15 most highly regulated gene-related sub-networks are listed. Some interesting subnetworks of genes emerge from SNEA analysis. There is a striking upregulation of several sub-networks involved in interferon signaling particularly at day 12 during the transition from growth inhibition into renewed tumor growth. At day 12, STAT2, ISGF3, IF1H1, IRF9, IFNA2 and interferon are all among the top 15 sub-networks of genes with regulated expression. At day 7, the major changes are seen in the E2F transcription factor family and in APC/C (the anaphase promoting complex). Day 7 and day 12 have 2 subnetworks in common with TNFSD10 and TP53. PDGF was highly regulated at the later time points (days, 7, 12 and 21). SNEA: Regulating cell processes Fig. 1. Tumor growth in response to irradiation. The graph shows the growth curves for control (solid line) and tumors irradiated with 15 Gy (dotted lines).
7, 12, and 21 respectively compared to the pooled controls. It is clear that most differences are seen at day 7 and day 12 after irradiation which coincides with the transition from growth inhibition to repopulation. Supplemental Tables 1–4 list the most significant gene changes on each day using a p < 0.01 and 1.5-fold cut-off.
The results of the SNEA analysis focusing on the top 15 most highly regulated cell process sub-networks are shown in Table 4. Again, interestingly, there are several cell processes that are deemed highly regulated at multiple time points. The cell processes of ‘‘cell invasion’’, ‘‘epidermal cell differentiation’’, ‘‘epithelial to mesenchymal transition’’, ‘‘keratinocyte differentiation’’, ‘‘skin barrier’’, and ‘‘viral reproduction’’ are all regulated at days 4, 12, and 21. Interestingly there is no ‘‘3 time point’’ cell process that is regulated on day 7 where the majority of process sub-networks are cell cycle or DNA repair related.
GSEA: Gene Ontology biological processes The results of the GSEA analysis focusing on Gene Ontology biological process are presented in Table 1. The table presents the top 15 most highly regulated biological process at each time point after radiation. Virus–host interaction is regulated at all 4 time points. Apoptotic process and cellular protein metabolic process are regulated at each time point except day 7. Cell cycle and gene expression are regulated at days 7, 12, and 21 but not day 4. As mentioned previously, the most abundant changes in gene expression were seen at day 7 and day 12. This is reflected in the changing pattern of gene expression in Table 2. At day 4 the GO categories include some expected biological processes in response to a damaging dose of radiation such ‘‘negative regulation of cell proliferation’’, ‘‘apoptotic process’’, ‘‘response to oxidative stress’’ and ‘‘antigen processing and presentation of peptide antigen via MHC class I’’. At day 7, the tumors are still experiencing a growth inhibition (Fig. 1), but there is a clear alteration in cell cycle-related gene expression with the GO categories of ‘‘cell cycle’’, ‘‘mitotic cell cycle’’, and ‘‘cell division’’ being the top three categories showing highly significant changes in regulation. By day 12, during the transition into renewed physical growth of the tumors, a different set of biological processes become dominant in the tumor which seem to involve a general increase in various aspects of metabolic activity, gene expression and protein synthesis within the tumors. By day 21, during the regrowth phase, the most regulated biological processes tended to become more varied representing many different cellular activities without an obvious pattern.
GSEA: Cell process pathways The results of the GSEA analysis focusing on cell process pathways are presented in Table 2 which presents the top 15 most highly regulated processes at each time point. The cell process pathway ‘‘co-translational ER protein import’’ was highly ranked at all times. There was also a high representation of histone and chromatin remodeling-related pathways. The histone and chromatin-related cell process pathways were mainly associated with day 4 and day 7.
Discussion The response of tumors to radiation has typically been described in terms of the 4 ‘‘R’’s of radiation biology namely repair, redistribution, reoxygenation and repopulation with a fifth R being attributed to radiosensitivity. While the 4(5) Rs have contributed enormously to our understanding of why fractionated therapy works (or does not work), they shed little light on the molecular mechanisms involved in the tumor response. Recently, the 5 ‘‘R’’s were shown to apply to the larger single doses [19] used in this investigation providing more substantiation for the model used in this study. A significant body of work has described the association of genome-wide microarray analysis with radiosensitivity in cultures of both normal cells [20,21] and tumor cells [21,22]. However, to our knowledge, no studies have investigated the temporal response to irradiation in a solid tumor model designed to interrogate the genes, pathways and processes involved in growth inhibition, transition to regrowth and repopulation. As expected, the cellular response to, and recovery from, radiation damage were complex and multidimensional although patterns did emerge. The early response, during the period of profound growth arrest (day 4), showed a wide array of GO biological processes including epidermis development and keratinization which might be expected in a radiation-damaged squamous cell tumor. Negative regulation of cell proliferation, response to oxidative stress and apoptosis were also represented as significantly regulated processes at this time point. In terms of GSEA, the early highly regulated cellular process pathways were dominated by histone modifications and chromatin remodeling. The changes in histones and chromatin are also expected in response to significant damage, repair and modifications initiated by the extensive damage cause by 15 Gy. In terms of individual genes, Supplemental Table 1 lists the most significantly altered genes on day 4 compared to the controls using a p < 0.01 from ANOVA. Complement component 3 (C3) was the most significantly altered gene with a 6.8-fold increase over controls. C3 is a multifunctional glycoprotein
Please cite this article in press as: Wilson GD et al. Gene expression changes during repopulation in a head and neck cancer xenograft. Radiother Oncol (2014), http://dx.doi.org/10.1016/j.radonc.2014.08.022
4
Gene expression changes during repopulation
Table 1 List of the top 15 most highly regulated categories of biological process (Gene Ontology) as identified by GSEA. Categories in bold with Day 4 vs Controls
⁄
are found at >1 time point.
Day 7 vs Controls
GO biological process
p-Value
GO biological process
p-Value
⁄
1.01E 1.08E 1.24E 2.35E 4.07E 8.57E 1.97E 2.56E 3.10E 6.08E 6.22E 7.04E 4.46E 5.33E 7.74E
⁄
1.60E 1.52E 1.11E 1.48E 1.50E 8.33E 2.99E 4.12E 5.41E 6.80E 9.55E 1.56E 4.24E 5.17E 6.94E
Epidermis development Sterol biosynthetic process ⁄ Keratinization Negative regulation of viral genome replication Cholesterol biosynthetic process Negative regulation of cell proliferation ⁄ Apoptotic process Response to oxidative stress Antigen processing and presentation of peptide antigen via MHC class I Type I interferon-mediated signaling pathway ⁄ Cellular protein metabolic process ⁄ Virus–host interaction Respiratory electron transport chain ⁄ Small molecule metabolic process Oxidation–reduction process
07 07 07 07 07 07 06 06 06 06 06 06 05 05 05
Day 12 vs Controls
Cell cycle Mitotic cell cycle Cell division ⁄ Gene expression DNA replication M phase of mitotic cell cycle Mitosis Chromosome segregation DNA repair Mitotic prometaphase Response to DNA damage stimulus G1-S transition of mitotic cell cycle ⁄ Virus–host interaction ⁄ Small molecule metabolic process Regulation of cell cycle
23 20 15 12 12 12 11 11 10 10 10 09 09 09 09
Day 21 vs Controls
GO biological process
p-Value
GO biological process
p-Value
⁄
1.01E 1.67E 6.33E 1.98E 9.66E 2.37E 4.36E 6.16E 7.20E 8.24E 1.81E 1.93E 3.10E 8.14E 1.90E
⁄
1.23E 9.22E 1.62E 1.70E 2.71E 1.67E 2.27E 2.93E 3.51E 9.05E 9.55E 2.37E 2.41E 4.75E 3.21E
Cellular protein metabolic process ⁄ Gene expression ⁄ Epidermis development ⁄ Virus–host interaction ⁄ Apoptotic process Response to virus ⁄ Translation ⁄ RNA splicing ⁄ RNA metabolic process Endoplasmic reticulum unfolded protein response ⁄ Keratinocyte differentiation Negative regulation of apoptotic process ⁄ Cell cycle Type I interferon-Mediated signaling pathway Defense response to virus
that plays a central role in the activation of the complement system and is involved in multiple biological processes including immune response, innate immunity and inflammatory response [23]. Several other genes linked to inflammatory response were also upregulated on day 4 including ICAM1, ABCA1, AKR1B1, XDH, SOD2, CTNNB1 and CREB1. At day 7 after irradiation the UT-SCC-14 tumors were still profoundly inhibited but there was a major shift toward cell cycle-related GO categories and DNA repair categories; in particular mitosis-associated processes featured prominently. This was also evident in the GSEA cell process pathways where cell cycle, sister chromatid adhesion, kinetochore assembly and chromatin condensation were highlighted as significantly regulated processes in addition to persistent chromatin remodeling and histone modifications. Again, SNEA emphasized cell-cycle related expression targets as the most highly regulated including several members of the E2F family, APC/C and CDC20. However, somewhat surprisingly there were very few cell cycle-related genes that showed significant alteration (Supplemental Table 2). Perhaps, two key genes that are significantly downregulated, PLK1 and CDC45, and one gene that is upregulated, GADD45A, in conjunction with a closer look at the most significantly regulated sub networks (Table 4) where DNA replication initiation and response to DNA damage are highly regulated would suggest there is a continuing inhibition of the cell cycle at day 7. PLK1 expression is directly associated with the progression through each of the four main cell cycle phases where it peaks at the G2?M transition, plateaus throughout mitosis, and has a sharp reduction upon mitotic exit [24]. CDC45 has an essential role in the regulation of the initiation and elongation stages of eukaryotic chromosomal DNA replication and is the main target for a CHK1-regulated DNA damage checkpoint [25]. GADD45A is a p53-
12 12 12 11 11 10 10 10 10 10 09 09 09 09 08
Epidermis development ⁄ Gene expression ⁄ Cell cycle Transforming growth factor beta receptor signaling pathway ⁄ Virus–host interaction ⁄ Translation Keratinization mRNA metabolic process ⁄ Apoptotic process ⁄ RNA splicing Wound healing ⁄ RNA metabolic process ⁄ Cellular protein metabolic process ⁄ Keratinocyte differentiation Transcription from RNA polymerase II promoter
12 12 11 11 10 09 09 09 09 09 09 08 08 08 07
and BRCA1-regulated stress-inducible gene that plays a central role in cellular response to a variety of DNA damage agents including radiation [26]. Other noteworthy changes that began to emerge at day 7 included the upregulation of stem cell–cell related cellular processes and genes including ALDH1A3, C-MET, WNT7A, PLAUR and ABCA1. TIMP2 and TIMP3 showed significant upregulation as did TGFA. Other genes that showed highly significant changes without obvious connection with radiation included RNASE7, SERPINA12, SERPINA3 and TSPAN1. Day 12 represents the transition from growth arrest to regrowth in this model and was characterized by a different set of highly regulated GO categories that included cellular protein metabolic process, gene expression, epidermis development, virus–host interaction and apoptotic process in the top five. GSEA identified the cellular process pathways of co-translational ER protein transport, translation, cell cycle, histone phosphorylation and protein folding to be the most significant. SNEA highlighted the related sub-networks of STAT2, IRF9, ISGF3, interferon and IFNA2. The prominence of the virus-response sub-networks at this important transition in tumor growth after radiation treatment was also supported by a plethora of interferon and cytokine/chemokinerelated genes showing significant upregulation (Supplemental Table 3). These included CXCL11, IL8, IFIT3, IFIT2, CXCL1, IL32, CXCL16, IL1RL1, IL6ST, IL28A, TGFBR2, TNFAIP2, IL31RA1, IRF1, TGFA, IL6 and IL1R2. Further examination of Table 1 also reveals three other GO categories, in addition to virus host interaction, that were highly regulated at day 12 consisting of response to virus, type 1 interferon-mediated signaling and defense response to virus. The viral response mechanisms involving the interferon/STAT1/STAT2/IRF9 axis and the transcription factor ISGF3 are known to provide resistance to DNA damage in addition to their anti-viral response [27].
Please cite this article in press as: Wilson GD et al. Gene expression changes during repopulation in a head and neck cancer xenograft. Radiother Oncol (2014), http://dx.doi.org/10.1016/j.radonc.2014.08.022
5
G.D. Wilson et al. / Radiotherapy and Oncology xxx (2014) xxx–xxx Table 2 List of the top 15 most highly regulated cellular process pathways as identified by GSEA. Categories in bold are found at >1 time point. Day 4 vs Controls
Day 7 vs Controls
Cell process pathways
p-Value
Cell process pathways
p-Value
Co-translational ER Protein Import Histone phosphorylation Histone ubiquitination Cell cycle Histone and DNA methylation TRRAP/TIP60 chromatin remodeling Protein folding NURF chromatin remodeling Secretory pathway: Golgi transport NURD chromatin remodeling Histone sumoylation ER-associated degradation CHRAC chromatin remodeling Translation rRNA transcription and processing
4.72E 7.34E 3.77E 3.99E 7.72E 1.17E 1.22E 2.15E 4.23E 4.47E 7.45E 1.00E 1.37E 1.49E 1.50E
Cell cycle Sister chromatid cohesion Histone and DNA methylation Histone phosphorylation Histone ubiquitination Co-translational ER protein import Kinetochore assembly INO80 chromatin remodeling Chromosome condensation TRRAP/TIP60 chromatin remodeling NURD chromatin remodeling Protein folding NURF chromatin remodeling Histone sumoylation Histone acetylation
1.30E 7.75E 2.13E 3.16E 3.56E 7.66E 9.97E 7.15E 4.53E 5.16E 8.95E 8.22E 1.21E 2.86E 3.31E
08 07 06 06 06 05 05 05 05 05 05 04 04 04 04
Day 12 vs Controls
22 16 13 12 12 12 12 11 10 10 10 09 08 08 08
Day 21 vs Controls
Cell process pathways
p-Value
Cell process pathways
p-Value
Co-translational ER protein import Translation Cell cycle Histone phosphorylation Protein folding ER-associated degradation Histone ubiquitination Apoptosis Desmosome assembly Nuclear envelope Presentation of endogenous peptide antigen TRRAP/TIP60 chromatin remodeling INO80 chromatin remodeling Chromosome condensation Histone acetylation
1.73E 09 2.7E 09 3.79E 08 2.38E 07 4.25E 07 4.36E 07 5.8E 06 7.51E 06 8.82E 06 9.27E 06 1.23E 05 1.91E 05 2.3E 05 2.31E 05 3.43E 05
Co-translational ER protein import Translation Cell cycle Protein folding Desmosome assembly mRNA transcription and processing Histone ubiquitination Histone acetylation Tight junction assembly (occludin) Histone sumoylation NURD chromatin remodeling Nuclear envelope rRNA transcription and processing NURF chromatin remodeling Histone and DNA methylation
5.84E 3.35E 1.03E 1.51E 7.18E 8.16E 1.29E 1.92E 6.45E 1.51E 4.87E 4.97E 5.86E 9.92E 1.07E
13 10 09 08 08 08 07 07 07 06 06 06 06 06 05
Table 3 The top 15 sub-networks of genes with expression regulated by a given entity. Categories in bold are regulated at >1 time point. Day 4
Median change
Day 7
Median change
Day 12
Median change
Day 21
Median change
Interferon MIR29C DDX58 MAP3K5 STAT2 MKNK1 PML IRF9 CYR61 miR-30 STAT1 Tg(KRT5-cre)1Tak BGN KRT10 ITG
(1.01) 1.13 1.28 (1.03) 1.21 1.19 1.12 1.28 1.09 1.01 (1.02) 1.44 1.21 (1.32) (1.01)
E2F APC/C S100A4 E2F2 CDC20 E2F3 Caspase MIR155 ITGAV Fzr1 TNFSF10 E2F1 ERBB2 PEBP1 TP53
(1.06) (1.13) 1.15 (1.07) (1.18) (1.14) 1.03 1.04 1.29 (1.34) (1.02) (1.04) 1.01 1.06 (1.02)
STAT2 DDX58 ISGF3 CYR61 HRAS IFIH1 PDGF IRF9 TP53 TNFSF10 MAP3K5 Interferon IFNA2 BSG S1PR2
1.38 1.67 1.86 1.14 1.03 1.70 1.03 1.67 1.00 1.04 1.03 1.04 1.11 1.10 1.28
PDGF IL17RA PROK1 S1PR2 HRAS LRP1 KLF5 CA12 CTGF TNC PRKCD MAPK6 PRDX6 NOV MIR29C
1.03 1.15 1.15 1.23 1.02 1.09 (1.03) 1.46 1.05 1.11 1.03 (1.56) 1.27 1.09 1.22
Indeed, Weichselbaum and colleagues [28] described an IFNrelated DNA damage resistance signature (IRDS) which was associated with resistance to chemotherapy and/or radiation resistance across a panel of different cancer cell lines. A similar finding was also found in a pair of nasopharyngeal cell lines one of which was induced to become radioresistant [29]; the most significant GO category was type I interferon-mediated signaling. Fig. 2 shows the sub-network of genes at day 12 that are expression targets of interferon emphasizing how these pathways are highly regulated compared to controls. Day 21 had fewer differentially regulated genes and a heterogeneous assortment of highly regulated GO categories; the most significant biological processes were epidermis development, gene
expression, cell cycle, TGFB receptor signaling pathway, virus host interaction and translation. The cellular processes identified by GSEA again included co-translational ER protein transport, translation, cell cycle, protein folding and several associated with histone modifications. SNEA highlighted a disparate group of sub-network of genes including PDGF, IL17RA, PROK1, SIPR2 and HRAS. Similarly, the function of the differentially expressed individual genes (Supplemental Table 4) was equally assorted with C3 and SERPINA3 prominent again as on day 12 as were the stem cell-associated genes HAS2 and ALDH1A3. Interferon signaling was not prominent during the regrowth period. This is essentially a descriptive study utilizing one model of HNSCC so the study has limitations. However, the observation that
Please cite this article in press as: Wilson GD et al. Gene expression changes during repopulation in a head and neck cancer xenograft. Radiother Oncol (2014), http://dx.doi.org/10.1016/j.radonc.2014.08.022
6
Gene expression changes during repopulation
Table 4 The top 15 sub-networks of genes regulating cell processes. Categories in bold are regulated at >1 time point. Day 4
Day 7
Day 12
Day 21
Keratinocyte differentiation Cell invasion Epidermal cell differentiation Response to drug Senescence Epithelial to mesenchymal transition Trophoblast migration Viral reproduction Skin barrier Biomineral formation Cell aging Adherens junction assembly Skin development Sterol biosynthesis Tubulogenesis
Mitosis S phase Response to DNA damage Genome instability Kinetochore assembly G1 phase DNA Damage DNA repair G2/M transition Spindle assembly DNA replication initiation Cytokinesis Chromosome segregation G1/S transition Genome stability
Cell invasion Keratinocyte differentiation Epidermal cell differentiation Epithelial to mesenchymal transition Autophagy Viral reproduction Oncogenesis Cell motility Invasive growth Skin barrier Protein degradation Wound healing Response to viruses Cell aging Cytoskeleton organization and biogenesis
Keratinocyte differentiation Epidermal cell differentiation Cell invasion Cell aging Epithelial to mesenchymal transition Desmosome assembly G1/S transition Oncogenesis Wound healing Skin barrier Keratinocyte proliferation Senescence Autophagy Cytokinesis Viral reproduction
Fig. 2. Sub-network of genes that are expression targets of interferon. Category identified by SNEA as highly regulated at day 12 versus controls (p-value <0.01, 1.5-fold cutoff). Genes in red are upregulated in the irradiated sample; blue are downregulated. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
interferon and cytokine related genes and signaling pathways were prominent during the transition from growth inhibition to regrowth compliment other studies that have implicated these pathways in radioresistance [28]. Further work will be required to understand the significance of the interferon-related signaling pathways in treatment response but they are likely to lead to new therapeutic targets, predictors of response and prognostic tools.
Conflict of interest There is no conflicts of interest for any of the authors. Acknowledgment We thank Dr. Reidar Grénman for generously supplying the UTSCC-14 cell line.
Please cite this article in press as: Wilson GD et al. Gene expression changes during repopulation in a head and neck cancer xenograft. Radiother Oncol (2014), http://dx.doi.org/10.1016/j.radonc.2014.08.022
G.D. Wilson et al. / Radiotherapy and Oncology xxx (2014) xxx–xxx
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.radonc.2014.08. 022. References [1] Bourhis J, Overgaard J, Audry H, et al. Hyperfractionated or accelerated radiotherapy in head and neck cancer: a meta-analysis. Lancet 2006;368:843–54. [2] Numico G, Franco P, Cristofano A, et al. Is the combination of Cetuximab with chemo-radiotherapy regimens worthwhile in the treatment of locally advanced head and neck cancer? A review of current evidence. Crit Rev Oncol Hematol 2013;85:112–20. [3] Bjork-Eriksson T, West C, Karlsson E, Mercke C. Tumor radiosensitivity (SF2) is a prognostic factor for local control in head and neck cancers. Int J Radiat Oncol Biol Phys 2000;46:13–9. [4] Nordsmark M, Bentzen SM, Rudat V, et al. Prognostic value of tumor oxygenation in 397 head and neck tumors after primary radiation therapy. An international multi-center study.. Radiother oncol 2005;77:18–24. [5] Dewhirst MW, Chi JT. Understanding the tumor microenvironment and radioresistance by combining functional imaging with global gene expression. Semin Radiat Oncol 2013;23:296–305. [6] Withers HR, Taylor JM, Maciejewski B. The hazard of accelerated tumor clonogen repopulation during radiotherapy. Acta Oncol 1988;27:131–46. [7] Moncharmont C, Levy A, Gilormini M, et al. Targeting a cornerstone of radiation resistance: cancer stem cell. Cancer Lett 2012;322:139–47. [8] Liang H, Deng L, Chmura S, et al. Radiation-induced equilibrium is a balance between tumor cell proliferation and T cell-mediated killing. J Immunol 2013;190:5874–81. [9] Begg AC. Predicting recurrence after radiotherapy in head and neck cancer. Semin Radiat Oncol 2012;22:108–18. [10] Starmans MH, Zips D, Wouters BG, Baumann M, Lambin P. The use of a comprehensive tumour xenograft dataset to validate gene signatures relevant for radiation response. Radiother Oncol 2009;92:417–22. [11] Akervall J, Nandalur S, Zhang J, et al. A novel panel of biomarkers predicts radioresistance in patients with squamous cell carcinoma of the head and neck. Eur J Cancer 2014;50:570–81. [12] Pramana J, Van den Brekel MW, van Velthuysen ML, et al. Gene expression profiling to predict outcome after chemoradiation in head and neck cancer. Int J Radiat Oncol Biol Phys 2007;69:1544–52.
7
[13] Huang J, Chunta JL, Amin M, et al. Detailed characterization of the early response of head-neck cancer xenografts to irradiation using (18)F-FDG-PET imaging. Int J Radiat Oncol Biol Phys 2012;84:485–91. [14] Subramanian A, Tamayo P, Mootha VK, et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A 2005;102:15545–50. [15] Nikitin A, Egorov S, Daraselia N, Mazo I. Pathway studio – the analysis and navigation of molecular networks. Bioinformatics 2003;19:2155–7. [16] Kotelnikova E, Ivanikova N, Kalinin A, Yuryev A, Daraselia N. Atlas of signaling for interpretation of microarray experiments. PLoS ONE 2010;5:e9256. [17] Novichkova S, Egorov S, Daraselia N. MedScan, a natural language processing engine for MEDLINE abstracts. Bioinformatics 2003;19:1699–706. [18] Yuryev A, Mulyukov Z, Kotelnikova E, et al. Automatic pathway building in biological association networks. BMC Bioinformatics 2006;7:171. [19] Brown JM, Carlson DJ, Brenner DJ. The tumor radiobiology of SRS and SBRT: are more than the 5 Rs involved? Int J Radiat Oncol Biol Phys 2014;88:254–62. [20] Kis E, Szatmari T, Keszei M, et al. Microarray analysis of radiation response genes in primary human fibroblasts. Int J Radiat Oncol Biol Phys 2006;66:1506–14. [21] Ding LH, Shingyoji M, Chen F, et al. Gene expression profiles of normal human fibroblasts after exposure to ionizing radiation: a comparative study of low and high doses. Radiat Res 2005;164:17–26. [22] Jeong J, Hong SJ, Ju YJ, et al. Temporal cDNA microarray analysis of gene expression in human hepatocellular carcinoma upon radiation exposure. Oncol Rep 2006;15:33–48. [23] Barratt-Due A, Pischke SE, Brekke OL, et al. Bride and groom in systemic inflammation – the bells ring for complement and Toll in cooperation. Immunobiology 2012;217:1047–56. [24] Cholewa BD, Liu X, Ahmad N. The role of polo-like kinase 1 in carcinogenesis: cause or consequence? Cancer Res 2013;73:6848–55. [25] Broderick R, Nasheuer HP. Regulation of Cdc45 in the cell cycle and after DNA damage. Biochem Soc Trans 2009;37:926–30. [26] Zhan Q. Gadd45a, a p53- and BRCA1-regulated stress protein, in cellular response to DNA damage. Mutat Res 2005;569:133–43. [27] Cheon H, Holvey-Bates EG, Schoggins JW, et al. IFNbeta-dependent increases in STAT1, STAT2, and IRF9 mediate resistance to viruses and DNA damage. EMBO J 2013;32:2751–63. [28] Weichselbaum RR, Ishwaran H, Yoon T, et al. An interferon-related gene signature for DNA damage resistance is a predictive marker for chemotherapy and radiation for breast cancer. Proc Natl Acad Sci U S A 2008;105:18490–5. [29] Guo Y, Zhu XD, Qu S, et al. Identification of genes involved in radioresistance of nasopharyngeal carcinoma by integrating gene ontology and protein-protein interaction networks. Int J Oncol 2012;40:85–92.
Please cite this article in press as: Wilson GD et al. Gene expression changes during repopulation in a head and neck cancer xenograft. Radiother Oncol (2014), http://dx.doi.org/10.1016/j.radonc.2014.08.022