Introduction
Gene expression profiling of
in vitro cellular responses of human fibroblasts and lymphocytes to radiation has demonstrated that cells undergo complex early transcriptional responses of a wide spectrum of genes from different gene ontologies [
1‐
4]. Microarray studies have demonstrated that the transcriptional response of human cells exposed to radiation
in vitro differs between radiation sensitive patients and controls. Therefore this approach has been explored as a predictive test of radiation sensitivity using late normal tissue effects as the endpoint of radiation sensitivity [
5‐
7].
The spectrum of DNA damage caused by bleomycin sulphate is similar but not identical to that caused by ionising radiation, hence its definition as a radiomimetic agent [
8]. The molecular and clinical responses after bleomycin sulphate and radiation are similar: both induce post-mitotic differentiation of fibroblasts inducing a senescent phenotype associated with increased collagen production [
9‐
11], activate cascades of profibrotic chemokines and cytokines and cause skin and pulmonary fibrosis in animal models and in the clinic [
12‐
14]. On this basis, the potential of using bleomycin sulphate rather than radiation for predictive testing is here tested in an exploratory study.
Results
Patient characteristics
Patient characteristics are shown in Table
1. Of the 8 radiation sensitive cases included, 3 cases had grade 3 scores of change in photographic breast appearance at 5 years i.e. marked radiation change. The remainder were scored as cases with moderate change (grade 2) for at least 3 successive years. All control patients had grade 1 scores i.e. no/minimal change in breast appearance.
Table 1
Clinical characteristics of patients incorporated into analysis.
108
| 61 | 50 Gy/25 | 1 | 1 | 3 | 3 | 3 |
112
| 52 | 50 Gy/25 | 1 | 1 | 1 | 9 | 9 |
90
| 62 | 42.9 Gy/13 | 2 | 2 | 2 | 2 | 2 |
75
| 62 | 42.9 Gy/13 | 9 | 1 | 1 | 1 | 1 |
132
| 57 | 50 Gy/25 | 2 | 2 | 2 | 2 | 2 |
158
| 54 | 50 Gy/25 | 1 | 1 | 1 | 1 | 1 |
135
| 56 | 42.9 Gy/13 | 2 | 2 | 2 | 2 | 2 |
144
| 57 | 39 Gy/13 | 1 | 1 | 1 | 1 | 1 |
137
| 50 | 42.9 Gy/13 | 1 | 2 | 2 | 2 | 2 |
148
| 55 | 42.9 Gy/13 | 1 | 1 | 1 | 1 | 1 |
138
| 63 | 42.9 Gy/13 | 2 | 2 | 2 | 3 | 3 |
126
| 63 | 42.9 Gy/13 | 1 | 1 | 1 | 1 | 1 |
115
| 54 | 42.9 Gy/13 | 3 | 3 | 2 | 2 | 2 |
123
| 59 | 42.9 Gy/13 | 1 | 1 | 9 | 9 | 1 |
98
| 62 | 50 Gy/25 | 3 | 3 | 3 | 3 | 3 |
106
| 44 | 50 Gy/25 | 1 | 1 | 1 | 1 | 1 |
Transcriptional response of cultured fibroblasts exposed to bleomycin sulphate
To identify the transcriptional response of cells exposed to bleomycin sulphate, paired SAM of bleomycin sulphate and mock treated samples was carried out using all 16 fibroblast cultures. Of the statistically significant differentially expressed genes (false discovery rate = 0), 973 genes were up-regulated and 923 genes were down-regulated in bleomycin sulphate treated compared to mock treated fibroblasts [see Additional File
3].
For fibroblast response to bleomycin sulphate, the highly enriched gene ontologies for up-regulated genes were cellular localization and cell death (Table
2) and for down-regulated genes included regulation of progression through cell cycle, mitotic phase of cell cycle and DNA damage response and repair (Table
3).
Table 2
Enriched gene ontology terms for genes up-regulated in bleomycin sulphate treated fibroblasts compared to controls.
Establishment of localization | 187 | 1.01E-08 |
Localization | 187 | 1.49E-08 |
Cellular physiological process | 514 | 7.47E-08 |
Secretion | 31 | 7.61E-08 |
Transport | 169 | 3.11E-07 |
Protein transport | 51 | 3.81E-07 |
Establishment of protein localization | 52 | 4.27E-07 |
Protein localization | 52 | 1.06E-06 |
Apoptosis | 49 | 1.96E-06 |
Programmed cell death | 49 | 2.17E-06 |
Cell death | 50 | 2.52E-06 |
Death | 50 | 3.04E-06 |
Secretory pathway | 24 | 4.45E-06 |
Intracellular transport | 50 | 7.70E-06 |
Establishment of cellular localization | 50 | 1.07E-05 |
Cellular localization | 50 | 1.33E-05 |
Regulation of apoptosis | 34 | 1.57E-05 |
Intracellular protein transport | 32 | 1.74E-05 |
Regulation of programmed cell death | 34 | 1.77E-05 |
Cell organization and biogenesis | 98 | 1.81E-05 |
Table 3
Enriched gene ontology terms for genes down-regulated in bleomycin sulphate treated fibroblasts compared to controls.
Cell cycle | 82 | 6.00E-16 |
Mitotic cell cycle | 40 | 4.27E-15 |
Mitosis | 33 | 2.77E-14 |
M phase of mitotic cell cycle | 33 | 4.09E-14 |
M phase | 36 | 4.38E-13 |
Cell division | 34 | 7.07E-13 |
DNA metabolism | 72 | 1.36E-12 |
Biopolymer metabolism | 196 | 2.85E-12 |
Cellular physiological process | 537 | 2.60E-11 |
DNA replication | 33 | 1.52E-10 |
Regulation of progression through cell cycle | 53 | 2.42E-10 |
Regulation of cell cycle | 53 | 2.65E-10 |
Nucleobase, nucleoside, nucleotide and nucleic acid metabolism | 219 | 1.93E-09 |
Spindle organization and biogenesis | 11 | 4.02E-09 |
Response to DNA damage stimulus | 34 | 1.73E-08 |
Response to endogenous stimulus | 35 | 2.45E-08 |
Primary metabolism | 395 | 5.41E-08 |
DNA-dependent DNA replication | 19 | 5.68E-08 |
DNA repair | 31 | 6.80E-08 |
Macromolecule metabolism | 265 | 1.18E-07 |
Differences between radiation sensitive patients and matched controls
SAM was used to try to identify transcriptional differences between fibroblasts isolated from radiation sensitive patients and controls. One case control pair was excluded from analysis as it was incorrectly matched for radiotherapy fractionation.
Comparisons between radiation sensitive cases and matched controls were made for mock treated fibroblast samples (i.e. not exposed to bleomycin sulphate) and for bleomycin sulphate treated fibroblast samples. No significant differentially expressed genes were identified in either comparison (data not shown).
The next approach taken was to calculate fold induction in transcript levels using gene expression ratios of log2 values for bleomycin sulphate treated compared to mock treated samples. Again, no significant differentially expressed genes were identified (data not shown).
Andreassen et al. reported statistically significant associations for 2 single nucleotide polymorphisms (SNPs) in TGF β1 (positions -509 and codon 10) and the risk of developing late normal tissue effects in the same patient population [
18]. Of the 7 case control pairs in the current study, both high risk alleles were present in 5 cases and in 3 controls. The 5 cases with both high risk alleles and 5 matched controls were selected for further analyses. Two out of 5 of these matched controls had both high risk alleles present. SAM analysis was performed on the 5 case control pairs using fold induction values. No significant differentially expressed genes were identified (data not shown). Further analysis using 3 selected cases with marked radiation change and all 7 control patients also did not identify significant differentially expressed genes (data not shown).
Discussion
In the current study, transcriptional profiling of dermal fibroblasts after exposure to bleomycin sulphate was carried out to determine whether differences in transcriptional response could be identified between patients with late normal tissue radiation effects and matched controls. This was a pilot study to determine if bleomycin sulphate could be used as an alternative to radiation, in this context. No differences were detected between the 2 patient groups. There are a number of possible explanations for this negative finding.
In this study, a score for late normal tissue effects was performed using photographic appearance. Cases had moderate/marked change in breast appearance and matched controls had no/minimal change. The case control selection may be a limitation of the current study. Although confounding factors such as breast size were taken into account, conventional planning techniques were used in this population leading to variations in dosimetry between cases and controls. The limitation of 5 years of follow up may have incorrectly classified radiation sensitive cases into the control group in those patients whose late normal tissue injury became manifest later. However, time to development of late normal tissue injury is a relevant parameter for judging radiation sensitivity. Other than patient 112 for whom data was missing after year 3, the available data showed no apparent injury for 4 control patients at 5 years and for 3 control patients at 10 years after radiation (Table
1).
One of the main limitations of the study was the sample size. This and the related issue of inter-sample variation may have contributed to the negative findings. An additional possible source of variation was that cells were not synchronised prior to treatment. The issue of inter-sample variation was further addressed in the microarray analysis by using fold induction values between drug and mock treated samples. The potential superiority of this approach is supported in another study of predictive testing of radiation sensitivity from the Danish cohort of breast cancer patients [
19]. In this study, when cDNA array analysis of basal gene expression was compared between two patient groups, defined as radiation 'sensitive' and radiation 'resistant', only 6 genes were identified as being differentially expressed, suggesting that the difference between untreated fibroblasts from the two groups is likely to be small [
6]. The authors selected 17 differentially expressed candidate genes between the two groups, identified in irradiated fibroblast samples, which were further analysed by quantitative real time polymerase chain reaction (Q-RT-PCR) [
19]. The study reported that using fold induction values better differentiated radiation 'sensitive' and radiation 'resistant' patients than using either untreated samples or radiation exposed samples alone. Fold induction takes into account background levels (i.e. the transcriptional profile of untreated samples) and thereby controls for genetic variation. However in the current study, a difference between radiation sensitive cases and controls was not detected even with this approach.
Bleomycin sulphate stimulates post-mitotic differentiation of fibroblasts inducing a senescent or 'post-mitotic' phenotype associated with increased collagen production characteristic of the terminal differentiation pathway stimulated by ionising radiation [
9,
10,
20]. In the current study, sparse cell cultures were treated for 6 hours with 10 μg/ml bleomycin sulphate on day 1 and analysis was performed at 72 hours to examine the transcriptional response of cells in the post-mitotic state. This dose of bleomycin sulphate has been previously shown to induce post-mitotic differentation in fibroblasts [
9]. Under these conditions, transcriptional changes of genes related to the expression of the differentiated phenotype, considered to be relevant to late normal tissue radiation injury, may be seen. However the fibroblast response to bleomycin sulphate did not confirm enrichment of relevant gene ontology categories. For example, for up-regulated genes, the EASE score for extracellular matrix was 3.6E-02 and for response to oxidative stress was 7.3E-02.
Published data report the dose levels of bleomycin sulphate required to produce cell cycle arrest but not cell cytotoxicity. Bleomycin sulphate is known to cause both G1/S and G2/M arrest [
21]. At high doses, extensive double strand DNA breaks and apoptosis occur [
22]. We aimed to use a dose at which post-mitotic differentiation and cellular response pathways were induced but without causing lethality with predominant death signals. Preliminary experiments carried out in this study using FACS analysis confirmed that the dose applied resulted in predominantly G2/M arrest without significant cell death [see Additional File
2]. Using these experimental conditions, gene expression data did indeed show changes in levels of genes relevant to cell cycle control. Cell death pathway activation was also seen, and this may have contributed to the inability to differentiate radiation sensitive and control groups.
In conclusion, a difference between radiation sensitive cases and matched controls was not detected in this population of breast cancer trial patients who had prospective scoring of late normal tissue effects. This suggests any difference is likely to be small or the variation between patients too great to detect a difference. Limitations of the clinical trial design and the experimental laboratory design could have been contributory.
Acknowledgements
We thank Yvonne Hey and Stuart Pepper at the Cancer Research UK Affymetrix Genechip Microarray Service, Paterson Institute for Cancer Research, for performing the Affymetrix Gene Arrays and for technical support and advice and Henrik Edgren at Institute for Molecular Medicine, University of Helsinki, Finland for help with the Affymetrix gene normalisation. We also thank Carsten Herskind at the Department of Radiation Oncology, University Medical Center Mannheim, University of Heidelberg, Germany for helping with the experimental design and Jorge Reis-Filho and Jo Haviland at the Institute of Cancer Research for useful discussion. This work was supported by Cancer Research UK Section of Radiotherapy (CRUK) grant number C46/A2131 and Breakthrough Breast Cancer, the Institute of Cancer Research, UK. We acknowledge the support of the Cancer Research UK Affymetrix Genechip Microarray Service and NHS funding to the NIHR Biomedical Research Centre.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
CBW participated in the study design and statistical analysis, carried out cell and RNA preparation and drafted the manuscript. KKS participated in study design, performed statistical analysis and helped to draft the manuscript. ALBD particiapted in aspects of the study design and statistical analysis. JRY was responsible for conceiving the case control design and participated in the study design. CMI participated in the study design and study coordination. All authors read and approved the final manuscript.