Background
Medical applications for radionuclides are rapidly developing. The β particle-emitting
131I is frequently included in therapy regimens of various thyroid disorders due to selective uptake of the isotope in thyroid tissue and is also administered bound to tumour-seeking agents for therapeutic and diagnostic purposes [
1‐
4]. The α particle-emitting
211At is a suitable therapeutic radionuclide due to, e.g., a nearly optimal therapeutic linear energy transfer value of emitted α particles of 98.8 keV/μm [
5].
211At-labelled tumour-seeking pharmaceuticals have been utilized both in humans and in animals [
6‐
8]. Similar to unbound
131I, selective uptake of unbound
211At also occurs in thyroid tissue [
9‐
11], and administration of free
131I and
211At or
131I- and
211At-labelled radiopharmaceuticals have been shown to result in thyroid irradiation [
6,
12]. Additionally, nuclear accidents often involve an atmospheric release of
131I, as was the case in connection with the Chernobyl accident, which resulted in an increased incidence of thyroid cancer in children [
13,
14].
Despite the risk of exposing thyroid tissue to 131I and 211At, the understanding of radiation-induced effects is far from complete and molecular biomarkers of absorbed dose or radiation-induced effects on thyroid tissue are yet to be identified. Biomarkers are useful to indicate achieved therapeutic effects or estimate risk exposure and evaluate the quality and severity of side effects. RNA microarray analysis is a semi-quantitative method to identify changes in genome-wide transcriptional patterns between two or more samples. The result is a transcriptional profile, i.e. a snapshot of the radiation-induced cellular activity at the mRNA level. This can be used to determine the impact of radiation on biological functions and canonical pathways, to predict upstream regulation of target molecules, and for biomarker discovery without the risk of bias in focusing on a specific set of signalling pathways only.
Few investigations on global gene expression effects of α and β particle irradiation have been performed in normal (thyroid) tissue in vivo. There are, however, a few in vitro studies on global gene expression in fibroblasts and cancer cells after α particle exposure [
15,
16]. Additionally, effects on gene expression for a set of pre-defined genes after α particle irradiation have been measured both in cancer cells in vitro and in vivo in xenografted tumours [
17‐
19]. Previously, we have published results showing substantial differences between transcriptional profiles in thyroid tissue in vivo after different
131I or
211At exposures, varying absorbed dose, dose rate and time after administration [
20‐
22]. We then identified potential biomarkers for each type of exposure separately and concluded that biological response to radiation is complex and that it is difficult to predict or extrapolate radiation-induced effects for other exposure parameters.
The aim of this work was to re-evaluate the previously obtained transcriptional response in thyroid after administration of 131I− and free 211At in mice from different exposure conditions to gain a better understanding of variations in transcriptional regulation on absorbed dose, dose rate, time after administration and radiation quality.
Discussion
In the present study, the values used to calculate absorbed dose were based on previously published
131I and
211At biodistribution data, where mice were simultaneously injected with both
131I and
211At, which allows for direct comparison of the absorbed dose per injected activity between
131I and
211At in the same animal [
26]. In the present study, radioactivity measurements of individual thyroid samples would have enhanced the certainty in absorbed dose calculations but was not possible since all excised thyroid tissue was needed to ensure sufficient amount of RNA for microarray analysis. There are several important differences in the characteristics of
211At and
131I exposure: i) the difference in mean range of the α and β particles emitted (65 and 400 μm), ii) the much higher mean energy released per decay from
211At compared with
131I (7000 and 190 keV), iii) the difference in LET of particles emitted from
131I and
211At (0.25 and 98.8 keV/μm, respectively) and iv) much shorter half-life for
211At than
131I (7.2 h and 8.0 day, respectively). Taken together,
211At irradiates more heterogeneously and with higher dose rates at similar absorbed dose levels compared with
131I (and the dose rate will decline faster for
211At compared to
131I). The effects of radiation quality on global gene expression should be further studied, and to our knowledge, the present study is the first to investigate such differences between
131I and
211At.
RNA microarray analysis was used to evaluate the impact of 131I and 211At exposure on global transcriptional regulation in normal mouse thyroid tissue. Regulated genes were associated with biological functions using previously published literature reports and various databases, in addition to upstream and downstream regulation analysis and canonical pathway analysis generated by Ingenuity Pathway Analysis (IPA) software. In total, 1144 genes were differentially regulated showing a large variation in number of genes per group. In general, hierarchical clustering divided groups that received high absorbed doses into one branch and groups receiving low absorbed doses into another. Thus, we hypothesize that the transcriptional profiles presented here may reflect intrinsic biological properties predictive of 131I and 211At absorbed dose levels at various time points.
At both 1 and 6 h, the number of regulated genes increased with absorbed dose. This was not the case for
131I or
211At exposure at 24 h, where a broader range of absorbed dose was tested. Furthermore, hierarchical clustering revealed distinct differences between the transcriptional profiles of both similar and different exposures, e.g. the transcriptional profiles for 1.4 Gy at 1 and 6 h after
211At administration were distinctly different although the absorbed dose was similar. This could in part be explained by the profound differences in dose rate; dose rate effects on the transcriptional response have previously been described in vivo following radionuclide administration [
22,
25,
29]. These findings indicate that variations in the radiation-induced response with absorbed dose will be reflected in the number of regulated genes, in addition to which specific genes are regulated, although not in a clear dose-dependent manner. Instead it is likely that changes in transcriptional patterns in a specific tissue will depend with varying degree on, but not excluded to, the following parameters: exposure time, injected activity, absorbed dose, dose rate, dose distribution (e.g. frequency of non-, single- or multi-hit cells) and radiation quality. Each unique setup of these parameters may then yield a specific response in the target tissue. In addition, cells are dynamic systems with complex regulatory networks that activate cascades of downstream regulation that is sensitive to type and frequency of incoming stimulus.
It is valuable to identify genes with exposure-specific expression as they may be used as biomarkers. Biomarkers are useful to better understand the mechanisms behind the radiation-induced response. A potential application of biomarkers for ionizing radiation exposure of the thyroid might be in biological dosimetry after exposure to relatively high doses, maybe in a triage setting. In the present study, kallikrein 1 (
Klk1) and 12 of 13 kallikrein 1-related (
Klk1b) peptidases in the mouse genome were frequently regulated with fold change values between −3.8 and 110. The expression of these genes generally increased with absorbed dose and time after injection of
131I or
211At; however, 32 Gy resulted in less upregulation compared with 11 Gy (24 h,
211At) and the highest dose rate used (1.4 Gy, 1 h,
211At) resulted in downregulation. It is likely that the expression of
Klk1 and
Klk1-related peptidases depends, to a different degree, on dose rate, absorbed dose and time after injection. In a study on the rat urine proteome 24 h after 10 Gy total body irradiation, the occurrence of kallikrein 1-related peptidase b24 precursor protein increased while the kallikrein-binding serine protease inhibitor A3K precursor decreased [
30]. In another study, the plasma kallikrein levels decreased with absorbed dose (0–19 Gy) at 2–24 h after local irradiation of the hind legs in tumour-bearing rats and controls [
31]. Additionally, we have shown that regulation of
Klk1 and
Klk1-related genes in mouse thyroids after
131I exposure does not show a circadian variation [
32]. We hypothesize that genes involved in the kallikrein network may be potential biomarkers of radiation exposure, but further research is warranted to elucidate the relationship between radiation exposure and kallikrein proteases and kallikrein inhibitor levels. The kallikrein genes have also been shown to contribute to the radiation-induced death of various species. After treatment with soy bean trypsin inhibitors (SBTI), the mortality rate in mice and chickens 14 days after exposure to 690 and 820 R (6.7 and 8 Gy to soft tissue), respectively, decreased from 100 to 50 % in mice and from 86 to 4 % in chickens [
33]. The authors suggested that the decrease in mortality rate after administration of SBTI originated from a radioprotective effect on the vascular system with less vascular leakage and that the protease inhibited was likely tissue pre-kallikrein. We suggest that the radioprotective role of SBTI should be further assessed.
Recurrently regulated genes might be potential biomarkers and show how different exposure types influence similar/related genes and biological functions, although maybe with different magnitude and/or direction of regulation. The 27 recurrently regulated genes in the present study were divided into six clusters according to the transcriptional pattern of each individual gene following a specific exposure. In cluster 1,
211At-induced regulation was dependent on both absorbed dose and time after exposure with monotonous change in regulation at 24 h, and 8.5 Gy
131I exposure resulted in very high upregulation. Genes in cluster 1 are related to muscular activity and/or calcium activity (
Atp2a1,
Eno3,
Pvalb,
Tnnc2,
Tnni2 and
Tnnt3). Notably, the thyroid gland contains parafollicular cells (C-cells) that produce the calcium homeostasis regulating hormone calcitonin. Cluster 2 contains genes related to various biological functions, e.g. cellular and tissue development and wound healing (
Ctgf) [
34], energy transduction (
Coq10b) and circadian rhythm (
Dbp,
Per1) [
35,
36]. These genes were generally upregulated and may be indicators of radiation exposure in general. Cluster 3 consisted of only one gene (
Mfsd2) that was up- and downregulated after
211At and
131I exposure, respectively, indicating a difference between radiation qualities (
Ctgf,
Per1,
S100a8,
S100a9, also showed a radiation quality dependency). No clear connection to the immune system, inflammation or the cytokine system was found for genes in clusters 1–3. However, the sole gene in cluster 4 (
Ltf), all genes in cluster 5 (
Ccl8,
Ly6g6d,
S100a8,
S100a9) and a majority of genes in cluster 6 (
Aoc3,
Ccl9,
Clec2d,
Cpa3,
Fstl1,
Scara3) were related to the immune system in various ways, and many related to both inflammation and the cytokine network [
37‐
47]. The cytokine encoded by
Ccl9 is associated with systemic inflammation and has increased expression in macrophages after exposure to triiodothyronine [
43]. Lactoferrin—in mice encoded by the
Ltf gene, solely expressed in cluster 4 and generally downregulated—has been patented as a radioprotective drug and increased survival in mice exposed to 10 Gy (whole-body, external irradiation) via an impact on, e.g., cytokine regulation [
38]. In cluster 5, genes were oppositely regulated when comparing
131I and
211At exposure, indicating a radiation quality-dependent immune response. In cluster 6, genes were downregulated at low absorbed doses and upregulated at high absorbed dose levels even though the shift from down- to upregulation occurred at a lower absorbed dose level for
211At compared with
131I. This suggests that the radiation-induced regulation of genes in cluster 6 is dependent on both radiation quality and absorbed dose. The 27 recurrently regulated genes can potentially be used to discriminate between several different exposure parameters, e.g. radiation quality, absorbed dose levels and time after administration, and might be considered as potential biomarkers for at least
131I and
211At exposure of thyroid. These results indicate a connection between specific exposures and biological responses, especially for the genes in clusters 4–6 that were clearly associated with immunological response, inflammation and the cytokine network. These recurrently regulated genes should be further studied to better understand their impact on radiation-induced biological responses and in particular the local and systemic effects that involve inflammation, the immune system and the cytokine network.
To assess systemic effects from
131I and
211At exposure, regulation of the 27 recurring genes was compared with transcriptional changes in the lungs, spleen, liver and kidney cortex and medulla in the same mice dissected in the present study [
23,
24]. These non-thyroidal tissues, that are exposed at a much lower absorbed dose level compared with thyroid, shared regulation of 19/27 and 6/27 recurring genes after
211At and
131I exposure, respectively. Additionally, we have previously shown that the transcriptional response in the lungs, spleen, liver and kidney cortex and medulla in mice administered
131I and
211At can partly be explained as a systemic response from radiation-induced effects on thyroid [
32]. One gene with potential biomarker properties is
Dbp. The
Dbp gene expression pattern changed in several non-thyroidal tissues after low absorbed dose level exposure to both
131I and
211At [
23,
24], in kidneys in mice both early and late after
177Lu-octreotate administration [
48], and in rat thyroids after
131I administration [
49].
A connection to thyroid cancer was detected for 5/27 recurring genes. PVALB has been suggested as an ideal biomarker to discriminate between benign and malignant thyroid cancer [
50].
ENO3 is another cancer-related gene, associated to the PAX8-PPARG fusion protein in thyroid follicular carcinomas, and upstream regulation of PPARs and PPAR-related pathways was detected in the present study [
51]. The level of CTGF correlated with metastasis, tumour size and clinical stage for papillary thyroid carcinoma in a previous study [
52]. Furthermore, undifferentiated thyroid carcinomas have been shown to be S100A8/9 immunopositive and both genes were associated with, e.g., inflammation-associated cancer and aggressive breast cancer [
41,
53,
54].
According to the Ingenuity canonical pathway and diseases and functions analysis tool, the transcriptional response after both 131I and 211At exposure was related to thyroid cancer signalling and various thyroid cancers, respectively. It is uncertain whether induction of thyroid cancer can be detected at the transcriptional level at these early times after initiation of radiation exposure. However, according to IPA, exposure to ionizing radiation activates thyroid cancer signalling by rearrangements of RET and/or NTRK, both present in some of the exposed groups. Unfortunately, all parts of the thyroid samples from mice in the three studies this work is based on were used for microarray analysis, why further studies of genomic rearrangements were not possible from the same samples.
Additionally, KLK3 (human denotation of KLK1 in mouse) is among the involved molecules in all groups that show an impact on thyroid cancer signalling. Since IPA uses human protein nomenclature for annotation of genes, the presence of KLK3 in the IPA analysis is likely a result of regulation of mouse Klk1 and Klk1-related genes in the present study.
The number of annotated genes involved in calcium signalling generally increased with absorbed dose for
211At exposure. Several of the genes associated with these molecules could be found in cluster 1 among the 27 recurrently regulated genes (
Tnni2,
Tnnc2,
Tnnt3 and
Atp2a1). Additionally, gene products of several other recurrently regulated genes are also calcium-related according to literature reports, but not associated with calcium signalling in the IPA canonical pathway analysis. For example, the gene products of
Clec2d and
Ogn both inhibit osteoclasts that can release Ca
2+ into the blood [
55,
56]. As previously mentioned, the thyroid gland also contains parafollicular cells that produce calcitonin, a hormone partly responsible for calcium homeostasis and an inhibitor of osteoclast activity. However, the impact on calcitonin levels from
131I and
211At exposure was not investigated in the present study. A relationship between calcium and radiation-induced response has been previously reported and it was shown that calcium was required for bystander-induced apoptosis in unirradiated keratinocytes [
57].
An impact on integrin-linked kinase (ILK) signalling was identified using an IPA canonical pathway analysis, and some genes associated with calcium signalling were also associated with ILK signalling in the present study. For
211At, a higher absorbed dose generally resulted in a response involving a higher number of annotated genes. For
131I exposure, ILK signalling was only statistically significant after 8.5 Gy, suggesting a difference in response due to radiation quality. In blood from mice administered with
137Cs, genes associated with integrin-signalling were found upregulated at days 2 and 3 and downregulated at days 20 and 30 (transcriptional level) [
58]. No clear temporal effect on integrin-signalling was seen during the somewhat shorter time range used in the present study. Interestingly, however, in the present study, we demonstrate that genes, e.g., associated with ILK signalling were generally upregulated at higher absorbed dose levels and downregulated at lower absorbed dose levels, suggesting different involvement of this signalling pathway at different absorbed dose levels. This was especially the case 24 h after
211At administration, but a similar trend was also seen after
131I administration. ILK signalling and radiation damage have been previously connected, and furthermore, ILK signalling partly controls cell adhesion and mediates prosurvival and antiapoptotic signalling after exposure to ionizing radiation [
59]. Additionally, several cell adhesion GO terms were identified when performing in-depth separate analysis of the transcriptional response of thyroid tissue from animals in the three experiments that constitute the present study [
20‐
22].
In the present study, the predicted upstream regulation of several peroxisomal proliferator-activated receptors (PPARs), and PPAR ligands and agonists was found. The PPARs are of interest in the radiation-induced biological response. In one study, administration of the PPARα agonist fenofibrate prevented some cognitive function impairment in young rats exposed to 40 Gy fractionated whole-brain irradiation [
60]. In another study, knockout of PPARα resulted in inhibition of radiation-induced apoptosis in the mouse kidney through regulation of
Nfkb and anti-apoptosis factors [
61]. Together, these results indicate that the PPAR network may play a role in radiation-induced biological response and that it may be targeted to modulate radiation damage in various tissues.