Molecular and cellular radiobiological effects of Auger emitting radionuclides

Abstract

Although the general radiobiologic principles underlying external beam therapy and radionuclide therapy are similar, significant differences in the biophysical and radiobiologic effects from the two types of radiation continue to accumulate. Here, I will address the unique features that distinguish the molecular and cellular radiobiological effects of Auger electron-emitting radionuclides consequent to (1) the physical characteristics of the decaying atom and its subcellular localisation, (2) DNA topology and (3) the bystander effect. Based on these experimental findings, I postulate that the ability of track structural simulations as primary tools in modelling DNA damage and cellular survival at the molecular level would be greatly enhanced when these contributions are factored in.

INTRODUCTION

In 1925, a young physicist named Pierre Victor Auger reported that when a cloud chamber was irradiated with low-energy X-ray photons, multiple electron tracks were observed⁽¹⁾. The author went on to correctly conclude that this event is a consequence of the ejection of inner-shell electrons from the irradiated atoms, the creation of primary electron vacancies within the shells of these atoms, a complex series of vacancy cascades, and the ejection of very low-energy electrons. It was later recognised that such low-energy electrons are also ejected by many radionuclides decaying by electron capture (EC) and/or internal conversion (IC), two processes that were shown to introduce primary vacancies in the inner electronic shells of the daughter atoms, which once filled by electrons from higher shells move towards the outermost shell. These transitions are accompanied by the emission of either characteristic atomic X-ray photons or low-energy monoenergetic electrons (collectively known as Auger electrons). Typically, an atom undergoing EC and/or IC emits multiple (e.g. 5 to ∼50) electrons whose range in water is somewhere below a nanometer and up to a few micrometres (Table 1). It is important to note that these low-energy, light, negatively charged particles travel in contorted paths. However, unlike energetic electron emitters (e.g. ¹³¹I, ⁹⁰Y), whose linear energy transfer (LET) is low (∼0.2 keV μm⁻¹) along most of their rather long linear path (millimetre-to-centimetre in tissue), a consequence of the ionisations occurring sparingly, the LET of Auger electrons is 10–100-fold higher (∼4 to ∼25 keV μm^–1), especially at very low energies (Figure 1) with the ionisations clustered within several cubic nanometres around the point of decay⁽²⁾. The ejection of these orbital electrons also leaves the decaying atoms transiently with a high positive charge and leads to the deposition of highly localised energy around the decay site^(3–5).

Table 1.

Auger electron emitters: physical properties.

Radionuclide	Electrons yield per decaying atom	Number of electrons with
		<0.5 µm range (%)	<2 nm range (%)
125I	20	20 (98)	8 (39)
123I	11	11 (98)	5 (40)
77Br	7	6 (95)	3 (51)
111In	15	15 (98)	8 (53)
193mPt	30	29 (97)	6 (21)
195mPt	36	33 (92)	7 (19)

Figure 1.

LET of electrons as function of their energy. Arrows indicate the range of electrons at each specified energy.

RADIOBIOLOGICAL EFFECTS OF AUGER ELECTRONS

Throughout the first half of the twentieth century, the scientific community showed little interest in pursuing the radiobiological effects of low-energy electrons. In the early 1960s, Carlson and White⁽⁶⁾ demonstrated that the decay of the Auger electron emitter iodine-125 (¹²⁵I)—covalently bound to an ethyl or a methyl group—leads to the fragmentation of these molecules. By the late 1960s and thereafter, various groups began to report on the radiobiological effects of this and other Auger electron emitters in prokaryotic and eukaryotic organisms. In these early studies, a direct correlation was demonstrated between double strand breaks (DSB) and phage viability when these were exposed to the decay of DNA-incorporated ¹²⁵I⁽⁷⁾. It was also shown that a single DSB was formed in bacterial DNA per decaying atom⁽⁸⁾. Despite the fact that these studies were carried out in prokaryotic cells, they soon led to the general belief that the same is true in mammalian cells, i.e. the decay of DNA-incorporated ¹²⁵I in mammalian cells produces 1 DSB/decaying atom.

DSB yield in naked DNA and chromatin

As mentioned above, internal emitters that undergo radioactive decay by EC and/or IC result in the emission of a surge of low-energy (<1 keV) Auger electrons. Since many of these electrons traverse a very short distance (few nanometres), the density of the hydroxyl radicals (^•OH) generated will be very high around the decaying atom and will decrease drastically as a function of length of their tortuous path. Accordingly, when the radionuclide atoms are uniformly distributed in medium, their decay will result in the formation of randomly dispersed ‘hot spots’ (volumes densely traversed by electrons and occupied by ^•OH and other radicals) and ‘cold spots’ (volumes sparsely traversed by electrons and deficient in ^•OH and other radicals). Dosimetric calculations have also supported these expectations. For example, the decay of ¹²⁵I has been shown to lead to the deposition of a very high dose (∼10⁹ cGy/decaying atom) in the immediate vicinity (∼2 nm³) of the decay site and that there is a sharp and significant drop in the energy deposited (from ∼10⁹ to ∼10⁶ cGy) as a function of increasing distance (few nanometres) from the decaying ¹²⁵I atom^{(2–5, 9–13)}. Consistent with these dosimetric expectations, early studies generally demonstrated that when ¹²⁵I atoms are positioned in close proximity to the naked DNA molecules (e.g. synthetic oligonucleotides, bacterial DNA, supercoiled plasmid DNA), either by the incorporation of an ¹²⁵I-labelled nucleoside analogue into DNA or the binding of an ¹²⁵I-labelled DNA groove binder or intercalator, its decay leads to the efficient (∼0.5–1 DSB/¹²⁵I decay) induction of DSB that are mainly caused by direct ionisations^(14–20). These findings were surprising since the ^•OH-radical scavenger DMSO had been shown to substantially (∼70 %) reduce the DSB yield in mammalian cell DNA post-¹²⁵I decay⁽²¹⁾. These latter studies suggested that the spatial organisation of chromatin in mammalian cells supports conditions suitable for the induction of DSBs by the attack of free radical clusters at sites in DNA that are somewhat distant from the location of the ¹²⁵I decay. Since the mean diffusion distance of ^•OH in water is ∼6 nm⁽²²⁾, it was proposed⁽²¹⁾ that chromatin structure (highly packed DNA) provides conditions for the formation of >1 DSB/¹²⁵I decay by indirect mechanism(s) and that a cluster of ^•OH produced by decay of ¹²⁵I may thus attack a DNA site present 100s of nucleotides away from the decaying atom but factually placed in close proximity to the decaying atom (for example DNA strands juxtaposed within a nucleosome or those present in two successive turns of a 30-nm DNA solenoid).

In an attempt to further understand the factors controlling (and the mechanisms underlying) DNA DSB induction, the role of DNA topology in the production of radiation-induced DSB was recently addressed ⁽²³⁾. In these studies, DSB yields following the decay of ¹²⁵I were quantified and compared in plasmid DNA molecules with three differing topologies: SC (supercoiled), relaxed circular (nicked, N) and linear (L) DNA (Table 2). Surprisingly, and contrary to the expectations based on experimental results obtained in mammalian cells⁽²¹⁾, these studies⁽²³⁾ showed that the number of DSB produced by the decay of ¹²⁵I in SC DNA (∼0.5 DSB/¹²⁵I decay) was ∼1/3 of that seen in N and L DNA (∼1.6–1.7 DSB/¹²⁵I decay). Additionally, the mechanism of DSB formation was also found to be affected by these differing topologies. For example, whereas 100 % of the DSB in SC DNA is a consequent of direct hits, ∼50 % of the DSBs formed in circular and linear plasmid DNA are inhibited by the ^•OH-radical scavenger DMSO and are therefore caused by indirect effects (Table 3).

Table 2.

Topology of different forms of plasmid DNA.

Property	SC	N	L
Double stranded	Yes	Yes	Yes
Shape	Circular	Circular	Linear
Curvature	+++++	++	±
Compaction	Yes	Relaxed	Relaxed
Torsional energy	Yes	No	No
Twist	Under-wound	Normal (10.4 bp turn−1)	Normal (10.4 bp turn−1)

SC: supercoiled; N: nicked; L: linear.

Table 3.

DSB yield in plasmid DNA.

Topology	DSB/decay of 125IEH
	No DMSO	10 % DMSO	•OH-mediated indirect (%)
SC	0.5	0.5	0
L	1.6	0.8	∼50
N	1.7	1.0	∼40

IEH, iodo(¹²⁵I)-ethoxyHoechst 33342.

The emission spectra of the two Auger electron emitters ¹²³I and ¹²⁵I are nearly identical⁽²⁴⁾. The former decays basically by EC with the emission of an average of 11 electrons while the latter isotope, which decays simultaneously by EC and IC, emits ∼20 electrons. Since the average energies deposited per decaying atom in microscopic spheres (2–10 nm) centred around the decaying site of ¹²⁵I is about two times that of ¹²³I⁽²⁴⁾, it was consequently anticipated that the DSB yield following the decay of ¹²³I in SC DNA would be (1) ∼1/2 that observed following the decay of ¹²⁵I and (2) higher in L DNA (i.e. per ¹²⁵I results). Our findings were mixed: while the DSB yield in SC DNA exposed to ¹²³I decays (∼0.2 DSB/decaying atom) was in line with dosimetric expectations (i.e. ∼1/2 that of the ∼0.5 DSB yield of ¹²⁵I), it was surprising to find that the exposure of L DNA to the decay of this low-energy emitter did not lead to the production of any DSB⁽²⁵⁾. Taken together, these studies lead to the following two unexpected conclusions:

• Energy deposition in plasmid DNA with differing topologies is not indicative of DSB yield; and

• DSB yield in DNA with similar topologies is (SC) or is not (L) proportional to the energy deposited in DNA.

Toxicity to mammalian cells

The tri-dimensional organisation of chromatin within the mammalian cell nucleus involves many structural level compactions (e.g. chromatin fibres, nucleosomes, double-stranded DNA). Since the dimensions of these DNA conformational states are within the range of the high-LET, low-energy, short-range electrons, the toxicity of Auger-emitting radionuclides is expected to be very high and depends critically on the intracellular position of the decaying atom. These expectations have been substantiated by the results of in vitro studies showing that the decay of DNA-incorporated Auger electron emitters is highly toxic and leads to a monoexponential decrease in survival^{(4, 10, 24, 26–29)}. In addition, it was also shown that the decay of Auger electron emitters (e.g. ⁵¹Cr, ⁶⁷Ga, ⁷⁵Se, ¹²⁵I, ²⁰¹Tl) within the cell cytoplasm^{(3, 5, 11, 30–32)}, affixed to cell plasma membranes^{(32, 33)}, or located extracellularly^{(3–5, 11, 30, 32, 34)} produces no extraordinary lethal effects, and the survival curves (1) resemble those observed with X ray (have a distinct shoulder) and (2) exhibit shallow slopes. For example, when ¹²⁵I is localised within the cytoplasm, the survival curve is of the low LET type and the number of decays needed to reduce survival is ∼80 times that of DNA-incorporated ¹²⁵I^{(4, 5)}.

Historically, dosimetric predictions and in vitro studies had shown that the dose to the cell nucleus was predictive of the radiotoxicity of Auger electron-emitting radionuclides. This conclusion was based on experiments showing that when mammalian cells were exposed to the decay of DNA-incorporated ¹²⁵I, ⁷⁷Br or ¹²³I and cell survival is plotted as a function of nuclear dose⁽³⁴⁾, a single monoexponential curve was representative of survival (Figure 2). More recently, however, it has become apparent that the dose to the cell nucleus is not necessarily indicative of toxicity. For example, when the radiotoxicity of the DNA-intercalator ¹²⁵I-acridine⁽³⁵⁾ and that of the DNA-adduct-forming ^195mPt-labelled trans-platinum⁽³⁶⁾ were assessed in mammalian cells and survival plotted as a function of radiation dose to the cell nucleus, the data points could all be fitted on a single monoexponential curve whose slope was lower than that for the DNA-incorporated ¹²⁵I, ¹²³I and ⁷⁷Br (Figure 2). Other studies have also shown that when cells are incubated with the DNA-minor-groove-binder ¹²⁵I-Hoechst 33342⁽³⁷⁾ or treated with a triplex-forming ¹²⁵I-labelled synthetic oligonucleotide⁽³⁸⁾, the survival curves are linear quadratic. In the latter case, the decays needed to decrease survival to the same degree was >400-fold greater in comparison to DNA-incorporated ¹²⁵I. These findings reiterate that nuclear localisation (1) does not necessarily lead to monoexponential decrease in survival only, and (2) is not always highly toxic to mammalian cells.

Figure 2.

Radiotoxicity of Auger electron emitters post-localisation within nuclei of mammalian cells as function of energy deposited in nucleus⁽³⁴⁾. IUdR and BrUdR: iodo- and bromo-deoxyuridines (thymidine analogues); I-AP: iodo-acetyl proflavine (DNA intercalator); Trans-Pt: trans-platinum (forms adducts with DNA).

In 2001, Yasui et al.⁽³⁹⁾ reported that the DSB yield per ¹²⁵I decay for ¹²⁵I-estrogen, which binds to specific sequences within the DNA of mammalian cells, is ∼8-fold higher than that obtained for DNA-incorporated ¹²⁵I but that the efficiencies for cell killing are the same for both agents. However, earlier studies by Panyutin and colleagues^{(38, 40)} had shown that the number of DSB produced per decaying ¹²⁵I atom is the same when this Auger electron emitter is either DNA incorporated or localised in the cell nucleus consequent to the uptake of ¹²⁵I-triplex-forming oligonucleotides. Taken together, it is clear that (1) nuclear localisation of an Auger electron-emitting radionuclide does not necessarily lead to high mammalian cell killing, and (2) DSB formation in nuclear DNA is not necessarily directly proportional to cell survival.

Bystander effects

Another judgement made by those examining the radiotherapeutic potential of low-energy electron emitters was the perceived need to radiotarget every tumour cell (a direct consequence of the short range of the emitted electrons and, therefore, the absence of cross-fire irradiation of neighbouring cells). This too has proved to be inaccurate, as the decay of such isotopes has recently been shown to lead to a bystander effect, i.e. a dose-independent inhibition/retardation and/or stimulation of mammalian cell growth in non-radiotargeted cells by signals(s) produced in Auger electron-labelled cells. For example, in vivo studies have shown that the growth of subcutaneously implanted human tumour cells was influenced by the presence of ¹²⁵I- or ¹²³I-labelled cells that had been previously mixed with unlabelled tumour cells. Despite the fact that the electron spectra of both radionuclides are identical, the injection of a mixture of unlabelled and ¹²⁵I-labelled cells in mice inhibited the growth of unlabelled, unirradiated cells⁽⁴¹⁾, whereas the mixture with ¹²³I-labelled cells enhanced the growth of unlabelled cells⁽⁴²⁾. Interestingly, the percentage decrease (¹²⁵I induced) and increase (¹²³I induced) in tumour growth were quite comparable for both isotopes (Figure 3). Finally, it should be noted that similar inhibitory (iBE) and stimulatory (sBE) bystander effects were observed when the radiolabelled cells were incubated in vitro with unlabelled cells^(41–45).

Figure 3.

Inhibitory (¹²⁵IUdR) and stimulatory (¹²³IUdR) bystander effects consequent to decay of either Auger electron emitter within human tumours grown subcutaneously . in mice^{(41, 42)}.

The bystander effect induced by radioactive decay has impacted the views of many investigators interested in assessing the dose-related risks and the therapeutic efficacy following the administration of radiopharmaceuticals to patients. Traditionally, dose estimations are carried out by averaging the radiation dose to cells within a tissue or tumour mass from radioactive atoms present on or within the cells (self-dose) and that from radionuclides present in/on other cells or in the extracellular fluids (cross-dose). Such absorbed dose estimates have played an important role in determining the amount of radioactivity to be administered to patients in diagnostic/therapeutic procedures. Since the decay of Auger electron-emitting radionuclides can lead to either an iBE or an sBE, i.e. the actual radiobiological response will be greater/less than that predicted by dosimetric estimates alone, it is essential to assess the bystander effect of all therapeutic radiopharmaceuticals and select those exhibiting an iBE.

TARGETED AUGER ELECTRON RADIOTHERAPY

The very high in vitro radiotoxicity observed with DNA-incorporated Auger electron emitters has been exploited in experimental radionuclide therapy. While many radiopharmaceuticals have been synthesised and their radiotoxic/therapeutic effects examined in vitro and in vivo, the thymidine analogue 5-iodo-2′-deoxyuridine (¹²³IUdR/¹²⁵IUdR) has been used in most of these studies and the results have been very promising^(46–49). For example, the intraperitoneal (i.p.) injection of ¹²⁵IUdR⁽⁴⁶⁾ or ¹²³IUdR⁽⁴⁷⁾ into mice with i.p. ovarian cancer has led to a 4–5-log reduction in tumour cell survival. Similarly, the intrathecal (i.t.) administration of therapeutic doses of ¹²⁵IUdR into rats with i.t. tumours significantly delay the onset of paralysis, especially when the radiopharmaceutical was co-administered with methotrexate (MTX), an antimetabolite that enhances IUdR uptake by DNA-synthesising cells. In the latter cases, this was exemplified by a 5–6-log tumour cell kill and the curing of ∼30 % of the tumour-bearing rats^{(48, 49)}. Consequent to these promising results, MTX and ¹²⁵IUdR (1.85 GBq) were recently administered i.t. to a patient with pancreatic cancer metastatic to the CNS who had failed to respond to conventional therapy⁽⁵⁰⁾. A dramatic drop in spinal fluid CA19.9 level was observed after the single treatment with the radiolabelled agent that was accompanied by clinical improvement (Figure 4). The findings observed suggest that this approach may be an effective treatment for neoplastic meningitis.

Figure 4.

Therapeutic response of CA19.9 antigen in patient with intrathecal tumour following i.t. administration of MTX and ¹²⁵IUdR⁽⁵⁰⁾.

In another set of studies, ¹²³IUdR has been infused via the hepatic artery into 16 colon cancer patients with hepatic metastases⁽⁵¹⁾. SPECT imaging demonstrated that tumour uptake reaches an average 7.9 % ID (range 2.11–17.61 % ID) at the end of the infusion and that ∼5 % ID (range 1.36–11.12 % ID) is retained by these tumours. In these patients, no significant uptake was detected in the bone marrow or in other normal dividing tissues. In a consequent study⁽⁵²⁾, six patients were pre-treated with 5-fluorouracil and l-folinic acid, a drug combination known to inhibit thymidylate synthetase. Here again, an increase in early tumour uptake—by an average 64 % versus the baseline ¹²³IUdR infusion study (from 9.1 to 14.9 % ID) and a sustained enhancement of tumour uptake at 18 and 42 h (average 72 % increased stable uptake, from about 5.8 to about 10 % ID)—was observed (Figure 5). These results demonstrate that it is possible to biochemically modulate in vivo the tumour incorporation of radiolabelled IUdR infused intra-arterially in patients with liver metastases from colorectal cancer.

Figure 5.

(A) Planar image obtained 42 h after infusion of ¹²³IUdR into the hepatic artery of a patient with massive liver metastatic disease from colorectal cancer (residual radioactivity adsorbed in the subcutaneous port and in the intra-arterial catheter is visible-arrows). ¹²³IUdR is incorporated in several metastatic tumour lesions, primarily in the left lobe⁽⁵¹⁾. (B) Metabolic modulation of ¹²³IUdR uptake in liver metastases. Patients were imaged after hepatic artery infusion of ¹²³IUdR both under baseline conditions and following biochemical modulation obtained by premedication with 5-fluorouracil and folinic acid 1 week later⁽⁵²⁾.

SUMMARY AND CONCLUSIONS

Auger electron emitters represent an attractive alternative to beta-particle emitters for cancer therapy. While substantial progress in the understanding of the biophysical and radiobiologic phenomena underlying the emission of Auger electrons has been made, analyses of recent findings have indicated that certain invalid and/or unjustifiable assumptions have been made by radiobiologists and dosimetrists that are not in line with the experimental data. The following is a list that summarises some of these:

1. DSB yield determinations in naked DNA (plasmid) do not necessarily reflect the DSB yields of mammalian cells;

2. DNA topology has a major—and difficult to predict—impact on DSB yield;

3. Current Monte Carlo simulations do not often accurately predict the DSB yield in naked DNA and in mammalian cells chromatin;

4. Nuclear localisation is not always a good indicator of cell toxicity;

5. DSB yields in mammalian cells are not always a good indicator of cell survival; and

6. Modeling of opposing BE—seen even when the decaying atoms have similar electron spectra and occurring consequent to a change in dose—continues to be difficult.

These shortcomings should remind one that radiobiologic responses (micro and macro) are often accompanied with various homeostatic responses favouring the overall survival of the cell and the organism. Consequently, it is essential to:

1. develop microdosimetric approaches that factor in the experimental findings (radiative and non-radiative) and adjust all ‘calculated’ dose estimates by a ‘Dose Modification Factor’, a radionuclide-specific constant that factors in hitherto not-so-well recognised biophysical processes (e.g., consequent to bystander effects);

2. develop experimental designs that can simultaneously address the radiobiological phenomena taking place at the molecular, cellular, tissue, organ and whole organism level;

3. factor in all such variables when a radiopharmaceutical that is radiolabelled with a low-energy electron emitter is being developed for therapy.

FUNDING

This work was supported by NIH R01 CA15523.