The in vivo fate of 225Ac daughter nuclides using polymersomes as a model carrier

Increasing attention is given to personalized tumour therapy, where α-emitters can potentially play an important role. Alpha particles are ideal for localized cell killing because of their high linear energy transfer and short ranges. However, upon the emission of an α particle the daughter nuclide experiences a recoil energy large enough to ensure decoupling from any chemical bond. These ‘free’ daughter nuclides are no longer targeted to the tumour and can accumulate in normal tissue. In this paper, we used polymersomes as model carrier to evaluate the retention of recoiling daughters of 225Ac in vivo, and assessed their suitability as therapeutic agents. Vesicles containing 225Ac were injected intravenously in healthy mice, and intratumourally in tumour-bearing mice, and the relocation of free 213Bi was assessed in different organs upon the injection [225Ac]Ac-polymersomes. The therapeutic effect of 225Ac-containing vesicles was studied upon intratumoural injection, where treatment groups experienced no tumour-related deaths over a 115 day period. While polymersomes containing 225Ac could be suitable agents for long-term irradiation of tumours without causing significant renal toxicity, there is still a significant re-distribution of daughter nuclides throughout the body, signifying the importance of careful evaluation of the effect of daughter nuclides in targeted alpha therapy.

HER-2/neu monoclonal antibody due to the relocation of recoiling daughter atoms to the kidney 11 . In another study by Kennel et al., the potential of killing lung tumours in vivo was clearly demonstrated but at the cost of radiotoxic effects associated with released daughter radionuclides which were likely partially responsible for the death of the animals 12 .
Encapsulation of the 225 Ac mother nuclide in nanocarriers can help retain the daughter atoms at the tumour site and thus limit damage to healthy organs. This approach has been investigated with a number of nanoparticles, including metal-based particles 13 , zeolites 14 and liposomes 15 . Woodward et al. demonstrated that LaPO 4 based nanoparticles are capable of containing 225 Ac daughter nuclides. They observed a near complete retention of 225 Ac in the nanoparticles, but found that about 50% of the daughter nuclides were released in vitro. Subsequently, they followed the in vivo release of 213 Bi in the lungs, liver, spleen and kidney. Elevated levels of 213 Bi were found in the kidneys, indicating some release of this daughter nuclide following nuclear recoil, despite its encapsulation in the nanoparticle 16 . 225 Ac-containing liposomes have been shown to be very efficient at selectively killing tumour cells in vitro 17 . Their efficacy has been tested in 3D tumour models as well as in vivo, introducing a diffusion-assisted approach for full tumour coverage. Here, the liposomes degrade upon entering the lower-pH tumour environment, releasing 225 Ac and its daughter nuclides and hence allowing for a better distribution in vitro. This study has shown promising in vivo results, with a significant decrease in tumour volume upon the intravenous administration of the liposomes 18 . The distribution of the daughter atoms in vivo was not assessed, nor was potential renal toxicity considered, however, very poor 213 Bi retention in 100 nm liposomes has been reported in an earlier in vitro study 19 , suggesting similar problems to occur in vivo.
The aim of our study is to assess the retention of the radioactive daughter atom 213 Bi in a carrier system which has demonstrated good retention of the 225 Ac mother nuclide, and determine the in vivo fate of free 213 Bi 20 . We have selected polymersomes as delivery vesicles for 225 Ac-based targeted α therapy 8,20,21 , which have shown to retain the daughter nuclides 221 Fr and 213 Bi to a certain extent 20 . This retention can be improved by forming small InPO 4 nanoparticles around the 225 Ac encapsulated within the polymersomes 21 . Nanocarriers are known to accumulate in the liver and spleen, whereas free 213 Bi accumulates in the kidneys, allowing for clearly distinguishable uptake characteristics between the polymersomes themselves and any free daughter nuclides. In this study, we evaluate the retention of 213 Bi, one of the daughters in the decay chain of 225 Ac, in vivo upon intravenous and intratumoural injection, and looked at the distribution of free 213 Bi in select tissues, This is especially important to consider when using long-circulating carriers like nanoparticles or antibodies for tumour targeting. We assessed the effect of the polymersome diameter as well as the presence of InPO 4 nanoparticles encapsulated within the vesicles on the retention and distribution of 213 Bi. Finally, the effect of [ 225 Ac]Ac-polymersomes and [ 225 Ac]AcDOTA control on tumour growth, proliferation, apoptosis, and double strand DNA breaks upon intratumoural injection are evaluated in BALB/c nude mice.

Results and Discussion
It is well known that the recoil energy experienced by the daughter atom is sufficient to break any chemical bond between the daughter atom and a targeting vector. This issue could partially be circumvented through the use of nanocarriers, such as polymersomes, which present a system which is capable of retaining at least part of the daughter nuclides. We have used polymersomes to encapsulate 225 Ac in the aqueous cavity of the vesicles containing either DTPA 20 or InPO 4 nanoparticles 21 . In this study, we compared the recoil retention of the 213 Bi daughter nuclide in vivo of both these polymersome formulations. Throughout this manuscript, 'free 213 Bi' indicates 213 Bi which is no longer encapsulated in the polymersomes following nuclear recoil.
Vesicle characterization and loading of radionuclides. Prior to using the polymersomes in the in vivo studies, they were fully characterized by both DLS and Cryo-TEM (Fig. 1). In all cases, relatively monodisperse Bi upon intravenous injection. While the 225 Ac mother nuclide is very well retained in both systems (retention of approximately 93%) 20,21 , this is not the case for the daughter radionuclides. The theoretical recoil distance of the daughter radionuclides is about 100 nm in water 8 , which means that the probability of the daughter nuclides to be being retained in polymersomes with an optimal diameter of around 100 nm is limited 23,24 . The use of larger polymersomes will increase the retention of daughter radionuclides, but this is known to reduce the circulation time and subsequent accumulation in tumour tissue 21 . Another factor influencing the retention of recoiled daughter is the location of the decaying radionuclide in the vesicle. For instance, if the mother nuclide 217 At (second daughter of 225 Ac) is located in the polymersome bilayer, the chance of the 213 Bi daughter recoiling out of the vesicle is much higher than when 217 At would be located in the centre. The loading of 225 Ac into polymersomes through co-precipitation with InPO 4 nanoparticles has been shown more successful in retaining daughter nuclides compared with 225 Ac loaded through chelation with DTPA 21 . When 225 Ac is co-precipitated with InPO 4 , the recoil distance decreases to about 30 nm, greatly increasing the probability of daughter retention.
To determine whether these predictions also hold within a more complex in vivo system, the distribution of 213 Bi after administration of [ 225 Ac]Ac-polymersomes was determined in mice. With a half-life of a little under an hour, the ingrowth of 213 Bi was determined through continuous measurement of the ex-vivo organ activity in time. These measurements allowed for extrapolation to the time of sacrifice and a direct comparison of the presence of 213 Bi and 225 Ac at organ level (see Fig. 2).
Unfortunately, given both the short half-life of 221 Fr (t 1/2 = 4.8 min) and the time it took to perform the biodistributions together with the relatively large distance between the animal facility and gamma counter, we were not able to accurately assess the distribution of 221 Fr. Like most nanocarriers, polymersomes accumulate mainly in the spleen and liver 25 , while free 213 Bi is known to accumulate mainly in the kidneys (40%), or is excreted via the urine (30%) 26 . To be able to measure the 213 Bi ingrowth with a certain degree of accuracy, it was important to keep the total number of samples small so that they could be measured frequently. To this end, it was decided to determine the recoil retention of 213 Bi in three organs of interest; the blood (injection site), spleen (polymersome accumulation) and kidneys (accumulation of free 213 Bi). Immediately upon sacrifice, the organs of interest were collected Figure 2. Schematic depictions of the amount of polymersomes ( ) containing 225 Ac and associated daughter nuclides, and free 213 Bi ( ) in three organs (blood, spleen and kidney), with underneath per organ the measured % ID/g activity of 213 Bi as function of the measurement time. Time t = 0 represents the moment of sacrifice, 4 h after the injection of 50 kBq 100 nm [ 225 Ac]Ac-polymersomes. The percentage of free 213 Bi daughters can be obtained by extrapolating back to t = 0, and dividing by the equilibrium activity. and continuously measured until the daughter nuclides were in equilibrium with the mother nuclide 225 Ac. Back-extrapolation of the obtained data enabled the determination of the amount of 213 Bi at the moment of sacrificing the animal, as this is the time point at which blood flow stops, stopping subsequent inter-organ redistribution of free daughters. The activity at this time ( ) therefore represents the activity of 213 Bi nuclides both from within the polymersome, as well as any redistributed free 213 Bi. At equilibrium, the 213 Bi activity ) comes only from the decay of 225 Ac in the polymersomes, and thus gives an accurate representation of the biodistribution of the polymersomes. The ratio of these two numbers indicates the distribution of free 213 Bi, and shows that the redistribution of daughter nuclides has a non-negligible effect on the dose distribution of α therapy to the different organs.
In Table 1, the activity of 213 Bi at the moment of sacrifice (t = 0) as well as in equilibrium with 225 Ac is given for selected organs. Furthermore, the ratio of the 213 Bi at t = 0 to the equilibrium activity of 213 Bi after in-growth to the mother nuclide activity is given. When this ratio is <1, it means that the organ of interest is releasing 213 Bi (in this case the blood and spleen), while for a ratio >1, the organ is accumulating free 213 Bi. The daughter nuclide distribution presented in Table 1 was measured 4 h after the injection of 225 Ac encapsulated in either DTPA-containing polymersomes, or co-precipitated in InPO 4 nanoparticles within the polymersomes. In both cases, a significant difference (p < 0.02) between initial 213 Bi presence and 213 Bi levels at equilibrium with 225 Ac was found in the blood. Significantly more (p < 0.05) 213 Bi was retained in the blood for the InPO 4 containing vesicles vs the DTPA ones (with ratios of 0.14 ± 0.07 and 0.06 ± 0.03 respectively). Based on the blood values, co-precipitation of 225 Ac with InPO 4 nanoparticles in polymersomes constitutes a definite improvement over DTPA containing polymersomes. However, the ratio of the redistributed 213 Bi at time of sacrifice to the equilibrium 213 Bi activity found in the spleen and to the kidneys is not significantly different (at a significance level of α = 0.05) when considering the two different polymersome formulations. At first glance this is surprising; a lower amount of free 213 Bi would be expected to be present in the mice injected with InPO 4 -containing polymersomes based on the blood values, which should have resulted in ratios closer to 1 for both the spleen and the kidneys. However, in the kidneys the accumulation of InPO 4 -containing polymersomes is nearly a factor two lower than for the DTPA-containing polymersomes (2.6 ± 1.1% ID/g vs 4.7 ± 0.8% ID/g respectively), which subsequently increases the ratio of free 213 Bi to the equilibrium activity of 213 Bi in polymersomes to similar levels. Free 213 Bi present in the kidneys at time t = 0 nearly exclusively originates from the release of daughter nuclides from the polymersomes in other organs. The amount of free 213 Bi in the kidneys at time of sacrifice therefore more accurately represents the amount of recoiled 213 Bi than the ratio does. There is significantly less free 213 Bi at t = 0 in the kidneys for the InPO 4 containing polymersomes (at a significance level of α = 0.05) which agrees with the differences observed in the blood. For the spleen, similar ratios between the DTPA and InPO 4 containing polymersomes were observed. This is most likely explained by the fact that macrophages within the spleen take up the polymersomes, and the macrophages themselves can than act as a secondary barrier against the escape of the 213 Bi daughter nuclides from the macrophages. This effect has been described by McDevitt et al., who used 225 Ac-labeled antibodies, and found that upon internalization in the cell, the cell aids in the retention of the daughter nuclides 27 .
In this study, we have assessed the retention of 213 Bi in the polymersomes upon intravenous injection, and looked at the distribution of free 213 Bi in organs of interest. However, whereas the daughter radionuclides are retained to at least some degree within the polymersomes themselves, they will always be released when chelated to a tumour-targeting antibody due to the recoil effect 28 , with amongst others the different chemical properties of the daughter nuclide making re-association very unlikely. Therefore, it is of great interest to compare the results obtained in our study to antibody-targeting studies assessing the recoil retention in vivo. Unfortunately, to date there are only very few studies looking at the free daughter distribution upon decay of 225 Ac attached to an antibody. McDevitt et al. recognized that the retention of the daughter alpha-emitters at the target site is critical to the success of the therapy, and therefore specifically focussed on the use of target cell-internalized 225 Ac constructs, where internalization by the tumour cell itself resulted in enhanced retention of the daughter nuclides 27 . The daughter nuclides 221 Fr and 213 Bi were very well retained at the tumour site at 2 days p.i. (88 ± 9% and 89 ± 2% respectively), although it has to be mentioned that these tumours were counted between 6 and 12 minutes after death and not extrapolated back to time of death, increasing the calculated tumour retention. Kidney values of 213 Bi show a significant accumulation of free 213 Bi, at least 3 times higher than at equilibrium. However, the equilibrium values of the 225 Ac-antibody constructs in the kidneys are much higher compared to the low uptake of [ 225 Ac]Ac-polymersomes found in our study, which significantly decreases the ratio of free 213 Bi to that in equilibrium with 225 Ac. Furthermore, their values were reported in cpm instead of % ID/g preventing the direct 213 Bi the 213 Bi activity at time of death (t = 0) or in equilibrium with the mother nuclide 225 Ac (t = eq). 225 Ac was encapsulated in polymersomes containing either the hydrophilic chelate DTPA or InPO 4 nanoparticles. The uncertainty corresponds to standard deviations based on five mice per polymersome type. www.nature.com/scientificreports www.nature.com/scientificreports/ comparison of free 213 Bi uptake. Another study by Jaggi et al. looked in detail at the distribution of 225 Ac daughter nuclides upon the intravenous injection of 225 Ac-labeled antibodies 9 . While in most cases they do not specifically assess the ratio of the free daughter nuclide to the equilibrium situation, they do show a kidney-to-femur ratio of 225 Ac which is much lower than that of 213 Bi (an estimated ratio of approximately 1 for 225 Ac vs 20 for 213 Bi) indicating significant redistribution of 213 Bi to the kidneys. Song et al. used 225 Ac-labeled anti-rat HER-2/neu monoclonal antibody and found a difference in kidney activity of 213 Bi at time of sacrifice of 2449.8 Bq/g to 887.0 Bq/g at equilibrium 11 , also concluding that there is significant uptake of free 213 Bi in the kidneys. Therefore, while all these studies show that there is indeed redistribution of free 213 Bi, confirming the results obtained in our study, the presented ratios cannot be compared directly.

Intratumoural injection.
To determine the distribution of free daughter nuclides after accumulation in the tumour, 225 Ac-containing polymersomes were intratumourally injected and the distribution of 213 Bi was quantified immediately upon sacrifice. In these experiments, 225 Ac was encapsulated into polymersomes through chelation with DTPA. Although the polymersomes where 225 Ac was co-precipitated with InPO 4 yielded better 213 Bi retention upon intravenous injection, polymersomes containing DTPA have been studied more extensively 20,22 , including their therapeutic potential in vitro 29 , and were thus selected for the intratumoural experiments presented in this paper. Again, free 213 Bi was expected to be transported by the blood to the kidneys, and hence the organs of interest in this study were the blood, kidneys and tumour. Polymersomes with a diameter of either 100 nm or 200 nm were intratumorally injected to study the effect of polymersome diameter on daughter nuclide retention. These larger vesicles were not used in the intravenous study, as polymersomes should ideally have a diameter below 100 nm 23 for optimal circulation times. Table 2 shows the retention of 213 Bi in the tumour tissue, giving both the amount of 213 Bi present in tissue at time of sacrifice (t = 0) as well as at equilibrium. The ratios of free 213 Bi to 213 Bi in equilibrium are also displayed in this table, where again a ratio is <1 signifies a release of 213 Bi in the tissue of interest, and a ratio >1 an accumulation of free 213 Bi. Clearly, the tumour is in all cases the only tissue releasing free 213 Bi, where the 213 Bi is subsequently transported by the blood to the kidneys amongst others. However, it has to be kept in mind that the ratios for both the blood and the kidneys appear exceedingly high due to near-zero amount of polymersomes present in these organs at equilibrium. Rather, for these organs the activity of free 213 Bi at time of sacrifice should be regarded as the main indicator of their free 213 Bi accumulation. The very low radionuclide activity in the blood shows that 213 Bi is transported by the blood but not retained there. There was no large difference in free 213 Bi uptake between the 100 nm and 200 nm vesicles in the blood, although in all cases kidney uptake of free 213 Bi was slightly lower for the 200 nm polymersomes. This corresponds to the retention of free 213 Bi in the tumour tissue, where although already quite well retained for 100 nm polymersomes, 213 Bi is nearly completely retained in the tumour when encapsulated in 200 nm vesicles. However, for both polymersome diameters, no significant increase in tumour retention of 213 Bi is seen over time, which indicates that the size of the polymersomes still influences the retention of the daughter nuclides suggesting that part of the vesicles has not yet been taken up by the tumour cells. This corresponds to earlier obtained results in in vitro tumour spheroids, where the polymersomes were still distributing themselves throughout the spheroid more than 4 days after the addition of fluorescently labelled polymersomes to the cell medium 29 . Looking specifically at potential kidney toxicity, the activity of 213 Bi at time of sacrifice is in all cases much higher in the tumour than in the kidneys, with tumour: kidney ratios of 9.9 for 100 nm polymersomes, and 29.5 for 200 nm polymersomes at 7 days p.i. Renal toxicity due to free 225 Ac daughter atoms which have recoiled out of the polymersomes was thus not be expected to be a problem for the therapeutic studies 20 .
A study by Woodward et al., examined the recoil retention of 213 Bi in LaPO 4 nanoparticles containing 225 Ac and showed that the release of 213 Bi from the target organ decreased over the course of 5 days 16 . They attributed this to the potential uptake of their nanoparticle conjugates by the endothelial cells of the lungs, which subsequently trapped any free 213 Bi within the cells, retarding its diffusion through the tissue. We did not observe a similar pattern in our study; the retention of 213 Bi in the tumour tissue is stable in time for both polymersome sizes. This difference in behaviour could be due to the different injection methodologies. Woodward    www.nature.com/scientificreports www.nature.com/scientificreports/ their nanoparticles intravenously allowing them to be taken up in the target tissue in time. Although they did not present any direct measurements of circulation time of the nanoparticles, the activity in the target organs is still increasing up to 48 h p.i., pointing to a long circulation time during which the nanoparticles are not completely taken up in the endothelial cells yet. On the other hand, our polymersomes were injected directly within the tumour tissue and hence did not circulate for an extended period of time, resulting in no significant increase in retention between 1 and 7 days p.i.

Biodistribution. Intravenous injection.
While 221 Fr emits gamma particles with energies suitable for SPECT imaging 16 , the activity used in this experiment is too low for proper imaging which is why we performed a biodistribution study. The distribution of 225 Ac complexed to DTPA in polymersomes was compared to that of 225 Ac co-precipitated with InPO 4 nanoparticles in polymersomes upon intravenous injection. Biodistribution data of the [ 225 Ac]Ac-polymersomes at 4 h p.i. can be seen in Fig. 3, for both 225 Ac coupled to DTPA or encapsulated in InPO 4 nanoparticles. Because the radionuclides are encapsulated within the aqueous core of the polymersomes, the outer surface fo the polymersomes remains the same, logically resulting in the similar organ uptake. The only organ displaying a significant difference in uptake between the two types of vesicles is the liver. At the moment, we have no good explanation for this difference in liver uptake. The two types of polymersomes (containing either InPO 4 or DTPA in the aqueous cavity) were composed of the same batch of block copolymers, prepared according to the same method, and injected in mice which had been randomly distributed between the two groups. Since the outer surface of the polymersomes was not changed, a difference in uptake was not expected, and further research would be required to fully understand this difference in uptake. The circulation time of these polymersomes in tumour bearing mice is known to be considerably shorter than in healthy mice, with circulation half-lives of 5 min and 117 min respectively 30 . The exceedingly short circulation half-life in tumour bearing mice prevents sufficient polymersome accumulation at the tumour site upon the intravenous injection of 111 In-containing polymersomes (0.44 ± 0.39% ID/g). We therefore decided to assess therapeutic efficacy of [ 225 Ac] Ac-polymersomes through intratumoural injection.

Intratumoural injection.
For the biodistribution and therapeutic study of intratumorally administered [ 225 Ac] Ac-polymersomes, vesicles with a diameter of 100 nm were used. Here, the biodistribution of both intratumourally injected 225 Ac complexed to DTPA in polymersomes as well as [ 225 Ac]AcDOTA was assessed at 1 and 7 days p.i. The reason for the use of [ 225 Ac]AcDOTA instead of 225 Ac-DTPA as a control for the polymersomes is its increased kinetic stability 31 . [ 225 Ac]AcDTPA has been shown to display substantial toxicity due to 225 Ac leakage, whereas [ 225 Ac]AcDOTA constitutes an substantial improvement 31 . While this lower stability of the [ 225 Ac] AcDTPA complex has been shown not to result in enhanced 225 Ac leakage from the polymersomes 20 , the more stable [ 225 Ac]AcDOTA complex was chosen for the in vivo studies to minimize toxicity due to unbound 225 Ac. In Fig. 4, the distribution of both [ 225 Ac]Ac-polymersomes and [ 225 Ac]AcDOTA in selected organs is displayed. When viewing these biodistribution results, it is important to keep in mind that the injected activity as well as volume were very small, due to which the activity measured in various organs was just slightly above background. As expected based on earlier results by Wang et al. 25 , the polymersomes are very well retained in the tumour tissue, while the [ 225 Ac]AcDOTA compound is rapidly cleared. The tumour retention of the polymersomes was found to be 244 ± 74% ID/g and 289 ± 130% ID/g at 1 and 7 days p.i. respectively, whereas less than 10% ID/g and 5% ID/g of the [ 225 Ac]AcDOTA was retained at the tumour site at both time-points respectively (Fig. 4). Uptake in most other organs is minimal for both compounds. In the [ 225 Ac]AcDOTA study, nearly no activity is found in any other organ, suggesting that the majority of the compound was excreted already at time of sacrifice. The only organ exhibiting some uptake of [ 225 Ac]AcDOTA are the kidneys at 1 day p.i., indicating that renal excretion is indeed the main pathway for [ 225 Ac]AcDOTA clearance. At 7 days p.i., renal uptake is present only in insignificant amounts. The [ 225 Ac]Ac-polymersomes are much better retained by the tumour tissue than [ 225 Ac] AcDOTA. Organs other than the tumour exhibiting uptake of [ 225 Ac]Ac-polymersomes are the liver and spleen, and to a lesser extend also the bone and muscle. The main cause for healthy tissue uptake can likely be found in release of the polymersomes themselves. Polymersomes which have diffused out of the tumour tissue will either be filtered out by Kupffer cells, which are responsible for the phagocytic activity of the liver, or by macrophages in the spleen which corresponds well with the uptake observed in Fig. 4 32 . The 225 Ac presence in the healthy organs could also, in part, be caused by 225 Ac which was released from the polymersomes. Both free 225 Ac as well as chelated 225 Ac have been shown to accumulate in the bones as well as in the liver 31,33,34 , and is temporarily also retained in the kidneys 33,35 . Davis et al. have shown that this kidney uptake is very time dependent and much smaller than the liver uptake, so even though kidney uptake in our study is minimal, there might be some free 225 Ac responsible for the organ uptake. However, in earlier work, the 225 Ac released from the polymersomes was found to be only 2% of the total amount of encapsulated 225 Ac over a period of 8 days, making the release of 225 Ac from the polymersomes in vivo unlikely 20 . These biodistribution results indicate that renal toxicity from 225 Ac-polymersome accumulation is unlikely to play a major role in inducing dose-limiting toxicity, though any health issues related to the kidneys and bone marrow should be carefully studied.

tumour growth and overall survival upon intratumoural injection. The tumour model chosen,
MDA-MB-231, is a well-vascularized tumour model. Upon injection of the compounds (the polymersomes as well as the DOTA and PBS), some blood was observed to leak out of the tumours. In Fig. 5, the tumour growth of the 8 mice per group bearing subcutaneous MDA-MB-231 tumours is presented. At 28 days after inoculation the tumours were injected with the various compounds. In 7/8 tumours injected with [ 225 Ac]Ac-polymersomes a complete inhibition of tumour growth was observed (Fig. 5A). Two out of eight tumours doubled in size (at day 17 after therapy administration), but decreased in size again over the course of the experiment. A similar tumour growth inhibition pattern was found in mice intratumourally injected with [ 225 Ac]AcDOTA; 6/8 tumours with inhibition of tumour growth (Fig. 5B). Five out of eight tumours doubled in size during the experiment (mean time after treatment administration was 67 ± 31 days), but four of these tumours decreased again during the study. Tumours in the two control groups (no intratumoural injection and tumours injected with PBS) showed a pronounced growth starting around 50 days after tumour inoculation (Fig. 5C,D). Seven out of eight tumours were doubled in size at an average of 35 ± 10 days after PBS administration. Six out of eight tumours of the non-injected mice did not show tumour growth inhibition at the end of the study. The average tumour doubling time was 18 ± 9 days after treatment administration, but two of these tumours decreased in size over the course of the experiment. The PBS injection is not expected to have any influence on tumour growth, though the injection itself could have damaged the tumour tissue resulting in regression of the tumour. Surprisingly, tumour regression was also observed in some of the animals which had not received any treatment at all (Fig. 5D), showing that some of these tumours may spontaneously regress. Despite the fact that some tumours from the control groups also showed regression, we could still demonstrate that [ 225 Ac]Ac-polymersomes significantly inhibited tumour growth as compared to the control groups. When time to tumour related decease is displayed in a Kaplan-Meier survival curve (Fig. 5E), the therapeutic efficacy of [ 225 Ac]Ac-polymersomes and [ 225 Ac]AcDOTA administration is evident as demonstrated by significantly (p < 0.05) improved survival as compared with the control groups.
The biodistribution study (Fig. 4)   www.nature.com/scientificreports www.nature.com/scientificreports/ is roughly equivalent to the dose given by 5 kBq 225 Ac in polymersomes to a 400 µm tumour spheroid as studied earlier 29 taking into account the polymersome uptake in the spheroids of only 0.10-0.15%. There, a significant reduction in tumour size was observed at 3-6 days after addition of the vesicles. In the current in vivo study, no significant increase in tumour size was observed after the injection of the polymersomes. The excellent response in the group injected with 50 kBq [ 225 Ac]AcDOTA, which, similarly to the [ 225 Ac]Ac-polymersome group, experienced no tumour related death over the treatment period, was somewhat unexpected. The biodistribution studies at 1 day p.i. show that at this point nearly all [ 225 Ac]AcDOTA has been cleared from the tumour site (Fig. 4). However, in previous in vitro studies, [ 225 Ac]Ac-polymersomes with an activity of only 0.1 kBq, already showed a decrease in tumour growth, indicating that very low amounts of activity can already influence tumour growth 29 . The dose received within the first few hours after injection of [225Ac]AcDOTA was thus sufficient to delay tumour growth, illustrating the effectiveness of α radionuclide therapy.

immunohistochemistry. To determine the effect of the intratumoural injection of the [ 225 Ac]
Ac-polymersomes, [ 225 Ac]AcDOTA, as well as PBS on the tumour tissue, tissue sections were stained with HE and γ-H2AX (Fig. 6). The HE histological stain was used to demonstrate different tissue structures. While hematoxylin stains nuclei blue, eosin stains the cytoplasm as well as tissue fibers and matrigel 36 . Representative tumour sections of each of the treated groups are presented in Fig. 6. To detect the presence of double-stranded breaks (DSBs) in the DNA, γ-H2AX staining was used. γ-H2AX foci in the tumours gives an indication of the www.nature.com/scientificreports www.nature.com/scientificreports/ distribution and effectiveness of the α therapy, as the high LET of α radiation causes double strand breaks in DNA 37 . The tumours visualised in Fig. 6 show a similar uniform distribution of γ-H2AX foci in both the PBS and [ 225 Ac]AcDOTA groups. The images show that the [ 225 Ac]Ac-polymersomes treatment groups demonstrate a larger degree of γ-H2AX foci than the control groups, with an increasing foci gradient towards the centre of the tumour especially at day 1. As the polymersomes were injected in the centre of the tumour, the largest amount of radiation damage logically occurs in the vicinity of the centre. At 7 days p.i. the vesicles have not yet uniformly distributed themselves throughout the tumour, which is in line with the distribution rate as observed in the 3D tumour spheroids 29 , and is expected to improve at later time-points. However, despite the observed therapeutic efficacy in the [ 225 Ac]AcDOTA group, the increase in DSBs was less pronounced. This could be due to the fast clearance of the [ 225 Ac]AcDOTA complex from the tumour, and a subsequent decline in γ-H2AX signal to detect the DSBs. While the maximum number of γ-H2AX foci is reached 30 min after irradiation, the signal subsequently continuously decreases, reaching half their maximum already 2.5 h p.i. 38,39 . Hence, all damage to the tumour caused by [ 225 Ac]AcDOTA will likely had happened within the first few hours p.i., whereas the treatment with [ 225 Ac]Ac-polymersomes clearly irradiated the tumour over an extended period of time. Renal toxicity was not observed in any of the treatment groups, confirming the hypothesis that the amount of free 213 Bi accumulated in the kidneys was not sufficient to cause long-term toxicity issues.

conclusion
The main objective of this paper was to study the distribution of free 213 Bi in vivo after either intravenous or intratumoural injection of 225 Ac encapsulated in polymersomes. It is well known that the recoil energy of the daughter nuclide upon α decay is sufficient to break any chemical bond, leading sometimes to renal toxicity caused by accumulation of 213 Bi when using conventional antibody or peptide targeting approaches. Polymersomes have been shown to retain 225 Ac daughter radionuclides to some extent 21 , and in this study we specifically studied the 213 Bi retention and free 213 Bi distribution in different tissues of mice. Two types of polymersomes were examined www.nature.com/scientificreports www.nature.com/scientificreports/ upon intravenous injection, where 225 Ac was either encapsulated through conjugation with DTPA, or by coprecipitation with InPO 4 nanoparticles contained within polymersomes. The benefit of the shorter recoil distance caused by the presence of the nanoparticle was shown, as more than twice the amount of 213 Bi was retained in the case of the polymersomes containing nanoparticles. Upon intratumoural administration, 213 Bi was retained in the tumour tissue to a large extend, with tumour: kidney ratios of 213 Bi of 9.9 and 29.5 for 100 nm and 200 nm diameter polymersomes respectively. The therapeutic potential of intratumorally injected [ 225 Ac]Ac-polymersomes was also studied, and although biodistribution studies showed very favourable retention of the polymersomes in the tumour tissue, whereas 225 Ac coupled to DOTA was rapidly excreted, both formulations inhibited tumour growth. No tumour-related deaths were observed in either treatment group, and immunohistochemical analysis of the treatment groups at 1 d and 7 d p.i. showed an increase in γ-H2AX foci in the [ 225 Ac]Ac-polymersomes group, indicating a larger degree of double-stranded breaks.
Concluding, while [ 225 Ac]Ac-polymersomes and its daughters are retained at least partly the tumour site upon intratumoural administration showing their effectiveness in α therapy, their circulation time and tumour uptake upon intravenous administration needs further optimization before they can be used to target metastasized tumours. Furthermore, despite the much better retention of daughter nuclides in polymersomes compared to targeting molecules, 213 Bi is still released and accumulates in the kidneys upon intravenous administration. This demonstrates how essential it is for any studies using 225 Ac as α therapeutic to take daughter nuclide distribution into account.

chemicals.
The block copolymer PBd-PEO with M w of 1900-900 g/mol was purchased from Polymer Source (Quebec, Canada). The 225 Ac was obtained from the Directorate for Nuclear Safety and Security (Karlsruhe, Germany). The PD10 size exclusion columns were obtained from GE Healthcare (Hoevelaken, the Netherlands). Instant Thin-Layer Chromatography (iTLC) strips were purchased from Varian (USA). For the immunohistochemical analysis, rabbit-anti-H2AX (Cell Signaling, art.nr. 9718), goat-anti-rabbit (Vector, art.nr. BA-1000), avidin-biotin (Vectastain, art.nr. PK-6100) and Bright DAB (Immunlogic, art.nr. BSO4-500) were used. All other chemicals were purchased at Sigma Aldrich. polymersome preparation and loading of 225 Ac. Polymersomes containing either DTPA as hydrophilic chelator, or KH 2 PO 4 as precipitating agent were prepared for the in vivo studies, according to earlier published procedures 21 . The vesicles used in the intravenous injection studies were prepared using the direct dissolution method 22 , where 10 mg/mL block copolymer was added to a 1 mM DTPA PBS solution at pH 7.4, or a 0.5 M KH 2 PO 4 PBS solution at pH 3. The solution was stirred at 300 rpm for a week, upon which the polymersomes were extruded to a diameter of 100 nm by passing them several times through polycarbonate filters with cut-off membrane of 800, 400, 200 and 100 nm. For the intratumoural experiments, the DTPA containing vesicles were prepared according to the solvent displacement method. In short, 1 mM DTPA in 1 mL PBS buffer solution (pH 7.4) was slowly added to 1 mL acetone containing 20 mg/mL block copolymer under continuous stirring. After evaporation of the acetone, 1 mL PBS was added to bring the final concentration to 10 mg/mL block copolymer. In all cases, before loading the polymersomes with the radionuclide 225 Ac, the remaining unencapsulated DTPA or KH 2 PO 4 was removed by passing it through a Sephadex G 25 size extrusion column (L × D = 30 × 1 cm). 225 Ac dissolved in pH 2 HCl together with 200 μL 10 mM HEPES buffer was added to a vial containing 0.1 mg A23187 dissolved in 100 μL CHCl 3 . Upon the evaporation of the CHCl 3 , 800 μL polymersome solution was added and incubated for 1 hour. A PD10 column was used for purification, and the 7 th 0.5 ml fraction was collected for further in vivo experiments at an 225 Ac concentration of either 250 kBq/mL for the intravenous experiments, or 2 MBq/mL for the intratumoural experiments. Loading efficiencies were calculated as the amount of activity eluted from the PD10 column in the polymersome-containing fractions divided by the total activity in the sample. The daughter nuclide 221 Fr (218 keV) was counted with the Wizard gamma counter (PerkinElmer) when equilibrium was reached and decay-corrected as a representative measurement of 225 Ac in the sample. Samples were counted for 1 minute each, with as counting window 170-270 keV for 221 Fr and 380-520 keV for 213 Bi.
DLS and cryo-teM measurements. Before being used for the in vivo studies, the vesicles were characterized by both dynamic light scattering (DLS) and cryogenic transmission electron microscopy (Cryo-TEM). The DLS consisted of a JDS Uniphase 633 nm 35 mW laser, a fibre detector, an ALV sp 125 s/w 93 goniometer and a Perkin Elmer photon counter, with an ALV-5000/epp correlator and software. A 0.01 mg/mL polymersome solution was placed in a toluene filled, temperature regulated bath (20 °C), where the intensity autocorrelation function was determined at 90°, and the data was fitted using the Contin method. The hydrodynamic radius was obtained using Einstein-Stokes equation.
Cryo-TEM images were obtained by depositing 4 μL of the 10 mg/mL polymersome solution on a holey carbon film (Quantifoil 1.2/1.3, Cu 200 mesh grids) supported on a TEM grid. The drop was blotted for four seconds with filter paper in order to obtain a thin layer on the grid, and subsequently vitrified by rapidly immersing in liquid ethane (Leica EM GP version 16222032). The sample was inserted into a cryo-holder (Gatan model 626) and then transferred to a Jeol JEM 1400 TEM. Images were obtained at an acceleration voltage of 120 keV. For the statistical analysis of the polymersome and nanoparticle diameters, about 30-50 images were made of each of the polymersome samples. The diameter of polymersomes and nanoparticles within those images were measured with ImageJ 40 .
[ 225 Ac]AcDotA radiolabelling. For the radiolabelling of [ 225 Ac]AcDOTA, 1.4 MBq 225 Ac was added to efficiency was determined with iTLC. A 2 µL droplet of the radiolabelled sample was placed on a 7 cm ITLC strip, with 0.1 M NaOH at pH 12 as mobile phase. The ITLC strip was subsequently imaged with the Phosphor-imager to determine the labelling efficiency, which was calculated as the fraction of the total activity which travelled with the mobile phase divided by the total activity on the sample. The sample was used without further purification, and diluted in PBS to an 225 Ac concentration of 250 kBq/mL for intravenous, and 2 MBq/mL for intratumoural studies. cell culture. The human breast cancer cell line MDA-MB-231 was cultured in RPMI-1640 (GIBCO, ThermoFisher Scientific, Waltham, MA, USA) media supplemented with 2 mM glutamine (GIBCO) and 10% fetal calf serum (Sigma-Aldrich Chemie BV), at 37 °C in a humidified atmosphere with 5% CO 2 . Cells were dissociated when 80-90% confluency was reached using 0.05% trypsin (w/v) in 0.53 mM EDTA (Life Technologies) and maintained as proliferating cultures. Mycoplasma contamination was evaluated every four months using a MycoAlert ™ mycoplasma detection kit (Lonza, Basel, Switzerland). After thawing, cells remained in culture for a maximum of six months.
Animal studies. All animal studies were approved by the Dutch central committee on animal research and the local ethical committee on animal research of the Radboud University under protocol 2015-0071, and performed according to the institutional guidelines. 6-8 weeks old female BALB/cAnNRj-Foxn1 nu /Foxn1 nu mice (Janvier Labs, France) were randomly tattooed for identification upon arrival. The mice were acclimatized for ≥4 days before any experimental procedure, and had unlimited access to food and water. Cages were replaced by clean cages every week, and the animals were housed with 5-6 mice per cage in a controlled environment (12 h dark/light cycle, 22 ± 1 °C, 55 ± 10% humidity). In all biodistribution studies, mice were sacrificed through CO 2 asphyxiation and the organs of interest were collected, weighed, and the activity of two of the daughter nuclides of 225 Ac, namely 221 Fr (218 keV) and 213 Bi (440 keV), were counted with the Wizard gamma counter (PerkinElmer) approximately 20 hours after sacrifice (when secular equilibrium was reached). The activity concentration for 225 Ac was estimated from the daughter activity of 221 Fr counted after equilibrium.

Daughter nuclide retention upon intravenous injection.
Retention of 213 Bi in the polymersomes was determined in vivo through the intravenous injection of 200 μL of the 3 mg/mL polymersome solution containing 50 kBq 225 Ac in the tail vain of five non-tumour bearing mice per group. Four hours post injection, the mice were sacrificed through CO 2 asphyxiation. Organs of interest for daughter retention (blood, spleen and kidneys) were rapidly collected and the 213 Bi activity was measured continuously on the Wizard gamma counter (PerkinElmer) for approximately 18 hours (Fig. 1). 213 Bi activity at the time of sacrifice was determined by fitting the data with equation 1 and extrapolating back to t = 0 20 : where A 1 the activity of 225 Ac, and A 2 (t) is the 213 Bi activity at time t with decay constant λ 2 . Figure 1 depicts the method of extrapolating the 213 Bi activity in each of the three organs of interest at time of sacrifice. Paired two-sided Student's t-tests were used to calculate the significance of the difference in organ uptake. , with V the volume of the tumour, W the width, L the length and H the height of the tumour. Tumour size was used to block-randomized the mice over the different treatment groups. When the average tumour volume was approximately 100 mm 3 , the mice were intratumourally injected with either 25 μL, 50 kBq [ 225 Ac]Ac-polymersomes, or 50 kBq [ 225 Ac]AcDOTA, and one of the control groups with 25 μL PBS, while the other group was left untreated.
Biodistribution and recoil retention. The ex vivo biodistribution and 213 Bi recoil retention were analysed at 1 and 7 days p.i. for the mice having been injected with [ 225 Ac]Ac-polymersomes, or [ 225 Ac]AcDOTA. The mice of the PBS and untreated groups were sacrificed for immunohistochemical analysis of the tumour and kidneys. Tumours and kidneys of all animals (3 per treatment group) were fixated in 4% paraformaldehyde in PBS overnight and embedded in paraffin to analyse double strand DNA breaks immunohistochemically. Retention of the daughter nuclide 213 Bi in the tumour, and accumulation in the kidneys, was assessed according to the procedure in the intravenous experiment, where 213 Bi activity at the time of sacrifice was determined by extrapolating the in-growth back to time t = 0.
Therapeutic evaluation. All groups consisted of 8 mice each, of which the tumour growth and overall survival was monitored up to 115 days after the start of treatment. Tumour volume was measured twice a week. Mice were taken out of the experiment when they reached one of the humane endpoints (tumour size > 2 cm 3 , ulceration or invasive tumour growth, other signs of clinical discomfort e.g. 15% weight loss in two days, dehydration). Technicians assessing the humane endpoints were blinded for the treatment groups.
immunohistochemistry. Tumours harvested at 1 and 7 days after the intratumoural injection were embedded in paraffin. Immunohistochemistry was performed using 4 µm thick tumour sections. For the haematoxylin and eosin (HE) staining, paraffin was removed by incubation with xylene, and subsequently sections were incubated with haematoxylin for 20 minutes and with eosin for 5 minutes, followed by dehydration and mounting in Permount.