*2.3. Ex Vivo Biodistribution and Atherosclerotic Plaque Targeting of [18F]AlF(RESCA)-cAbVCAM1-5*

The biodistribution of [18F]AlF(RESCA)-cAbVCAM1-5 is summarised in Figure 4A and Table S1. Uptake in various organs and tissues is expressed as injected activity per gram (%IA/g). Constitutively VCAM-1 expressing organs such as the spleen (1.01 ± 0.34 %IA/g), lymph nodes (0.55 ± 0.15 %IA/g) and thymus (0.32 ± 0.09 %IA/g) showed specific uptake. These values were significantly lower when an excess of unlabelled Nb was co-injected (respectively 0.34 ± 0.14 %IA/g, 0.33 ± 0.22 %IA/g and 0.22 ± 0.06 %IA/g). In corroboration with the imaging data, high bone uptake was observed, which could not be reduced by competition (1.13 ± 0.33 vs. 0.96 ± 0.33 for the control). Other organs and tissues, except the kidneys (14.00 ± 3.75 %ID/g), showed no uptake of the tracer. Analysis of the dissected aortas and gamma counting confirmed the specific lesion targeting with [18F]AlF(RESCA)-cAbVCAM1-5 Nb as seen on the PET/CT images. Uptake in the aortas of ApoE−/<sup>−</sup> mice was 2.15 <sup>±</sup> 0.06 times higher (*p* < 0.03) as compared to the control group (Figure 4B). This was further confirmed by autoradiography of the dissected aortas even though some background in the blocked group is visible due to non-specific binding (Figure 4C).

**Figure 4.** (**A**) Ex vivo biodistribution profile of [18F]AlF(RESCA)-cAbVCAM1-5 Nb in ApoE−/<sup>−</sup> mice and ApoE−/<sup>−</sup> mice co-injected with a 90-fold excess of unlabelled Nb (blocking condition). (**B**) Ex vivo analysis of excised atherosclerotic aorta, showing significantly higher uptake (2.15 ± 0.06 times; *p* < 0.03) of the [18F]AlF(RESCA)-cAbVCAM1-5 Nb compared to the control group (blocking condition). (**C**) Confirmation of the uptake by ex vivo autoradiography. The number of asterisks in the figures indicates the statistical significance (\* *P* < 0.05).

### **3. Discussion**

In the present study, as a proof of concept, a preclinical PET scanner capable of achieving sub-mm spatial resolution was used to image VCAM-1 expression in atherosclerotic lesions of ApoE−/<sup>−</sup> mice using an Al18F(RESCA)-labelled Nb.

The lead compound cAbVCAM1-5 has previously been reported to target both the murine and human VCAM-1 receptor with nanomolar affinity, and was originally validated as technetium-99m ( 99mTc)-labelled tracer for SPECT imaging by Broisat et al. [14]. Although new generation clinical SPECT cameras are emerging [21], PET imaging remains the preferred imaging modality to quantitatively image molecular markers in the clinic. In addition, short-lived isotopes commonly used for PET imaging, such as fluorine-18 (18F) or gallium-68 (68Ga), match the short biological half-life of Nbs, allowing to decrease the radiation burden for the patients. To this end, the cAbVCAM1-5 Nb was radiolabelled with 18F via the N-succinimidyl 4-[18F]-fluorobenzoate ester ([18F]SFB) prosthetic group [22]. Despite excellent in vivo results such as lower kidney retention and lower specific organ uptake, the production of [18F]SFB remains a time-consuming multi-step procedure. In an attempt to facilitate its clinical

translation, analogues of the cAbVCAM1-5 Nb labelled with radiometals via chelation were studied by Bala et al. [23] 68Ga, however, has some disadvantages such as a very short half-life of 68 min, limiting the number of patients per synthesis, and requires a 68Ge/ 68Ga generator which is not available at every hospital. More recently, a new bifunctional chelator, RESCA-tetrafluorophenyl ester (TFP-RESCA) has been developed by Cleeren et al. [18,19]. Contrarily to most fluorination methods, the RESCA chelator, allows fast radiolabelling of biomolecules with [18F]AlF at RT in aqueous solution.

We applied this strategy by coupling the RESCA chelator to the cAbVCAM1-5 Nb via its lysine residues, and an excellent radiolabelling yield with [18F]AlF was obtained (>75% RCY in less than 60 min) (in comparison, radiolabelling of Nbs using [18F]SFB have global yields ranging between 5 and 15% for a procedure time of 180 min) [24].

PET imaging of atherosclerotic lesions in the aortic arch of mice could be performed 2.5 h after intravenous (IV) administration of the radiolabelled Nb. Ex vivo analysis confirmed the imaging data, showing a significantly higher uptake of the tracer in excised aortas (0.42 ± 0.08 %IA/g) compared to the blocking condition (0.20 ± 0.02 %IA/g). This uptake, however, is lower than the previously reported uptake with [18F]FB-labelled cAbVCAM1-5 Nb (1.18 <sup>±</sup> 0.36 %IA/g) [22].

The non-invasive detection of small atherosclerotic lesions could be improved using a preclinical PET scanner of the latest generation (such as the β-CUBE), yielding significantly higher plaque-to-brain and plaque-to-heart ratios. This confirms the importance of PET scanners with sub-mm spatial resolution and high sensitivity to evaluate novel tracers for atherosclerotic plaque detection and characterisation.

Uptake of the radiolabelled cAbVCAM1-5 Nb in constitutively VCAM-1 expressing organs such as the spleen and the thymus was expected since this was already observed in previous studies [14,22,23]. This is attributed to specific targeting because uptake could be prevented by an excess of the cold analogue. Contrarily, bone uptake was noticeably high for both experimental groups. Although some specific uptake due to the expression of VCAM-1 in the bone marrow could be expected [25], the signal was observed in both groups, indicating the non-specific character of the uptake. The PET/CT images accurately co-localised the signal with bone structures (e.g., skull, limb bones, and vertebral column and sternum) and is most likely a result of uptake of degradation products derived from the tracer. The nature of this degradation product could either be formation of an active metabolite, or most likely in vivo decomplexation of [18F]AlF, or defluorination, considering that radiometals and free fluorine tend to accumulate in the bone structures [26]. A hypothesis could be that after glomerular filtration and reabsorption in the proximal tubuli, the [18F]AlF-RESCA complex is degraded in the lysosomes where the Nb is internalised. This results in the release of [18F]AlF or [18F]F<sup>−</sup> that returns into blood circulation and accumulates in the bone structures. As this instability appears to be lower in the case of other proteins [20,27], it would be interesting to further investigate the reasons behind the degradation in the case of Nbs.

In the context of atherosclerosis imaging, this bone uptake is particularly undesirable for two reasons. First of all, imaging must be performed at a time point that ensures the lowest achievable blood background signal, while keeping a significant signal in the VCAM-1-low-expressing targeted plaque. Even though Nbs are cleared very quickly from the blood, previous studies showed that imaging 2.5 to 3 h after tracer injection for the cAbVCAM1-5 Nb resulted in an optimal target-to-blood ratio [14] which unfortunately correlates with the formation of radio-metabolites observed during in vivo biodistribution studies. Secondly, [18F]NaF is already being investigated in the clinic to image active calcification in atherosclerosis [28]. Thus, if free [18F]F<sup>−</sup> or [18F]AlF is entering the blood, it would become unclear which biological process is being imaged.

#### **4. Material and Methods**

All reagents and solvents were purchased from Sigma–Aldrich (Overijse, Belgium) or VWR (Oud-Heverlee, Belgium). The RESCA bifunctional derivative was synthetised as described in patent WO/2016/065435 [18]. All buffers used for derivatisation and labelling of the Nb were prepared in metal-free water (Fluka, Honeywell, Brussels, Belgium), chelexed (chelex, 100 sodium form (Sigma Aldrich, Overijse, Belgium), 2 g/L, overnight shaking at RT) and filtrated with a 0.2 μm PES membrane filter (VWR). 18F was produced on site using a cyclotron (Cyclone KIUBE, IBA, Ottignies-Louvain-la-Neuve, Belgium) by irradiation of H218O with 18-MeV protons. Radioactivity was measured using an ionisation chamber-based dose calibrator (Veenstra Instruments, Joure, The Netherlands).
