Background
Delivery of biologics to the central nervous system (CNS) has been a major challenge. This is partly due to the fact that the CNS is physically separated from the periphery by several barriers, including the blood–brain barrier (BBB), a monolayer of endothelial cells supported by astrocytes and pericytes [
1]. It is estimated that only 0.1% of circulating macromolecules is able to reach the brain parenchyma, which severely limits the use of biologics to treat CNS-related diseases [
2].
Transcytosis pathways involved in the delivery of essential nutrients have been explored for delivery of drugs to the brain. Nutrients are able to cross the BBB via specific receptors expressed on the luminal side of brain endothelial cells via receptor-mediated transcytosis (RMT). So-called Trojan Horse approaches exploit this mechanism. Therapeutic biologics are coupled to receptor-targeting entities such as peptides or antibodies binding to these nutrient receptors to shuttle them to the brain [
3,
4]. Regardless intense research of RMT [
5] during the past 30 years, only two drug candidates successfully completed phase 1–2 clinical trials [
6,
7]. Moreover, currently there are no therapeutic proteins approved for clinical use that cross the BBB to exert their effect [
8].
Nanobodies (also called single-domain antibodies or VHHs) could be an interesting addition to the existing Trojan Horse delivery methods. Nanobodies are derived from the variable domain of the heavy-chain-only antibodies found in camelids [
9,
10]. They recognize antigens with similar affinities and specificity as monoclonal antibodies and can be easily fused to a wide variety of compounds [
11‐
15]. Nanobodies have been used in various fields, ranging from therapeutic to biochemistry applications [
10,
16‐
18].
Many reports have claimed access of nanobodies to the brain. Due to their small size and often cationic charge, nanobodies are able to fuse with the negatively charged cell membrane which can lead to brain uptake via adsorptive-mediated transcytosis [
19]. This mechanism entirely relies on the charge of the brain-penetrating entity. Fusion to a therapeutic compound changes this charge which then alters crossing efficiency [
20‐
22]. Since small changes or fusions of various entities to nanobodies that reach the brain via adsorptive-mediated transcytosis can alter their crossing capabilities, such nanobodies are not suitable as a Trojan Horse delivery systems. Instead, nanobodies that utilize RMT to deliver drugs to the CNS could be more successful Trojan Horses.
One of the most investigated RMT targets is the transferrin receptor (TfR), which is highly expressed on brain endothelial cells [
5,
23]. Anti-TfR monoclonal antibodies deliver therapeutics to the brain in rats [
24,
25], mice [
26,
27], monkeys [
4] and humans [
7]. Also Tfr antibody-fragments, such as single-chain variable fragments [
28‐
30], Fab fragments [
31] or dual variable domain immunoglobulins [
32], are able to facilitate blood to brain transport. Very recently, two studies have showed that engineering the Fc fragment of a monoclonal antibody to target the TfR resulted in more brain uptake of various therapeutic proteins [
33,
34]. This previous research opens the door for the discovery of brain-penetrating, anti-TfR nanobodies. In order to select potential Trojan Horse nanobodies (meaning nanobodies reaching the brain via RMT and not by adsorptive-mediated transcytosis), unequivocal preclinical evidence for transfer to the brain has to be delivered.
A reliable method to demonstrate CNS target-engagement is inhibition of beta-secretase 1 (BACE1). BACE1 is an enzyme that cleaves the amyloid precursor protein (APP) in neuronal endosomes leading to the generation of amyloid-β peptides (Aβ) [
35,
36]. Here, brain-penetrating moieties fused to a BACE1 inhibiting entity are intravenously injected into animal species, after which the brains are harvested and homogenized. A decrease in central Aβ levels, measured by ELISA, indicates BBB crossing [
33,
37‐
40]. While this proves good evidence for functional brain targeting this method requires for each measurement point the use of multiple mice [
41]. Moreover the further processing of brain extracts and ELISAs are time consuming and expensive, prohibiting large screening efforts.
In order to improve the robustness of preclinical in vivo CNS research and reduce the number of animals needed for proof-of-concept, a method is needed that demonstrates brain uptake by target engagement, is unambiguous with regard to brain target, and, finally, allows for reuse of laboratory animals. Here, we explore whether it is possible to use a nanobody to reach the brain using new nanobodies raised against the mouse Transferrin receptor TfR (mTfR), a known receptor-mediated transcytosis target. As a readout, we coupled the nanobodies to NT, a neuropeptide that elicits hypothermia after binding to the NT receptor (NTSR1) expressed in the CNS [
42]. Since this receptor is a G-protein-coupled receptor (GPCR) located in the cell membrane of hypothalamic neurons [
43], it can interact with molecules in the interstitial fluid. In contrast to intracerebroventricular administered NT, intravenously administered NT is not able to elicit a hypothermic response [
44]. Therefore, the observed hypothermic effect after intravenous injection of nanobody-NT fusions is direct evidence of BBB transport facilitated by that particular nanobody. Moreover, this method assesses the brain penetrating capabilities of the generated nanobodies after a single IV injection and without the need to sacrifice the animal, making it possible to use the animal multiple times. The developed method led to the discovery of the first nanobodies that can deliver a cargo (NT) to the brain via the TfR.
Discussion
Here, we describe the discovery and characterization of the first anti-mTfR nanobodies that enter the CNS from the periphery via receptor-mediated transcytosis demonstrated by a novel, robust in vivo validation method. This method allows animal re-use and relies on target-engagement by targeting the NTSR1 expressed on hypothalamic neurons. Nanobodies coupled to NT that are able to penetrate the brain will activate the NTSR1, which causes a drop in body temperature. This hypothermic effect has been described previously as a secondary effect following intravenous administration of ANG2002, which consists of NT fused to a brain-penetrant peptide Angiopep-2, targeting the LDL receptor–related protein-1 [
50]. Next to the central hypothermic effect, Demeule et al
. also observed in rats a blood pressure reduction following peripheral NT administration. Potentially, this peripheral effect of NT might also be present in mice and could potentially influence the body temperature of the animals. However, no drop in body temperature was observed for our non-BBB crossing controls which were fused to NT. Therefore, the observed temperature drop is centrally mediated and makes NT an ideal tool to validate BBB-crossing of agents like shown by this study. Even though the hypothermic effect of NT has been described before, this is the first time this robust and unambiguous model is used to rank multiple nanobodies in terms of their brain-penetration efficiency. These first anti-mTfR nanobodies are useful tools to study drugs that are targeted to brain targets and are unable to reach the CNS on their own, in a non-invasive way. Moreover, a brain-penetrating nanobody fused to NT itself could be a potential drug candidate in various diseases where body temperature lowering could be beneficial [
52‐
55]. An example is induced hypothermia in acute ischaemic stroke, where a quick, but short duration of hypothermia is beneficial on infarct size [
56].
In order to select brain-penetrating nanobodies by receptor-mediated and not adsorptive-mediated transcytosis, an acidic peptide was incorporated between the nanobody and NT sequence, which resulted in only neutral or acidic pI values (Fig.
2a). Next, their brain-penetrating capacities were determined. Out of the 7 nanobodies tested in vivo, only one was able to reach the brain parenchyma. There are several explanations possible why most of the anti-mTfR nanobodies do not cross in vivo
. For instance their epitopes might be shielded in vivo by transferrin which has a micromolar plasma concentration and a nanomolar affinity for its receptor [
57]. It is known that binding to the apical domain of the TfR can lead to brain uptake of monoclonal antibodies [
31]. Here we show this also to be the case for the anti-mTfR nanobodies, which aligns with other agents reaching the brain via the TfR [
4,
31,
58].
Next, the hypothermic effect was assessed for dose dependent effects. Different amounts of NT were injected in the lateral ventricles of mice. Higher doses resulted in deeper and longer drops in body temperature (Fig.
3a). Subsequently, Nb62 was injected intravenously with increasing dose. Again, the amount of injected nanobody corresponded with the hypothermic effect. The negative control did not elicit a drop in body temperature at the highest dose, showing there is no passive transport of acidic nanobodies at a dose of 500 nmol/kg. These experiments provide evidence that higher concentrations of brain-penetrating nanobody in the blood leads to more nanobody entering the brain and consequently a bigger NT effect.
Subsequently, we performed a limited structure–function analysis of Nb62 using our new in vivo screening method to identify more efficient brain-penetrating nanobodies. Here, mutants of Nb62 were generated with different binding affinities, since it is known that affinity for the mTfR affects brain uptake [
27,
59]. By using site-directed mutagenesis, single site saturation libraries were generated where each amino acid in the CDR3 region of Nb62, which is most probably involved in antigen binding [
60], was substituted to all other amino acids. The nanobodies, fused to the smallest active fragment of neurotensin (NT8-13), were expressed and purified, followed by the generation of binding curves by ELISA (Fig.
4a). The single amino acid substitutions resulted in a batch of nanobodies with different binding profiles. Next, the nanobodies were intravenously injected. As can be seen from Fig.
4b, different body temperature drop profiles were observed, indicating differences in brain uptake. Generally, we see that the nanobodies with the strongest binding have the lowest brain uptake. This is in accordance with literature regarding bispecific antibodies targeting the TfR, where it is shown that high affinity monoclonal antibodies are being located to the lysosomes for degradation [
61]. In that paper, however, only two monoclonal antibodies were analyzed. All of the analyzed mutants in Fig.
4 induced a significant decrease in body temperature compared to the negative control, indicating they all penetrated the brain to some extent.
Finally, we tested different administration routes for injection of Nb62. IV, IP and SC injections give different PK profiles [
62,
63], as can be seen from the shapes of the temperature curves in Fig.
5. By comparing the three administration routes, CNS delivery is highest upon IP injection, while a more continuous delivery is reached after SC injection. This is in line with literature where they rely on TfR targeting to deliver an anti-tumor necrosis factor decoy receptor antibody to the brain [
63]. These differences in CNS delivery profiles can be contributed to the short plasma half-life of 10–20 min of nanobodies due to fast renal clearance [
64,
65]. An immediate high blood concentration following IV injection might saturate the TfRs present at the BBB, which will deliver the VHHs to the brain parenchyma. Upon recycling to the luminal side of the plasma membrane, most of the VHH has been cleared from the bloodstream, resulting in the IV profile observed in Fig.
5. IP injection might also saturate the TfRs, but due to the sustained release it is possible for the recycled TfRs to be saturated again. This would lead to a double dose reaching the brain compared to IV, which is indicated by the doubling of the temperature drop and duration of the effect (Fig.
5). SC delivery leads to the lowest plasma concentration [
63] which would not saturate the TfRs. However, the sustained release leads to a prolonged brain uptake compared to IV delivery. This is indicated by the plateau of the SC injection profile (Fig.
5).
The short half-life of nanobodies can be interesting for applications such as imaging [
22] and PET/CT assessments [
66], and can be prolonged for therapeutic applications by fusion to proteins where the nanobody adopts the half-life of the fusion protein, such as the Fc portion of a monoclonal antibody [
67,
68]. Ultimately, next to being a tool to assess the brain-penetrating potential of novel RMT targets, nanobody-NT fusions have the potential to improve the speed of cooling acute stroke patients compared to conventional methods. Here, the short plasma half-life would be beneficial, since an inverse relation between the duration of hypothermia and infarct size is observed [
56].
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.