The choroid plexus: a door between the blood and the brain for tissue-type plasminogen activator

Apart from BSA coupled to AlexaFluor⁴⁸⁸ or AlexaFluor⁵⁵⁵, which were commercially available (Life Technologies), recombinant proteins were conjugated to AlexaFluor⁴⁸⁸ or AlexaFluor⁵⁵⁵ as follows. For tPA, Actilyse^® (Boehringer Ingelheim) was first dialyzed in a 0.3 M bicarbonate buffer, pH 8.4, at 4° C for 48 h, in order to remove excipients, in particular arginine (the dialysis buffer was changed every twelve hours). Next, recombinant proteins were mixed with 1 mg of AlexaFluor⁵⁵⁵ (λ_excitation: 555 nm; λ_emission: 592 nm) or AlexaFluor⁴⁸⁸ (λ_excitation: 494 nm and λ_emission: 517 nm) Succinimidyl Ester (Life Technologies), diluted in 100μL of dimethyl sulfoxide, for 6 h at room temperature and with constant stirring. In order to remove excess AlexaFluor, the solution was dialyzed in a bicarbonate buffer first, and then in a phosphate buffered saline containing arginine hydrochloride to avoid protein precipitation.

HEK293/FRT cells were used as the eukaryotic system for production of recombinant proteins with the Flp-in system (ThermoFisher scientific). After selection with hygromycin B, stable cells were placed in production without serum (60 petri dishes per protein, diameter 10 cm), culture media were changed every 4 days for 12 days. The PCDNA3.1-Rattus tPA delta finger construct was made from the PCDNA3.1-Rattus-tPA vector. The latter is constructed like this: ATG-6His-BamhI-tPA-Stop-XhoI. tPA delta finger was amplified by PCR from amino acid 79 of tPA of PCDNA3.1-Rattus tPA (beginning of the EGF like domain) to the stop and cloned between BamHI and XhoI of the PCDNA3.1-Rattus tPA construct. We thus obtained ATG-6His-BamhI-tPA delta finger-Stop-XhoI. The human albumin cDNA was purchased from GenScript (vector pcDNA3.1 + /C-(K)DYK-Albumin, #OHu18744). We amplified the cDNA of the finger domain by PCR from a PCDNA5.1-Rattus tPA construct (finger: 36- > 78), the domain was amplified with ATG, the signal peptide for secretion and the 6his tag of the PCDNA5.1-Rattus tPA construct. It was then cloned between NheI and HindIII of the pcDNA3.1 + /C-(K)DYK-Albumin vector. Finally we obtained an amino terminal cloning: ATG-6His-Finger-Albumin-Stop.

The 6xHistidine fused to the N-terminal of ΔF-tPA and of F-BSA allows purification of recombinant proteins by high performance liquid chromatography using Nickel-Nitrilo-triAcetic Acid affinity columns (Ni–NTA, GE Healthcare) and a mobile phase imidazole gradient. Positive elutions were pooled, concentrated and dialyzed in phosphate buffer. The ΔF-tPA and F-BSA proteins were then coupled to AlexaFluor⁴⁸⁸ and AlexaFluor⁵⁵⁵, respectively, as described above, except that there was no initial dialysis.

Eight weeks old male Swiss mice (35–45 g, Centre Universitaire de Ressources Biologiques, Normandy University, Caen, France) were housed at 21 °C in a 12-h light/dark cycle with food and water ad libitum. All procedures with anaesthesia were performed by an initial exposure to 5% isoflurane, followed by a maintaining phase of 1.5–2% isoflurane 30%O₂/70%N₂O.

Anesthetized mice were placed in a stereotaxic frame. Recombinant proteins were injected in the tail vein, in 200µL of saline (0.9% NaCl) over approximately 5 s.

Tissue sections were rinsed and rehydrated in PBS and are then incubated in a PBS solution containing 0.25% of Triton X-100 overnight at room temperature and in a humid atmosphere, with an anti-TTR antibody to label the CPECs (1:1000; Abcam 9015) or anti.

CD-31, a marker of the endothelium (1:1000; BD Pharmingen 553,370). The next day sections were incubated in PBS containing 0.25% Triton ™ X-100 and Fab'2 fragments directed against the host species primary antibody (1:1000; Jackson ImmunoResearch) for 90 min at room temperature and humid atmosphere. Sections were covered with an assembly medium (Fluoromount-G^®, SouthernBiotech) containing, or not, DAPI and a coverslip. The slides were then observed and microphotographed with an inverted confocal microscope (Leica TCS SP8) to be analysed with Fiji Is Just ImageJ. For each animal, a dozen images were acquired on 5 Sects. (4 of the brain and one of the cerebellum). On each image, the CPECs were manually counted, thanks to TTR immunostaining, as well as the spots of fluorescence within CPECs.

A subset of slices was also stained with an antibody against NeuN (mouse monoclonal antibody MAB377; 1:500; Merck Millipore, Alexa⁶⁴⁷-conjugated secondary antibody donkey anti-mouse, 1:800, Jackson ImmunoResearch) to determine if tPA⁵⁵⁵ can be detected in the brain parenchyma (30 min after intravenous administration).

After addition of loading buffer (Sucrose, SDS, Bromophenol Blue), the CSF samples (5 µl) were subjected to electrophoresis (SDS-PAGE) under non-reducing conditions (200 V, 50 min) on a 10% polyacrylamide gel. The fluorescence signal was measured in the gel using an ImageQuant™ LAS 4000 camera (GE Healthcare), and the signal quantification was done with Fiji is just ImageJ. Alternatively, tPA was immunodetected after transfer onto a PVDF membrane, using a goat anti-tPA antibody (1:500) and a peroxydase-conjugated secondary antibody (1/160000).

The CPECs were isolated based on a protocol described in rats by Monnot & Zheng [30], with some modifications. After cervical dislocation, preceded by deep anaesthesia with 5% isoflurane (in 30% O₂ / 70% N₂O), and recovery of the brains of ten 8-week-old Swiss mice, the plexuses from 3rd and 4th ventricles and lateral ventricles were dissected under a binocular lens. The plexuses were diluted in 1 ml of ice-cold HBSS (Hank's Balanced Salt Solution, Sigma) and underwent enzymatic digestion with pronase (Pronase E from Streptomyces griseus, Serva Electrophoresis) at 2 mg/ml for 20 min at 37 °C and 5% CO₂. Digestion was stopped by adding 4 mL of ice-cold HBSS. The plexus clusters were centrifuged at 4 °C for 5 min at 800g, the supernatant, which mainly contains debris, was removed, the pellet suspended in 6 ml of ice-cold HBSS, centrifuged again at 800 g for 5 min. The pellet was then suspended in 2 ml of growth medium (sodium pyruvate 110 μg/ml, glutamine 2 mM, fetal calf serum 10%, epidermal growth factor 0.01 mg/ml and 1% of diluted penicillin/streptomycin in Essential Medium Modified by Dulbecco, DMEM) at 37 °C then mechanically dissociated using a syringe with a 21G needle. The dissociated cells were firstly seeded in a 35 mm dish and incubated at 37 °C and 5% CO₂ for 4 h so that the fibroblasts and macrophages adhere to the plastic. After 4 h, CPECs in suspension in the dish were collected and seeded in 24-well plates previously treated with type I collagen from rat tail (Cultrex ™) at 0.1 mg/mL in acetic acid 0.02 M. After 48 h, the plates were rinsed with PBS and the medium was replaced by medium supplemented with cytosine arabinoside (Ara-C). Cells were then rinsed, and the medium changed every other day until use after 8 days in vitro. The continued presence of the inhibitor of DNA synthesis Ara-C specifically inhibits the growth of contaminating cells such as stromal fibroblasts. Unlike CPECs, these cells express nucleoside transporters unable to distinguish ribose and arabinose residues and thus incorporate Ara-C into their genomic DNA, which results in specific suppression of their growth [25].

Cultured CPECs were rinsed twice with PBS at 37 °C, then maintained for 1 h in a serum-free medium, and treated with tPA⁵⁵⁵ (at the indicated doses/durations). After treatment, cells were rinsed three times with cold PBS. Proteins were extracted at 4 °C in a 50 mM Tris–HCl buffer, pH 7.4, 150 mM NaCl and 0.5% Triton ™ X-100 (TNT buffer) supplemented with protease and phosphatase inhibitors 1 µg/mL. After centrifugation (12000g at 4 °C for 20 min), the supernatants were collected and then assayed using the Pierce BCA Protein assay kit. After adding loading buffer (Sucrose, SDS, Bromophenol Blue), the total proteins were separated according to their molecular mass by electrophoresis (SDS-PAGE) on a 10% polyacrylamide gel under non-reducing conditions (200 V, 50 min). The fluorescence of tPA⁵⁵⁵ was detected directly in the gel with an ImageQuant™ LAS 4000 camera, and the signal quantification was done with Fiji Is Just ImageJ.

LRP1 Stealth RNAi™ oligoribonucleotides were transfected at a concentration of 20 nM into CPECs cultures using Lipofectamine 2000 (2 µg, Invitrogen). Sequences were: siLRP1_01 sense UGGCUGACGGGAAACUUCUACUUUG; siLRP1_01 antisense CAAAGUAGAAGUUUCCCGUCAGCCA; siLRP1_02 sense CACACCCAUUUGCCGUGACACUGU; siLRP1_02 antisense UACAGUGUCACGGCAAAUGGGUGUG; siLRP1_03 sense CCAAGGUGUGAGGUGAACAAGUGUA; siLRP1_03 antisense UACACUUGUUCACCUCACACCUUGG.

Stealth RNAi™ siRNA Negative Control, Medium GC negative control duplex (Invitrogen™) was used as the control condition (scramble). After 48 h, cells were rinsed 3 times with PBS, and total proteins were recovered. The efficacy of extinction was checked by western blot: after adding loading buffer (Sucrose, SDS, Bromophenol Blue), the total proteins were separated according to their molecular mass by electrophoresis (SDS-PAGE) on a 10% polyacrylamide gel under non-reducing conditions (200 V, 50 min). Then, Proteins were transferred (120 min, 200 mA) onto a cellulose membrane in a humid environment (24 mM Tris, 192 mM glycine, 20% ethanol). Membranes were incubated in a 2% TTBS-BSA solution (Tween Tris Buffered Saline, 10 mM Tris, 200 mM NaCl, 0.1% Tween—2% BSA, pH 7.4) for 1 h and then incubated overnight at 4 °C with the primary anti-LRP1 antibody (Abcam 92,544) diluted 1:1000 in the TTBS-BSA 2% buffer. Membranes were incubated with the secondary antibody coupled to peroxidase, diluted 1:50,000 in 2% TTBS-BSA, for 1 h at room temperature. The detection was performed using the Pierce ECL Plus Western blotting substrate kit, membranes were then placed in the ImageQuant ™ LAS 4000 camera and the signal quantification done with Fiji Is Just ImageJ.

Whole CPs were collected from mice (8-weeks-old Swiss males) and were quickly washed with PBS at 37 °C. They were then exposed for the indicated time to the specified Alexa-conjugated protein. After treatment PCs were rinsed three times with PBS and fixed with 4% paraformaldehyde in PBS 0.1 M and pH 7.4, for 15 min. PCs were mounted in Fluoromount-G® then micro-photographed with an inverted confocal microscope (Leica TCS SP8) to be then analysed with Fiji Is Just ImageJ. The various proteins were quantified by their fluorescence within the CPECs, which were delimited manually.

Results are expressed as mean ± the Standard Error of the Mean (SEM). The samples being unpaired and having low numbers a Mann–Whitney U test was used. Statistical analyses were performed using Mann–Whitney test with GraphPad software. Data were considered statistically different if probability values (p) were at least < 0.05.

The potential transport of tPA through the CPs and its underlying mechanisms were investigated by tracking human recombinant tPA tagged with AlexaFluor⁵⁵⁵ (tPA⁵⁵⁵) or AlexaFluor⁴⁸⁸ (tPA⁴⁸⁸). We first used a model of primary culture of mouse CPECs (based on [30]. After 8 days, cultured CPECs display characteristics identical to those of the CPs in situ (Fig. 1A), namely a cuboid morphology, the expression of the marker transthyretin (TTR) and the presence of tight junctions (occludin staining). Cultured CPECs were able to uptake extracellular tPA⁵⁵⁵ (40 nM) in a time-dependent process (visible as increasing numbers of intracellular Alexa⁵⁵⁵ punctiform stainings, Fig. 1B). Measures of tPA⁵⁵⁵ levels in protein extracts from cell monolayers resolved by electrophoresis (Fig. 1C) confirmed these observations, with a sharp rise up to 5 min after initiating exposure, and an ongoing increase up to 60 min (after 5 min, the amount of internalized tPA⁵⁵⁵ already corresponded to 41.84% of what was internalized after 60 min). CPECs were also exposed for 15 min to doses of tPA⁵⁵⁵, increasing from 20 to 400 nM. Intracellular fluorescent dots accumulated when increasing doses of tPA⁵⁵⁵ (Fig. 1D). In-gel quantifications confirmed that the uptake of tPA⁵⁵⁵ in CPECs rose with the amount of tPA⁵⁵⁵ applied in the medium (after exposure to 20 nM, the amount of internalized tPA⁵⁵⁵ corresponded to 11.72% of what was internalized with 400 nM) (Fig. 1E). To support these findings, freshly isolated explants of mouse CPs were exposed to tPA⁵⁵⁵ and BSA⁴⁸⁸, used as a control (Bovine Serum Albumin is also a circulating protein, with a molecular weight comparable to that of tPA, as shown on Fig. 2A). CPECs exhibited strong accumulation of tPA⁵⁵⁵, but barely no signal for BSA⁴⁸⁸ (Fig. 2B, C). Inverting fluorophores on our tracked proteins gave similar results (Fig. 2D–F). In addition, we observed that 30 min after their intravenous co-injection (50 µg each), while tPA⁵⁵⁵ was detected in CPECs, Alexa⁴⁸⁸ extravasated out of choroidal vessels, but remained in the stroma without being taken up by CPECs (Additional file 1: Figure S1). Together, these in vitro and ex vivo results show that, CPECs uptake tPA, in a time- and dose-dependent manner, but do not uptake BSA.

Then, mice were injected intravenously with 50 µg of tPA⁵⁵⁵ and 50 µg of BSA⁴⁸⁸, and CSF and brains were harvested 10, 30, 60 or 120 min later (Fig. 3A). In CPs, confocal microscopy (Fig. 3B) revealed dots of tPA⁵⁵⁵ as early as from 10 min, being more numerous at 30 and 60 min, and then virtually absent 120 min after intravenous injection. A z-stack of confocal images clearly shows that dots of tPA⁵⁵⁵ were located inside CPECs (Fig. 3C). We also found that tPA was internalized in clathrin-coated endocytotic vesicles, as it was co-localized with the marker AP2 (Pearson’s coefficient of 65%; Additional file 1: Figure S2). We did not find any BSA⁴⁸⁸ in CPs after intravenous injection. We quantified the number of TTR immune-positive cells with dots of tPA⁵⁵⁵ as an index of the internalization of tPA by CPECs (Fig. 3D). This index followed a bell curve shape, peaking 30 min post-injection. After two hours, tPA was no longer detected in the CPs, suggesting a passage of tPA⁵⁵⁵ from the blood to the CSF. Accordingly, an immunoblot analysis on CSF samples harvested at different time points, revealed the appearance of tPA in the CSF starting between 30 and 60 min post intravenous injection (Fig. 3E). Since the antibody used is specific for human tPA and does not recognize murine tPA, this confirms that vascular tPA crosses the CPs and reaches the CSF. Of note, when administered in the ventricle, tPA⁵⁵⁵ was also detected in the CPECs, suggesting that CPs also allow the clearance of tPA from the CSF (Additional file 1: Figure S3). We also found that after an intravenous injection, the decline in tPA⁵⁵⁵ levels in the blood parallels the increase in tPA⁵⁵⁵ levels in the CSF (Additional file 1: Figure S4). Moreover, to ensure that the observed transport occurs directly from the choroidal circulation to the CPECs, and not via indirect pathways, we administered tPA⁵⁵⁵ in the internal carotid and demonstrated that very shortly after (5 min), dots of tPA⁵⁵⁵ were detected in the stroma and in CPECs (Additional file 1: Figure S5). Importantly, in barrier-protected parenchymal areas, we could detect tPA⁵⁵⁵ in neurons, suggesting that once in the CSF, tPA can enter the parenchyma (Fig. 3F). In barrier-free areas, both tPA⁵⁵⁵ and BSA⁴⁸⁸ reached the brain parenchyma (exemplified here in the median eminence: 30 min after injection, fluorescent BSA and tPA diffused outside the vascular system, Additional file 1: Figure S6).

Overall, these results show that after an intravenous injection, tPA is internalized by CPECs and is then transported to the CSF. This suggests that CPs are a potential route of entry for tPA into the brain parenchyma.

Transcytosis/uptake of tPA has been attributed in various cell types to members of the LRP family, in particular to LRP1 [3, 4, 14]. A 3D projection of confocal images acquired after intravenous injection of tPA⁵⁵⁵ in mice and coupled to the immunodetection of LRP1 on TTR positive cells supports the hypothesis that LRP1 could be responsible for tPA transport by CPECs. Indeed, we found dots of tPA⁵⁵⁵ close to LRP1 immunolabellings at the basolateral pole of CPECs (Fig. 4A). In cultured CPECs, that also express LRP1 (Fig. 4B), we first found that tPA⁵⁵⁵ internalization was likely an active mechanism. Indeed, when exposed to 40 nM of tPA⁵⁵⁵ for 15 min, CPECs protein extracts displayed fluorescent signals when exposure was performed at 37 °C but not when performed at 4 °C (data not shown). To evaluate the participation of a member of the LRP family, we compared the internalization of tPA by CPECs in the absence or presence of RAP (Receptor Associated Protein, 500 nM), a competitive LRP antagonist (Fig. 4C–D). CPECs co-exposed to tPA⁵⁵⁵ and RAP presented fewer tPA⁵⁵⁵ dots than cells exposed to tPA⁵⁵⁵ alone (Fig. 4C). This result was confirmed by electrophoresis, showing a 41.22% reduction in tPA⁵⁵⁵ internalization induced by RAP co-treatment (Fig. 4D). To evaluate the participation of LRP1 in the uptake of tPA by the CPECs, we knocked-down its expression with a siRNA strategy. The extinction of LRP1 expression was 58.77% effective (Fig. 4E). When compared to control CPECs or CPECs transfected with a scramble siRNA, CPECs transfected with LRP1 siRNA internalized 32.55% less tPA⁵⁵⁵ (Fig. 4F). Importantly, in vivo, we confirmed the role of LRP1 in transporting tPA in CPECs. Indeed, there was around 60% loss of tPA⁵⁵⁵-related fluorescence in CPECs when RAP (90 µg) was co-administered intravenously with tPA⁵⁵⁵ (50 µg), compared to tPA⁵⁵⁵ alone (Fig. 4G–H).

Among its five functional domains (Fig. 5A), the finger domain of tPA is known to be the one interacting with LRP1. To assess its involvement in the passage of CPECs, we produced a tPA devoid of its finger domain (ΔF-tPA), and alternatively, we grafted the finger domain of tPA to albumin (F-BSA) (Fig. 5E). Both mutant proteins were also tagged with an Alexa Fluor for their tracking (Fig. 5B, F). Ex-vivo, CP explants were exposed to 40 nM of tPA⁵⁵⁵ and with increasing doses of ΔF-tPA⁴⁸⁸ (40 nM, 200 nM or 400 nM) for 15 min (Fig. 5C, D). As expected, tPA⁵⁵⁵ was internalized by CPECs whereas ΔF-tPA was not, even with a tenfold higher concentration. The finger domain of tPA is therefore necessary for internalization by CPECs. Interestingly, fusing the finger domain of tPA to BSA conferred the new ability to BSA to be internalized by CPECs ex vivo (Fig. 5G, H).

These ex vivo observations were confirmed in vivo, in mice injected intravenously with 50 µg of tPA⁵⁵⁵/ ΔF-tPA⁴⁸⁸ or with 50 µg of F-BSA⁵⁵⁵/ BSA⁴⁸⁸ (Fig. 6A). A confocal microscopy analysis (Fig. 6B) revealed dots of tPA⁵⁵⁵ in TTR positive cells, 30 and 60 min after injection. However, virtually no dots were detected for ΔF-tPA⁴⁸⁸. We quantified the number of TTR immune-positive cells with dots of tPA⁵⁵⁵ and dots of ΔF-tPA⁴⁸⁸ as an index of their internalization by CPECs (Fig. 6C). This index was null for ΔF-tPA⁴⁸⁸ and peaked 30 min post-injection of tPA⁵⁵⁵, with 73.09% more cells containing tPA⁵⁵⁵ than at 60 min. Interestingly, the confocal microscopy analysis (Fig. 6D) revealed dots of F-BSA⁵⁵⁵ in TTR positive cells, 30 and 60 min after IV injection, while no dots were detected for BSA⁴⁸⁸. The corresponding quantification confirmed that BSA per se, cannot be uptaken by CPECs, but that grafting the finger domain of tPA then allows the internalization of the fusion protein (Fig. 6E). This index followed a bell curve shape, with a peak of internalization of tPA⁵⁵⁵ reached 30 min post-injection, with 56.84% more cells containing tPA⁵⁵⁵, than at 60 min and no detection of ΔF-tPA⁴⁸⁸. Also important, as confirmed by electrophoresis of corresponding CSF samples, F-BS^A555 can cross the BCSFB, like tPA⁵⁵⁵ (Fig. 6F, G). The finger domain of tPA is thus an interesting strategy to allow therapeutic fusion proteins reaching the CSF and the brain parenchyma.