All chemicals and reagents were purchased from Merck (Soeborg, Denmark.) unless otherwise indicated.

Wild-type (WT) Sprague Dawley rats were purchased from Janvier Labs. The Sorl1 ^−/− (KO) Sprague Dawley rats were custom-made by Horizon Discovery for breeding at the Animal Facility at Aarhus University. Sorl1 ^−/− has a seven base pair deletion in exon 3, resulting in a stop codon which leads to the subsequent complete destruction of the Vps10p-D folding in the SorLA protein. No mRNA splice variants or protein fragments were detected by RT-PCR or Western blot analysis. The KO-rats were kept in standard cages (Scanbur) in a pathogen-free and temperature-controlled environment. Cages were cleaned weekly and supplied with nesting and bedding material, a wooden stick, and a tunnel. Chow (Altromin 1324, Brogaarden, Denmark) and water were provided ad libitum. The light was provided during daytime in 12 h light/12 h dark cycle. Animals were treated according to the European and Danish animal experimentation legislation (directive 2010/63/EU). Animal experiments were permitted by the Danish Animal Experiments Inspectorate under permission number 2017-15-0201-01192.

Primary BECs, astrocytes, and pericytes were isolated and cultured, as reported previously (Nakagawa et al., 2009; Toth et al., 2022). For BECs and pericytes, cell cultures of brain microvessels were isolated from the forebrain of 3-week-old rats. The meninges were removed, and the grey matter was minced into small pieces. 1 mg/ml collagenase CLS2 (Worthington, United States) in Dulbecco’s modified Eagle medium (DMEM) for 1.5 h at 37°C was used for tissue digestion. The myelin contamination was removed by centrifugation (1,000×g, 20 min) in 20% bovine serum albumin–DMEM. The remaining microvessels were further digested with 1 mg/ml collagenase-dispase (Roche) in DMEM for 1 h. Brain capillaries were separated on a 33% continuous Percoll gradient (1,000×g, 10 min) and collected. The isolated microvessels were frozen in 10% dimethyl sulfoxide - fetal bovine serum mixture until further use.

Brain microvessel fragments contain pericytes as well as BECs. The survival and proliferation of the pericytes were ensured by specific culture conditions. Pericytes typically overgrow BECs during 2 weeks of culturing in DMEM supplemented with 10% FBS and antibiotics in uncoated culture dishes. (Nakagawa et al., 2007). The culture medium was changed twice a week.

Cultures of astrocytes were prepared from 1–2 days-old newborn rats. The brains were mechanically dissociated into cortical pieces and plated in poly-L-lysine coated T75 flasks (Thermo Fisher Scientific, Roskilde, Denmark). The cells were grown for a minimum of 3 weeks in DMEM containing 1% penicillin/streptomycin, and 10% FBS. The culture was shaken and washed with phosphate buffered saline (PBS) twice a week to remove contaminating microglias. After 3 weeks, astrocytes were frozen 10% dimethyl sulfoxide - fetal bovine serum mixture until further use. Before co-culture, astrocytes were thawed and grown as in 12-well plate. In 2 weeks, the culture became confluent, and the medium was changed every 3 days from now on.

For BECs, T75 flasks were coated with collagen IV (100 μg/ml) and fibronectin (100 μg/ml) 10% plasma-derived bovine serum (First Link Ltd., Wolverhampton, United Kingdom). Basic fibroblast growth factor (1 ng/ml), heparin (15U), insulin-transferrin-selenium, and gentamicin (5 μg/ml) were added to the DMEM/F12 medium. To the media of the towed rat brain capillaries, puromycin (4 μg/ml) was supplemented for the first 4 days to obtain a pure culture of BECs and remove contaminating cells. On the fifth day, the cells were passaged onto 1.12 cm² polycarbonate Transwell® filter inserts with .4 μm pore size (Costar, Corning, Kennebunk, ME, United States) at a density of 1 × 10⁵ cells/cm². On the following day, the non-contact co-culture BBB models were established. The inserts of the BECs were placed in the 12-well culture dishes of astrocytes. On the third day of co-culture, media of BECs and astrocytes were supplied with differentiation factors: 550 nM hydrocortisone, 250 μM 8-(4-chlorophenylthio)adenosine-3`,5`-cyclic monophosphate (cAMP) and 17.5 μM RO-201724. On the next day, permeability experiments were carried out, and the samples were collected for further investigation. The tightness of the in vitro BBB models was checked visually and by measuring TEER.

This study used three different models: WT-BBB, KO-BBB, and a mixed genotypic BBB model (mixed-BBB). The WT-BBB model was used as a control, which contained WT-BECs on the top of WT-astrocytes. Similarly, in the KO-BBB model, KO-BECs were co-cultured with KO-astrocytes. For the mixed-BBB model, the inserts of the WT-BECs were placed on top of KO-astrocytes. The mixed- or the KO-models were created and maintained parallelly with the WT-BBB model. All the cell cultures and BBB models were held in a humidified incubator with 5% CO₂ at 37°C.

Astrocytes and pericytes were cultured in T75 flasks while BECs were grown on 75 mm diameter polycarbonate Transwell® inserts (Costar, Corning) in co-culture with astrocytes as described in ‘In vitro BBB models’. Before harvesting all the cell cultures, they were washed twice with ice-cold PBS. NucleoSpin® RNA Isolation Kit (Macherey-Nagel, Düren, Germany) was used to extract total RNA by following the manufacturer’s instructions. Material from one insert or one T75 flask was used to generate each RNA sample, and the cells for the 3 parallel samples originate from 3 different cell isolations. NanoDrop 2000c UV-VIS Spectrophotometer (ThermoFisher Scientific, Waltham, MA, United States) was used to measure mRNA purity. For further analysis, mRNA samples with a 260/280 ratio over 2.00 were used. SDSC BiologyWorkbench and NCBI retrieved from Thermo Fisher Scientific were used to design primers. To perform reverse transcription of mRNA, High capacitycDNA reverse transcription kit (4368814; from Thermo FisherScientific) was used according to the manufacturer’s protocol. For reverse transcription of mRNA to cDNA, PTC-200 ThermalCycler (MJ Research) was applied with Riboloc R1 (Thermo Scientific) RNAse inhibitor. The following primer pairs were used: SorLA fwd 5′-AAG​GTG​CAC​AAC​ACC​AAT​GA-3′, rvs 5′-GTG​CCC​AAG​AAT​CAC​ACC​TT-3,′ product size 117 bps; Glyceraldehyde-3-Phosphate Dehydrogenase (GAPDH) fwd 5′-AGA​CAG​CCG​CAT​CTT​CTT​GT-3’, rvs 5′-CTT​GCC​GTG​GGT​AGA​GTC​AT-3′, product size 207 bps; ß-actin fwd 5′-TGT​CAC​CAA​CTG​GGA​CGA​TA-3′, rvs 5′-CTG​GGT​CAT​CTT​TTC​ACG​GT-3′, product size 139 bps. Primer efficiencies were calculated by the equation E = 10 (−1/slope). The slope was created from a ten-fold serial dilutions calibration curve. All the primer pairs have regular melting curves, and their efficiency range was between 1.9 and 2.1. Light Cycler 96 Instrument (Roche, Basel, Switzerland) was used for the RT-qPCR analysis. The used amount of cDNA corresponds to .31 μg RNA per reaction. 20 μl reaction mixture consisted of 2 μl cDNA, 5 μl primer (a mix containing 2 μM of forward and reverse primers), 3 μl water (PCR-grade), and 10 μl mastermix. Cycling conditions were as follows: pre-incubation at 95°C for 600 s, 45 cycles 95°C for 10 s, 55°C for 10 s, and 72°C for 20 s. To verify the specificity of amplicons, melting curves were determined for all samples. All samples were run in triplicates. All PCR data were normalized to the geomean of reference genes of β-actin and GAPDH. Genes were considered not to be expressed above 35 average cycle-threshold value.

Cell lysates was prepared by washing the cells three times in PBS and lysed in ExB lysis buffer (150 mM NaCl, 20 mM MgCl₂, 20 mM CaCl₂, 100 mM HEPES, 1% TritonX-100, complete^TM protease inhibitor). From each cell lysate a minimum of 5 μg of was loaded on a 4–12% polyacrylamide gel (GeneScript, New Jersey, United States). The protein samples were transferred to a nitrocellulose membrane (Invitrogen iBlot) and the membrane was blocked in 5% skimmed milk, .01 M Tris-HCl, .15 M NaCl, and .1% Tween 20, pH 7.6 in buffer solution and incubated with primary antibodies overnight at 4°C. For primary antibodies, mouse anti-LR11 (SorLA) antibody (612,633, 1:1,000, B.D. Transduction Laboratories San Diego, CA, United States) and mouse anti-β-actin antibody (A5441, 1:1,000) were used. Next, the samples were treated with secondary horseradish peroxidase-conjugated anti-mouse IgG antibodies (1:2000) for 1 h at room temperature. The bands were detected using ECL (G.E. Healthcare) or SuperSignal (Thermo Scientific, Rockford, United States) according to the manufacturer’s protocol and visualized using LAS 4000 (Fujifilm). To improve clarity for representative Western blot images, contrast and brightness adjustment were applied equally across the entire picture.

EVOM resistance meter using EndOhm-12 cup system (World Precision Instruments Inc, Sarasota, FL, United States) was used to measure TEER. TEER values were calculated relative to the surface area (Ω × cm²) and shown in each experiment as a percentage of the WT-group.

To evaluate the permeability across endothelial cell layers, a small molecular weight marker, sodium fluorescein (376 Da) was used. BECs have grown to a confluency of 100% on Transwell inserts as described above. For the permeability experiments, the culture medium was replaced by DMEM phenol red-free medium (Gibco, Life technologies). In the upper chambers (luminal compartment) the 500 ml medium was supplemented with 10 μg/ml sodium fluorescein and 1% bovine serum albumin. The abluminal (lower) compartment contained the 1.5 ml medium without any supplements. The plates were gently shaken (150 rpm) on a horizontal shaker in a 37°C incubator with 5% CO₂. To minimize luminal to abluminal transport, every 20 min the inserts were transferred to a new well for 1 h. The concentrations of the marker molecules in samples from the abluminal (lower) compartments were determined by fluorescent spectrophotometry (emission: 525 nm, excitation: 440 nm) with Enspire multimode plate reader (PerkinElmer Inc., Lillestrøm, Norway). Flux across coated inserts without cells was also measured. The apparent permeability coefficient (P_app) was calculated by the following equation (Youdim et al., 2003). P a p p c m s = ∆ C A ∗ V A A ∗ C L ∗ ∆ t

Δ[C]_A—concentration difference of the tracer in the abluminal compartments at the end of the experiment.

V_A—volume of the abluminal compartment (1.5 ml).

A—surface area available for permeability (1.1 cm²).

[C]_L—concentration of the tracer in the luminal compartments at beginning of the experiment.

Δt—length of the experiment (1 h).

The P_app values of the samples are compared in each experiment to mean value of the WT-group and shown as its percentage.

For immunostaining, astrocytes were grown on coated cover glasses while BECs were cultured on polycarbonate Transwell® filter inserts in co-culture with astrocytes as described above. All steps of immunocytochemistry were carried out at room temperature. Confluent cell layers were fixed with 4% paraformaldehyde in PBS for 20 min. The samples were permeabilized in .2% Triton X-100 in PBS for 10 min and blocked in 2% bovine serum albumin in PBS for 20 min. Primary and secondary antibodies were diluted in 1.5% bovine serum albumin and .05% Triton X-100. The following primary antibodies were applied for 1 h: mouse anti-LR11 antibody (612,633, 1:300 BD Transduction Laboratories), rabbit anti-claudin 5 antibody (SAB4502981, 1:300), rabbit anti-ZO-1 antibody (61–7300, 1:300; Invitrogen, Carlsbad, CA, United States), rabbit anti-occludin antibody (71–1,500, 1:300; ThermoFischer Scientific, Rockford, IL, United States), mouse anti-glial fibrillary acidic protein antibody (GA5; 3670, 1:200; Cell Signaling Techonology, Danvers, MA, United States). The samples were treated with secondary antibodies (Alexa Fluor-488 conjugated anti-mouse (1:400) and Alexa Fluor-568 conjugated anti-mouse or anti-rabbit (1:400)) for 1 h. For nuclei staining, Hoechst 32,528 (.5 μg/ml) was diluted in distilled water and applied for 10 min as a separate step on the samples. To remove unbound antibodies, samples were washed three times for 5 min in .02% Triton-X100 in PBS in between the stages. As the last step, samples were mounted on glass slides with Dako fluorescence mounting medium (Dako, Glostrup, Denmark). Confocal images were taken using an inverted Olympus IX-83 fluorescent microscope with Yokogawa confocal spinning unit, and Andor iXon Ultra 897 camera, Olympus UPlanSApo, 60x/1.20 NA water lens. More than three images of randomly selected fields of each sample were acquired. On the compared representative images shown in this paper, the brightness and contrast adjustments were applied to the same extent in ImageJ software.

Line intensity scan analysis was performed on images of BECs in Fiji software as described by Toth et al. (Toth et al., 2022). Minimum 3–5 cells were measured on each image by drawing lines in the cytoplasm and the cell border. The intensity parameters were counted along the lines. At the cell border and in the cytoplasm, the measured fluorescent intensity values were compared between the WT- and the mixed-groups from the same experiment. The values are shown as a percentage of the WT-groups (100%) from each experiment.

All studies were conducted at least three times using a minimum of three samples for each group. All curve fitting and statistical analyses were performed in Graph Pad Prism version 7.0 (GraphPad Software, La Jolla, CA, United States). Values were compared using Student’s t-test. Data are reported as means ± standard error of the mean (SEM). Changes were considered statistically significant at p ≤ .05. Student’s t-test was performed on the normalized data.

To investigate what role SorLA plays at the BBB, we first analyzed the expression of Sorl1 gene in the cell types constituting the brain capillaries from wild-type rats. The Sorl1 gene was highly expressed in astrocytes. However, its expression level was negligibly low in BECs and pericytes (Figure 1A). To further validate Sorl1 gene`s expression, we investigated the protein level of SorLA in cultured BECs, astrocytes, and pericytes isolated from wild-type rats. In agreement with the gene expression data, Western blot showed that SorLA is present in the cell lysate of astrocytes, and it is not detectable in BECs and pericytes (Figure 1B). Accordingly, SorLA was only detectable in WT-, but not in KO-astrocytes (Figure 1C). As the next step, we characterized the subcellular localization of SorLA by immunocytochemistry (Figure 1D). Positive staining for SorLA could be observed only in the WT- and not in the KO-astrocytes. SorLA localized in vesicles of WT-astrocytes with a solid perinuclear placement. Both astrocytes expressed glial fibrillary acidic protein and showed a multipolar shape with long processes.

Astrocytes play a vital role in maintaining the BBB’s integrity. Therefore, we examined the tightness of the endothelial monolayer in vitro (Figure 2). As illustrated by Figure 2A, WT-BECs were co-cultured with WT-astrocytes; KO-BECs were maintained with KO-astrocytes. The TEER values reflect the tightness of the endothelial monolayer and indicate the permeability of the intercellular junctions for sodium ions (Srinivasan et al., 2015). The tightness of the KO-endothelial monolayer (1,047 ± 343 Ω × cm²) was three-fourth of the WT-endothelial monolayer`s (1,373 ± 419 Ω × cm²) (Figure 2B). Accordingly, The KO-BBB model was leaky for sodium-fluorescein paracellular permeability marker (Figure 2C). We measured 1.7 times higher permeability (3.41 ± 1.49 × 10^–6 cm/s) across the KO-endothelial layer than across the WT-endothelial layer (1.86 ± .63 × 10^–6 cm/s).

SorLA is expressed only in astrocytes and not in BECs (Figures 1A, B). To prove that the KO-astrocytes compromise the integrity of the endothelial monolayers, we constructed a mixed-BBB model where WT-BECs were co-cultured with KO-astrocytes (Figure 2D). The tightness and permeability of the mixed models were similar to the KO-BBB models. The TEER values of the endothelial layer in the mixed-BBB model were 20% lower (1,105 ± 115 Ω × cm²) than the WT ones (1,389 ± 145 Ω × cm²) (Figure 2E). Accordingly, the measured permeability for sodium-fluorescein in the mixed-BBB model (3.57 ± 1.79 × 10^–6 cm/s) was 1.8 times higher than in the WT-BBB model (1.86 ± .63 × 10^–6 cm/s) (Figure 2F).

Since SorLA is expressed only in astrocytes (Figure 1), we continued to investigate how KO-astrocytes compromise the integrity of the barrier using the mixed BBB model. Altered tightness and permeability (Figures 2E, F) can indicate changes in the junctional proteins of the BECs (Lochhead et al., 2020). Therefore, we have investigated the localization of the most well-characterized tight junction proteins, namely claudin-5, occludin, and ZO-1 (Figure 3) in the BECs of mixed- and WT-BBB models (Figure 2D). At the cell border and in the cytoplasm, the fluorescence intensity of the junctional proteins was measured to quantify the observed differences. The immunofluorescent staining of claudin-5 was weaker in the BECs of the mixed-BBB model (Figure 3A). Accordingly, the measured intensity values both at the cell border and in the cytoplasm were three-fourth of the BECs` from the WT-BBB model. The other two tight junction proteins also showed lower intensity in the mixed-BBB model. However, this change did not prove to be significant (Figures 3B, C). In any of the staining, we could not observe any morphological changes in the BEC when they were co-cultured with KO-astrocytes (Figure 3). Overall, the results indicated that a disturbance in claudin-5 caused the observed BBB leakage (Figure 2).