The effects of co-culture with U251 and/or SH-SY5Y cells on the integrity of hCMEC/D3 and BBB function.

(A): Schematic diagram of the establishment process of the triple co-culture BBB model. (B): Four types of BBB models: h: hCMEC/D3 cells mono-culture model; h+S: double co-culture with SH-SY5Y cells BBB model; h+U: double co-culture with U251 cells BBB model; h+S+U: triple co-culture model. (C): The transendothelial electrical resistance (TEER) of four models, and the TEER values in day 6 were compared. Blank: the inserts without seeding cells. n=4. (D): The TEER of hCMEC/D3 and U251 cells monolayer. U: U251 cells mono-culture model. n=4. (E and F): The apparent permeability coefficient (Papp, × 10-6 cm/s) of fluorescein (NaF) and FITC-Dextran 3–5 kDa (FITC-Dex) of four BBB models. n=4. (G): The Papp (× 10-6 cm/s) of NaF and FITC-Dex across the blank inserts, and hCMEC/D3 or U251 mono-culture models. n=4. (H): The cell density of hCMEC/D3 cells after mono/co-culturing. n=3. (I and J): The mRNA levels of tight junction proteins, adherent junction proteins, and transporters. n=4. (K and L): The protein expression levels of claudin-5 (CLDN-5), ZO-1, occludin in hCMEC/D3 cells. n=4. (M and N): The protein expression levels of VE-cadherin (VE-Cad), β-catenin, and BCRP in hCMEC/D3 cells. n=4. (O and P): The correlations between the Papp (× 10-5 cm/s) of NaF and claudin-5 expression (O), or VE-cadherin expression (P). (Q and R): The correlation between Papp (× 10-5 cm/s) of FITC-Dex and claudin-5 expression (Q), or VE-cadherin expression (R). The above data are shown as the mean ± S.E.M. Statistical significance was determined using one-way ANOVA with Fisher’s LSD test or Welch’s ANOVA test. The simple linear regression analysis was used to examine the presence of a linear relationship between two variables.

Figure 1—source data1 (Raw and annotated data files for western blot)

Figure 1—source data2 (Summary of original data and analysis)

Neurons and astrocytes upregulated claudin-5 and VE-cadherin expression in hCMEC/D3 cells due to GDNF secretion.

(A): Effects of conditioned medium (CM) on claudin-5 and VE-cadherin expression. Con: the normal medium; S-CM: the CM from SH-SY5Y cells; U-CM: the CM from U251 cells; US-CM: the CM from SH-SY5Y cells co-culture with U251 cells. (B): The mRNA expression levels of neurotrophic factors in hCMEC/D3, U251, and SH-SY5Y cells. (C): Concentrations of GDNF, bFGF, IGF-1, and TGF-β in the CMs. H-CM: the CM from hCMEC/D3 cells. (D-G): Effects of GDNF (D), IGF-1 (E), bFGF (F), and TGF-β (G) on the expression of claudin-5 and VE-cadherin. The dosages have been marked in the figure. (H and I): Effects of anti-GDNF antibody on the upregulation of claudin-5 and VE-cadherin expression induced by US-CM (H) or 200 pg/mL GDNF (I). (J): Effects of 200 pg/mL GDNF and US-CM on claudin-5 and VE-cadherin expression in primary rat brain microvascular endothelial cells. (K and L): Effects of 3 μM RET tyrosine kinase inhibitor SSP-86 (SPP), and 5 μM Src family kinases inhibitor PP2 on the upregulation of claudin-5 and VE-cadherin induced by 200 pg/mL GDNF (K) and US-CM (L). (M and N): Effects of SPP on the TEER (M), the permeability of NaF, and FITC-Dex (N) of the hCMEC/D3 mono-culture BBB model treating 200 pg/mL GDNF. The TEER values of day 6 were compared. (O and P): Effects of PP2 on the TEER (O), the permeability of NaF, and FITC-Dex (P) of the hCMEC/D3 mono-culture BBB model treating 200 pg/mL GDNF. The TEER values of day 6 were compared. (Q and R): Effects of SPP on the TEER (Q), the permeability of NaF, and FITC-Dex (R) of the triple co-culture BBB model. The TEER values of day 6 were compared. (S and T): Effects of PP2 on the TEER (S), the permeability of NaF, and FITC-Dex (T) of the triple co-culture BBB model. The TEER values of day 6 were compared. The above data are shown as the mean ± S.E.M.; n=4. Statistical significance was determined using one-way ANOVA with Fisher’s LSD test or Welch’s ANOVA test.

Figure 2—source data1 (Raw and annotated data files for western blot)

Figure 2—source data2 (Summary of original data and analysis)

GDNF induced claudin-5 and VE-cadherin expression in hCMEC/D3 cells by activating the PI3K/AKT and MAPK/ERK signaling.

(A): Effects of 3 μM LY294002 (LY) on the levels of claudin-5, VE-cadherin, and p-AKT/AKT in hCMEC/D3 cells stimulated by 200 pg/mL GDNF. (B): Effects of 2 μM U0126 (U0) on the levels of claudin-5, VE-cadherin, and p-ERK/ERK in hCMEC/D3 cells stimulated by 200 pg/mL GDNF. (C): Effects of 5 μM SP600125 (SP) on the levels of claudin-5, VE-cadherin, and p-JNK/JNK in hCMEC/D3 cells stimulated by 200 pg/mL GDNF. (D): Effects of 2 μM SB203580 (SB) on the levels of claudin-5, VE-cadherin, and p-p38/p38 in hCMEC/D3 cells stimulated by 200 pg/mL GDNF. (E): Effects of anti-GDNF antibody on the GDNF-induced p-AKT/AKT and p-ERK/ERK ratios. (F): Effects of 3 μM LY on the levels of claudin-5, VE-cadherin, and p-AKT/AKT in hCMEC/D3 cells stimulated by US-CM. (G): Effects of 2 μM U0 on the levels of claudin-5, VE-cadherin, and p-ERK/ERK in hCMEC/D3 cells stimulated by US-CM. (H): Effects of 5 μM SP on the levels of claudin-5, VE-cadherin, and p-JNK/JNK in hCMEC/D3 cells stimulated by US-CM. (I): Effects of 2 μM SB on the levels of claudin-5, VE-cadherin, and p-p38/p38 in hCMEC/D3 cells stimulated by 2 US-CM. (J): Effects of anti-GDNF antibody on the US-CM-induced p-AKT/AKT and p-ERK/ERK ratios. The above data are shown as the mean ± S.E.M.; n=4. Statistical significance was determined using one-way ANOVA with Fisher’s LSD test or Welch’s ANOVA test.

Figure 3—source data1 (Raw and annotated data files for western blot)

Figure 3—source data2 (Summary of original data and analysis)

GDNF induced the claudin-5 expression in hCMEC/D3 cells by activating the PI3K/AKT/FOXO1 pathway.

(A and B): Effects of US-CM and GDNF on the phosphorylated FOXO1 (p-FOXO1)/FOXO1 ratio, total FOXO1 expression (A), cytoplasmic p-FOXO1, cytoplasmic FOXO1, and nuclear FOXO1 expression (B). (C and D): The expression levels of total and nuclear FOXO1 (C), claudin-5, and VE-cadherin (D) in hCMEC/D3 cells transfected with FOXO1 siRNA (siFOXO1). NC: negative control. (E): Effects of FOXO1 overexpression (FOXO1-OE) and GDNF on the expression levels of claudin-5, total FOXO1, and nuclear FOXO1. FOXO1-NC: negative control plasmids. (F): Effects of LY and U0 on GDNF-induced alterations of total p-FOXO1/FOXO1 ratio, cytoplasmic p-FOXO1, cytoplasmic FOXO1, and nuclear FOXO1 expression. (G): Effects of LY on the claudin-5 expression upregulated by siFOXO1. The above data are shown as the mean ± S.E.M.; n=4. Statistical significance was determined using one-way ANOVA with Fisher’s LSD test.

Figure 4—source data1 (Raw and annotated data files for western blot)

Figure 4—source data2 (Summary of original data and analysis)

GDNF induced VE-cadherin expression in hCMEC/D3 cells by activating the PI3K/AKT/ETS1 and MAPK/ERK/ETS1 pathways.

(A and B): Effects of US-CM and GDNF on total (A) and nuclear (B) ETS1 expression. (C and D): Effects of LY and U0 on 200 pg/mL GDNF-induced total (C) and nuclear (D) ETS1 expression. (E and F): Expression levels of total (E) and the nuclear ETS1 (F) in hCMEC/D3 cells after knocking down ETS1 with siRNA (siETS1). (G-I): Effects of GDNF and siETS1 on the expression of VE-cadherin and claudin-5. The above data are shown as the mean ± S.E.M.; n=4. Statistical significance was determined using one-way ANOVA with Fisher’s LSD test.

Figure 5—source data1 (Raw and annotated data files for western blot)

Figure 5—source data2 (Summary of original data and analysis)

The deficiency of brain GDNF in mice increased the permeability of BBB and reduced claudin-5 and VE-cadherin expression in mice brains.

(A): Experimental configuration of AAV-GFP (shNC) or AAV-shGdnf (shGdnf) intracerebroventricular injection. (B): Effects of brain-specific Gdnf silencing on the expression levels of GDNF, claudin-5, and VE-cadherin in the brains. (C-E): Effects of brain-specific Gdnf silencing on NaF levels in plasma (C), brain (D), and the ratio of brain to plasma (E). (F-H): Effects of brain-specific Gdnf silencing on FITC-Dex levels in plasma (F), brain (G), and the ratio of brain to plasma (H). (I-K): The expression ratios of p-AKT/AKT (I), p-ERK/ERK (J), and p-FOXO1/FOXO1 (K) in the brains of Gdnf silencing mice. (L): The expression level of ETS1 in the brains of Gdnf silencing mice. The above data are shown as the mean ± S.E.M.; n=6. Statistical significance was determined using the unpaired t-test or unpaired t-test with Welch’s correction.

Figure 6—source data1 (Raw and annotated data files for western blot)

Figure 6—source data2 (Summary of original data and analysis)

In vitro/in vivo correlation (IVIVC) assay of BBB permeability.

(A): The comparison of the estimated permeability coefficient-surface area product (PS Mono) recalculated from Papp, Mono with the observed in vivo PS values (PS Obs). (B): The comparison of the estimated permeability coefficient-surface area product (PS Triple) recalculated from Papp, Triple with the observed in vivo PS values (PS Obs). The solid line represents a perfect prediction, and the dashed lines represent the 0.5-2 folds of their observations. The PS Obs values were determined by in situ brain perfusion in rodents, which were collected from the literature.

The mechanism of neurons and astrocytes induced the integrity of brain endothelial cells.

Neurons but also astrocytes trigger the activation of PI3K/AKT and MAPK/ERK pathways in brain endothelial cells by GDNF secretion, which in turn regulates transcription factors of claudin-5 (FOXO1) and VE-cadherin (ETS1) to promote claudin-5 and VE-cadherin expression and leads to the enhancement of BBB integrity. Meanwhile, with the increase in barrier integrity, the in vitro BBB model also obtained a stronger in vivo correlation.

Key Resources Table

Initial concentrations in donor chamber and chromatographic conditions of prazosin, verapamil, and lamotrigine

The summary of mass charge ratio, extraction, initial concentrations in donor chamber

Primer sequences for qPCR for indicted genes

The target sequences for siRNA or shRNA

The unbound fraction in brain (fu, brain), the observed PS Obs, and the predicted PS (PS Pre), Papp across the hCMEC/D3 mono-culture model (Papp, Mono) and triple co-culture model (Papp, Triple) of the tested drugs. n=4.