In Vitro Analysis of TWEAK/Fn14 Axis in the Blood–Brain Barrier Models during Oxygen–Glucose Deprivation and Reoxygenation

Abstract

Upregulation of tumor necrosis factor-like weak apoptosis-inducing factor (TWEAK) and its receptor fibroblast growth factor-inducible 14 (Fn14) was observed in stroke patients and murine models, contributing to neuronal apoptosis and blood–brain barrier (BBB) disruption. This study aimed to investigate the TWEAK/Fn14 signaling axis in cerebral ischemia and reperfusion using different in vitro oxygen–glucose deprivation (OGD) durations and cellular models. Western blot and RT-qPCR were used to evaluate TWEAK/Fn14 expression in monocultures, co-cultures, and triple-cultures of human immortalized endothelial cells, pericytes, and astrocytes. Six OGD conditions were tested: 4, 8, and 16 h, with or without 24 h reoxygenation. BBB model integrity was evaluated by analyzing occludin, zonula occludens-1, and VE-cadherin. A significant, duration-dependent downregulation of Fn14 was observed in monocultures after OGD (up to 85%, p < 0.05–p < 0.001), with partial recovery after 24 h reoxygenation (p < 0.05). TWEAK levels remained stable with minor fluctuations. Similar Fn14 reductions were seen in co- and triple-cultures (p < 0.01), followed by recovery. Endothelial biomarkers exhibited an initial stress response post-OGD, followed by recovery during reoxygenation. In conclusion, TWEAK remains stable during ischemia without immune cells, while Fn14 is downregulated during OGD and recovers after reoxygenation, indicating time-dependent roles in ischemic response and repair. The findings indicate a time-dependent regulation of Fn14 under ischemic conditions in vitro, highlighting its role in BBB stress and recovery. Nevertheless, further preclinical studies are needed to establish its therapeutic potential.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12035-026-05691-5.

Introduction

Dysfunction of the blood–brain barrier (BBB), a highly selective barrier essential for maintaining brain homeostasis and protecting the central nervous system from harmful substances, is a key factor in various neurological disorders, including stroke, Alzheimer’s disease, multiple sclerosis, and traumatic brain injury [1–3]. Depending on the disorder, BBB dysfunction may act as an early pathological event contributing to disease onset or arise secondarily due to ongoing neuroinflammation and vascular damage [4, 5]. Given its crucial role in brain health, understanding the mechanisms of BBB disruption is vital for developing targeted interventions to preserve or restore its function, potentially slowing disease progression and improving patient outcomes.

The BBB is a complex structure composed of endothelial cells, pericytes, and astrocytes. Endothelial cells are bound together by tight junction proteins, such as claudin-5, occludin, and zonula occludens-1 (ZO-1). This structure restricts paracellular flux of solutes and permits a highly selective transcellular transport of molecules. Together with neurons and microglia, the components of the BBB form the neurovascular unit [3]. This unit is further supported by extracellular matrix components, which provide structural integrity and contribute to the regulation of BBB function [6]. Tumor necrosis factor-like weak apoptosis-inducing factor (TWEAK, encoded by TNFSF12) is a type II transmembrane protein of the tumor necrosis factor superfamily. It is expressed in neurons and microglia of the central nervous system, as well as in perivascular astrocytes and endothelial cells [7–9]. Following proteolytic cleavage, soluble TWEAK (sTWEAK) binds to its receptor, fibroblast growth factor-inducible 14 (Fn14, encoded by TNFRSF12A), activating mitogen-activated protein kinases (MAPK), nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB), and phosphoinositide 3-kinase/protein kinase B (PI3K/AKT) downstream signaling pathways. This signaling axis regulates cell proliferation, differentiation, apoptosis, neovascularization, and inflammatory responses [10–14]. CD163 was also found as a secondary TWEAK scavenger receptor exclusively on cells of the monocyte-macrophage lineage [11, 15].

Previous studies have linked TWEAK to neurodegenerative disorders, stroke, traumatic brain injury, and myocardial infarction, with its elevated expression playing a key role in cellular and inflammatory responses in these conditions [16–19]. Both sTWEAK and its receptor Fn14 have been found to be rapidly upregulated after ischemic injury, suggesting their involvement in post-stroke inflammation and tissue remodeling. Specifically, increased expression of TWEAK and Fn14 has been observed in human patients following ischemic stroke, both in circulating serum and in post-mortem brain tissue [10, 20–22]. In experimental models, Fn14 is induced in primary cell cultures (primary cortical neurons and astrocytes) subjected to oxygen–glucose deprivation (OGD) in vitro and in rodent models of middle cerebral artery occlusion, where its elevation persists for several days after reperfusion [23–26]. It has also been demonstrated that TWEAK interacts with CD163 regulating tissue regeneration following ischemic injury [27].

Intracerebral administration of recombinant TWEAK in non-ischemic wild-type mice leads to the activation of NF-κB and MMP-9, resulting in increased BBB permeability. Notably, this effect is absent in Fn14-deficient mice, indicating that TWEAK directly influences the structure and permeability of the neurovascular unit [26]. Furthermore, the inhibition of TWEAK signaling, either through treatment with an Fn14-Fc decoy receptor or through the genetic deletion of Fn14, significantly improves the neurovascular unit integrity and reduces BBB permeability following cerebral ischemia [28].

In vitro studies using hCMEC/D3 cultures (an immortalized endothelial cell line) showed that TWEAK modulates cytokine expression, cell adhesion molecules, tight junction proteins, and matrix metalloproteinases (MMPs), leading to increased BBB permeability [9]. These findings align with prior research emphasizing TWEAK’s role in regulating immune cell infiltration in the central nervous system during experimental autoimmune encephalomyelitis [29]. Additionally, under OGD conditions, recombinant TWEAK has been shown to induce cell death via NF-κB activation, poly (ADP-ribose) polymerase (PARP-1) cleavage, and caspase-3 activation in wild-type neurons. However, this pro-apoptotic effect is not observed in neurons deficient in either Fn14 or TWEAK, further confirming the role of TWEAK/Fn14 signaling in ischemia-induced neuronal damage [30].

Despite these findings, the precise role of the TWEAK/Fn14 axis in different phases of ischemia, before, during, and after the event, remains unclear. While some studies suggest that TWEAK may contribute to post-ischemic neuroinflammation and exacerbate secondary brain damage, others indicate that it may also be involved in tissue repair and neurovascular remodeling. This dual role highlights the need for further research to elucidate the exact timing and mechanisms through which TWEAK/Fn14 signaling influences ischemic injury. Understanding these dynamics could provide valuable insights into potential therapeutic strategies aimed at modulating this pathway to minimize brain damage while promoting recovery.

This study explored the TWEAK/Fn14 signaling axis in cerebral ischemia using in vitro OGD models, employing Western blot analysis and RT-qPCR to evaluate both protein levels and gene expression. Experimental conditions included monocultures of human immortalized endothelial cells, brain pericytes, and astrocytes; co-cultures of endothelial cells with either pericytes or astrocytes; and a triple-culture model comprising all three cell types to enable a comprehensive analysis of cell type specific responses. Six experimental groups were defined according to the duration of OGD: three groups underwent OGD for 4, 8, or 16 h, while the remaining three experienced the same OGD durations, each followed by a 24 h reoxygenation period to mimic ischemic stroke and reperfusion phases. Finally, we evaluated the expression of key BBB-associated tight junction proteins, including occludin, ZO-1, and VE-cadherin, under these conditions.

Materials and Methods

A schematic overview of the culture setup, experimental timeline, and analysis procedures is provided in Fig. 1A.

Fig. 1.

A Schematic overview of the culture and analysis procedures. Representative images of endothelial cells (B), astrocytes (C), and pericytes (D) after 4, 8, and 16 h of OGD or normoxia (control) (scale bar = 200 µm)

Cell Culture

Immortalized human brain endothelial cells (hCMEC/D3, Merck Millipore) [31, 32], human brain immortalized astrocytes (P10251, Innoprot), and human immortalized pericytes (CL 05008CLTH, CELTHER) were used in the experiments after confirming the absence of mycoplasma contamination via PCR. Cells were cultured in pre-coated flasks with: i) Collagen-I (1:7 in PBS, Thermo Fisher Scientific) for endothelial cells, ii) Poly-L-lysine (0.015 µL PLL/µL sterile water, Innoprot) for astrocytes, and iii) 0.5% gelatin (Sigma-Aldrich) for pericytes. Microvascular endothelial cell growth medium (Pelo Biotech, GmbH) was used for endothelial cells, pericytes, and co- and triple-culture models, while astrocytes were maintained in Astrocyte Medium (Innoprot) with the supplements recommended by the manufacturer. The culture medium was refreshed every 48 h, and cells were detached using TrypLE (Thermo Fisher Scientific) when necessary.

Generation of 2D BBB Model

In monoculture experiments, endothelial cells, astrocytes, and pericytes (all at 24,000 cells/cm²) were cultured separately under standard conditions (5% CO₂, 37 °C, in a humidified incubator) in 6-well plates (clear polystyrene, Corning). For co-culture and triple-culture models, endothelial cells (8000 cells/cm²) were seeded on the apical side of 0.4µm pore-size Transwell inserts (#3450, Corning) in 12-well plates (clear polystyrene, Corning). In co-culture, pericytes or astrocytes (both at 24,000 cells/cm²) were plated at the bottom of the well. In triple-culture, a 1:1 mixture of astrocytes and pericytes (both at 12,000 cells/cm²) was plated at the bottom, enabling cross-talk between cell types via soluble factors, better mimicking physiological BBB conditions. The 12-well plates were specifically selected to allow transepithelial electrical resistance (TEER) measurements. Cells were maintained under standard culture conditions for 3–5 days, depending on the specific cell combination, until confluence was achieved [32].

OGD Experiments

Once the cells reached confluence (visually confirmed using phase-contrast microscopy), quiescence was induced to promote the characteristic endothelial phenotype. This was achieved by replacing the culture medium with 1% fetal calf serum (FCS), without vascular endothelial growth factor (VEGF), and incubating the cells for 24 h prior to the experiment. This procedure was applied to all cultures except pericyte and astrocyte monocultures, which were maintained in their specific media as described earlier [32–34].

For the OGD experiments, cells were incubated in a hypoxic chamber set to 1% oxygen using a deoxygenation system. Dulbecco’s Modified Eagle Medium (DMEM) with low glucose and without pyruvate, glutamine, or phenol red (Thermo Fisher Scientific) was used. The selected conditions aimed to mimic the ischemic penumbra, as it is a region potentially treatable after a stroke with reduced but non-zero cerebral blood flow [35]. Two experimental groups were established: i) the OGD group and ii) the reoxygenation group. In both groups, cells were exposed to OGD for 4, 8, or 16 h.

In the OGD group, samples were collected immediately after OGD exposure. In the reoxygenation group, immediately after OGD, the medium was replaced without any intermediate washing steps, and the cells were subjected to an additional 24 h reoxygenation phase in a standard incubator. Endothelial cells, co-cultures, and triple cultures were reoxygenated using microvascular endothelial cell growth medium supplemented with 1% FCS and no VEGF, while astrocyte and pericyte monocultures were maintained in their respective media.

Based on previous experiments with mono-cultures and co-cultures, 4 h of OGD was determined to be the optimal condition for the triple-culture system.

In all experiments, control groups for both OGD and OGD with reoxygenation were maintained in parallel under normoxic conditions for the corresponding durations. Medium changes were performed identically to those in the experimental groups to ensure consistent handling and timing across all conditions. The expression and levels of TWEAK, Fn14, and other in vitro BBB markers in response to OGD conditions was evaluated in both the OGD and reoxygenation groups.

Western Blot Analysis

After OGD experiments, cells were lysed using RIPA buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 0.1% SDS, 0.5% sodium deoxycholate, 1% NP-40) supplemented with a protease inhibitor cocktail (cOmplete Protease Inhibitor Cocktail, Roche). The samples were mixed with Laemmli buffer (750 mM Tris–HCl, pH 6.8, 5% SDS, 40% glycerol, and 80 mM DTT) and denatured at 95 °C for 10 min. Proteins were then separated on a 10–12% SDS-PAGE gel and transferred overnight onto a PVDF membrane using a Mini Trans-Blot Electrophoretic Transfer Cell (Bio-Rad).

The membrane was blocked with 5% non-fat dry milk and incubated overnight with primary antibodies targeting the following proteins: (i) anti-TWEAK (1:1000, #4437, Cell Signaling Technology),(ii) anti-TWEAKR (1:1000, #4403, Cell Signaling Technology), (iii) anti-ZO-1 (1:500, #33–9100, Thermo Fisher Scientific), (iv) anti-VE-cadherin (1:500, #sc-52751, Santa Cruz),(v) anti-occludin (1:500, #33–1500, Thermo Fisher Scientific), and (vi) anti-β-actin (1:25,000, #A5316, Sigma-Aldrich) used as a loading control.

After incubation with primary antibodies, the membrane was washed and then incubated with secondary antibodies, anti-mouse (#7076, 1:3000, Cell Signaling Technology) or anti-rabbit (#7074, 1:3000, Cell Signaling Technology), overnight at 4 °C. Protein detection was carried out using an enhanced chemiluminescence solution and was visualized with an Image FluorChem FC2 system (Cell Biosciences), employing AlphaView Software (Version 1.3.0.7, Innovatech Corporation). Densitometric analysis were performed using Image Lab Software (Version 6.0.1.34, Bio-Rad) to quantify protein levels.

Quantitative Real-Time PCR (RT-qPCR)

Total RNA was extracted using the NucleoSpin® RNA Isolation Kit (Machery-Nagel), following the manufacturer’s instructions. Subsequently, 500–1000 ng of total RNA was reverse transcribed using the High Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific). RT-qPCR was performed using TaqMan probes for the following target genes: (i) OCLN (Hs00170162_m1), (ii) CLDN5 (Hs00533949_s1), (iii) CDH5 (Hs00901465_m1), (iv) TJP1 (Hs01551871_m1), (v) TNFRSF12A (Hs00171993_m1), and (vi) TNFSF12 (Hs00387540_g1). Calnexin (CANX) (Hs01558409_m1) was used as the housekeeping gene after preliminary validation for stable expression across OGD conditions and was compared with GADPH. PCR reactions were carried out using the TaqMan® Fast Advanced Master Mix on the Quant Studio 7 flex Fast Real-Time PCR System (Thermo Fisher Scientific). Gene expression levels were analyzed using the comparative Ct (ΔΔCt) method.

sTWEAK ELISA Analysis

Cell culture media from the co-culture and triple-culture OGD experiments were collected and immediately centrifuged at 12,000 g for 10 min to remove any cellular debris. The supernatants were then stored at − 80 °C until further analysis. The concentration of sTWEAK was quantified using the Human TWEAK (TNFSF12) ELISA Kit (#EHTNFSF12, Thermo Fisher Scientific). The assay was carried out following the manufacturer’s instructions, with a detection sensitivity of 40 pg/mL. Each sample was analyzed in duplicate, and appropriate controls were included.

Measurement of Transendothelial Electrical Resistance (TEER)

After performing the OGD experiments, TEER was measured to assess the integrity of the endothelial cell monolayer. TEER was measured using the Electrical Resistance System (Millicell-ERS-2; Millipore Bedford, MA), following the manufacturer’s instructions, which allows for the precise calculation of electrical resistance across the cell layer. Empty Transwell inserts were used as internal controls to account for any background resistance from the membrane itself. This measurement provides valuable insights into the disruption or maintenance of the BBB properties following OGD and can indicate alterations in endothelial permeability.

Permeability Assay and Apparent Permeability Coefficient (P_app)

The P_app was calculated to evaluate the effects of OGD on endothelial cells in both co-culture and triple-culture systems. This assay quantifies the transport of a compound, such as fluorescein, across the endothelial monolayer, providing valuable insights into changes in permeability under OGD conditions. Cell culture medium containing 376 Da fluorescein (10 µM) was added to the upper chamber. Every 20 min of incubation at 37 °C for up to 1 h, samples were collected from the lower chamber, and fluorescence was measured in triplicate using a microplate reader (Tecan) at 485/535 nm. A known fluorescein concentration was used to construct a concentration curve and to extrapolate the fluorescence concentrations. After fluorescence detection, the P_app was determined using the Eq. (1) as follows:

$$P_{\mathit{app}}=\frac{V_r\times C_r}{A\times C_o\times t}$$ 1

where, V_r = volume of the receiver chamber; C_r = final concentration of fluorescein in the receiver chamber; A = surface area of the cell monolayer in the Transwell insert; C₀ = initial concentration of fluorescein in the donor chamber; t = total duration of the experiment.

This method allows assessing compound transport across the cellular barrier, providing a quantifiable measure of how OGD influences endothelial permeability. A higher value indicates greater permeability, while a lower value suggests a more restrictive barrier. By monitoring changes in P_app, this assay helps to understand the impact of OGD on the integrity of the BBB in our in vitro models.

Statistical Analysis

All data were collected and organized in a custom database created in Microsoft Excel 2016 (Microsoft Corporation) and subsequently analyzed using GraphPad Prism 9 (GraphPad Software). Data are presented as the mean ± standard error of the mean (SEM). Experimental groups were defined based on the duration of OGD, with or without reoxygenation. All experiments were performed in triplicate (n = 3 independent biological replicates) with technical duplicates where applicable, though some conditions had n = 2 due to technical constraints. The Shapiro–Wilk test was performed to assess the normality of the data distribution. For normally distributed data, comparisons between groups were made using t-tests with Welch’s correction applied when standard deviations were unequal. For non-parametric data, the Mann–Whitney test was used to determine statistical differences. Statistical significance was considered when p < 0.05 (*p < 0.05; **p < 0.01; ***p < 0.001). All graphical representations were generated using GraphPad Prism 9, ensuring clear and accurate presentation of the data.

Results

Monocultures of Endothelial Cells, Astrocytes, and Pericytes

Inverted phase-contrast microscopy was employed to visualize morphological changes in monocultures of endothelial cells (Fig. 1B), astrocytes (Fig. 1C), and pericytes (Fig. 1D) under OGD conditions compared with controls. In control samples, cells displayed a typical round morphology and uniform distribution. In contrast, OGD exposure induced notable structural changes, including increased cell death, disrupted distribution, and a significant loss of confluence, effects that intensified with longer OGD durations.

Western blot analysis revealed a marked downregulation of Fn14 levels in all monoculture models following OGD exposure (p < 0.001). Specifically, Fn14 levels decreased by approximately 85% in endothelial cells (Fig. 2A–B), 60% in astrocytes (Fig. 3A–B), and 64% in pericytes (Fig. 4A–B) compared with their respective control conditions. After reoxygenation for 24 h under normoxic conditions, Fn14 protein was totally recovered across all three cell types. These findings suggest a time-dependent regulatory response to ischemic stress and subsequent reoxygenation. In the particular case of endothelial cell monoculture, more than a recovery in Fn14 expression was observed. Reoxygenation after 8 or 16 h of OGD induced higher Fn14 levels than the initial levels prior to OGD (p < 0.05).

Fig. 2.

Endothelial cells were subjected to different OGD durations with or without 24h reoxygenation (RO) (n = 3). A Representative Western blot showing TWEAK, Fn14, VE-cadherin, ZO-1, and β-actin expressions. The quantification of protein levels was normalized to β-actin, and relative mRNA expression was normalized to CANX. B Fn14 protein levels, C FN14 mRNA expression, D TWEAK protein levels, E TWEAK mRNA expression, F VE-cadherin protein levels, G ZO-1 protein level, H occludin protein levels. I CDH5 (VE-cadherin) mRNA expression, J TJP1 (ZO-1) mRNA expression, and K relative OCLN (occludin) mRNA expression; columns represent mean values, and the vertical lines denote the standard error of the mean (SEM). *p < 0.05, **p < 0.01, and ***p < 0.001 between the columns are indicated by horizontal lines

Fig. 3.

Astrocyte cells were subjected to different OGD durations, with or without 24h reoxygenation (RO) (n = 3). A Representative Western blot showing TWEAK, Fn14, VE-cadherin, ZO-1, and β-actin expressions. The quantification of protein levels was normalized to β-actin, and relative mRNA expression was normalized to CANX. B Fn14 protein levels, C FN14 mRNA expression, D TWEAK protein levels, and E TWEAK mRNA expression; columns represent mean values, and vertical lines denote the standard error of the mean (SEM). *p < 0.05 and **p < 0.01 between the columns are indicated by horizontal lines

Fig. 4.

Pericyte cells were subjected to different OGD durations, with or without 24h reoxygenation (RO) (n = 3). A Representative Western blot showing TWEAK, Fn14, VE-Cadherin, ZO-1, and β-actin expressions. The quantification of protein levels was normalized to β-actin, and relative mRNA expression was normalized to CANX. B Fn14 protein levels, C FN14 mRNA expression, D TWEAK protein levels, and E TWEAK mRNA expression; columns represent mean values, and vertical lines denote the standard error of the mean (SEM). *p < 0.05 and **p < 0.01 between the columns are indicated by horizontal lines

The measurement of Fn14 mRNA levels further supports these findings, showing a consistent pattern of downregulation during OGD followed by a recovery, or an enhancement in endothelial cells (Fig. 2C for endothelial cells, Fig. 3C for astrocytes, and Fig. 4C for pericytes). These results highlight the transient yet reversible nature of Fn14 expression in response to ischemic stress and recovery, which may have implications for its role in BBB model integrity and neurovascular repair mechanisms.

No significant differences in TWEAK protein levels were observed across all monocultures, and the minor variations detected were reversed following reoxygenation (Figs. 2D, 3D, and 4D). However, distinct cell-type-specific trends in TWEAK mRNA emerged in response to varying OGD durations. Endothelial cells exhibited a reduction in TWEAK expression following 8 h of OGD (p < 0.05, Fig. 2E), while astrocytes showed a similar decline after 16 h of OGD exposure (p < 0.01, Fig. 3E). In contrast, pericytes displayed an opposite response, with a notable increase in TWEAK expression following 8 h of OGD (p < 0.01, Fig. 4E). This suggests that although transcriptional regulation of TWEAK is occurring in response to OGD, it is not translated into corresponding changes at the protein level, indicating possible post-transcriptional regulation or limited protein synthesis activity under these conditions.

In the endothelial cultures, BBB junction markers, occludin (OCLN), ZO-1 (TJP1), and VE-cadherin (CDH5) showed no significant changes at either the protein (Fig. 2F–H) or mRNA level (Fig. 2I–K) across conditions. ZO-1 mRNA exhibited a slight upward trend with prolonged OGD, suggesting a potential compensatory response. In contrast, occludin protein and mRNA progressively declined with increasing OGD duration. Although a small increase in occludin mRNA was observed after 4 h of OGD, it decreased with longer OGD and after reoxygenation, indicating a gradual compromise of tight junction integrity under sustained ischemic stress.

Co-Cultures of Endothelial Cells–Pericytes and Endothelial Cells–Astrocytes

A co-culture of endothelial cells and pericytes or astrocytes was studied following the same procedures as in the monocultures. The expression of Fn14 and TWEAK was measured in the cell lines independently at both protein and RNA levels.

In endothelial cell–pericyte co-cultures, Fn14 protein (Fig. 5A–B and D) and gene expression (Fig. 5F and H) were significantly reduced in both endothelial cells and pericytes after OGD exposure (p < 0.05, p < 0.001). This reduction indicates a pronounced downregulation of Fn14 under ischemic conditions. Notably, 24 h reoxygenation restored or even enhanced Fn14 expression directly related to the OGD duration (p < 0.05), suggesting a reversible, adaptive regulatory mechanism aimed at re-establishing homeostasis.

Fig. 5.

Endothelial cells and pericytes were subjected to different OGD durations, with or without 24h reoxygenation (RO) (n = 3). A Representative Western blot showing TWEAK, Fn14, VE-cadherin, ZO-1, and β-actin expressions in endothelial cells and pericytes. The quantification of protein levels was normalized to β-actin, and relative mRNA expression was normalized to CANX. The quantification of Fn14 and TWEAK protein levels in endothelial cells (B–C) and pericytes (D–E); relative FN14 and TWEAK mRNA expression in endothelial cells (F–G) and pericytes (H–I); the quantification of additional protein levels of VE-cadherin (J), ZO-1 (K), and occludin (L); and relative mRNA expression of CDH5 (VE-cadherin) (M) TJP1 (ZO-1) (N), and OCLN (occludin) (O). Columns represent mean values, and vertical lines denote the standard error of the mean (SEM). *p < 0.05, **p < 0.01, and ***p < 0.001 between the columns are indicated by horizontal lines

In contrast, TWEAK protein levels (Fig. 5C and E) and gene expression (Fig. 5G and I) remained unchanged across experimental conditions. Although no consistent trend emerged relative to OGD duration or reoxygenation, a transient increase in TWEAK mRNA was observed in endothelial cells following OGD periods and in pericytes after 4 h of OGD (52%). This early induction may represent a short-lived stress response, though further investigation is needed to determine the functional implications of this possible transient increase.

Regarding key endothelial cell biomarkers, both protein (Fig. 5J–L) and gene expression (Fig. 5M–O) followed similar patterns; however, statistically significant differences were observed only at the gene expression level: (i) VE-cadherin exhibited a significant increase after 4 h of OGD (p < 0.05), with a notable but non-significant elevation (58%) after 8 h. Its expression subsequently declined following the reoxygenation phases. (ii) ZO-1 displayed a comparable trend, although statistical significance emerged only after reoxygenation, suggesting a delayed regulatory response. (iii) Occludin mirrored the expression pattern of VE-cadherin (p < 0.05), showing an initial upregulation during OGD followed by a decrease post-reoxygenation.

In endothelial cell–astrocyte co-cultures, Fn14 protein levels (Fig. 6A–B and D) significantly decreased in both endothelial cells and astrocytes after OGD (p < 0.01) and partially recovered after 24 h of reoxygenation under normoxic conditions. RT-qPCR data (Fig. 6F and H) confirmed this pattern at the mRNA level.

Fig. 6.

Endothelial cells and astrocytes were subjected to different OGD durations, with or without 24h reoxygenation (RO) (n = 3). A Representative Western blot showing TWEAK, Fn14, VE-cadherin, ZO-1, and β-actin expressions in endothelial cells and astrocytes. The quantification of protein levels was normalized to β-actin, and relative mRNA expression was normalized to CANX. The quantification of Fn14 and TWEAK protein levels in endothelial cells (B–C) and astrocytes (D–E); relative FN14 and TWEAK mRNA expressions in endothelial cells (F–G) and astrocytes (H–I); the quantification of protein levels of VE-cadherin (J), ZO-1 (K), and occludin (L); and relative mRNA expression of CDH5 (VE-cadherin) (M), TJP1 (ZO-1) (N), and OCLN (occludin) (O). Columns represent mean values, and vertical lines denote the standard error of the mean (SEM). *p < 0.05 and **p < 0.01 between the columns are indicated by horizontal lines

Although TWEAK protein (Fig. 6C and E) or mRNA (Fig. 6G and I) did not show significant changes, an increase was observed in astrocytes after 4 or 8 h of OGD at 50%. This was followed by levels returning to baseline after 24 h reoxygenation, suggesting a transient induction.

The expression of endothelial junctional biomarkers closely resembled that observed in the co-culture with pericytes (Fig. 6J–O). After 16 h of OGD, the levels of both VE-cadherin and ZO-1 increased, followed by a decline after reoxygenation. In contrast, occludin exhibited a pronounced reduction of approximately 80% after OGD, with no recovery observed following reoxygenation. This suggests that while VE-cadherin and ZO-1 may exhibit compensatory mechanisms to maintain junctions, occludin appears to be more vulnerable to ischemic stress, which could contribute to increased BBB permeability.

In summary, both co-culture models showed comparable patterns: Fn14 decreased during OGD and recovered after 24 h of reoxygenation, while TWEAK exhibited OGD-independent regulation.

Triple-Culture Models of Endothelial Cells–Pericyte–Astrocytes

Given the comparable trends observed in TWEAK and Fn14 levels across monocultures and co-cultures in response to varying OGD durations, the triple-culture study was designed with a single OGD condition of 4 h, followed by its corresponding 24h reoxygenation phase. This decision was based on prior findings indicating that 4 h of OGD represents an optimal condition for investigating BBB model integrity while minimizing excessive cellular damage, allowing for a clearer evaluation of the effects of TWEAK/Fn14 signaling within the complex triple-culture system.

Both Western blotting (Fig. 7A–B and D) and RT-qPCR (Fig. 7F and H) analyses revealed a decrease in Fn14 levels under OGD conditions (p < 0.01), followed by a recovery after 24 h of reoxygenation under normoxic conditions (p < 0.05) in the three types of cells. Fn14 downregulation was less pronounced in astrocytes–pericytes than in endothelial cells, 52% vs. 82%, respectively. On the other hand, no significant differences were detected in TWEAK protein levels (Fig. 7C and E) or gene expression (Fig. 7G and I). These findings indicate that while Fn14 expression appears to be more sensitive to OGD-induced stress, TWEAK levels may exhibit a delayed response to ischemic conditions.

Fig. 7.

Endothelial cells–astrocytes–pericytes were subjected to 4 h of OGD with or without 24h reoxygenation (RO) (n = 3). A Representative Western blot showing TWEAK, Fn14, VE-Cadherin, ZO-1, and β-actin expressions in endothelial cells and astrocytes–pericytes. The quantification of protein levels was normalized to β-actin, and relative mRNA expression was normalized to CANX. The quantification of Fn14 and TWEAK protein levels in endothelial cells (B–C) and astrocytes–pericytes (D–E); relative FN14 and TWEAK mRNA expression in endothelial cells (F–G) and astrocytes–pericytes (H–I); the quantification of protein levels of VE-cadherin (J), ZO-1 (K), and occludin (L); and relative mRNA expression of CDH5 (VE-cadherin) (M), TJP1 (ZO-1) (N), and OCLN (occludin) (O). Columns represent mean values and vertical lines denote the standard error of the mean (SEM). *p < 0.05 and **p < 0.01 between the columns are indicated by horizontal lines

Regarding endothelial cell biomarkers (Fig. 7J–O), no significant differences were observed. We highlight a noticeable elevation in the mRNA expression levels of all three evaluated markers following OGD exposure, which translates into a trend to elevated protein levels of VE-cadherin and occludin, suggesting a potential stress response. This trend to increase was followed by a subsequent recovery to baseline levels after reoxygenation, indicating a possible stabilization of endothelial integrity under normoxic conditions.

sTWEAK ELISA

Analysis of cell culture media from both co-culture and triple-culture systems did not yield detectable signals for sTWEAK. This suggests that sTWEAK secretion may be below the detection threshold under the experimental conditions used or that its presence in the media is minimal, potentially due to rapid uptake, degradation, or limited release by the cultured cells.

TEER and Permeability Assay

We evaluated the TEER of the in vitro BBB model among the tested conditions (endothelial cells–astrocytes, endothelial cells–pericytes, and endothelial cells–astrocytes–pericytes) to more accurately reflect the multicellular architecture of the BBB model. Although there were no significant differences, we observed a decreasing trend in TEER values compared with the controls for all cell groups and all OGD time points evaluated (Fig. 8A–C). It was observed that the endothelial cells–pericytes group exhibited the greatest reduction in TEER values, with a mean decrease of approximately 25%. Notably, reoxygenation restored TEER values toward baseline levels. In particular, a significant improvement was observed following 24 h of reoxygenation after 4 h of OGD in the endothelial cell–pericyte group, indicating a substantial recovery of barrier integrity under these conditions.

Fig. 8.

Relative TEER measurements of endothelial cells–astrocytes (n = 2) (A), endothelial cells–pericytes (n = 3) (B), and endothelial cells–astrocytes–pericytes (n = 3) (C) after 4, 8, and 16 h of OGD, with or without 24h reoxygenation. Apparent permeability (P_app) was measured in endothelial cell–astrocyte–pericyte co-cultures at 20, 40, and 60 min following 4 h of OGD, with or without 24h reoxygenation (D), in endothelial cell–pericyte (E–G) and endothelial cell–astrocyte (H–J) co-cultures at 20, 40, and 60 min after 4 h (E and H), 8 h (F and I), and 16 h (G and J) of OGD, with or without 24h reoxygenation. Columns represent mean values and vertical lines denote the standard error of the mean (SEM). *p < 0.05 between the columns are indicated by the horizontal lines

Based on the Papp P_app results, no statistically significant differences were observed across the different culture configurations studied: (i) endothelial cells–astrocytes, (ii) endothelial cells–pericytes, and (iii) endothelial cells–pericytes–astrocytes. These configurations were evaluated at 20, 40, and 60 min after 4, 8, and 16 h of OGD conditions, with and without 24h reoxygenation (Fig. 8D–J).

Although Papp P_app values exhibited variations, particularly following OGD exposure, these differences did not reach statistical significance. Notably, the triple-culture model, which includes endothelial cells, pericytes, and astrocytes, demonstrated the greatest resilience to OGD conditions. Papp P_app values in this model remained relatively stable across all time points and experimental conditions, indicating a strong resistance to OGD-induced barrier disruption. In contrast, the endothelial cell–pericyte co-culture group exhibited the largest variations in Papp P_app, with increases of up to 50%. These findings suggest that the inclusion of all three cell types contributes to a more robust and physiologically relevant BBB model capable of mitigating OGD-induced permeability changes.

Discussion

This study investigates the TWEAK/Fn14 signaling axis using in vitro BBB models under OGD conditions. The experimental design includes monocultures of human immortalized endothelial cells, pericytes, and astrocytes, as well as co-cultures of endothelial cells–pericytes and endothelial cells–astrocytes, along with triple-culture models incorporating the three cell types. Six experimental conditions were established: OGD exposure (4, 8, or 16 h) and 24 h reoxygenation after OGD aimed to mimic the conditions that occur during an ischemic stroke and the reperfusion phase. Additionally, TEER and key BBB tight junction proteins, including occludin, ZO-1, and VE-cadherin, were evaluated to assess BBB model integrity and response to ischemic conditions.

To the best of our knowledge, this is the first study to characterize the TWEAK/Fn14 axis under specific OGD conditions, across these particular cell culture configurations (monoculture, co-culture, and triple-culture), and to utilize the selected cell types. Different cell types in the BBB respond to OGD in distinct ways over time. This study allowed us to evaluate both independent and combined conditions, offering a deeper understanding of the complex interactions between the TWEAK/Fn14 signaling axis and various BBB components. By investigating how endothelial cells, pericytes, and astrocytes individually and together respond to ischemic stress, we gained valuable insights into how the TWEAK/Fn14 pathway is modulated by OGD–reoxygenation.

By incorporating different exposure times, we were able to assess how BBB function and other cellular processes evolve during both the acute (early) and subacute (later) phases of ischemia: i) 4 h represents the early phase of ischemia, where cells begin to experience stress, but irreversible damage has not yet occurred; ii) 8 h represents the middle phase, where ischemic damage intensifies, and cellular responses like inflammation and apoptosis become more pronounced, this time frame is crucial for understanding how the BBB responds as ischemia progresses; iii) 16 h corresponds to a more advanced stage of ischemia, where there is significant cellular damage and a higher likelihood of irreversible changes, this stage allows for the evaluation of the extent of BBB breakdown and the mechanisms that lead to severe damage.

Studying the effects of reoxygenation allowed us to assess the role of TWEAK/Fn14 signaling in the recovery or deterioration of BBB model integrity following the restoration of blood flow. Reoxygenation induces complex cellular processes, including inflammation, oxidative stress, and repair mechanisms, that may modulate TWEAK/Fn14 activity and influence BBB function. A 24h reoxygenation period was selected to capture both early and delayed molecular responses, as key pathways involved in inflammation and tight junction regulation often require several hours to become detectable at the transcriptional and protein levels.

Our main findings reveal that (i) human brain endothelial cells, human brain immortalized astrocytes, and human immortalized pericytes express both TWEAK and Fn14 at the protein and mRNA levels, confirming the presence of this signaling axis across the key cellular components of the BBB in immortalized models. The analysis of Fn14 expression revealed that the regulation after OGD was maintained across different cell types. These findings support the hypothesis that the TWEAK/Fn14 axis may play a crucial role in BBB dynamics and neurovascular interactions, particularly under ischemic conditions. (ii) A significant decrease in Fn14 protein levels was observed following all OGD conditions. In monocultures, the greatest decreases were found in endothelial cells, and this trend was also reflected in both co-cultures and triple-cultures. (iii) Fn14 expression showed an OGD time-dependent recovery after 24 h of reoxygenation under normoxic conditions. Remarkably, in endothelial cell monoculture, this recovery can be increased to an enhanced expression for longer periods of OGD. This recovery/stimulation was attenuated in co-cultures and triple-cultures, indicating that the presence of additional cell types may influence the degree of Fn14 restoration after reoxygenation. (iv) We did not observe significant alterations in TWEAK expression under the OGD conditions tested or during the subsequent reoxygenation phases. This stability suggests that TWEAK levels are relatively unaffected by ischemic stress and recovery, indicating a potentially constitutive expression pattern in the studied cell types. (v) The levels of biomarkers associated with the BBB indicate a deregulation linked to recovery and maturation processes, primarily observed during the 4–8 h of OGD periods. This suggests that key structural components of the BBB undergo dynamic changes in response to ischemic stress, potentially as part of an adaptive or compensatory mechanism that allows the maintenance of the junction protein equilibrium, which in turn would maintain BBB functionality.

Previous studies have reported an upregulation of Fn14 following OGD. However, this increase has primarily been observed when post-OGD recovery periods under normal culture conditions were introduced [22–25]. Our study demonstrated that OGD conditions significantly reduced Fn14 expression across all three studied cell types. This downregulation suggests that oxygen and nutrient deprivation suppress Fn14 mRNA transcription, but causality remains unconfirmed without direct pathway studies (for example, using Fn14 knockdown or Fn14-Fc decoy). This finding aligns with previous research on Fn14 protein stability, which has a short lifespan of about 70 min, much shorter than the typical 24h protein turnover period [36]. Following reoxygenation, Fn14 levels exhibited a substantial increase, suggesting a potential adaptive response to normoxic recovery rather than active BBB repair, which our data do not directly demonstrate.

The observed upregulation of Fn14 post-reoxygenation may reflect a physiological adaptation to prior cellular stress and could be associated with the recruitment of additional sTWEAK molecules, which activate pathways related to proliferation, angiogenesis, inflammation, and cell death, all crucial for cellular homeostasis and repair [14]. While a similar Fn14 expression pattern was observed across co-cultures and the triple-culture model, the reduction in Fn14 levels in astrocytes–pericytes was notably less pronounced compared to that in endothelial cells within the triple-culture system, indicating a higher susceptibility of monocultures to ischemic stress. This differential response may be attributed to interactions among the three principal BBB cell types via soluble factors. Supporting this idea, previous studies suggest that more complex BBB models exhibit improved BBB marker expression and functionality [37]. Further investigations are needed to determine whether these intercellular interactions influence Fn14 dynamics and whether they contribute to BBB resilience under ischemic conditions.

TWEAK is primarily synthesized by leukocytes, and therefore, its upregulation, observed in the in vitro and in vivo models, could not be reproduced in our models because they lack immune cells [37, 38]. However, both astrocytes and endothelial cells synthesize TWEAK, although their expression and levels are lower than those in hematopoietic cell lineages [9, 39]. Protein analysis confirmed the presence of TWEAK in endothelial cells, astrocytes, and pericytes, although it remained low and was challenging to detect in both cell lysates and conditioned media. For this reason, even though TWEAK is synthesized by endothelial cells, astrocytes, and pericytes, we can conclude that its transcription and translation are not affected by OGD-reoxygenation phases, highlighting a potential alternative response to those previously described.

Finally, we assessed key BBB markers, including occludin, ZO-1, VE-cadherin and claudin-5. Although an initial stress response could tend to disbalance their gene expression, particularly in occludin, no significant differences were detected in their protein expression levels. A finding further supported by functionality measurements like TEER and the P_app [33, 40]. These results suggest that, under the tested conditions, OGD did not significantly compromise BBB model integrity. However, these results are consistent with the previously described observation that hCMEC/D3 cells present low junctional tightness (measured by TEER), probably as a consequence of the reduced expression of the main TJ protein claudin-5 [30, 41].

While previous studies have proposed TWEAK/Fn14 blockade as a therapeutic strategy for BBB dysfunction, our data challenge some aspects of this hypothesis. Specifically, (i) Fn14 downregulation during OGD suggests that  oxygen and glucose deprivation directly suppresses its mRNA transcription, contrary to assumed stress-induced upregulation. (ii) Fn14 recovery after reoxygenation may indicate a role in cellular adaptation (angiogenesis, inflammation, and cell survival), but our data do not confirm BBB repair. (iii) The attenuated Fn14 response in co- and triple-cultures suggests that intercellular communication modulates Fn14 expression and could influence BBB resilience in ischemia. (iv) In contrast to Fn14, TWEAK expression remained low and stable under OGD conditions and after reoxygenation in endothelial cells, astrocytes and pericytes.

This study has several limitations that temper its conclusions. First is the exclusive use of immortalized cell lines. While immortalized lines offer advantages in reproducibility and experimental standardization, they may exhibit altered metabolic responses, altered expression of transporters and tight junction proteins, and attenuated inflammatory signaling compared to primary cells. Particularly, as mentioned before, hCMEC/D3 may not fully recapitulate the barrier properties, stress responses or gene expression of primary human brain cells. Importantly, however, unpublished data from our laboratory demonstrate similar Fn14 downregulation patterns during OGD and recovery after reoxygenation in primary endothelial cells. This may reflect a conserved response across endothelial cell types. Nevertheless, validation in primary human brain endothelial cells, astrocytes, and pericytes remains essential to confirm the translational relevance of these findings. Second, the absence of immune cells represents a relevant limitation of our in vitro system. While the main sources of TWEAK are monocytes, macrophages, and microglia, it is also synthesized by astrocytes and endothelial cells. For this reason, we can conclude that TWEAK synthesis in endothelial cells, astrocytes, and pericytes is not affected by ischemia–reperfusion. However, it would be interesting to study the response of the TWEAK/Fn14 pathway in the specific case of immune cells and subsequently integrate immune cells into the BBB model to evaluate TWEAK/Fn14 dynamics and their effects on the main cell types of the brain. Third, the OGD model is effective for simulating hypoxia, though it lacks hemodynamic factors (e.g., shear stress). Fourth, the sample size was limited, with the majority of conditions including only n = 3 and some n = 2. This limitation may reduce the sensitivity to detect subtle effects. Nevertheless, consistent patterns were observed, particularly the downregulation of Fn14 following OGD and its upregulation after reoxygenation. These findings should be validated through additional independent experiments to fully characterize the remaining responses. Fourth, the lack of direct TWEAK/Fn14 pathway manipulation prevents causal inferences about Fn14 role in ischemic responses. Nevertheless, our data provide important descriptive insights, showing consistent patterns of Fn14 downregulation during OGD and upregulation after reoxygenation, which establish a strong foundation for future mechanistic studies.

Future studies incorporating primary cells, microfluidic models, functional assays, and immune cell co-cultures could address these limitations and enhance translational relevance. In particular, the immune cell response to OGD should be further investigated and integrated with studies involving other main brain cell types. This approach could clarify whether the neuroprotection observed in vivo after pathway inhibition results from immune cell modulations rather than direct regulation of TWEAK/Fn14 in other cell types. Furthermore, additional research should clarify Fn14’s dual role in ischemia and recovery, its modulation by astrocyte–pericyte interactions, the influence of immune cells on TWEAK production, and the optimal timing of Fn14-targeted therapies to improve BBB model integrity while minimizing inflammation and injury during the stroke and reperfusion phases.

Conclusions

We identified distinct regulatory patterns of the TWEAK/Fn14 axis using OGD models in both co-cultures and triple-cultures. Our findings indicate that TWEAK remains stable in the absence of immune cells, while Fn14 suppression during ischemia and during recovery post-reoxygenation suggests a dynamic, time-dependent role in BBB responses and potential recovery. Future therapeutic strategies should consider a phase-specific modulation of Fn14, rather than broad inhibition, to balance its roles in BBB damage and recovery.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

Organization and design of the study (RIR, MB). Data acquisition and processing (ASV, MLAA, PH). Statistical analysis and graphical presentation (JC, RIR, ASV). Manuscript drafting (ASV, RIR, MB, EA). Critical revision and execution of the project (PH, JC, EA). Supervision, review and critique (JC, MLA, PH, RIR, MB, EA). All authors read, reviewed and agreed on the manuscript version.

Funding

This research was funded by the Xunta de Galicia (Axencia Galega de Innovación: IN607A2022-03), Axencia Galega para a Xestión do Coñecemento en Saúde (SA304D-PRIS-T-2024/03), and the Instituto de Salud Carlos III (ISCIII) (ISCIII/PI21/01256/, PI24/00813, co-financed by the European Union, DTS23/00103 and AC21_2/00014, under the frame of EuroNanoMed III). This work, integrated into the Framework of PERTE for Cutting-Edge Health, has been co-financed by the Spanish Ministry of Science, Innovation and Universities with funds from the European Union NextGenerationEU; the Recovery, Transformation and Resilience Plan (PRTR-C17.I1); and the Autonomous Community of Galicia within the framework of the Biotechnology Plan Applied to Health. Furthermore, R. Iglesias-Rey received funding support (CP22/00061) from the Miguel Servet Program of Instituto de Salud Carlos III and was co-financed by the EU. The sponsors did not participate in the study design, collection, analysis, or interpretation of the data, in writing the report, or in the decision to submit the paper for publication.

Data Availability

Data will be made available on request.

Declarations

Competing interests

The authors declare no competing interests.