Background

Brain endothelial cells (ECs) are critical components of the blood brain barrier (BBB). Increased BBB permeability leads to the development of tissue swelling, inflammatory cell infiltration and subsequently exaggerate injury in ischemic stroke [1]. Therefore, protection of ECs and BBB function should be an important strategy for reducing ischemic injury. On the other hand, endothelial progenitor cells (EPCs) have been suggested to participate in EC protection, repair and angiogenesis [2]. Transplantation of EPCs is a promising cell therapy for ischemic diseases such as acute myocardial infarction and stroke [3–5]. Our previous studies have shown that EPC infusion promotes angiogenesis in mouse ischemic stroke models [5, 6]. EPCs released angiogenic growth factors, such as vascular endothelial growth factor (VEGF) and insulin-like growth factor, could be responsible for the beneficial effect of EPC conditioned medium on the viability of H2O2-compromised human umbilical vein ECs [7, 8]. Currently, we do not know whether EPCs can protect cerebral ECs against hypoxia/reoxygenation (H/R)-injury.

Transplantation of neural progenitor cells (NPCs) has also been shown to be effective for treating ischemic stroke in animal models [9, 10]. In addition to generating neurons, grafted NPCs could promote angiogenesis in a rodent stroke model [11]. A recent report suggests that co-culture with NPCs decreases the passive permeability of brain ECs [12]. Collectively, these studies indicate a crosstalk between NPCs and ECs. However, it is unclear whether NPCs and EPCs have synergistic effects on EC protection.

The PI3K/Akt signal pathway participates in various cellular processes such as cell survival and proliferation [13]. Previous studies have shown that activation of the PI3K/Akt signal pathway promotes neuron survival [14, 15], cardiac microvascular EC migration [16], and axonal outgrowth compromised by oxygen-glucose deprivation [17, 18]. It is unknown whether this pathway is involved in the mechanism of the benefits of NPCs and EPCs.

The aims of this study were to elucidate whether EPCs and NPCs synergistically protect brain ECs from H/R-induced injury and to explore whether the effects are mediated by the PI3K/Akt signal pathway.

Results

NPCs and EPCs were successfully generated from human inducible pluripotent stem cells

As shown in Fig. 1, the human inducible pluripotent stem cells (iPSCs) grew as colonies staining positively for pluripotent markers, Sox2 and Oct3/4. The generated NPCs grew as neurospheres after 7-day neural induction, and expressed neural progenitor markers pax6 (98 ± 1 %) and nestin (96 ± 1.5 %), but not expressed Oct3/4, indicating a high differentiation efficacy. The generated NPCs had ability of differentiating into neurons, which was evidenced by expressing neuron specific marker Tuj1.

Fig. 1
figure 1

Characterization of EPCs and NPCs from human iPSCs. A1, iPSCs grew on feeder-free matrigel. Scale bar: 2000 μm. A2-A3, iPSCs expressed Sox2 and Oct3/4. B1, NPCs grew as neurospheres after 7 days neural differentiation. B2-B4, generated NPCs positively expressed neural progenitor markers nestin and Pax6, and were negative for pluripotent stem cell marker Oct3/4. C, generated neurons positively stained neuron specific marker Tuj1. D1-D3, purified EPCs positively expressed endothelial progenitor markers CD34 and KDR, but were negative for Oct3/4. Black curves: isotype control; red curves: antibodies. D4, tube formation of EPCs. DAPI counterstained cell nucleus. Scale bar: 200 μm from A2-D4

Full size image

After 7-day EPC induction, approximately 48 ± 2.1 % of cells positively expressed endothelial progenitor marker CD34. In order to get a pure population of EPCs, we used CD34-conjugated microbeads to enrich the generated EPCs. The CD34-conjugated microbeads purified cells positively expressed CD34 (96 ± 2.1 %) and KDR (95 ± 1.8 %). As expected, the purified EPCs did not express Oct3/4. In addition, the generated EPCs had tube formation ability as revealed by matrigel assay.

Co-culture with EPCs and NPCs synergistically protected ECs from H/R-induced apoptosis and compromised viability via activating the PI3K pathway

After exposed to the hypoxic condition for 6 h, ECs were co-cultured with EPCs and/or NPCs for 24 h, followed with apoptotic assay or MTT assay. Results (Fig. 2) showed that co-culture with EPCs and NPCs exerted a greater effect on decreasing H/R-injured EC apoptosis than that co-culture with EPCs or NPCs separately did (vehicle vs. EPCs or NPCs, p < 0.05; EPCs and NPCs vs. EPCs or NPCs, p < 0.05). Similarly, the EC viability was also synergistically increased by co-culture with EPCs and NPCs (vehicle vs. EPCs or NPCs, p < 0.05; EPCs and NPCs vs. EPCs or NPCs, p < 0.05). The synergistic effects on reducing EC apoptosis and improving EC viability were achieved by an increase of approximately 24 and 28 %, respectively.

Fig. 2
figure 2

EPCs and NPCs promoted the survival of H/R-injured ECs via activating the PI3K pathway. MTT assay and PI/FITC-Annexin V apoptosis assay were conducted on H/R-injured ECs co-cultured with EPCs and/or NPCs for 24 h as described in Material and Methods. A1, representative morphology images showing the viability of ECs. A2, summarized data showing EC viability which is synergistically increased when co-cultured with the combination of EPCs and NPCs than that co-cultured with EPCs or NPCs alone. B1, representative flow plots of EC apoptotic rate. B2, summarized data of the apoptotic rate of ECs, showing that the combination of EPCs and NPCs offers better anti-apoptotic effect than EPCs or NPCs alone. Block the PI3K pathway could diminish the beneficial effects of EPCs and/or NPCs. And the PI3K pathway upstream blockers, SU1498 and K252a, reduced these effects of EPCs and NPCs. *p < 0.05, vs. Normoxia; #p < 0.05, vs. vehicle. Data are expressed as mean ± SEM, n = 6/group/measurement. LY294002: PI3K inhibitor; SU1498: VEGFR2 inhibitor; K252a: TrkB inhibitor

Full size image

Moreover, our data showed that the PI3K inhibitor (LY294002) pre-treatment could completely abolish the abovementioned effects of EPCs and/or NPCs, suggesting that the beneficial effects of EPCs and NPCs are mediated by the PI3K pathway. To define the contribution of VEGFR2 and TrkB (PI3K upstream molecules) to these effects, the respective inhibitors SU1498 and K252a were pre-added in the co-culture system. Our results revealed that blockade of the VEGF/VEGFR2 and BDNF/TrkB signals reduced the effects of EPCs and NPCs.

Co-culture with EPCs and NPCs synergistically decreased the oxidative stress of H/R-injured ECs via activating the PI3K pathway

As shown in Fig. 3, ROS production was decreased in H/R-injured ECs co-cultured with EPCs or NPCs (vehicle vs. EPCs or NPCs, p < 0.05). Moreover, co-culture with EPCs and NPCs decreased ROS production to a larger extent than that with EPCs or NPCs alone did (EPCs and NPCs vs. EPCs or NPCs, p < 0.05). The synergistic effect on decreasing ROS production was obtained by about 18 % increase.

Fig. 3
figure 3

EPCs and NPCs decreased ROS production via activating the PI3K pathway. A, ROS production, showing that ROS over-production was much decreased in H/R-injured ECs co-cultured with EPCs and NPCs than that co-cultured with EPCs or NPCs alone. LY294002 could diminish the anti-oxidative effect of EPCs and/or NPCs, and the combination of SU1498 and K252a reduce such effect. *p < 0.05, vs. Normoxia; #p < 0.05, vs. Vehicle. Data are expressed as mean ± SEM, n = 6/group/measurement. LY294002: PI3K inhibitor; SU1498: VEGFR2 inhibitor; K252a: TrkB inhibitor

Full size image

As expected, pre-treatment with PI3K inhibitor, LY294002, abolished the anti-oxidative effect of EPCs and/or NPCs on H/R-injured ECs. Pre-treatment with a combination of the PI3K upstream blockers SU1498 and K252a reduced the most anti-oxidative effects of EPCs and NPCs on H/R-injured ECs. All of these data indicate that the anti-oxidative effect of EPCs and NPCs is mediated by the PI3K signal pathway.

H/R-compromised tube formation ability of ECs was synergistically improved by co-culturing with EPCs and NPCs via activating the PI3K signal pathway

We further assessed whether co-culture with EPCs and/or NPCs altered the tube formation function of ECs exposed to H/R. The results (Fig. 4) showed that EPCs or NPCs alone increased the tube formation ability of H/R-injured ECs (vehicle vs. EPCs or NPCs, p < 0.05). Moreover, co-culture with EPCs and NPCs exhibited a synergistic effect on improving the tube formation ability of ECs compromised by H/R (EPCs and NPCs vs. EPCs or NPCs, p < 0.05). The synergistic effect of EPC plus NPC co-culture on improving the tube formation ability was increased by approximately 19 %.

Fig. 4
figure 4

EPCs and NPCs improved the angiogenic function of H/R-injured ECs via activating the PI3K pathway. a, representative plots of tube formation. Scale bar: 200 μm. b, summarized data of EC tube formation, showing that EPC and NPC co-culture offers synergistically effects on improving EC function compared to EPCs or NPCs alone. And such synergistic effect could be blocked by LY294002, or partially abolished by the combination of SU1498 and K252a. *p < 0.05, vs. Normoxia; #p < 0.05, vs. Vehicle. Data are expressed as mean ± SEM, n = 6/group/measurement. LY294002: PI3K inhibitor; SU1498: VEGFR2 inhibitor; K252a: TrkB inhibitor

Full size image

In order to elucidate the possible role of the PI3K pathway in the effect of EPCs and/or NPCs on EC tube formation, the pathway specific inhibitor LY294002 was used in the co-culture study. Our results showed that PI3K inhibition entirely abolished this effect of EPCs and NPCs. Similarly, to further explore whether VEGF/VEGFR2 and BDNF/TrkB signals could be responsible to trigger the activation of the PI3K pathway, we pre-added their respective inhibitors SU1498 and K252a into the co-culture system. As we expected from the data of apoptotic and MTT assays, blockade of the VEGF/VEGFR2 and BDNF/TrkB signals reduced the effect on tube formation.

The endothelial permeability was improved by co-culturing with EPCs and NPCs

Under physiological conditions, the endothelial membrane is impermeable to macromolecules (mass weight around 70 k Dalton) [19]. We performed permeability assay to evaluate whether co-culture of EPCs and/or NPCs could improve the barrier function of ECs compromised by H/R. As expected, H/R injury increased trans-endothelial permeability to FITC-conjugated dextran (mass weight around 10 k Dalton). Co-culture of EPCs or NPCs decreased the flux of FITC-dextran, and EPCs combined with NPCs was more effective in decreasing the FITC-dextran flux through the EC monolayer (Fig. 5a).

Fig. 5
figure 5

EPCs and NPCs modulated the permeability and VEGF and BDNF secretion of H/R-injured ECs. a fold change of FITC-dextran flux, showing that the combination of EPCs and NPCs has better effects than EPCs or NPCs alone on improving the endothelial barrier function of H/R-injured ECs. b, c the levels of VEGF and BDNF in culture medium of normoxic cultured ECs, EPCs and NPCs, as well as hypoxic ECs co-cultured with EPCs, or NPCs, or both EPCs and NPCs. The summarized data showing that the levels of VEGF and BDNF were much increased in H/R-injured ECs co-cultured with EPCs and NPCs than that co-cultured with NPCs or EPCs alone. *p < 0.05, vs. Normoxia, #p < 0.05, vs. vehicle. Data are expressed as mean ± SEM, n = 6/group/measurement

Full size image

Co-culture with EPCs and NPCs complementarily elevated the levels of VEGF and BDNF in the conditioned medium of ECs exposed to H/R

In order to explore the mechanisms underlying the protective benefits of EPCs and NPCs, we performed ELISA assay to determine the levels of VEGF and BDNF in the culture medium. As shown in Fig. 5b, c, we found that co-cultured with EPCs alone increased the VEGF level, but not the BDNF level in the EC culture medium, whereas, co-cultured with NPCs alone raised the BDNF level, not the VEGF level. Moreover, co-culture with EPCs and NPCs increased the levels of both VEGF and BDNF in the EC medium, suggesting a complementary effect.

The expression of VEGFR2 was upregulated and ratios of p-Flk1/VEGFR2 and p-Akt/Akt were increased in H/R-injured ECs co-cultured with EPCs and NPCs

As shown in Fig. 6a, co-culture with NPCs alone or with EPCs and NPCs similarly increased the expression level of VEGFR2 in H/R-injured ECs, whereas, co-cultured with EPCs alone did not significantly change the expression of VEGFR2 in ECs, indicating that interaction of NPCs with ECs.

Fig. 6
figure 6

Co-culture with EPCs and NPCs activated the PI3K/Akt signal pathway on H/R-injured ECs. a, VEGFR2 expression was significantly upregulated in H/R-injured ECs co-cultured with NPCs or the combination of EPCs and NPCs. b the protein expression ratio of p-Flk1/VEGFR2 was significantly increased in H/R-injured ECs co-cultured with EPCs or NPCs, with a higher ratio in ECs co-cultured with the combination of EPCs and NPCs. c the protein expression ratio of p-Akt/Akt was increased in H/R-injured ECs co-cultured with EPCs and NPCs, and this effect was blocked or reduced when ECs were pre-treated with PI3K inhibitor LY294002 or VEGFR2 inhibitor SU1498 or TrkB inhibitor K252a. *p < 0.05, vs. Normoxia; #p < 0.05, vs. vehicle. Data are expressed as mean ± SEM, n = 6/group/measurement

Full size image

Western blot results demonstrated that the expression ratios of p-Flk1/VEGFR2 and p-Akt/Akt in H/R-injured ECs were increased by co-culture with EPCs or NPCs alone, with a greater increase when co-cultured with both EPCs and NPCs (Fig. 6b, c). Data showed that the net increase of the synergistic effect on up-regulating the expression ratio of p-Akt/Akt was approximately 30 %. As expected, the PI3K inhibitor LY294002 abolished the phosphorylation of Akt, suggesting that the PI3K/Akt signal pathway is activated in ECs co-cultured with EPCs and NPCs. A combination of SU1498 and K252a decreased the phosphorylation of Akt, reflecting that it at least partially depends on the upstream molecules VEGFR2 and TrkB (Fig. 6b).

Discussion

In the present study, we showed that EPCs and NPCs produced from human iPSCs had synergistic beneficial effects on H/R-injured brain ECs. The major findings include: i) Co-culture with EPCs and NPCs synergistically protected ECs from H/R-induced apoptosis and dysfunction; ii) The levels of VEGF and BDNF in the medium of ECs co-cultured with EPCs and NPCs were increased; iii) Co-culture with NPCs up-regulated VEGFR2 expression and its phosphorylation on ECs; iv) Blockade of the VEGFR2 and Trkb or PI3K/Akt pathway inhibited or abolished the protective effects of EPCs and NPCs (Fig. 7).

Fig. 7
figure 7

Proposed molecular mechanism for the protective effect of EPCs and NPCs on H/R-injured brain ECs. Co-culture with EPCs and NPCs synergistically increased the survival ability, decreased the oxidative stress and improved the angiogenic and barrier functions of H/R-injured EC, via activating the PI3K/Akt signal pathway that mainly depended on the progenitor paracrine (VEGF and BDNF) mediated signals. EPCs: endothelial progenitor cells; NPCs: endothelial progenitor cells; VEGF: vascular endothelial growth factor; BDNF: brain derived neurotrophic factor; VEGFR2: vascular endothelial growth factor receptor 2; TrkB: tyrosin kinase B; PI3K: phosphatidylinositol-3-kinase; H/R: hypoxia/reoxygenation; EC: endothelial cells; ROS: reactive oxygen species

Full size image

ECs are unique and critical in maintaining normal BBB function [20]. Impairment of BBB occurs in the early stage of ischemic brain injury, leading to subsequent brain swelling and inflammatory responses [21]. Thus, protecting brain ECs from H/R-induced injury will theoretically alleviate brain tissue damage in ischemic stroke. Nevertheless, there is no clinically effective strategy to protect ECs against H/R-induced injury in acute ischemic stroke. Transplantation of stem cells has been shown to accelerate the functional recovery of ischemic stroke by promoting angiogenesis and neurogenesis [22]. Indeed, others and our studies have demonstrated that engrafted EPCs or NPCs can alleviate acute ischemic injury and promote angiogenesis and neurogenesis in an ischemic stroke mouse model [ELISA assay of VEGF and BDNF

For determining the baseline level of VEGF and BDNF in the culture medium of EPCs and NPCs, we collected their respective culture medium before the co-culture experiments. After co-culture with EPCs and/or NPCs, the conditional medium of H/R-injured ECs in various groups was also collected. The levels of trophic factors VEGF and BDNF in the culture medium were determined with ELISA kits (R&D systems) by following the manufacturer’s instructions. ECs cultured in normoxia served as a control. ECs in the vehicle group were cultured with EC culture medium only.