Active Endothelial Barrier Repair Occurs at 48 Hours

Endothelial permeability increased progressively between 0 and 24 hours, associated with increasing numbers of apoptotic ECs, and decreased progressively between 24 and 96 hours, associated with decreasing numbers of apoptotic ECs and increasing numbers of proliferating ECs (Figure 1A).³⁰ On the basis of these observations, we defined 0 to 24 and 24 to 96 hours, respectively, as endothelial barrier injury and repair phases. At 48 hours, endothelial permeability decreased rapidly, associated with the highest level of EC proliferation (Figure 1A).³⁰ We considered 48 hours as the active barrier repair phase and focused our subsequent studies on 48 hours.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Endothelial barrier recovery is associated with increased resident endothelial cell (REC) and bone marrow–derived endothelial progenitor (BMDEPC) proliferation. A, Endothelial permeability in lungs was measured using Evan blue dye (EBD) leakage index at indicated time (hours) after saline (Con) or lipopolysaccharide (LPS; 5 mg/kg IP) injection. Mean±SEM of 6 mice per group. *P<0.05, compared with control. B–D, Lung cryosections were prepared at indicated time (hours) after LPS and 4 hours after 5-bromo-2-deoxyuridine (BrdU) injection, stained with BrdU+reverse tetracycline transactivator (rtTA), BrdU+green fluorescence protein (GFP) or GFP antibodies, and nuclei counterstained with 4,6-diamidino-2-phenylindole (DAPI). The number of BrdU⁺/rtTA⁺ proliferating RECs (B), BrdU⁺/GFP⁺ proliferating BMDEPCs (C), or GFP⁺ recruited BMDEPCs (D) was counted and expressed as a percentage of total cells as revealed by DAPI nuclear staining. Mean±SEM of 5 mice per group. *P<0.05, compared with controls.

Endothelial Barrier Recovery Is Associated With Increased REC and BMDEPC Proliferation

If RECs and BMDEPCs play important roles in endothelial barrier repair, the number of proliferating RECs or BMDEPCs should increase at the active repair phase. To track proliferating RECs or BMDEPCs in vivo, we generated EC-rtTA-GFP-BM chimeras by transplanting lethally irradiated EC-rtTA mice³¹ with bone marrows (BMs) from Tie2-GFP mice (Tables I and II in the online-only Data Supplement). These chimeras overexpress the reverse tetracycline transactivator (rtTA) on RECs and green fluorescent protein (GFP) on BMDEPCs. Fluorescence-activated cell sorting (FACS) analysis of BM mononuclear cells from donors and chimeras 2 months later confirmed that 95% of BM cells in the chimeras were donor BM origin (Figure IA in the online-only Data Supplement). FACS analysis of peripheral blood mononuclear cells further confirmed the high degree of BM chimerism in the EC-rtTA-GFP-BM mice (data not shown).

The endothelial-specific Tie2 promoter drives GFP expression in endothelial lineage cells. However, Tie2 was reported to be expressed on a subset of monocytes/macrophages.³² To clarify the percentage of monocytes in the GFP⁺ cell population in the lungs, we phenotyped GFP⁺ cells from lungs of Tie2-GFP donor mice 48 hours after lipopolysaccharide challenge. We found that 99% of the GFP⁺ cells are CD45⁻/CD31⁺/eNOS⁺ endothelial lineage cells, and <1% of them are CD45⁺/CD31⁺ hematopoietic cells (Figure II in the online-only Data Supplement). Our finding is consistent with a previous report.¹⁵ Thus, GFP⁺ cells in the EC-rtTA-GFP-BM mice represent principally BMDEPCs.

Two months after BM transplantation, mice were injected with saline or lipopolysaccharide and then with bromodeoxyuridine (BrdU) to label proliferating cells in vivo. Proliferating RECs or BMDEPCs, or recruited BMDEPCs, were identified by BrdU/rtTA or BrdU/GFP double staining, or by GFP staining, and counted. Time-course analyses showed that endothelial barrier recovery was associated with remarkably increased REC and BMDEPC proliferation (Figure 1). In particular, the high numbers of proliferating RECs and BMDEPCs at 48 hours were concordant with 48 hours being active barrier repair phase. No significant numbers of proliferating RECs and BMDEPCs were detected at 12 (injury phase) or 96 hours (barrier function recovered; Figure 1). ALI was associated with an increased recruitment of GFP⁺ BMDEPCs in lungs (Figure 1D). However, BMDEPC recruitment was not correlated with barrier recovery but correlated with barrier injury (Figure 1A versus 1D), suggesting that ALI stimulates BMDEPC recruitment.

RECs Participate in Endothelial Barrier Repair

If RECs participate in endothelial barrier repair, these cells should proliferate in situ on endothelial layer at the active repair phase to give rise to new ECs. Furthermore, the REC-derived daughter ECs should significantly increase in lungs after recovery from injury. EC-rtTA-GFP-BM mice that overexpress rtTA only on RECs (Table II in the online-only Data Supplement) were injected with BrdU at 44 hours after lipopolysaccharide injection to label proliferating cells. Lungs were harvested at 48 hours or at 8 weeks after lipopolysaccharide injection to track the location of proliferating RECs or to quantify the REC-derived new ECs in lungs. We visualized endothelial layer by immunofluorescence staining of lung sections with rtTA or CD31 antibody. We identified proliferating RECs by BrdU and rtTA double immunofluorescence staining. Confocal microscopic examination revealed that BrdU⁺/rtTA⁺ proliferating RECs were localized on the endothelial layer of microvessels (Figure 2A). The BrdU⁺/rtTA⁺ proliferating RECs coexpressed EC marker, CD31, and were localized on the CD31⁺ endothelial layer but were not localized on the aquaporin-5 (Aqu5)⁺ epithelial layer (Figure 2A). This result provides histological evidence that RECs proliferate in situ on endothelial layer at the active barrier repair phase.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Resident endothelial cells (RECs) participate in endothelial repair. A, RECs proliferate in situ on the endothelial layer at the active repair phase. Lung sections from mice 48 hours after lipopolysaccharide (LPS) injection were stained with antibodies against proliferative marker, 5-bromo-2-deoxyuridine (BrdU), REC marker, reverse tetracycline transactivator (rtTA), EC marker, CD31, and alveolar epithelial cell marker, aquaporin-5 (Aqu5), and nuclei counterstained with TO-PRO-3 dye (Pro-3). 3D projections (A1–A6) or single images (A7–A10) of confocal z-stacks are shown. A1, BrdU⁺ staining (green) detects proliferating cells (light blue nuclei); blue, Pro-3 nuclear staining. A2, rtTA⁺ staining (red) detects RECs and visualizes the endothelial layer. A3, Merge of A1 and A2 shows BrdU⁺/rtTA⁺ RECs (arrow indicated) localized on rtTA⁺ endothelial layer of alveolar microvessels. A4 and A5, Orthogonal view (X–Y, X–Z, and Y–Z) of the boxed area in A3 at higher magnification confirms colocalization of BrdU⁺ and rtTA⁺ signals and colocalization of BrdU⁺ and Pro-3⁺ stainings. Note, the blue nuclear staining in A4 or the red rtTA staining in A5 was omitted for clarity. A6 and A7, BrdU⁺/CD31⁺ RECs (arrow indicated) are localized on CD31⁺ endothelial layer of alveolar microvessels. A8–A10, Higher magnification of the boxed area in A7 is shown. A8, BrdU (green) and CD31 (red) double stain shows that BrdU⁺ proliferating REC is localized on CD31⁺ endothelial layer (red). A9, BrdU (green) and Aqu5 (blue) double stain shows that BrdU⁺ proliferating REC is not localized on Aqu5⁺ epithelial layer (blue). A10, Merge of A8 and A9 confirms that BrdU⁺ REC is localized on the endothelial layer (red) between 2 epithelial layers (blue). Scale bars, 40 µm (A1, A2, A3, A6 and A7), 8 µm (A4 and A5), and 3 µm (A8, A9, and A10). Fluorescence-activated cell sorting pictures (B) and bar graph (C) show an increased number of REC-derived ECs, defined as CD45⁻/CD31⁺/rtTA⁺/BrdU⁺ cells, in lungs of mice 8 weeks after LPS injection, compared with saline-injected mice (Con). Mean±SEM of 5 mice per group. *P<0.05, compared with control.

FACS analysis showed that the number of REC-derived new ECs (CD45⁻/CD31⁺/BrdU⁺/rtTA⁺) was ≈22-fold higher in lungs of EC-rtTA-GFP-BM mice 8 weeks after lipopolysaccharide-induced injury, compared with lungs from mice 8 weeks after saline injection (Figure 2B and 2C). These results provide cytological evidence for REC’s participation in endothelial barrier repair.

BMDEPCs Contribute to Endothelial Barrier Repair

BMDEPC incorporation into endothelial layer is a critical step in BMDEPC-mediated endothelial repair. To seek histological evidence of BMDEPC engraftment, we stained lung sections from mice 48 hours after lipopolysaccharide injection with antibodies against BMDEPC marker, GFP, EC markers, CD31 and Ve-cadherin (Ve), or alveolar epithelial cell marker, Aqu5. Confocal microscopic examination identified GFP⁺/CD31⁺ BMDEPCs localized on the CD31⁺ endothelial layer of lung microvessels (Figure 3A). The GFP⁺ BMDEPCs were also Ve⁺ and localized on the Ve⁺ endothelial layer but not localized on the Aqu5⁺ alveolar epithelial layer (Figure 3A). This result provides evidence for BMDEPC engraftment into the endothelial layer in the lung at the active repair phase.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Bone marrow–derived endothelial progenitor cells (BMDEPCs) contribute to endothelial repair. A, BMDEPCs engraft into microvessel wall. Lung sections from mice 48 hours after lipopolysaccharide (LPS) injection were stained with antibodies against BMDEPC marker, green fluorescent protein (GFP), endothelial cell (EC) markers, CD31 and Ve-cadherin (Ve), alveolar epithelial cell marker (Aqu5), and nuclei counterstained with Pro-3 dye. 3D projections (A1 to A6) or single images (A7 to A10) of confocal z-stacks are shown. A1, GFP⁺ (green) staining identifies BMDEPCs. A2, CD31⁺ (red) staining visualizes the endothelial layer. A3, Merge of A1 and A2 shows engrafted (GFP⁺/CD31⁺) BMDEPCs (yellow cells) localized on CD31⁺ endothelial layer (red). A4 and A5, 3D (A4) or orthogonal (A5) projection of the boxed area in A3 at higher magnification confirms GFP/CD31 coexpression on the same individual BMDEPC. For clarity, the blue nuclear staining was omitted. A6 and A7, GFP⁺/Ve⁺ BMDEPCs (arrow indicated) are localized on Ve⁺ endothelial (red) but not on Aqu5⁺ epithelial (blue) layer. High magnification of the boxed area in A7 is shown in A8–A10. A8, GFP (green) and Ve (red) double staining shows that GFP⁺ BMDEPCs express Ve (yellow cells) and are localized on Ve⁺ endothelial layer. A9, GFP (green) and Aqu5 (red) double staining shows that GFP⁺ BMDEPCs do not express Aqu5 (green cells) and are not localized on Aqu5⁺ epithelial layer. A10, GFP (green), Ve (red), and Aqu5 (blue) triple staining confirms that GFP⁺ BMDEPCs (yellow cells) are localized on endothelial (red) but not on epithelial (blue) layer. Scale bars, 40 µm (A1, A2, A3, A6, and A7) and 8 µm (A4, A5, A8, A9, and A10). B and C, Fluorescence-activated cell sorting pictures (B) and bar graph (C) show increased numbers of GFP⁺/CD31⁺ cells in the CD45⁻/CD31⁺/eNOS⁺ EC population in lungs of mice 8 weeks after LPS injection, compared with lungs of mice 8 weeks after saline injection (Con), indicating an increased BMDEPC engraftment. Mean±SEM of 5 mice per group. *P<0.05, compared with control mice. eNOS indicates endothelial nitric oxide synthase.

We quantified the engrafted BMDEPCs (CD45-/CD31⁺/eNOS⁺/GFP⁺) in lungs from mice 8 weeks after saline or lipopolysaccharide injection, a time point when organ inflammation and inflammation-associated BMDEPC recruitment has been subsided. FACS analysis showed that the number of CD45⁻/CD31⁺/eNOS⁺/GFP⁺ cells was ≈121-fold higher in lipopolysaccharide-injected than in saline-injected lungs (Figure 3B and 3C), indicating an increased BMDEPC engraftment in injured lungs. These results provide cytological evidence for BMDEPC engraftment into lung microvessels and suggest that BMDEPC contribute to endothelial barrier repair.

Different Contributions of RECs Versus BMDEPCs to Endothelial Barrier Repair

We next compare the relative contributions of RECs versus BMDEPCs to endothelial barrier repair. We counted the numbers of proliferating RECs and BMDEPCs in lungs at 48 hours and the numbers of REC- and BMDEPC-derived new ECs in lungs 8 weeks after lipopolysaccharide-induced injury. At 48 hours, the number of proliferating RECs was 2.4-fold higher than proliferating BMDEPCs in lipopolysaccharide-challenged lungs, although both proliferating RECs and BMDEPCs were remarkably higher than control lungs (Figure 4A and 4B). In lungs, 8 weeks after lipopolysaccharide-induced injury, total number of new ECs, defined as CD45⁻/CD31⁺/BrdU⁺ cells, accounted for ≈2% of total lung cells, which is in agreement with our previous observation that apoptotic ECs accounted for ≈2% of total lung cells in this mouse model of ALI.³⁰ Among the new EC subpopulation, 83% were rtTA⁺ (REC-derived), but only 9.5% were GFP⁺ (BMDEPC-derived; Figure 4C and 4D).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Resident endothelial cells (RECs) and bone marrow–derived endothelial progenitor cells (BMDEPCs) contribute differently to endothelial barrier repair. A, Representative micrographs of reverse tetracycline transactivator (rtTA)+5-bromo-2-deoxyuridine (BrdU; RECs) or green fluorescence protein (GFP)+BrdU (BMDEPCs) double staining of lung sections from EC-rtTA-GFP-BM mice 48 hours after saline (Con) or lipopolysaccharide (LPS) injection. The number of rtTA⁺/BrdU⁺ proliferating RECs was significantly higher than the number of GFP⁺/BrdU⁺ proliferating BMDEPCs (arrows indicated). Scale bar, 75 µm. B, Bar graph shows higher number of proliferating RECs (red bars) than proliferating BMDEPCs (black bars) in lungs at 48 hours post LPS. Mean±SEM of 5 mice per group. *P<0.05, compared with BMDEPCs at 48 hours. C and D, Fluorescence-activated cell sorting pictures (C) and bar graph (D) show that lung cells from EC-rtTA-GFP-BM mice 8 weeks after LPS injection have a remarkably higher number of REC-derived (rtTA⁺/CD31⁺) than BMDEPC-derived (GFP⁺/CD31⁺) ECs in the CD45⁻/CD31⁺/BrdU⁺ new EC population. Mean±SEM of 5 mice per group. *P<0.05, compared with BMDEPCs. E, Bar graph shows higher number of REC-derived new ECs (CD45⁻/CD31⁺/BrdU⁺/rtTA⁺) than engrafted BMDEPCs (CD45⁻/CD31⁺/eNOS⁺/GFP⁺) in lung cells from EC-rtTA-GFP-BM mice 8 weeks after LPS injection. Mean±SEM of 5 mice per group. *P<0.05, compared with engrafted BMDEPCs.

BMDEPCs may replace apoptotic/dead ECs by differentiation.⁹ We compared the number of REC-derived ECs (CD45⁻/CD31⁺/BrdU⁺/rtTA⁺ cells) to the number of total engrafted BMDEPCs (CD45⁻/CD31⁺/eNOS⁺/GFP⁺ cells), which includes BrdU⁺ and BrdU⁻, BMDEPC-derived ECs, in lungs of EC-rtTA-GFP-BM mice (Table II in the online-only Data Supplement) 8 weeks after lipopolysaccharide injection. REC-derived ECs constituted 22%, but the total engrafted BMDEPCs constituted only 3.7% of the total CD45⁻/CD31⁺ EC population (Figure 4E). Thus, both histological and FACS analyses revealed that RECs are a major source and BMDEPCs is a complementary source of new ECs in endothelial barrier repair in the lungs.

Targeted Inhibition of REC or BMDEPC Intrinsic NF-κB Activity Suppresses REC or BMDEPC Proliferation

We previously demonstrated that doxycycline-induced I-κBαmt expression inhibited NF-κB in endothelial lineage cells but not in other cell types in the EC-I-κBαmt mice.³¹ Using the EC-I-κBαmt mice as recipients or donors, we generated EC-I-κBα-WT-BM or WT-EC-I-κBα-BM chimeras (Table II in the online-only Data Supplement). We confirmed the high level of donor BM engraftment (>95%) in the chimeras (Figure IB and IC in the online-only Data Supplement). EC-I-κBα-WT-BM or WT-EC-I-κBα-BM mice overexpress I-κBαmt on RECs or BMDEPCs and enabled us to selectively block REC or BMDEPC intrinsic NF-κB activity through doxycycline-induced I-κBαmt expression, which leads to the suppression of REC or BMDEPC proliferation.

At 48 hours post lipopolysaccharide, tissue NF-κB activity is reduced to a low level. Techniques (immunofluorescence and immunohistochemistry) capable of revealing EC-selective NF-κB inhibition are not sensitive enough to detect the subtle change in NF-κB activity caused by doxycycline-induced I-κBmt expression. We verified that treatment of the chimeras with doxycycline induces I-κBαmt expression and inhibits NF-κB activity. We showed that injection of EC-I-κBα-WT-BM mice with doxycycline at 36 hours post lipopolysaccharide induced high level of I-κBαmt mRNA expression (Figure III in the online-only Data Supplement) and repressed lung tissue level of vascular cell adhesion molecule protein, a widely used marker of endothelial NF-κB activity, at 48 hours (Figure 5A).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Targeted suppression of resident endothelial cell (REC) and bone marrow–derived endothelial progenitor cell (BMDEPC) proliferation at barrier repair phase augments endothelial permeability. Wild type (WT), EC-rtTA-GFP-BM (rtTA-GFP), EC-I-κB-WT-BM (I-κB-WT), WT-EC-rtTA-BM (WT-rtTA), and WT-EC-I-κBα-BM (WT-I-κB) mice (Table II in the online-only Data Supplement) were injected with saline (Con) or lipopolysaccharide (LPS; 48 h) and then with doxycycline (Dox; 0.5 mg/mouse IP) 36 hours after LPS injection. At 48 hours after LPS, lung cryosections were prepared or tissue level of vascular cell adhesion molecule (VCAM)-1 protein determined or lung endothelial permeability measured using Evan blue dye (EBD) index. A, Western blot photographs show that Dox-induced I-κBαmt expression at 48 hours reduces the lung tissue level of VCAM-1 protein, a nuclear factor-κB (NF-κB)–dependent gene product, in EC-I-κB-WT-BM mice, indicating an inhibition of REC NF-κB activity. TA-C, EC-rtTA-GFP-BM control, I-κB-C, EC-I-κB-WT-BM control, TA48, EC-rtTA-GFP-BM 48 hours, and I-κB48, EC-I-κB-WT-BM 48 hours. Representative of 3 independent experiments. B and C, Bar graphs show that targeted blockade of REC or BMDEPC intrinsic NF-κB activity suppresses REC or BMDEPC proliferation. Lung cryosections were stained with 5-bromo-2-deoxyuridine (BrdU)+reverse tetracycline transactivator (rtTA) antibodies and nuclei counterstained with DAPI. The number of BrdU⁺/rtTA⁺ proliferating RECs (B) or BrdU⁺/rtTA⁺ proliferating BMDEPCs (C) was counted and expressed as a percentage of total BrdU⁺ cells. Mean±SEM of 5 mice per group. *P<0.05, compared with Con groups. #P<0.05, compared with 48-hour group of rtTA-GFP or Wt-rtTA mice. D, Suppression of REC or BMDEPC proliferation by REC- or BMDEPC-targeted NF-κB blockade was associated with an impeded endothelial barrier recovery. Inhibition of REC (white versus black bars) or BMDEPC (white versus grey bars) proliferation was associated with an augmented endothelial permeability. Mean±SEM of 8 mice per group. *P<0.05, compared with Con groups. #P<0.05, compared with WT-48 hour group.

Targeted inhibition of REC or BMDEPC intrinsic NF-κB activity suppressed REC or BMDEPC proliferation. At 48 hours after lipopolysaccharide and 12 hours after doxycycline injection, the number of proliferating RECs or BMDEPCs in lung sections of EC-I-κBα-WT-BM or WT-EC-I-κBα-BM mice (Table II in the online-only Data Supplement, with NF-κB inhibition) was significantly lower than that in lung sections of EC-rtTA-GFP-BM or WT-EC-rtTA-BM mice (Table II in the online-only Data Supplement, without NF-κB inhibition; Figure 5B and 5C).

Suppression of REC or BMDEPC Proliferation at Active Repair Phase Was Associated With an Augmented Endothelial Permeability

At 48 hours after lipopolysaccharide and 12 hours after doxycycline injection, wild-type (WT) mice, in which REC or BMDEPC proliferation was not inhibited, displayed a moderate increase in endothelial permeability that is consistent with endothelial barrier repair phase (Figure 5D). EC-I-κBα-WT-BM or WT-EC-I-κBα-BM mice, in which REC or BMDEPC proliferation was inhibited, exhibited a significantly augmented endothelial permeability (Figure 5D). Blockade of REC intrinsic NF-κB activity caused a 42% reduction in REC proliferation and 41% augmentation of endothelial permeability in the EC-I-κBα-WT-BM mice (Figure 5B versus 5D), suggesting that augmentation of endothelial permeability and suppression of REC proliferation are causally related.

Proliferation-Independent Mechanisms Do Not Contribute to the Augmented Endothelial Permeability Caused by Endothelial NF-κB Blockade at 48 Hours

In addition to controlling EC proliferation, NF-κB mediates endothelial permeability by inducing inflammatory gene expression²⁶ and by disrupting interendothelial junctions, as results of myosin light chain (MLC) phosphorylation and junction protein internalization and downregulation.^33–36 To clarify whether these proliferation-independent mechanisms may contribute to the augmented endothelial permeability caused by endothelial NF-κB blockade at 48 hours, we compared lung tissue levels of NF-κB–regulated cytokines, tumor necrosis factor-α, interleukin-1β, interleukin-6, and chemokine (C-X-C motif) ligand 1, between mice at 6 and 48 hours to assess organ inflammatory status at 48 hours, and between WT and EC-I-κBαmt mice to examine the effects of endothelial NF-κB blockade. We used EC-I-κBαmt mice (Table I in the online-only Data Supplement), the recipients or donors of EC-I-κBα-WT-BM or WT-EC-I-κBα-BM chimeras (Table II in the online-only Data Supplement), for these studies. The use of EC-I-κBαmt mice enabled us to evaluate the effects of inhibiting REC and BMDEPC intrinsic NF-κB activities simultaneously. We also compared tissue levels of phospho–MLC-2 and membrane-bound VE-cadherin, 2 major markers of disruption of interendothelial junctions,^33–36 between the 2 time points and the 2 groups of mice.

Lung tissue levels of the 4 cytokines at 48 hours were higher than that of controls but were many-fold lower than that at 6 hours, and, importantly, they were not affected by EC-restricted NF-κB inhibition (Figure IVA–IVD in the online-only Data Supplement). At 48 hours post lipopolysaccharide, lung tissue levels of phospho–MLC-2 and membrane-bound and cytoplasmic VE-cadherin proteins were all at control levels and were not affected by EC-restricted NF-κB inhibition (Figure IVE and IVF and Figure V in the online-only Data Supplement). By contrast, the lung tissue level of phospho–MLC-2 was several-fold higher at 3 hours, and the level of membrane-bound or cytoplasimic VE-cadherin protein was several-fold lower or higher at 6 hours than controls and at 48 hours, indicating increased MLC-2 phosphorylation and VE-cadherin internalization at 6 hours but not at 48 hours. All these changes were abrogated by EC-restricted NF-κB blockade (Figure IVE and IVF and Figure V in the online-only Data Supplement). This result excludes the possible involvement of NF-κB–mediated, proliferation-independent mechanisms in the augmentation of endothelial permeability caused by endothelial NF-κB blockade at 48 hours.