Prenatal VEGF Nanodelivery Reverses Congenital Diaphragmatic Hernia-associated Pulmonary Abnormalities

Abstract

Rationale: Congenital diaphragmatic hernia (CDH) results in lung hypoplasia. In severe cases, tracheal occlusion (TO) can be offered to promote lung growth. However, the benefit is limited, and novel treatments are required to supplement TO. VEGF (vascular endothelial growth factor) is downregulated in animal models of CDH and could be a therapeutic target, but its role in human CDH is not known. Objectives: To investigate whether VEGF supplementation could be a suitable treatment for CDH-associated lung pathology. Methods: Fetal lungs from patients with CDH were used to determine pulmonary morphology and VEGF expression. A novel human ex vivo model of fetal lung compression recapitulating CDH features was developed and used to determine the effect of exogenous VEGF supplementation. A nanoparticle-based approach for intrapulmonary delivery of VEGF was developed by conjugating it on functionalized nanodiamonds, which was tested in experimental CDH in vivo. Measurements and Main Results: VEGF expression was downregulated in the distal pulmonary epithelium of human CDH fetuses in conjunction with attenuated cell proliferation. The compression model resulted in impaired branching morphogenesis similar to CDH and downregulation of VEGF expression in conjunction with reduced proliferation of terminal bud epithelial progenitors; these could be reversed by exogenous supplementation of VEGF. Prenatal delivery of VEGF with the functionalized nanodiamond VEGF platform in CDH fetal rats resulted in lung growth and pulmonary arterial remodeling that was complementary to that achieved by TO alone with appearances comparable to healthy controls. Conclusions: This innovative approach could have a significant impact on the treatment of CDH.

Keywords: VEGF; alveolar epithelium; congenital diaphragmatic hernia; mechanical compression; nanoparticles.

Figure 1.

(A) Schematic illustration of the study design based on the analysis of the early-stage congenital diaphragmatic hernia (CDH) fetal lung sections to assess early hypoplasia and VEGF (vascular endothelial growth factor) impairment. (B) Hematoxylin and eosin staining analysis of postmortem CDH human lung tissue reveals impaired epithelial development compared with control healthy tissue at matching developmental stages. Scale bars, 100 μm. (C) Morphometric analysis of CDH and healthy lung tissue at 18–22 postconception weeks reveals decreased values of volume densities of air space in CDH tissues compared with healthy tissue at the same developmental stages. A similar trend can be observed for the mean linear intercept of air space. (D) Histologic analysis of VEGFA, KDR, Ki67, and NKX2-1 markers in normal (n = 3 donors) and CDH (n = 2 donors) fetal lung tissue in the range of 18–22 postconception weeks. Scale bars, 50 μm. (E) Quantification of epithelial VEGFA and KDR expression reported in D shows a significant decrease for both markers in CDH compared with healthy samples. Correspondingly, the percentage of Ki67-positive epithelial cells is significantly lower in CDH compared with healthy samples, with no significant differences in the number of NKX2-1–positive cells (unpaired t test, ***P < 0.001). (F) Immunofluorescence analysis of E-cadherin in normal and CDH fetal lung tissue. Scale bars, 100 μm. (G) Quantification of epithelial nuclear density based on the results in E shows no significant differences between CDH and healthy tissue (unpaired t test). Data in the scatter plots are presented as mean ± SD. Dots represent expression values from different tips from all biological replicates. a.u. = arbitrary units; %Lmair = linear intercept of air space; ns = not significant; pcw = postconception weeks; %Vair = volume densities of air space.

Figure 2.

(A) Experimental setup of the ex vivo model of mouse fetal lung compression during pseudoglandular stage. Freshly isolated mouse fetal lungs (gestational day 12.5) were embedded in Matrigel drops and cultured in air–liquid interface. The next day (Day 0), a 20% coumarin HCC-conjugated eight-arm polyethylene glycol gel solution was dispersed into the Matrigel drop and selectively cross-linked under a two-photon microscope to surround the lungs and inhibit their growth over a total of 2 days. Control lungs had the same amount of polyethylene glycol gel solution added to the Matrigel, but this was not cross-linked with the two-photon laser. (B) Atomic force microscopy analysis confirms that the stiffness of the gel in culture conditions after 2 weeks from printing is approximately 20-fold higher compared with Matrigel. (C) Representative time-lapse imaging of lung growth under physical confinement (compression) compared with control. Scale bars, 100 μm. (D) Quantification of left lobe surface area (left) and terminal buds (right) in control and compressed lungs after 2 days of mechanical confinement for the following biological replicates: control (n = 6) and compression (n = 4). Mann-Whitney test: *P < 0.05, **P < 0.01, and ***P < 0.001. Data are presented as mean ± SD. Dots represent biological replicates. (E) Representative hematoxylin and eosin staining of control (n = 2) and compressed (n = 2) lungs after 2 days of mechanical confinement. Scale bars, 100 μm. (F) EdU (5-ethynyl-2-deoxyuridine) incorporation assay after 2 days of mechanical confinement for the following biological replicates: control (n = 3) and compression (n = 3) (unpaired t test, *P < 0.05). Data are presented as mean ± SD. Dots represent single-tip quantification from all biological replicates. (G) Volcano plot of bulk RNA sequencing analysis of compressed versus control lungs after 2 days of mechanical confinement. Differentially expressed genes have been identified with fold change >1.1 and P < 0.05. Differentially expressed gene analysis shows downregulation of Vegfa in compressed lungs. (H) Gene network analysis of overexpressed genes in the compressed lungs reveals upregulation of categories related to inflammatory and immune response. Ecad = E-cadherin; PEG-HCC = coumarin HCC-conjugated eight-arm polyethylene glycol gel.

Figure 3.

(A) Experimental setup of the ex vivo model of human fetal lung compression during the pseudoglandular stage. (B) Representative time-lapse imaging of human lung tissue growth under physical confinement (compression) compared with control (n = 4 biological replicates; scale bars, 100 μm). (C) Quantification of human lung tissue surface area (left) and terminal buds (right) in control and compressed lungs after 6 days of mechanical confinement for the following biological replicates: control (n = 3) and compression (n = 3). Mann-Whitney test, *P < 0.05, and ***P < 0.001. Data in the scatter plots are presented as mean ± SD. Dots represent different technical replicates from all biological replicates. (D) Finite element method (FEM) analysis validation based on the time-course experiment in Figure E2E. Comparison of normalized area of the lung sample over time for experimental data and numerical values from FEM analysis. FEM analysis was developed by adopting two values of the Young’s modulus to account for the uncertainty about the stiffness of the human lung tissue. (E) Pressure field prediction. Representative pressure field (in kPa) for control and compressed lung fragments at Day 6, assuming a Young’s modulus of 1.5 kPa for the human lung tissue. (F) Uniform Manifold Approximation and Projection (UMAP) visualization of 32,630 fetal lung cells from three biological replicates. Leiden clustering revealed 18 distinct cell types within mesenchymal, epithelial, endothelial, and hematopoietic fractions. (G) Magnified UMAP plot of the epithelial compartment cells. Distal tip cells (tips⁺) were identified by coexpression of known tip markers (SOX9⁺, TESC⁺); the remaining cells comprise non-tip epithelium (tips⁻). (H) GO enrichment analysis reveals differential expression of several gene profiles in tips⁺ cells of the congenital diaphragmatic hernia model. (I) Relative expression of 11 proliferation-associated genes and 3 VEGF-family genes between control and congenital diaphragmatic hernia model tips. (J) Experimental setup of the human ex vivo model to test exogenous VEGF supplementation. After 7 days of culture in basal medium, compressed human lung fragments were treated with recombinant VEGF for an additional 7 days using the specific KDR/Flk1 (VEGF receptor 2) inhibitor SU5416 as a negative control. (K) Representative pictures of EdU (5-ethynyl-2-deoxyuridine) proliferation assay after 7 days of VEGF treatment (compression + VEGF) compared with control, compression only, and compressed samples treated with the VEGF-specific inhibitor SU5416. Immunostaining for E-cadherin was used to identify epithelial cells. Scale bars, 100 μm. (L) Quantification of the number of E-cadherin/EdU double-positive cells after 7 days of VEGF treatment for the following biological replicates: control (n = 2) and compression (n = 2) (unpaired t test, ***P < 0.001). Data are presented as mean ± SD. Dots represent values from different epithelial tips from different technical replicates of the same biological replicate. ECAD = E-cadherin; GO = Gene Ontology; ns = not significant.

Figure 4.

(A) Experimental design of in vivo intervention studies in fetal rats using VEGF-loaded nanodiamonds (ND-VEGF). There were eight experimental groups: healthy control (non–congenital diaphragmatic hernia [CDH], olive oil gavage-fed mothers; n = 10), sham intervention (surgery but no injection or tracheal occlusion [TO] in CDH fetuses from nitrofen-fed mothers; n = 18 underwent sham surgery), PBS + TO (intratracheal vehicle injection at gestational day 19 [E19] followed by TO in CDH fetuses; n = 17 underwent PBS solution injection followed by TO), free VEGF + TO (free VEGF injection followed by TO; n = 40), ND + TO (unconjugated ND-NH₂ injection followed by TO; n = 35), ND-VEGF + TO (ND-VEGF injection followed by TO; n = 34), SU5416 + TO (KDR/Flk1 inhibitor SU5416 injection followed by TO; n = 11), and ND-VEGF+SU5416 + TO (ND-VEGF and SU5416 coinjection followed by TO; n = 13). Animals in the groups with ND-VEGF administration received 13.4 ± 1.9 μg ND and 100 ng VEGF, whereas animals in the ND + TO and VEGF + TO groups received the same amounts of ND-NH₂ and free thiolated VEGF in 50 μl of PBS solution. The concentration of the KDR/Flk1 inhibitor SU5416 in the injection solution used in SU5416 + TO and ND-VEGF + SU5416 + TO groups was 120 μg/ml. (B) Confirmation of the presence of diaphragmatic defect in E18 fetal rats using micro-ultrasound (pink outline, lung; purple outline, heart; white outline, liver). (C) Representative confocal microscopy image demonstrating localization of ND-VEGF in airways at gestational day 21 following gestational-day 19 intratracheal administration of TO, with evidence of uptake/retention of VEGF by E-cadherin–expressing alveolar epithelial cells (E-cadherin, green; VEGF, red; DAPI, blue; magnification columns 1–3, ×20; magnification column 4, ×63; scale bars, 100 μm). E = gestational day.

Figure 5.

(A) Representative images of gestational day 21 (E21) lungs from rats in the intervention study (left to right: sham, PBS solution +  tracheal occlusion [TO], VEGF + TO, nanodiamond (ND)-VEGF + TO). (B) Summary of lung–to–body weight ratio data (expressed as a percentage; *P < 0.0001, healthy [n = 8] vs. sham [n = 11], PBS + TO [n = 9], ND + TO [n = 14], VEGF + TO [n = 10], SU5416 + TO [n = 7] and ND-VEGF + SU5416 + TO [n = 8]; ^†P < 0.0001 vs. PBS + TO and ND-VEGF + TO [n = 12]; ^‡P < 0.001 vs. ND + TO and VEGF + TO; ^§P < 0.01 vs. ND-VEGF + SU5416 + TO; ▪P < 0.05 vs. SU5426 + TO; •P < 0.0001 vs. PBS + TO, ND + TO, VEGF + TO, SU5426 + TO, and ND-VEGF + SU5416 + TO; one-way ANOVA with Bonferroni post hoc tests). (C) Summary of total protein content data (expressed as μg/mg of lung tissue; ★P < 0.0001 healthy [n = 8] vs. sham [n = 5], PBS + TO [n = 7], ND + TO [n = 12], VEGF + TO [n = 5], SU5416 + TO [n = 7], and ND-VEGF + SU5416 + TO [N = 8]; ^†P < 0.0001 vs. PBS + TO and ND-VEGF + TO [n = 9]; ^‡P < 0.001 vs. ND + TO and VEGF + TO; ^§P < 0.01 vs. ND-VEGF + SU5416 + TO; ▪P < 0.05 vs. SU5426 + TO; •P < 0.0001 vs. PBS + TO, ND + TO, VEGF + TO, SU5426 + TO, and ND-VEGF + SU5416 + TO; one-way ANOVA with Bonferroni post hoc tests). (D) Representative images of sections of E21 rat lungs from animals included in the in vivo intervention studies. Tissues were stained with hematoxylin and eosin and were used for subsequent airway morphometry analyses (magnification, ×20; scale bars, 100 μm). (E) Summary of mean linear intercept of wall transection length data (★P < 0.0001, healthy [n = 6] vs. sham [n = 11]; ^†P < 0.01 vs. VEGF + TO [n = 10], ND + TO [n = 14], SU5426 + TO [n = 7], and ND-VEGF + SU5426 + TO [n = 8]; ^‡P < 0.05 vs. PBS + TO [n = 9]; ^§P < 0.0001 vs. ND-VEGF + TO [n = 11]; ▪P < 0.001 vs. PBS + TO; •P < 0.01 vs. VEGF + TO, ND + TO, SU5416 + TO, and ND-VEGF + SU5416 + TO; ♦P < 0.001 vs. ND + TO, SU5416 + TO, and ND-VEGF + SU5416 + TO; □P < 0.01 vs. VEGF + TO; ⴲP < 0.05 vs. PBS + TO; one-way ANOVA with Bonferroni post hoc tests) and mean linear intercept of parenchymal airspace data (★P < 0.0001, healthy [n = 7] vs. sham [n = 11]; ^†P < 0.01 vs. VEGF + TO [n = 10], ND + TO [n = 12], SU5426 + TO [n = 6], and ND-VEGF + SU5426 + TO [n = 6]; ^‡P < 0.05 vs. PBS + TO [n = 8]; ^§P < 0.0001 vs. ND-VEGF + TO [n = 9]; ▪P < 0.001 vs. PBS + TO; •P < 0.01 vs. VEGF + TO, ND + TO, SU5416 + TO, and ND-VEGF + SU5416 + TO; ♦P < 0.001 vs. ND + TO, SU5416 + TO, and ND-VEGF + SU5416 + TO; □P < 0.01 vs. VEGF + TO; ⴲP < 0.05 vs. PBS + TO; one-way ANOVA with Bonferroni post hoc tests). (F) Representative images of lung sections from in vivo intervention studies stained for SPC (surfactant protein C; brown) using immunohistochemistry (IHC; magnification, ×20; scale bars, 100 μm). (G) Summary of SPC⁺ IHC index (ratio of the number of SPC⁺ cells over total cell number in predetermined areas of lung parenchyma; expressed as percentage) data (★P < 0.0001, healthy [n = 7] vs. sham [n = 7], PBS + TO [n = 6], ND + TO [n = 9], SU5416 + TO [n = 6], and ND-VEGF + SU5416 + TO [n = 10]; ^†P < 0.01 vs. VEGF + TO [n = 7]; ^‡ND-VEGF + TO [n = 9]; P < 0.0001 vs. sham, PBS + TO, VEGF + TO, ND + TO, SU5416 + TO, and ND-VEGF + SU5416 + TO; one-way ANOVA with Bonferroni post hoc tests). (H) Representative images of sections of E21 rat lungs from animals included in the in vivo intervention studies. Miller’s elastic staining was performed, and stained sections were used for subsequent vascular morphometry analyses (peripheral arterioles 30–50 μm in diameter; magnification, ×20; scale bars, 100 μm). (I) Summary of medial thickness data (expressed as a percentage of the external diameter of the blood vessel) (★P < 0.0001, healthy [n = 7] vs. sham [n = 7]; ^†P < 0.01 vs. VEGF + TO [n = 7], ND + TO [n = 10], SU5426 + TO [n = 7], and PBS + TO [n = 8]; ^‡P < 0.05 vs. ND-VEGF +  SU5426 + TO [n = 8]; ^§P < 0.0001 vs. ND-VEGF + TO [n = 9]; ▪P < 0.001 vs. PBS + TO, VEGF + TO, ND + TO, SU5416 + TO, and ND-VEGF +  SU5416 + TO; •P < 0.01 vs. ND + TO; ♦P < 0.001 vs. PBS + TO and ND-VEGF + SU5416 + TO; □P < 0.01 vs. ND-VEGF + SU5426 + TO; one-way ANOVA with Bonferroni post hoc tests) and summary of adventitial thickness data (expressed as a percentage of the external diameter of the blood vessel). (J) Representative images of lung sections from in vivo intervention studies stained for thrombomodulin (brown) using IHC. (K) Summary of thrombomodulin⁺ IHC index (ratio of the number of thrombomodulin⁺ cells over total cell number in predetermined areas of lung parenchyma; expressed as percentage) data (★P < 0.0001, healthy [n = 7] vs. sham [n = 7], PBS + TO [n = 6], ND + TO [n = 9], SU5416 +  TO [n = 6], and ND-VEGF + SU5416 + TO [n = 10]; ^†P < 0.01 vs. VEGF + TO [n = 7]; ^‡P < 0.001 ND-VEGF + TO [n = 9] vs. sham, PBS + TO, ND + TO, SU5416 + TO, and ND-VEGF + SU5416 + TO; one-way ANOVA with Bonferroni post hoc tests). Lma = linear intercept of air space; Lmw = linear intercept of wall transection length.

Figure 6.

(A) Experimental design of in vivo intervention studies in fetal rabbits using VEGF-loaded nanodiamonds (ND-VEGF). There were four experimental groups: healthy control (non–congenital diaphragmatic hernia; n = 6), sham intervention (surgery but no injection or tracheal occlusion/tracheal occlusion [TO] in congenital diaphragmatic hernia fetuses; n = 6 underwent sham surgery), ND-VEGF(low) + TO (ND-VEGF with low-dose VEGF injection followed by TO; n = 6 underwent intratracheal injection followed by TO), and ND-VEGF(high) + TO (ND-VEGF with high-dose VEGF injection followed by TO; n = 7). Animals in the groups with ND-VEGF(low) administration received 375 ± 16 μg ND and 1 μg VEGF, whereas animals in the ND-VEGF(high) group received 375 ± 27 μg ND and 5 μg VEGF. (B) Summary of fetal survival data. (C) Summary of lung–to–body weight ratio data (expressed as a percentage) (★P < 0.05, sham [n = 6] vs. healthy [n = 6]; ^†P < 0.01 vs. ND-VEGF[low]  + TO [n = 6]; one-way ANOVA with Bonferroni post hoc tests; n = 6 for the ND-VEGF[high] group). (D) Summary of mean linear intercept of wall transection length data (★P < 0.05, ND-VEGF[high] + TO [n = 6] vs. sham [n = 6]; one-way ANOVA with Bonferroni post hoc tests; n = 6 for the ND-VEGF[low] group). (E) Summary of mean linear intercept of parenchymal airspace data (★P < 0.01, ND-VEGF[high] + TO [n = 6] vs. healthy [n = 6] and ND-VEGF[low] + TO [n = 6]; ^†P < 0.05, sham [n = 6] vs. healthy and ND-VEGF[low] + TO; one-way ANOVA with Bonferroni post hoc tests). (F) Representative micro–computed tomography images (three-dimensionally reconstructed) of the pulmonary vasculature of fetal rabbits in the ND-VEGF(low) + TO and ND-VEGF(high) + TO groups. Such images were used for assessment of pulmonary vascular branching. (G) Summary of pulmonary vascular branching (at least fifth order; peripheral) data (expressed as percentage of total; ★P < 0.001, ND-VEGF[high] + TO [n = 4] vs. healthy [n = 6] and ND-VEGF[low] + TO [n = 3]; ^†P < 0.01, sham [n = 6] vs. healthy and ND-VEGF[low] + TO; one-way ANOVA with Bonferroni post hoc tests). (H) Summary of pulmonary artery acceleration/ejection time data obtained by pulmonary echocardiography (★P < 0.05, sham [n = 6] vs. healthy [n = 6], ND-VEGF[low] + TO [n = 8] and ND-VEGF[high] + TO [n = 3]; one-way ANOVA with Bonferroni post hoc tests). (I) Summary of medial thickness data (expressed as a percentage of the external diameter of the blood vessel; n = 6 for all experimental groups). AT/ET = acceleration time/ejection time; E = gestational day; LBWR = lung–to–body weight ratio; Lma = linear intercept of air space; Lmw = linear intercept of wall transection length.