Collagen fiber regulation in human pediatric aortic valve development and disease

Figures

• 

  Figure 1. Histology analysis identifies ECM mixing and collagen dominance in pCAVS valves at a young end-stage. (A) Examples Movat’s Pentachrome staining between normal and age matched pCAVS samples. Blue stain corresponds to glycosaminoglycan (GAG) regions, dark purple to elastin, and yellow to collagen. Normal valves show traditional trilayer structure, while pCAVS valves show lack of trilayer and areas of ECM mixing. (B) Quantification of Movat’s Staining via color deconvolution analysis. Blue = GAG, purple = Elastin, yellow = collagen. Shown as a percent of total Movat’s stain area. (C) Multinomial logistical regression analysis of Movat’s Pentachrome Color Deconvolution data shows pCAVS valves are collagen dominant at a younger age than normal patients. Variables: ECM dominant category (GAG, elastin, or collagen); Patient Category (normal or pCAVS); Age (0–18 years). (D–F) Quantification of Elastin, GAG, and Collagen (respectively) Movat’s Pentachrome color deconvolution results (Fig. 1B) comparing two pediatric age groups: young child (Newborn to < 6 years of age) and Child-Adolescent (6–18 years of age). **p-value 0.008, MWU test. pCAVS GAG decrease is concomitant with collagen…

• 

  Figure 2. Pediatric CAVS demonstrate high deposition of immature collagen compared to age-matched controls. (A,B) Herovici staining of SKU DB27, a 17-year-old normal sample with regions of Hinge, Belly, and Free Margin 10 × zoom shown (B). Blue = immature. Purple-mature. Normal immature collagen production is localized to the ventricularis, while pCAVS immature production becomes increasing dysregulated towards the free margin. (C,D) Herovici staining of SKU DB16, an age-matched pCAVS sample with regions of Hinge, Belly, and Free Margin 10 × zoom shown in (D). (E) Patient specific breakdown of Herovici color deconvolution analysis. pCAVS patients show increased immature collagen production compared to age matched controls. (F) Boxplot showing quantification of Herovici color deconvolution, showing larger heterogeneity amongst pCAVS patients. Scale Bar: 300 μm. Normal n = 4; pCAVS n = 11; AVI n = 5.

• 

  Figure 3. Pediatric end-stage CAVS valves have a greater number of collagen fibers with localized regions of collagen density. (A) Collagen fiber visualization of select pCAVS and normal valves. From left to right: Second Harmonic Generation (SHG); SHG multichannel; Polarized picrosirius red (PSR); PSR Regions on Increased Fiber Density (RIFD) Masque; Movat’s Pentachrome. Movat’s Pentachrome: blue = GAGs, yellow = collagen, and purple = elastin. Scale Bar = 200 μm. (B) Boxplot quantification of picrosirius red color deconvolution analysis, showing percent thick stain over total PSR stain. (C) Patient-specific data of PSR color deconvolution analysis, showing percent thick stain over total PSR stain. pCAVS and AVI patients show larger heterogeneity across age groups compared to normal. (D) Boxplot quantification of collagen RIFD areas normalized to total valve area. *p-value 0.024, MWU test. pCAVS and AVI patients have significantly more regions of high collagen fiber density compared to normal patients. (E) Boxplot showing quantification of total number of collagen fibers as measured by CT-FIRE analysis. *p-value 0.014, MWU test. pCAVS valves have significantly more collagen fibers than…

• 

  Figure 4. Pediatric CAVS valves have reduced HYP, including in collagen-ECM binding sites. (A) A log2(fold change) scatterplot of pCAVS versus Normal peptides. Red squares are hydroxyproline containing peptides, gray squares are unmodified peptides. Bold lines indicate X + 2 and X − 2. Top 10 collagen peptides are annotated with corresponding sequence and protein shown (A, right). P(X) where X is between 0–1, indicated HYP probability at that site. Significantly regulated collagen subtypes differ between pCAVS and normal, with non-fibril types (multiplexin, FACIT) differentially regulated in normal. (B) Representative hydroxyproline site mapping for collagen 1A1, collagen 1A2, and collagen 3A1, shown for Normal and pCAVS patients (DB27 and DB16, respectively). Normal sample contained more HYP sites than age matched controls. Unique HYP sites in Normal (top, red) are found in glycoprotein vi and integrin binding motifs (black and blue arrows, respectively). HYP is required in these motifs for binding. (C) Quantification of hydroxyproline state of proteins in all peptides identified in normal (black) or pCAVS (gray). (D) Hydroxyproline percent in peptides containing glycoprotein vi,…

• 

  Figure 5. Extracellular matrix proteomics implicates collagen binding and collagen organization pathways. (A) Aortic valve extracellular matrix proteins identified by the collagen-targeting proteomics. A total of 18 collagen proteins were identified with COL1A1, COL1A2 and COL6A3 being the most abundant collagens. Protein level quantification, which includes all peptides both unmodified and modified, revealed significant differences based on valve diagnoses in COL1A1, COL1A2, COL5A1, BGH3, FN1, and COL5A2. *Mann–Whitney U adj. p-value < 0.05 normal AV compared to pCAVS, †Mann–Whitney U adj. p-value < 0.05 normal AV compared to AVI, ‡Mann–Whitney U adj. p-value < 0.05 AVI compared to pCAVS. (B) Gene Ontology Functional classification of non-collagen type proteins identified in the proteomics dataset. (C) Gene Ontology Fold Enrichment analysis of collagen type function by non-collagenous proteins identified by proteomics, Fisher’s exactp-value ≤ 1.84E–5, false discover rate ≤ 6.66E–3. Proteins associated with collagen binding had the highest degree of fold enrichment, followed by collagen fibril organization. (D) Upstream analysis of the ECM proteome by Ingenuity Pathways Analysis to…

• 

  Figure 6. Proteomics based protein–protein interaction networks identify BAMBI as a potential master regulator of the collagen interactome in pCAVS. (A) Protein–Protein Interaction network showing relationship BAMBI (Supplemental Figure S5) relative to the ECM proteome and key valvular development proteins (Supplemental Table S6). (B) Quantification of qRT-PCR expression data on BAMBI (normalized to GAPDH), between patient categories. One-way ANOVA shows significant differential expression between patient categories (*p-value 0.013), with significant reduction in pCAVS (MWU *p-value 0.045) and trending reduction in aCAVS (MWU p-value 0.053) compared to normal pediatric patients. (Normal n = 2; pCAVS n = 12; aCAVS n = 5) C. Quantification of qRT-PCR expression data on BAMBI (normalized to GAPDH) of CAVS patients, compared between pediatric age categories (Infant n = 2; Young Child n = 3; Child n = 2; Adolescent n = 5; Adult n = 6). A suggestive trending decrease of BAMBI expression as a function of pediatric age categories is seen (ANOVA p-value 0.096) (Table 1; Supplemental Table S7). (D) DAB stain area quantification of BAMBI IHC of select valves in the cohort. (E) Representative…