Fig 2: The collagen cross-linking enzymes PLOD2 and LOXL2 are co-expressed at sites of active fibrogenesis in IPF.(A–F) Expression of LOX, LOXL1, LOXL2, LOXL3, LOXL4, and PLOD2 in healthy alveolar septae, IPF alveolar septae and IPF fibroblast foci (n = 10 individual healthy and IPF donors). Relative expression levels are calculated as Fragments Per Kilobase of transcript per Million mapped reads (FPKM). Bars represent standard geometric means. **p < 0.01; ****p < 0.0001 by Tukey’s multiple comparisons test. (G) Scatterplot of paired fibroblast foci data from (C) and (F) were plotted to compare expression of PLOD2 and LOXL2 (Spearman rank correlation coefficient r = 0.63, p = 0.04). (H) Representative image of mRNA expression of PLOD2 (red chromagen) and LOXL2 (green chromagen) in IPF lung tissue (n = 7 donors) using RNAscope RNA in-situ hybridisation. A fibroblastic focus is identified by * and arrows identify co-expression pattern. Left scale bar 100 µm, right scale bar 20 µm.

Fig 3: HIF pathway activation promotes PLOD2 and LOXL2 gene expression relative to fibrillar collagen expression.Lung fibroblasts from IPF donors (n = 3 across two independent experiments) were cultured in the presence or absence TGFß1, DMOG, combined TGFß1 and DMOG, or vehicle control for 48 hr. (A, B) Relative gene expression of PLOD2 (A) and LOXL2 (B) using the ??Ct method. Bars indicate geometric means. Data are mean ± s.d. **p < 0.01; ***p < 0.001; ****p < 0.0001 by Dunnett’s multiple comparisons test. (C) PLOD2 and LOXL2 protein levels. ß-actin was used as a loading control. (D) Protein expression of LOXL2 in conditioned media. Ponceau S staining showing total protein levels. The full blots are shown in Figure 4—source data 1. Bars in graph indicate geometric means. Data are mean ± s.d. **p < 0.01; ***p < 0.001; ****p < 0.0001 by Dunnett’s multiple comparisons test. (E, F) Expression of PLOD2 and LOXL2 from (A and B) was divided by COL1A1 expression (shown in Figure 4—figure supplement 2) to calculate proportionate expression changes of cross-linking enzymes relative to collagen fibrillogenesis gene expression. Bars indicate geometric mean. Grouped analysis was performed using Dunnett’s multiple comparison test. * p < 0.05, ** p < 0.01, *** p < 0.001, ****p < 0.0001. Figure 4—source data 1.Full membrane scans for western blot images for Figure 4a and b.

Fig 4: Hypoxia mimetics strongly promote PLOD2 and LOXL2 expression in lung fibroblasts.(A–B, D) Relative gene expression using the ??Ct method of PLOD2, LOXL2, and COL1A1 in healthy lung fibroblasts over a 72-hr time course in the presence of EGF, TGFß1, the hypoxia mimetic DMOG, Wnt3a, Wnt5a, or vehicle control. n = 3 independent experiments. Bars indicate geometric means. *p < 0.05; ***p < 0.001; ****p < 0.0001 by Dunnett’s multiple comparisons test. (C) PLOD2 and LOXL2 protein levels at 72 hr. ß-actin loading control. The full blots are shown in Figure 2—source data 1. (E) Representative immunofluorescence images of healthy lung fibroblasts with indicated treatment stained for LOXL2 (red), PLOD2 (green), and DAPI (blue). Scale bar 50 µm. Figure 2—source data 1.Full membrane scans for western blot images for Figure 2c.

Fig 5: Pseudohypoxia and loss of FIH activity promotes HIF pathway signalling in IPF fibroblasts and increases tissue stiffness.(A) HIF GSVA scores calculated in human lung fibroblasts derived from control or patients with interstitial lung disease (scleroderma lung or a usual interstitial pneumonia / IPF pattern) (GSE40839). Data are mean ± s.d. ***p < 0.001; ****p < 0.0001 by Dunnett’s multiple comparisons test. (B) HIF GSVA scores calculated in human bronchoalveolar lavage derived mesenchymal stromal cells from patients with stable and progressive IPF (GSE73854). Data are mean ± s.d. *p < 0.05 by the unpaired t test. (C) PLOD2, LOXL2, HIF1ß, FIH, and ß-tubulin protein levels in lung fibroblasts from patients with IPF transfected with indicated siRNA. ß-tubulin was used as a loading control. The full blots are shown in Figure 6—source data 1. (D) Diagram explaining the HIF1a CAD reporter assay in E and F. In brief, the FIH asparaginyl hydroxylase hydroxylates HIF1a CAD, inhibiting its binding with CBP/p300 and decreasing luciferase activity. When FIH is inhibited, the non-hydroxylated HIF1a CAD can bind with CBP/p300 increasing luciferase activity. (E) HIF1a CAD reporter assays in normal human lung fibroblasts (control fibroblasts) or IPF lung fibroblasts (IPF fibroblasts). Values represent the relative fold increase of firefly luciferase in relation to Renilla luciferase, normalised against control (1.0). Data are mean ± s.d. n = 3 samples per group. **p < 0.01 by unpaired t test. (F) HIF1a CAD reporter assays in control fibroblasts with indicated treatment (hydrogen peroxide (T-hydro), DMOG, or vehicle control). Values represent relative fold of firefly luciferase in relation to Renilla luciferase, normalised against control (1.0). Data are mean ± s.d. n = 3 samples per group. (G and H) Control lung fibroblasts (n = 3 donors, two experiments per donor) were used in the 3D model of fibrosis in the presence of IOX2 and/or DM-NOFD or vehicle control as indicated. (G) Total mature trivalent (PYD+ DPD) collagen cross-links determined by ELISA. n = 6 samples from three donors. (H) Tissue stiffness measured from parallel-plate compression testing (n = 6 samples from three donors) determined by Young’s modulus and represented as proportion of control. * p < 0.05, ** p < 0.01, *** p < 0.001, ****p < 0.0001 by Dunnett’s multiple comparisons test. Figure 6—source data 1.Full membrane scans for western blot images for Figure 6.