Fibroblast-enriched endoplasmic reticulum protein TXNDC5 promotes pulmonary fibrosis by augmenting TGFβ signaling through TGFBR1 stabilization

Abstract

Pulmonary fibrosis (PF) is a major public health problem with limited therapeutic options. There is a clear need to identify novel mediators of PF to develop effective therapeutics. Here we show that an ER protein disulfide isomerase, thioredoxin domain containing 5 (TXNDC5), is highly upregulated in the lung tissues from both patients with idiopathic pulmonary fibrosis and a mouse model of bleomycin (BLM)-induced PF. Global deletion of Txndc5 markedly reduces the extent of PF and preserves lung function in mice following BLM treatment. Mechanistic investigations demonstrate that TXNDC5 promotes fibrogenesis by enhancing TGFβ1 signaling through direct binding with and stabilization of TGFBR1 in lung fibroblasts. Moreover, TGFβ1 stimulation is shown to upregulate TXNDC5 via ER stress/ATF6-dependent transcriptional control in lung fibroblasts. Inducing fibroblast-specific deletion of Txndc5 mitigates the progression of BLM-induced PF and lung function deterioration. Targeting TXNDC5, therefore, could be a novel therapeutic approach against PF.

Fig. 1. TXNDC5 was upregulated in IPF lungs/lung fibroblasts and correlated with fibrogenic genes.

a Transcript analysis showed significant upregulation of TXNDC5 mRNA in human IPF (n = 26, central and lateral lung from 13 biologically independent samples), compared with control (n = 18, central and lateral lung from 9 biologically independent samples), lung tissues. b Immunoblot analysis showed a marked increase in TXNDC5 and αSMA protein levels in human IPF (n = 5 biologically independent samples), compared with control (n = 3 biologically independent samples), lung tissues. c Protein expression level of TXNDC5 and αSMA were also increased in IPF, compared with control, lung fibroblasts following TGFβ1 treatment (n = 5 biologically independent samples per group). d Re-analyses of microarray data from IPF human lung fibroblasts (GSE40839) revealed strong positive correlation between the expression level of TXNDC5 and that of genes encoding fibrogenic proteins (COL1A1, ELN, ACTA2) in human lung fibroblasts (n = 13 biologically independent samples per group) (Data are presented as mean ± SEM, P value determined using two-tailed Mann–Whitney U test. Source data are provided as a Source Data file).

Fig. 2. TXNDC5 was upregulated in BLM-induced PF and enriched in lung fibroblasts in vivo.

a Transcript expression levels of Txndc5 and fibrogenic protein genes (Col1a1, Col3a1, Eln, Fn and Ctgf) were markedly increased in the lung tissues from BLM-(n = 4 biologically independent animals per group), compared with PBS- (Sham, n = 3 biologically independent animals), treated WT mice on Day 21. b Immunohistochemical (IHC) staining on the serial sections of mouse lungs showed a strong upregulation of αSMA and TXNDC5 in the lung tissues from BLM-, compared with PBS- (Sham), treated WT mice on Day 21 (n = 9 fields examined over 3 biologically independent animals per group). c IF staining for TXNDC5 on lung sections from sham-operated and BLM-treated Col1a1-GFP^Tg mice. BLM treatment significantly increased TXNDC5 expression and the number of GFP-positive cells in the mouse lungs on Day 21 (n = 3 biologically independent animals). There was a high degree of co-localization of TXNDC5 with GFP-positive, collagen producing lung fibroblasts (white arrows and inset) (data are presented as mean  SEM, P value determined using two-tailed Mann–Whitney U test. Source data are provided as a Source Data file. BLM bleomycin).

Fig. 3. Knockout of Txndc5 markedly reduced the severity of pulmonary fibrosis induced by BLM.

a, b Representative microCT images of lung tissues from WT and Txndc5^−/− mice with and without BLM treatment on Day 21. MicroCT imaging showed marked parenchymal destruction and volume reduction in WT mouse lungs following BLM treatment (a). Lung volume reduction and fibrotic changes (white areas in b) were significantly attenuated in Txndc5^−/− mice. c Quantification of lung volume by microCT in WT and Txndc5^−/− mice with and without BLM treatment on Day 21 (WT sham n = 5, WT BLM n = 3, Txndc5^−/− BLM n = 5 biologically independent animals). d Picrosirius red staining (top panels) of lung sections from WT and Txndc5^−/− mice 21 day after intra-tracheal administration of BLM. Images in the lower panels were magnified from the inset of the photomicrographs in the upper panels. Middle panel showed the representative fluorescence images of lung sections from Col1a1-GFP^Tg and Col1a1-GFP^Tg*Txndc5^−/− mice treated with BLM or PBS (Sham). Quantitative results of picrosirius red- (WT sham n = 4, WT BLM n = 6, Txndc5^−/− BLM n = 8 biologically independent animals) and GFP-positive areas (n = 9 fields examined over 3 biologically independent animals per group) determined from above were shown on the right. e Hydroxyproline content was markedly increased in the mouse lungs from WT mice treated with BLM on Day 21. Global deletion of Txndc5 significantly reduced hydroxyproline content in the mouse lungs following BLM treatment (n = 5 biologically independent animals per group). f Representative pressure-volume curves (WT sham n = 4, WT BLM n = 7, Txndc5^−/− BLM n = 7, Txndc5^−/− BLM n = 11 biologically independent animals) and (g, h) lung function parameters determined using flexiVent FX system in WT and Txndc5^−/− mice 21 days following sham procedure or BLM treatment(g: WT sham n = 4, WT BLM n = 6, Txndc5^−/− BLM n = 7, Txndc5^−/− BLM n = 11 biologically independent animals, h: WT sham n = 5, WT BLM n = 7, Txndc5^−/− BLM n = 7, Txndc5^−/− BLM n = 12 biologically independent animals) (Data are presented as mean ± SEM, P value determined using two-tailed Mann–Whitney U test. Source data are provided as a Source Data file. n.s. non-significant, BLM bleomycin).

Fig. 4. Global deletion of Txndc5 did not alter inflammatory response to BLM treatment.

a Representative photomicrographs of modified Giemsa staining of the bronchoalveolar lavage fluid (BALF) from WT and Txndc5^−/− mice 14 days after intra-tracheal BLM or PBS (Sham) instillation (top panel). Quantification of total protein content and number of inflammatory cell (macrophage, neutrophil and lymphocytes) in the BALF from each group of the experimental animals (bottom panel) (WT sham n = 3, WT BLM n = 5, Txndc5^−/− sham n = 3, Txndc5^−/− BLM n = 5 biologically independent animals). b Transcript expression levels of pro-inflammatory cytokines, Il6 and Il1b, quantified in the lung tissues from WT and Txndc5^−/− mice 7 days and 14 days after BLM or PBS (Sham) treatment (n = 5 biologically independent animals per group). c Immunofluorescence staining of TXNDC5 (green) and CD11b (red) of lung sections from WT mice 14 days following BLM treatment. TXNDC5 was not present in CD11b-positive, inflammatory cells (Data are presented as mean ± SEM, P value determined using two-tailed Mann–Whitney U test. Source data are provided as a Source Data file. n.s. non-significant, BLM bleomycin).

Fig. 5. TXNDC5 is both essential and sufficient for HPF activation, proliferation, and ECM production.

Protein (n = 6 biologically independent samples per group) (a) and transcript (n = 6 biologically independent samples per group) (b) expression levels of COL1A1, fibronectin, elastin, periostin, and αSMA/ACTA2 were markedly increased in control (shScr) human pulmonary fibroblasts (HPF) following TGFβ1 (10 ng/ml) treatment. TGFβ1-induced upregulation of these fibrogenic proteins/genes was significantly attenuated in HPF with TXNDC5 depletion (shTXNDC5). c TGFβ1 treatment significantly increased the cellular proliferation activity in shScr-, but not in shTXNDC5-, transduced HPF (n = 24 biologically independent samples per group). ECM proteins (COL1A1 and fibronectin), markers for fibroblast activation (periostin and αSMA) (n = 9 biologically independent samples per group) (d) and fibroblast proliferation activity (n = 9 biologically independent samples per group) (e) were markedly increased in HPF transduced with TXNDC5 (TXNDC5 OE), compared with empty, expression vector (Data are presented as mean ± SEM, P value determined using two-tailed Mann–Whitney U test. Source data are provided as a Source Data file. n.s. non-significant, ctrl control, KD knockdown, OE overexpress).

Fig. 6. TXNDC5 modulates TGFBR1 expression in human lung fibroblasts and mouse lungs.

a Immunoblots showed that TGFBR1, but not TGFBR2, was markedly upregulated in control HPF (shScr) following TGFβ1 treatment. TXNDC5 knockdown (shTXNDC5) prevented TGFBR1 upregulation induced by TGFβ1 treatment completely (n = 6 biologically independent samples per group). b Forced TXNDC5 expression led to marked upregulation of TGFBR1, but not TGFBR2, protein in HPF (n = 12 biologically independent samples per group). c TGFBR1 and COL1A1 proteins were both markedly upregulated in the lung tissues from WT, but not Txndc5^−/−, mice 21 days following BLM treatment (WT sham n = 4, WT BLM n = 5, Txndc5^−/− sham n = 3, Txndc5^−/− BLM n = 5 biologically independent animals). d Representative IF staining and quantification (e) of TGFBR1 in Col1a1-GFP^Tg and Col1a1-GFP^Tg*Txndc5^−/− mouse lungs with PBS (Sham) or BLM treatment on day 21 (n = 9 fields examined over 3 biologically independent animals per group). TGFBR1 was marked increased and showed strong co-localization with GFP-positive lung fibroblasts in BLM-treated mouse lungs. Global deletion of TXNDC5 prevented the upregulation of TGFBR1 in mouse lungs following BLM treatment (Data are presented as mean ± SEM, P value determined using two-tailed Mann–Whitney U test. Source data are provided as a Source Data file. n.s. non-significant, KD knockdown, OE overexpress, BLM bleomycin, TGFBR1 TGFβ receptor type 1, TGFBR2 TGFβ receptor type 2).

Fig. 7. TXNDC5-induced pulmonary fibroblast activation and ECM production requires TGFBR1.

a Forced expression of TXNDC5 in HPF led to increased SMAD3 phosphorylation, fibroblast activation (as evidenced by increased periostin levels) and ECM (COL1A1 and fibronectin) production, all of which were abolished completely by the treatment of LY364947 (10 μM for 48 h), a TGFBR1 kinase inhibitor (n = 6 biologically independent samples per group). b Knockdown of TGFBR1 reversed TXNDC5 overexpression-induced COL1A1, periostin, and αSMA expression (n = 4 biologically independent samples per group) (Data are presented as mean ± SEM, P value determined using two-tailed Mann–Whitney U test. Source data are provided as a Source Data file. n.s. non-significant, KD knockdown, OE overexpress, shTGFBR1 TGFBR1 knockdown with shRNA).

Fig. 8. TXNDC5 promotes fibrogenesis by enhancing TGFBR1 protein stability via its PDI activity.

a A cycloheximide chase assay performed in HPF with TXNDC5 knockdown (shTXNDC5) showed accelerated degradation of TGFBR1 protein, comparing to control (shScr) (n = 4 biologically independent samples per group). b Overexpression of TXNDC5 slowed down TGFBR1 protein degradation significantly in HFP (n = 7 biologically independent samples per group). c TXNDC5 depletion-induced downregulation of TGFBR1 protein was partially reversed by the treatment of proteasome inhibitor MG132 (20μM for 48 h) (n = 7 biologically independent samples per group). d, e Overexpression of WT, but not AAA mutant (see text for details), TXNDC5 protein in HPF led to significant upregulation of TGFBR1, ECM (fibronectin and COL1A1) proteins and fibroblast activation markers (αSMA and periostin) (n = 8 biologically independent samples per group) (Data are presented as mean ± SEM, P value determined using two-tailed unpaired t tests. Source data are provided as a Source Data file. OE overexpress, TXNDC5 AAA TXNDC5 mutant lacking its PDI enzyme activity).

Fig. 9. TGFβ1 induces TXNDC5 expression in HPF through ER stress-dependent ATF6 activation.

a TGFβ1 treatment in HPF led to increased ER stress levels, as reflected in upregulated ER stress markers BiP and ATF6α(N) (activated ATF6). Co-treatment with 4-PBA (2 mM), an ER stress inhibitor, reversed TGFβ1-induced increases in BiP and ATF6α(N) (Bip, ATF6α(P): n = 9 biologically independent samples per group, ATF6α(N): n = 12 biologically independent samples per group). b TXNDC5 mRNA was increased in response to TGFβ1 treatment, which was significantly attenuated by 4-PBA (n = 7,6,6 biologically independent samples). c Knockdown efficiency of lentiviral vectors carrying ATF6-targeted shRNA in HPF (n = 6 biologically independent samples per group). d TXNDC5 transcript was significantly increased in control (shScr), but not in ATF6-knockdown (shATF6), HPF following TGFβ1 treatment (n = 6 biologically independent samples per group). e Schematic illustration of the human TXNDC5 promoter luciferase reporter construct, which contains an ATF6-binding site. Deletion of the ATF6-binding site (TGACGTGG, + 642 to +653, ΔATF6) markedly reduced the transcriptional activity of the TXNDC5 promoter in response to TGFβ1 stimulation (WT TXNDC5: n = 16 in ctrl, 17 in TGFβ1 biologically independent samples, ΔATF6: n = 10 in ctrl, 11 in TGFβ1 biologically independent samples) (Data are presented as mean ± SEM, P value determined using two-tailed Mann–Whitney U test. Source data are provided as a Source Data file. n.s. non-significant, ctrl control, shATF6: ATF6 knockdown with shRNA).

Fig. 10. Fibroblasts-specific Txndc5 deletion lessened the progression of pulmonary fibrosis.

a Illustration of experimental design for deletion of Txndc5 in pulmonary fibroblasts. Tamoxifen (80 mg/kg i.p. every other day) was administered between 7–21 days after BLM treatment. b Picrosirius red staining (left panels) and second harmonic generation (SHG) images (right panels) of lung sections from Col1a2-cre/ERT2 (Col1a2-cre) and Col1a2-Cre/ERT2*Txndc5^fl/fl (Txndc5^cKO) mice 7 day and 21 day after intra-tracheal administration of BLM. The quantitative results of picrosirius red-(Col1a2-cre, n = 3 in D0 and D7, n = 5 in D21; Txndc5^cKO, n = 3 in D0 and D7, n = 6 in D21 biologically independent animals) (c) and SHG- (Col1a2-cre, n = 3; Txndc5^cKO, n = 3 in D0 and D7, n = 5 in D21 biologically independent animals) (d) positive areas showed rapid progression of PF in Col1a2-cre, but not in Txndc5^cKO, lungs after BLM instillation. For each of the lung sections scanned for SHG, additional two-photon-excited fluorescence (TPEF) imaging was obtained to show the outline of the imaged tissue. e Hydroxyproline content was similarly increased in Col1a2-cre and Txndc5^cKO on Day 7 (before Tamoxifen injection) following BLM treatment. Hydroxyproline content continued to increase in the mouse lungs of Col1a2-cre mice between 7 and 21 days post-BLM treatment (Col1a2-cre, n = 3 in D0, n = 4 in D7, n = 5 in D21; Txndc5^cKO, n = 3 in D0 and D7, n = 7 in D21 biologically independent animals). The hydroxyproline content remained stable in BLM-treated Txndc5^cKO mouse lungs after tamoxifen-induced Txndc5 deletion. Representative lung function parameters (Col1a2-cre, n = 5 in D0, n = 3 in D7, n = 6 in D21; Txndc5^cKO, n = 4 in D0, n = 3 in D7, n = 7 in D21 biologically independent animals) (f) and pressure-volume curves (on Day 21, sham: n = 3, BLM: n = 6 biologically independent animals) (g) determined using flexiVent FX system in Col1a2-cre and Txndc5^cKO mice following sham procedure or BLM treatment. h Schematic summary of the proposed profibrotic mechanisms by which TXNDC5 contributes to the pathogenesis of pulmonary fibrosis (Data are presented as mean ± SEM, P value determined using two-tailed Mann–Whitney U test. Source data are provided as a Source Data file. BLM bleomycin). Illustrations in a and h were created by T.H.L.