Journal: iScience

Article Title: A mechanistic computational model of the HIF signaling pathway in endothelial cells

doi: 10.1016/j.isci.2026.116195

Figure Lengend Snippet: Diagram of the HIF signaling pathway and the upregulation of VEGF Under normoxia, PHDs mark HIF α for degradation mediated by VHL. Under hypoxia, PHDs’ hydroxylation of HIF1 α is reduced, and HIFs are stabilized and internalized. In the nucleus, HIF1 α dimerizes with HIF1 β and interacts with HREs, promoting the transcription of growth factors, such as VEGF mRNA, to stimulate angiogenesis and rescue homeostasis. Diagram was created with BioRender.com .

Article Snippet: Our results are consistent with previous reports of PHD2-dominant regulation of HIF1α and PHD3-preferential regulation of HIF2α, demonstrated in ECs: Takeda and Fong (2007) demonstrated that siRNA-mediated PHD2 knockdown enhanced HIF1 α and VEGFA mRNA expression under hypoxia, and Loinard et al. (2009) showed that selective shRNA silencing of PHD3 had a comparatively stronger effect on HIF2 α -dependent targets , and also in other cell types., These findings highlight the complementary and non-redundant roles of PHD2 and PHD3 in regulating hypoxia signaling and demonstrate the model’s ability to mechanistically resolve their contributions to HIF dynamics, as well as its potential to test isoform-specific regulatory mechanisms that may be challenging to isolate experimentally.

Techniques:

Journal: iScience

Article Title: A mechanistic computational model of the HIF signaling pathway in endothelial cells

doi: 10.1016/j.isci.2026.116195

Figure Lengend Snippet: Model reproduces HIF1α, HIF2α protein, and VEGFA mRNA dynamics in HUVECs under hypoxia Model predictions (lines) are compared against published time-course data (points) from independent datasets. , , Goodness-of-fit metrics (Pearson R, NRMSE, and runs test p value) are shown for each species.

Article Snippet: Our results are consistent with previous reports of PHD2-dominant regulation of HIF1α and PHD3-preferential regulation of HIF2α, demonstrated in ECs: Takeda and Fong (2007) demonstrated that siRNA-mediated PHD2 knockdown enhanced HIF1 α and VEGFA mRNA expression under hypoxia, and Loinard et al. (2009) showed that selective shRNA silencing of PHD3 had a comparatively stronger effect on HIF2 α -dependent targets , and also in other cell types., These findings highlight the complementary and non-redundant roles of PHD2 and PHD3 in regulating hypoxia signaling and demonstrate the model’s ability to mechanistically resolve their contributions to HIF dynamics, as well as its potential to test isoform-specific regulatory mechanisms that may be challenging to isolate experimentally.

Techniques:

Journal: iScience

Article Title: A mechanistic computational model of the HIF signaling pathway in endothelial cells

doi: 10.1016/j.isci.2026.116195

Figure Lengend Snippet: Model predicts oxygen-dependent HIF responses and dynamics under hypoxia-reoxygenation (A) Peak concentrations of HIF1α and HIF2α obtained from simulations at different oxygen levels (O 2 , %). (B) Simulated VEGFA mRNA levels at 6, 24, and 48 h across oxygen conditions. (C) Time to peak of HIF1α and HIF2α as a function of oxygen concentration. (D) Temporal profiles of O 2 , HIF1α, HIF2α, PHD2, PHD3, and VEGFA mRNA under cyclic hypoxia-reoxygenation. Oxygen levels were varied using step changes between normoxia and hypoxia over the simulation period. Red arrows indicate the time points corresponding to peak HIF1α and HIF2α levels during hypoxic intervals.

Article Snippet: Our results are consistent with previous reports of PHD2-dominant regulation of HIF1α and PHD3-preferential regulation of HIF2α, demonstrated in ECs: Takeda and Fong (2007) demonstrated that siRNA-mediated PHD2 knockdown enhanced HIF1 α and VEGFA mRNA expression under hypoxia, and Loinard et al. (2009) showed that selective shRNA silencing of PHD3 had a comparatively stronger effect on HIF2 α -dependent targets , and also in other cell types., These findings highlight the complementary and non-redundant roles of PHD2 and PHD3 in regulating hypoxia signaling and demonstrate the model’s ability to mechanistically resolve their contributions to HIF dynamics, as well as its potential to test isoform-specific regulatory mechanisms that may be challenging to isolate experimentally.

Techniques: Concentration Assay

Journal: iScience

Article Title: A mechanistic computational model of the HIF signaling pathway in endothelial cells

doi: 10.1016/j.isci.2026.116195

Figure Lengend Snippet: PHD activity inhibition differentially regulates HIF signaling dynamics (A and B) Peak levels of HIF1α (A) and HIF2α (B) as a function of increasing PHD activity inhibition (0%–100%) under hypoxic conditions. (C) VEGFA mRNA levels at 48 h in response to partial PHD inhibition, showing a nonlinear increase that is most pronounced under combined PHD2/3 inhibition. (D) Temporal dynamics of PHD2, PHD3, HIF1α, HIF2α, and VEGFA mRNA under global (combined) PHD2 and PHD3 inhibition at different levels (0%, 30%, 60%, and 90%). (E) Temporal dynamics under selective inhibition of PHD2 or PHD3 (60% and 90%). Colors denote different inhibition conditions; where curves overlap, distinct line styles are used for clarity.

Article Snippet: Our results are consistent with previous reports of PHD2-dominant regulation of HIF1α and PHD3-preferential regulation of HIF2α, demonstrated in ECs: Takeda and Fong (2007) demonstrated that siRNA-mediated PHD2 knockdown enhanced HIF1 α and VEGFA mRNA expression under hypoxia, and Loinard et al. (2009) showed that selective shRNA silencing of PHD3 had a comparatively stronger effect on HIF2 α -dependent targets , and also in other cell types., These findings highlight the complementary and non-redundant roles of PHD2 and PHD3 in regulating hypoxia signaling and demonstrate the model’s ability to mechanistically resolve their contributions to HIF dynamics, as well as its potential to test isoform-specific regulatory mechanisms that may be challenging to isolate experimentally.

Techniques: Activity Assay, Inhibition

Journal: iScience

Article Title: Stromal peroxidasin drives early tumor growth in breast cancer

doi: 10.1016/j.isci.2026.116078

Figure Lengend Snippet: Significantly differentially expressed ECM molecules in early, mid and late-stage mammary tumor samples from the PyMT mouse model of breast cancer in comparison to healthy fat pad tissue (A) PCA of individual samples ( n = 4 or 5 mice per group). Principal components 1 and 2 account for 81.6% of the variability in the dataset. (B) Heatmap of significantly differentially deposited matrisomal elements between groups within the dataset (left). Missing values are shaded in gray. Four major clusters were identified based on Euclidian hierarchical clustering of proteins (middle) and the matrisome categorical annotation for these clustered matrisomal elements based on the matrisome project is shown (right). Expression of PXDN in each sample (cluster 3) is highlighted in red.

Article Snippet: Lentiviral PXDN shRNA knockdown constructs and a Scrambled control construct containing GFP sequences were obtained from OriGene (TL513178V).

Techniques: Comparison, Expressing

Journal: iScience

Article Title: Stromal peroxidasin drives early tumor growth in breast cancer

doi: 10.1016/j.isci.2026.116078

Figure Lengend Snippet: Association of PXDN expression with patient overall survival Significance between Kaplan-Meier survival curves were calculated with the log-rank test. (A) Overall survival of 1,082 patients from the TCGA breast cancer cohort stratified by median mRNA expression of PXDN. (B) Overall survival of 1,082 patients from the TCGA breast cancer cohort separated by disease stage, and stratified by median mRNA expression of PXDN (dotted line = low PXDN expression, solid line = high PXDN expression) and stage of disease (stage I, red; II, gold; III, green; IV, blue; or unknown, purple). (C) Extraction of data from (B) showing patients with stage II breast cancer at the time of diagnosis stratified by median mRNA expression of PXDN (dotted line = low PXDN expression, solid line = high PXDN expression) ( n = 615). (D) Examples of PXDN IHC staining intensity and corresponding intensity scores for the stromal and epithelial compartments of tumors ( n = 334). Scale bars, 50 μm. (E) Overall survival of 311 invasive ductal carcinoma patients from the CREA tumor microarray cohort stratified by high vs. low Allred-scores for PXDN IHC staining of epithelial compartments of tumors. (F) Overall survival of 309 invasive ductal carcinoma patients from the CREA tumor microarray cohort stratified by high vs. low Allred-scores for PXDN IHC staining of stromal compartments of tumors. (G) Overall survival of invasive ductal carcinoma patients from the CREA tumor microarray cohort stratified by combined stromal and epithelial Allred-scores for PXDN IHC staining of cores. Log-rank p values between each curve of (G) are listed in .

Article Snippet: Lentiviral PXDN shRNA knockdown constructs and a Scrambled control construct containing GFP sequences were obtained from OriGene (TL513178V).

Techniques: Expressing, Extraction, Biomarker Discovery, Immunohistochemistry, Microarray

Journal: iScience

Article Title: Stromal peroxidasin drives early tumor growth in breast cancer

doi: 10.1016/j.isci.2026.116078

Figure Lengend Snippet: Single cell analysis of breast tumor cells expressing PXDN (A–C) Murine PyMT tumor single cell data (11,490 cells) obtained from Valdés-Mora et al. (A) UMAP of the major cell types identified in the murine single cell data. (B) UMAP of PXDN expression across cells. (C) Violin plots of fold change in PXDN expression according to cell type in the murine dataset. (D and E) Human breast cancer single cell and spatial data containing 130,246 cells from 26 invasive breast cancer patients. (D) UMAP of the major cell types identified in the human single cell data. (E) UMAP of PXDN expression across human cells. (F) Boxplots of fold change in PXDN expression according to cell type in the human dataset.

Article Snippet: Lentiviral PXDN shRNA knockdown constructs and a Scrambled control construct containing GFP sequences were obtained from OriGene (TL513178V).

Techniques: Single-cell Analysis, Expressing, Single Cell

Journal: iScience

Article Title: Stromal peroxidasin drives early tumor growth in breast cancer

doi: 10.1016/j.isci.2026.116078

Figure Lengend Snippet: PXDN affects cell proliferation and spheroid formation dynamics in CAFs Differences between shScr CAFs and sh#1 CAFs, or between WT CAFs and OE CAFs, were calculated using Student’s t tests and p values are shown on each graph. p values <0.05 were considered significant. ns = not significant. (A and C) RNA expression of PXDN in shScr and sh#1 (A), or WT and OE (C) CAFs. Error bars show the standard deviation of six independent experiments. (B and D) Representative western blots showing PXDN levels in conditioned media (top) from shScr and sh#1 CAFs (B) or WT and OE CAFs (D), with densitometry quantification of three biological replicate western blots (bottom). Densitometry measurements were normalized to total protein loading (measured by Ponceau S stain) and calculated relative to control (shScr or WT) protein expression. (E and F) 2D proliferation of shScr and sh#1 CAFs (E) or WT and OE CAFs (F) as measured by Alamar blue assay five days of culture (left). Differences between CAF lines were directly compared on day 5 (right). Error bars indicate the standard error of the mean for three biological replicates. (G–L) 3D growth of shScr (blue), sh#1 (red), WT (purple), and OE (gold) CAFs as spheroids over the course of 16 days. (G and I) Spheroid area and (K and L) representative images over the course of 16 days. Scale bars, 400 μm. Dotted lines indicate media changes on days 4, 8, and 12. Error bars indicate the standard error of the mean for three biological replicates. (H and J) The day of spheroid growth at which spheroids reached their smallest size before expanding. Error bars indicate the standard error of 24 spheroids across three biological replicates.

Article Snippet: Lentiviral PXDN shRNA knockdown constructs and a Scrambled control construct containing GFP sequences were obtained from OriGene (TL513178V).

Techniques: RNA Expression, Standard Deviation, Western Blot, Staining, Control, Expressing, Alamar Blue Assay

Journal: iScience

Article Title: Stromal peroxidasin drives early tumor growth in breast cancer

doi: 10.1016/j.isci.2026.116078

Figure Lengend Snippet: PXDN inhibits the ability of CAFs to remodel collagen matrices (A and C) Representative images of collagen matrices contraction over the course of 12 days when seeded with shScr or sh#1 CAFs (A) or WT or OE CAFs (C). Scale bars, 1 cm. (B and D) Area of matrices during contraction over the course of 12 days (left), with differences in final area of matrices compared at day 12 (right). Error bars indicate the standard error of the mean for three biological replicates. p values were calculated using Student’s t tests. p values <0.05 were considered significant.

Article Snippet: Lentiviral PXDN shRNA knockdown constructs and a Scrambled control construct containing GFP sequences were obtained from OriGene (TL513178V).

Techniques:

Journal: iScience

Article Title: Stromal peroxidasin drives early tumor growth in breast cancer

doi: 10.1016/j.isci.2026.116078

Figure Lengend Snippet: Impact of CAF-produced PXDN on cancer cell proliferation and motility Differences between shScr CAFs and sh#1 CAFs, or between WT CAFs and OE CAFs, were calculated using Student’s t tests and p values are indicated on each graph. p values <0.05 were considered significant. ns = not significant. (A) Schematic of generation of CAF-generated CDMs, onto which cancer cells were seeded. (B and C) Proliferation rates of cancer cells seeded on CDMs produced by sh#1 or shScr CAFs (B), or by WT and OE CAFs (C) as measured by Alamar blue at day 7. Error bars represent the standard deviation of eight replicates. (D and F) Mean velocity of cancer cells when seeded onto CDMs produced by shScr or sh#1 CAFs (D) or by WT or OE CAFs (F). (E and G) Distance traveled by cancer cells when seeded onto CDMs produced by shScr or sh#1 CAFs (E) or by WT or OE CAFs (G). A minimum of ten cells were tracked per biological replicate, across three biological replicates. Error bars indicate the standard deviation.

Article Snippet: Lentiviral PXDN shRNA knockdown constructs and a Scrambled control construct containing GFP sequences were obtained from OriGene (TL513178V).

Techniques: Produced, Generated, Standard Deviation

Journal: iScience

Article Title: Stromal peroxidasin drives early tumor growth in breast cancer

doi: 10.1016/j.isci.2026.116078

Figure Lengend Snippet: Effect of CAF PXDN expression on breast tumor progression in vivo Differences between shScr CAFs and sh#1 CAFs, or between WT CAFs and OE CAFs, were calculated using Student’s t tests and p values are indicated on each graph. p values <0.05 were considered significant; ns = not significant. (A and B) Latency of tumor formation from the time cells were implanted into mice along with shScr or sh#1 CAFs (A) or with WT or OE CAFs (B) until tumors reached 50 mm 3 in size. (C and D) Final weights of tumors upon collection from mice implanted with cancer cells and shScr or sh#1 CAFs (C) or with WT or OE CAFs (D). (E) Representative images of α-SMA staining in tumors from shScr or sh#1 CAF tumors. Scale bars, 100 μm. (F) Quantification of α-SMA staining in shScr and sh#1 CAF tumors. Error bars indicate the standard deviation. (G) Representative images of α-SMA staining in tumors from WT or OE CAF tumors. Scale bars, 100 μm. (H) Quantification of α-SMA staining in WT and OE CAF tumors. Error bars indicate the standard deviation.

Article Snippet: Lentiviral PXDN shRNA knockdown constructs and a Scrambled control construct containing GFP sequences were obtained from OriGene (TL513178V).

Techniques: Expressing, In Vivo, Staining, Standard Deviation

Journal: iScience

Article Title: Stromal peroxidasin drives early tumor growth in breast cancer

doi: 10.1016/j.isci.2026.116078

Figure Lengend Snippet: Effect of PXDN inhibition on breast tumor progression using AZD5904 (A) Inhibition of PXDN by a range of concentrations of AZD5904 measured on live cells using a modified Amplex red activity assay. Resorufin fluorescence was monitored continuously for 1 h. Error bars indicate standard error of the mean across three biological replicates, each consisting of eight technical replicates. (B) Final resorufin fluorescence readings at 60 min were used to determine to IC50 inhibitory concentration of AZD5904. (C) Kaplan-Meier curve of mouse survival measured from the start of treatment with AZD5904 or vehicle (when tumors reached a detectable size of 50 mm 3 ) until tumors reached a maximum ethical size of 1 cm by 1 cm and mice were sacrificed. Survival curves were significantly different, with a log-rank p value of 0.0088. (D) Final weights of tumors upon collection from mice in the AZD5904 in vivo study. Error bars indicate the standard deviation. Student’s t test showed no statistical significance between vehicle and AZD5904 treatment groups. (E) Time taken for tumors to reach the detectable size of 50 mm 3 at which point treatment started. Error bars indicate the standard deviation. Student’s t test showed no statistical significance between vehicle and AZD5904 treatment groups.

Article Snippet: Lentiviral PXDN shRNA knockdown constructs and a Scrambled control construct containing GFP sequences were obtained from OriGene (TL513178V).

Techniques: Inhibition, Modification, Activity Assay, Fluorescence, Concentration Assay, In Vivo, Standard Deviation