Absence of Glutathione Peroxidase 4 Affects Tumor Angiogenesis through Increased 12/15-Lipoxygenase Activity¹

Cell Line and Reagents

The 4-hydroxy-tamoxifen (TAM)-inducible GPx4 knockout murine embryonic fibroblasts (MEFs) were used as described by Seiler et al. [7]. All cell culture reagents were purchased from Invitrogen (Karlsruhe, Germany). MEFs were cultured at 37°C under 5% CO₂ and 5% O₂ in Dulbecco's modified Eagle medium supplemented with 10% fetal bovine serum, 2 mM l-glutamine, 100 U/ml penicillin, and 100 µg/ml streptomycin. To maintain the viability of the cells after induction of GPx4 deletion by adding TAM, cells were always maintained in the presence of (±)-6-hydroxy-2,5,7,8-tetramethylchromane-2-carboxylic acid (Trolox; 1 µM, #56510; Sigma-Aldrich, Deisenhofen, Germany), a water-soluble analog of α-tocopherol (see MTT Assay for Cell Viability section). Tumor cells were harvested near confluence by brief trypsination in 0.25% trypsin/EDTA solution, washed several times, and placed in sterile phosphate-buffered saline (PBS) shortly before implantation. The pan-COX-inhibitor indomethacin (I7378), the pan-LOX-inhibitor NDGA (N5023), and the 12/15-LOX-specific inhibitors Baicalein (#465119) and PD146176 (P4620) as well as 4-hydroxy-tamoxifen (H7904) were purchased from Sigma-Aldrich. pBJ3Ω c-Myc and pUC EJ6.6 Ha-ras^V12 were a kind gift from Dr. Hartmut Land. Each time before performing an in vitro or in vivo assay, GPx4-deleted cells were thoroughly washed with PBS to remove Trolox. Furthermore, an aliquot of the cells was checked for GPx4 expression by quantitative polymerase chain reaction (PCR; see below). As a second control, another aliquot was plated under normal cell culture conditions without adding Trolox, and cell death was analyzed within the next 48 to 72 hours (see below).

Transformation of Conditional GPx4-Deficient MEFs

For oncogenic transformation, GPx4-deficient MEFs were cotransduced with c-Myc- and ha-ras^V12-expressing lentiviruses. Soft agar assays were conducted to select for transformed cells. The assay was performed in 60-mm dishes (BD Falcon; Becton Dickinson, Heidelberg, Germany) containing two layers of soft agar. The top and bottom layers were 0.3% and 0.4% agarose, respectively, in Dulbecco's modified Eagle medium supplemented with 10% fetal bovine serum. A total of 2 x 10⁴ of the c-Myc- and ha-ras-infected cells were added to the top layer before pouring. Single-cell colonies were isolated approximately 2 to 3 weeks after culturing at 37°C and 5% CO₂. Single-cell clones were individually expanded, and GPx4 deletion was tested later in each of the established cell lines by adding TAM (±Trolox) followed by analysis of GPx4 expression (see quantitative reverse transcription [RT]-PCR) and cell death, respectively (see MTT Assay for Cell Viability section).

Immunoblot Analysis

Cell lysates and tumor tissue lysates, protein concentration determination, and immunoblot analysis of GPx4 were performed as described previously [7]. Ha-ras^V12 and c-Myc expression was confirmed using anti-Ha-ras (#3965; Cell Signaling Technology, New England Biolaboratories, Frankfurt am Main, Germany) and anti-c-Myc antibodies (sc-788; Santa Cruz Biotechnology, Heidelberg, Germany).

MTT Assay for Cell Viability

Cells were plated in 96-well plates at 1 x 10⁴ cells per well and incubated overnight. Cells were treated either with Trolox (a water-soluble analog of α-tocopherol; #56510; Sigma-Aldrich), with the pan-COX inhibitor indomethacin, the general LOX inhibitor NDGA and the 12/15-LOX-specific inhibitors Baicalein and PD146176 for 72 hours at various concentrations. Then, 10 µl of 5 mg/ml 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT; M2003; Sigma-Aldrich) in PBS was added to 100 µl of culture media. Plates were incubated for 4 hours, and the formazan precipitate was redissolved in 200 µl of 0.04N HCl in isopropanol. The absorbance at 540 nm was recorded using a Sunrise microplate absorbance reader (Tecan, Crailsheim, Germany).

Three-dimensional BD Matrigel Assay

Three days before starting the assay, deletion of GPx4 was induced by adding TAM to the cell culture medium. To ensure survival of the cells, Trolox, which fully complements GPx4 deficiency in vitro (see MTT Assay for Cell Viability section), was given as a supplement to the medium.

Two types of frozen Matrigel (Matrigel as well as Growth Factor-reduced Matrigel; #356234 and #35623; Becton Dickinson GmbH) were thawed overnight at 4°C, and 200 µl of each undiluted Matrigel was placed on 24-well plates and incubated for 1 hour at 37°C. Thereafter, tGPx4 knockout as well as control cells were thoroughly washed with PBS to remove Trolox. Cells were counted and 5 x 10⁴ cells each were reconstituted in 20 µl of culture medium ± TAM. Finally, cells were plated on the BD Matrigel layer, coated with 200 µl of undiluted Matrigel, and incubated for an additional hour at 37°C. Normal cell culture medium ± TAM was added on the top layer. In parallel, the same amount of wild-type and knockout cells was plated on uncoated 24-well plates ± TAM, and GPx4 deletion was verified by analyzing cell death within the next 48 to 72 hours.

The average number of spheroids per well was determined after 14 days of growth in both types of Matrigel. Pictures were taken from triplicate experiments.

RNA Isolation, Reverse Transcription, and Quantitative RT-PCR

Cells were kept 72 hours under the conditions described above adding either TAM and Trolox or Trolox alone to the cell culture medium. Total RNA was extracted with the Qiagen RNA isolation kit (RNeasy Mini Kit #74106; Qiagen, Hilden, Germany), and residual genomic DNA was removed by the RNase-Free DNase Set (#79254; Qiagen). RNA was reverse-transcribed into complementary DNA as described before [21]. Real-time PCR was performed with Lightcycler in a reaction volume of 10 µl using SYBR Green I (LC-FastStart DNA Master^Plus, #03 515 885 001; Roche, Mannheim, Germany) according to the manufacturer's protocol. Primers were used at a final concentration of 0.5 µM each. An initial denaturation step at 95°C for 10 minutes was followed by 40 cycles of denaturation (95°C, 1 second), annealing (65°C, 10 seconds), and extension (72°C, 20 seconds). Standard curve was prepared using four dilutions of complementary DNA obtained from mouse embryonic fibroblasts. Relative messenger RNA levels were normalized to β-actin messenger RNA levels in the same samples. Primer sequences are as follows: GPx4 A forward (5′-GCA ACC AGT TTG GGA GGC AGG AG-3′), GPx4 B reverse (5′-CCT CCA TGG GAC CAT AGC GCT TC-3′); β-actin forward (5′-CTA CGA GGG CTA TGC TCT CCC-3′) and reverse (5′-CCG GAC TCA TCG TAC TCC TGC-3′).

In Vivo Tumor Model

For in vivo implantation studies, cells were kept for at least 72 hours under TAM and Trolox, and the knockout was verified as described above. Before injecting the cells, cells were washed thoroughly with PBS to remove TAM and Trolox. After cell counting, 5 x 10⁶ c-Myc/Ha-ras^V12-transformed GPx4-deficient or GPx4-proficient MEFs diluted in 100 µl of sterile PBS were implanted subcutaneously into the flanks of C57BL/6 or severe combined immunodeficient (SCID) mice, respectively. Tumors were collected 16 days after implantation (n = 8 mice per group). Tumor volume was calculated by caliber measurement according to the following formula: V (mm³) = (length x width x high) x π/6. For the in vivo pharmacological studies, mice were treated with Baicalein (1.5 mg/kg in Tween 20) orally every second day beginning at day 3 after tumor cell implantation.

Mice were kept under standard conditions with food and water ad libitum (ssniff, Soest, Germany). All animal experiments were performed in compliance with the German Animal Welfare Law and have been approved by the institutional committee on animal experimentation and the government of Upper Bavaria.

Hematoxylin-Eosin Histology

Five-micrometer sections of 4% (wt./vol. in PBS) paraformaldehyde-fixed and paraffin-embedded tumor material were stained as described [22].

Immunohistochemistry and Image Analysis

Tumor tissue samples were snap frozen in liquid nitrogen and stored at -80°C. Ten-micrometer frozen sections were prepared using a cryotome (HM 560; Microm, Walldorf, Germany) and mounted on poly-L-lysine-coated glass slides (Menzel, Braunschweig, Germany). Immunostaining was performed using a phospho-histone H3 antibody (1:200; # 9701; Cell Signaling Technology, New England Biolaboratories) for proliferating cells, a CD31 antibody (1:800; BM 4086B; Acris Antibodies, Hiddenhausen, Germany) for endothelial cells, an a-smooth muscle cell antibody (1:200, A2547; Sigma-Aldrich) to mark endothelial coverage by vascular smooth muscle cells/pericytes and the ApopTag Peroxidase In Situ Apoptosis Detection Kit (S7100; Serologicals Corporation, Millipore, Schwalbach, Germany) for the detection of apoptotic nuclei according to the manufacturers' recommendations. Immunoperoxidase staining was performed using the Vector ABC Kit and Vector DAB Kit (Vector Laboratories, Linaris, Wertheim-Bettingen, Germany).

Quantification of Proliferation and Apoptosis

The number of proliferating or apoptotic cells, respectively, was determined in a genotype-blinded fashion in 10 random microscopic visual fields at 200-fold magnification and quantified using pixel-based thresholds in a computer-assisted image analysis software (KS400 Imaging System; Carl Zeiss Vision, Hallbergmoos, Germany) defined as positively stained area per total area.

Quantification of Blood Vessel Density

The number of total blood vessels was determined in a genotype-blinded fashion by counting the number of blood vessels in 10 random microscopic visual fields at 200-fold magnification. Blood vessel density, defined as percentage of CD31-positive staining per unit area, was quantified using pixel-based thresholds in a computer-assisted image analysis software (KS400 Imaging System; Carl Zeiss Vision).

Quantification of Pericyte-Covered Tumor Vessels

Tumor sections were double-stained with anti-CD31 antibody for endothelial cells, and anti-α-smooth muscle actin (αSMA) antibodies for smooth muscle cells/pericytes. The number of α-smooth muscle cell-covered vessels was determined in a genotype-blinded fashion in 10 random microscopic visual fields at 200-fold magnification.

In Vivo Fluorescence Microscopy

Experiments were performed in male C57BL/6 mice (8–10 weeks old, 25–35 g, n = 4 mice per group). A dorsal skinfold chamber was implanted under anesthesia (subcutaneous injection of 75 mg of ketamine hydrochloride/25 mg xylazine per kilogram body weight) as described [23]. Animals were allowed to recover from the chamber implantation surgery for 2 to 3 days before tumor cells were inoculated into the skinfold chamber preparation.

Before measurement of vessel diameter and vessel density at days 7, 10, and 14 after tumor cell implantation, animals were placed in a polycarbonate tube and were injected intravenously with 0.1 ml of a fluorescein isothiocyanate (FITC)-labeled dextran solution (5% in 0.9% NaCl, M_W = 200,000, FD20; Sigma-Aldrich).

The chamber preparation was mounted on an individually designed microscope stage (Axioplan; Zeiss, Jena, Germany), and tumor vascularization was observed by using a 20x objective (LD acroplan; Zeiss). Epi-illumination was achieved with a 100-W mercury lamp with a fluorescence filter for FITC (excitation, 450–490 nm; emission, 515–525 nm). Images of microvessels in five areas of the tumor were acquired by using a CCD camera (AVC D7; Sony, Tokyo, Japan), displayed on a video monitor and recorded on digital tapes by using a video recorder (SVO-9500MD; Sony) for offline analysis.

Flat-Panel Volume Computed Tomography Imaging

Mice were imaged with flat-panel volume computed tomography (fpVCT), a nonclinical volume computed tomography prototype (GE Global Research, Niskayuna, NY) as described [24,25]. In brief, mice were anesthetized with vaporized isoflurane at 0.8% to 1% concentration, centered on the fpVCT gantry axis of rotation, and placed perpendicularly to the z-axis of the system. At distinct time points, iodine-containing contrast agent, 150 µl of Isovist 300 (Bayer-Schering, Berlin, Germany) per mouse, was applied intravenously approximately 30 seconds before scan, and for better demonstration of small blood vessels, the blood pool agent eXia 160 (Binito Biomedical Inc, Ottawa, Canada) was administered intravenously 90 seconds before each scan as described by Jannasch et al. [25]. All data sets were acquired with the same protocol: 500 views per rotation, 4 seconds rotation time, 360 used detector rows, tube voltage of 80 kilovolt (peak), and a current of 100 mA. A modified Feldkamp algorithm was used for image reconstruction resulting in isotropic high-resolution volume data sets (512 x 512 matrix, with an isotropic voxel size of approximately 100 µm). For tumor segmentation and volume estimation, data sets were analyzed with voxtools 3.0.64 Advantage Workstation 4.2 (GE Healthcare, UK).

Enzyme-Linked Immunosorbent Assay for Vascular Endothelial Growth Factor A, Placenta Growth Factor, and Basic Fibroblast Growth Factor

Lysates from GPx4 knockout and control tumors as well as cell culture supernatants from both types of cells were prepared, and enzyme-linked immunosorbent assays (ELISAs) for vascular endothelial growth factor (VEGF; MMV00), placenta growth factor (PlGF; MP200), and basic fibroblast growth factor (bFGF, DFB50; all purchased from R&D Systems GmbH, Wiesbaden-Nordenstadt, Germany) were performed as described in the accompanying Quantikine ELISA protocols. All ELISA results were read on a Sunrise microplate absorbance reader (Tecan) at an absorbance of 450 nm, with wavelength correction at 550 nm, and calculated according to the manufacturers' recommendations. Total protein concentrations of tumor homogenates and supernatants were measured with the BCA Protein Assay Kit (#23227; Pierce, Fisher Scientific GmbH, Schwerte, Germany).

Statistical Analysis

Statistical analysis was performed using SigmaStat software. The statistical significance of the results was tested by Student's t test. P < .05 was considered statistically significant. Results were expressed as mean ± SD.

Transformation of Inducible GPx4 Knockout MEFs with the Oncogenes c-Myc and Ha-ras^V12

To investigate the impact of GPx4 depletion on tumorigenesis, we took advantage of our recently established TAM-inducible GPx4 knockout fibroblasts (Figure 1A) [7]. These cells were cotransduced with c-Myc- or Ha-ras^V12-expressing lentiviruses. To establish stably transformed cell lines, colony formation assays were performed, which allowed to isolate single-cell colonies after 2 to 3 weeks of growth in soft agar. Single-cell clones were individually expanded, and the expression of both oncogenes c-Myc and Ha-ras^V12 was confirmed by immunoblot analysis (Figure 1B). All the following experiments were performed using the c-Myc- or Ha-ras^V12-transformed TAM-inducible GPx4 knockout fibroblasts (tGPx4).

As shown previously, inducible depletion of GPx4 in nontransformed MEFs caused rapid cell death within 48 hours after knockout induction, which could be prevented either by a-tocopherol or by 12/15-LOX-specific inhibitors. Water-soluble antioxidants and 5-LOX- and COX-specific inhibitors were shown to be ineffective in preventing cell death [7]. To address whether transformed GPx4 knockout cells behave differently with respect to cell proliferation and cell survival, the knockout was induced by TAM in the presence and absence of Trolox, an a-tocopherol analog, and various LOX and COX inhibitors. Cell viability was determined by MTT assay 72 hours after TAM treatment. As illustrated in Figure 1, C and D, cell death of transformed GPx4 knockout cells could be prevented by Trolox, NDGA, Baicalein, and the 12/15-LOX-specific inhibitor PD146176, but not by the COX inhibitor indomethacin. Hence, in terms of cell viability in vitro, transformed GPx4-null cells do not substantially deviate from the non-transformed parental cell lines.

tGPx4-Deficient MEFs Form Tumor Spheroids in Three-dimensional Extracellular Matrix

One characteristic of tumor cells is their ability to form three-dimensional structures in reconstituted extracellular matrix (ECM) such as BD Matrigel, a solubilized basement membrane preparation extracted from the Engelbreth-Holm-Swarm mouse sarcoma [26]. Interaction of tumor cells with the ECM is thought to be a critical event in controlling angiogenesis and tumor growth. To compare the (tumor cell) behavior of GPx4-deficient and -proficient transformed MEFs, cells were plated in between two layers of BD Matrigel to obtain a three-dimensional culture assay.

Unexpectedly, tGPx4-null cells embedded in three-dimensional BD Matrigel survived and indeed formed tumor cell colonies (Figure 2B). Such colonies were similar in appearance and number (17.53 ± 6.51; Figure 2, B and E) compared with the colonies formed by control cells (17.60 ± 2.67; Figure 2, A and E). To test whether survival of GPx4-deficient MEFs was dependent on the growth factor content of BD Matrigel, cell survival and colony formation were tested in parallel in BD Growth Factor-reduced Matrigel. No significant difference in the number of colonies could be found between the two types cells, although the overall number of colonies was reduced under the conditions of limited growth factor availability (12.77 ± 7.07 control colonies vs 13 ± 5.24 knockout colonies; Figure 2F). In contrast, a strong reduction in cell number of tGPx4-null cells seeded on uncoated plates was observed after 48 hours (Figure 2D), whereas wild-type cells survived (Figure 2C).

Transformed GPx4-Deficient MEFs Form Tumors with Higher Mitotic Activity but Also an Increased Number of Apoptotic Cells

Because transformed GPx4-deficient cells died under normal cell culture conditions, but survived and formed colonies when plated in ECM-like BD Matrigel, we asked whether tGPx4-deficient cells generate tumors in vivo. To this end, tGPx4-deficient MEFs and the corresponding noninduced control cells were subcutaneously injected into the flanks of C57BL/6 mice. Tumor growth was monitored every second day, and tumors were harvested and analyzed at day 16 after tumor cell injection. Surprisingly, GPx4-deficient and GPx4 control tumors revealed no significant difference with regard to morphologic appearance (Figure 3, A and D), tumor volume (1399.49 ± 767.21 vs 1427.86 ± 1088.49 mm³; Figure 3E) and tumor mass (1.77 ± 0.72 vs 1.62 ± 1.58 g; Figure 3F). To exclude the possibility that tumors arose from noninduced wild-type cells in the TAM-treated group, GPx4 expression in tumor tissue was analyzed by immunoblot analysis with a GPx4-specific antibody. As illustrated in Figure 3G, tumor lysates showed strongly reduced GPx4 expression, indicating that tumors were indeed derived from tGPx4-null cells.

Interestingly, immunohistochemical staining using the proliferation marker phospho-histone H3 revealed an increase in the proliferation rate of GPx4-deficient tumors (8.08 ± 2.07%) compared with control tumors (6.82 ± 2.08%; Figure 3, H–J). Whereas tumors from both groups displayed regions of necrosis, the portion of apoptotic cells was higher in GPx4-deficient tumors (3.85 ± 0.78%) than in control tumors (2.30 ± 1.22%) as detected by TUNEL staining (Figure 3, K–M).

To quantitate the precise tumor growth and progression over time in a noninvasive fashion, we applied the high-resolution imaging technique, fpVCT. By virtual isolation of the tumor from the acquired isotropic high-resolution fpVCT data sets, tumor volumes were calculated automatically by considering all three dimensions, allowing the generation of precise growth kinetics for each tumor over time. Single fpVCT scans of the entire mice after application of an iodine-containing contrast agent were performed three times per week beginning at day 3 after tumor cell inoculation. In addition, at day 7 and day 14 after tumor cell implantation, we applied the blood pool contrast agent eXia 160 to illustrate the intratumoral vascular network as well as the vessel recruitment from the host in response to the developing tumor. Evaluation of GPx4-deficient tumors and control tumors over time by fpVCT imaging yielded no differences in tumor size and progression. Furthermore, contrastmedium-containing blood vessels with sizes greater than 150 µm in diameter around and within the tumors were determined. The observed recruitment of host vessels by the tumors as well as the characteristics of large diameter tumor vessels was comparable within the two groups (Figure 4, A–D).

Increased Number of Microvessels in GPx4-Deficient Tumors

GPx4-deficient tumors displayed a higher rate of apoptotic cells, nevertheless an increase in tumor cell proliferation. Therefore, no significant alterations about tumor size, tumor weight, or growth rate could be determined between GPx4-deficient tumors and control tumors. In the next step, we performed CD31 immunostaining to evaluate tumor vascularization. Analysis of serial sections for the CD31-positive area demonstrated a markedly increased number of small microvessels in GPx4-deficient tumors (total vessel number, 364 ± 119.21), whereas control tumors displayed decreased total vessel numbers (279.73 ± 83.79; Figure 5, A–D). However, it is noteworthy that, in control tumors, the mean vessel area was clearly augmented compared with GPx4 knockout tumors (mean ±SEM: 100.26 ± 10.35 vs 83.94 ± 12.69 µm²).

To analyze the density of effectively perfused tumor vessels in vivo, tGPx4-deficient MEFs and control cells were implanted into the dorsal skinfold chamber. Tumor cells were implanted 3 days after chamber preparation, and in vivo fluorescence microscopy was performed at days 7, 10, and 14 after intravenous administration of FITC-dextran, respectively. At days 7 and 10, no differences were observed in vessel diameter and vessel density between GPx4-deficient and control tumors. However, at day 14, GPx4-null tumors showed a significantly higher functional microcapillary density (262.96 ± 13.72) than control tumors (202.26 ± 13.97; Figure 5, E–G), whereas control tumors had a significantly increased tumor vessel diameter and caliber size (15.93 ± 0.60 vs 9.95 ± 0.47; Figure 5, E, F, and H).

Decreased Number of αSMA-Covered Vessels

Tumor vessels were also characterized with respect to smooth muscle cell/pericyte coverage by staining tumor sections with anti-CD31 and anti-αSMA antibodies; smooth muscle cell-positive vessels were defined as CD31-positive structures associated with αSMA-positive cells. Analysis of double-stained sections revealed a statistically significant reduction in αSMA-positive vessels in GPx4 knockout tumors when compared with control tumors (7.21 ± 1.81 vs 12.25 ± 0.76; Figure 6, A–C).

VEGF-A, PlGF, and bFGF Levels of Tumors Derived from tGPx4 Deficient Cells Were Similar to Control Tumors

Because GPx4 knockout tumors displayed a higher microvessel density that was associated with poor mural cell coverage, we wondered if this might be due to an increased endothelial growth factor production by GPx4 knockout tumors. Therefore, we quantified the amount of VEGF-A, PlGF, and bFGF in tumor tissue lysates. Surprisingly, all endothelial growth factor levels obtained from control tumors (VEGF-A, 479.61 ± 220.55 pg/ml; PlGF, 221.41 ± 151.27 ng/µg; bFGF, 272.63 ± 125.33 pg/µg) did not differ from the values obtained from GPx4 knockout tumors (VEGF-A, 555.63 ± 432.29 pg/µg; PlGF, 257.04 ± 83.82 ng/µg; bFGF, 233.03 ± 67.26 pg/µg). Because the SD in tumor tissue samples was high, we also analyzed cell culture supernatants after induction of GPx4 deletion (see below).

Increased bFGF and Decreased VEGF-A Levels in Supernatants of GPx4-Deleted tMEFs

Quantification of VEGF-A, PlGF, and bFGF levels in supernatants of GPx4-deleted tMEFs and control cells revealed a decrease in VEGF-A protein expression levels (729.4 ± 29.38 pg/µg in control supernatants vs 204.8 ± 6.13 pg/µg in GPx4 knockout supernatants), no change in the levels of PlGF (2.49 ± 1.29 ng/µg in control vs 2.13 ± 0.76 ng/µg in GPx4 knockout supernatants), as well as a strong increase in bFGF expression levels (14.6 ± 6.37 pg/µg in control vs 38.4 ± 8.66 pg/µg in GPx4 knockout supernatants). The increase of bFGF is in accordance with data from the literature describing that the 12/15-LOX product 13-HODE is an inducer of bFGF [27].

Baicalein Treatment of Tumor-Bearing Mice Reverses the Vascular Alterations in GPx4-Deficient Tumors

Baicalein, a specific LOX inhibitor, was originally described as an inhibitor of platelet-type 12-LOX, but later also reported to inhibit human 15-LOX [28]. Moreover, Baicalein has already been successfully used for treating neurological disorders in mice [29,30]. Because cell death of tGPx4 knockout cells could be prevented by Baicalein in vitro, we sought to address whether the effects of GPx4 disruption and perturbed 12/15-LOX activity might be reverted by Baicalein administration in vivo. For this reason, GPx4 knockout and GPx4 control tumor-bearing mice were treated daily by intraperitoneal injections of Baicalein. Immunohistologic analyses of Baicalein-treated tumor tissue revealed a clear reduction in microvessel density (untreated GPx4 knockout tumors 8.40 ± 0.45% vs treated knockout tumors 6.54 ± 0.85%; Figure 7, A–C), which was accompanied by an increase in tumor vessel diameters and mural cell coverage evaluated by αSMA-positive vessels (untreated GPx4 knockout tumors 7.21 ± 1.81 vs treated knockout tumors 9.22 ± 2.55; Figure 7D). Tumor vessel density and smooth muscle cell/pericyte coverage in Baicalein-treated control tumors revealed no difference compared with the untreated control group (tumor vessel density 8.79 ± 0.29% in untreated control tumors vs tumor vessel density 8.40 ± 0.42% in Baicalein-treated control tumors).