Endothelial cell growth medium (EGM), extracellular matrix gel (ECM), TRIzol, RIPA buffer, collagenase A, and AATF antibody were procured from Sigma-Aldrich (St. Louis, USA). Dulbecco’s modified eagle’s medium (DMEM), giemsa stain, hematoxylin, 8μm-transwell inserts, and 24- well plates were procured from HiMedia laboratories (India). Lipofectamine 3000, penicillin/streptomycin, glutamine, FBS were from Invitrogen (USA); WST-1 cell proliferation assay kit from Takara Bio Inc. (Shinga, Japan); antibodies to IgG, PEDF, pAkt, pErk1/2, pFAK, Akt, Erk, and control and AATF shRNA from Santa Cruz Biotechnology, Inc. (CA, USA); antibodies to β-actin from Cell Signaling Technologies (USA); verso cDNA synthesis kit, and DyNamo Colorflash SYBR green kit from Thermo Fisher Scientific (USA); western bright ECL HRP substrate from Advansta (USA); western blotting materials were from BioRad; primers from Integrated DNA Technologies (IDT) (Iowa, USA), and all other reagents were obtained from Thermo Fisher Scientific or Sigma.

Human HCC tissues (n=50) and normal tissues (n=15) were procured from the National Tumor Tissue Repository (NTTR), Tata Memorial Hospital, Mumbai, India. Normal liver tissue adjacent to the tumor will be considered as control. The study was approved by the institutional ethics committees of JSS Medical College, JSS AHER, Mysore, Karnataka, India (JSSMC/IEC/090721/28NCT/2021-22), and Tata Memorial Hospital (Inclusion Criteria: Patients with clinical or histologically documented hepatocellular carcinoma. All stages of HCC were considered; Exclusion Criteria: Patients with non-neoplastic lesions of the liver were excluded from the study). Clinical characteristics and biochemical parameters were evaluated. The demographic and clinicopathological data of the HCC subjects are described in Supplementary Table 1 .

The human umbilical cord was procured from Shree Devi Nursing Home, Mysore. Informed consent was obtained from the subjects for the use of cells in clinical research. The study was approved by the institutional ethics committee at JSS Medical College, JSS AHER, Mysore, Karnataka, India. (JSSMC/IEC/260822/37NCT/2022-23).

The freshly collected human umbilical cord (40-60 cm long, tied at both ends) from the maternity hospital was washed using 0.09% saline. After confirming the cord is devoid of hematic and physical damage, it is incubated with 2 mg/ml collagenase A solution at 37⁰C. The digested cells were collected into a tube and centrifuged at 750 g for 10 min at 4⁰C. The cells were resuspended with endothelial cell growth medium and cultured by incubating at 37⁰C in 5% CO₂. All procedure was carried out in the cell culture hood with laminar airflow under aseptic condition. HUVECs were used from passages 3-5 to ensure their endothelial characteristics and were freshly isolated to perform experiments in duplicate or triplicate ( Supplementary Figure 1B ).

Human HCC cells- QGY-7703 (kind donation from Dr. Devanand Sarkar, Virginia Commonwealth University, USA) and Hep3B (obtained from American Type Culture Collection (ATCC), USA) were cultured in Dulbecco’s modified eagle’s medium containing 4.5 g/L glucose and supplemented with 10% fetal bovine serum, L-glutamine, and 100 U/ml penicillin-streptomycin incubated at 37⁰C in 5% CO₂. The authentication of QGY-7703 cell line was done by short tandem repeats (STR) profiling.

Stable clones expressing AATF shRNA in QGY-7703 cells were prepared as described previously (32). Firstly, the optimal antibiotic concentration for selecting the stable cell colonies was determined using different concentrations of puromycin (1-10 μg/ml). Control and AATF shRNA plasmid containing puromycin resistance gene was transfected to QGY-7703 cells according to the manufacturer’s protocol. Individual colonies were isolated, expanded and maintained in 1 μg/ml Puromycin. Stable knockdown of AATF in QGY-7703 cells was confirmed by qRT-PCR and western blot. We have used two clones of AATF control and AATF knockdown cells in the experiments.

Human normal and HCC tissues were fixed in a 4% (v/v) formaldehyde solution in phosphate buffered saline for 16 h. After formalin fixation, tissues were processed and embedded in standard paraffin blocks. Subsequently, tissue sections of 5 μm thickness were cut from each paraffin block and stained with hematoxylin and eosin (H&E). A pathologist at JSS Hospital performed the histological grading of HCC according to the World Health Organization (WHO) classification: well differentiated, moderately differentiated, or poorly differentiated HCC (33). The various stages of HCC were determined according to the classification criteria of the American Joint Committee on Cancer (AJCC) TNM [primary tumor features (T), presence or absence of nodal involvement (N), and distant metastasis (M) staging systems (34).

The formalin-fixed, paraffin-embedded (FFPE) tissue sections were deparaffinized and rehydrated following the treatment with xylene and a series of ethanol concentrations. After antigen retrieval with citrate buffer (pH 6) at 94°C for 15 min, followed by washing with water, the sections were incubated with 3% hydrogen peroxide for 10 min. After incubating the slides with blocking buffer for 1 hr at room temperature, the slides were incubated with AATF antibody (1:100 dilution) overnight at 4°C in a humidified chamber. The signals were developed using the polyexcel HRP/DAB detection system-one step (PathnSitu, Biotechnologies) as per the manufacturer’s protocol, and the nucleus was stained using hematoxylin. All the immunohistochemistry images were taken using an Olympus BX53 microscope. The images were quantified using Image J software.

To examine the effect of the biologically active components secreted by HCC cells on angiogenesis, conditioned media was prepared from control and AATF knockdown clones of QGY-7703 cells and then treated on HUVECs. The AATF control and knockdown clones were cultured until cells reached 80-85% confluency. The cells were washed with PBS and then incubated in serum-free DMEM medium for 24 hours. The conditioned medium was collected and centrifuged at 2500 rpm for 5 min at 4⁰C to remove the dead cells and cell debris. The conditioned media was aliquoted and stored at -80⁰C. The amount of protein in the conditioned media was determined by Bradford’s protein assay ( Supplementary Figure 3B ). For all the experiments, the volume of conditioned media was normalized to have the same protein concentration as the control and AATF knockdown cells.

To neutralize the PEDF in the conditioned media secreted by the control and AATF knockdown QGY-7703 cells, the conditioned media was treated with anti-PEDF antibody. A non-specific antibody was used as an isotype control. The antibodies were used at a concentration of 5 μg/ml.

The concentration of PEDF was measured in the conditioned media collected from control and knockdown QGY-7703 cells by using ELISA kit (R&D Systems, Minnesota, USA) according to the manufacturer’s protocol.

The HUVECs were seeded in 96 well plates at a density of 10⁴ cells per well and cultured up to 70% confluency. The cells were treated with the conditioned media of control and AATF knockdown QGY-7703 cells for 24 h, 48 h, and 72 h. At the end of the treatment, the proliferation of HUVECs was evaluated using a premix WTS-1 cell proliferation assay mix (Takara Bio Inc., Japan). The absorbance was measured at 450 nm on a multi-mode plate reader (EnSpire™ Multimode Plate Reader, Perkin Elmer) according to the manufacturer’s protocol.

The HUVECs were seeded at a density of 5 X10⁵ cells per well in 6 well plates and allowed to attain 70-80% confluency. A scratch was made using a 1 ml pipette tip across the centre of the well, and the medium was removed to get rid of the detached cells. The HUVECs were incubated with conditioned media of control and knockdown cells with or without IgG and PEDF antibodies (Santa Cruz). The migration ability of the HUVECs was evaluated by measuring the gap widths at time intervals of 0 and 24 h. Images were acquired with a Zeiss Primovert inverted microscope and analyzed for the measurement of gap distance using ImageJ software.

Transwell inserts of 8 μm pore size were coated with matrigel matrix gel and placed into a 24-well plate. The HUVECs were suspended in the serum-free endothelial cell growth medium at the density of 5X10⁴ cells and seeded into each pre-coated transwell inserts. In the lower chambers of 24 well plates were added conditioned media from control and knockdown cells with or without IgG and PEDF antibodies. Cells were incubated at 37⁰C for 24 h to analyze the invasive ability of the cells. After the incubation, the non-invasive cells in the precoated transwell inserts were removed with a sterile PBS-soaked cotton swab. The invasive cells at the bottom of the transwell inserts were fixed with paraformaldehyde and stained with Giemsa stain. Images were captured using a Zeiss Primovert inverted microscope and analyzed by comparing the number of cells that had crossed the membrane between the control and knockdown conditioned media groups.

Fertilized chicken eggs were sterilized and pre-incubated at 37⁰C in 85% humidity. After 7 days of incubation, a small hole was made on the broad side of the shell, and carefully, a window of 1 cm² was created. The embryos were treated with the conditioned media of control and knockdown cells. The window was sealed, and the eggs were incubated for 48 h in the humidified incubator at 37⁰C. Images were photographed using a Nikon digital camera, and angiogenesis is quantified by comparing the number of blood vessels between the control and knockdown groups.

The protease activity of the metalloproteases was detected by gelatin zymography. 7.5% of polyacrylamide gels containing 0.1% gelatin with a Tris-glycine running buffer were used to separate proteins. After electrophoresis, gels were washed in 2.5% TritonX-100 prepared in 50 mM Tris-HCl of pH 7.5 for 1h. Later, gels were incubated in developing buffer (1% TritonX-100, 50 mM Tris-HCl pH 7.5 along with 5 mM Calcium chloride and 1 μM zinc chloride) for 24 h at 37⁰C. After the incubation, gels were stained using coomassie blue for 1 h and destained using a destaining solution. Images were taken using the Gel Doc system (Genesys) and analyzed in ImageJ software.

Total RNA from the cells and frozen liver tissues was isolated using the TRIzol method. The quality and concentration of the RNA were determined by a nanodrop spectrophotometer. The RNA was reverse transcribed by following the manufacturer’s instructions using the Verso cDNA synthesis kit. The real-time PCR reactions were carried out using the DyNamo Colorflash SYBR Green kit with 0.5 mM primers (IDT), and 50 ng of cDNA in a 20 μl reaction volume. The real-time PCR reactions were performed using the Rotor-Gene Q5plex HRM System (Qiagen). The relative quantification of the mRNA fold change was calculated as 2^-ΔΔCt and was expressed normalized with endogenous control β-actin. The primers for the qRT-PCR were validated, and the sequences of the primers used in this study are provided in Supplementary Table 2 .

The lysates were prepared by homogenizing the human liver tissues and HCC cells in RIPA buffer containing protease/phosphatase inhibitors. The supernatant was collected after the homogenized tissue or cell samples were centrifuged at 13000 rpm for 10 minutes at 4⁰C. The protein concentration was determined by using the Bio-Rad protein assay dye reagent (Bio-Rad) of Bradford’s protein estimation method. A 30-50 μg of protein was loaded to separate the proteins in the SDS-PAGE and transferred onto a nitrocellulose membrane for all western blots. The membranes were blocked using 5% nonfat skim milk for an hour at room temperature and probed with specific primary antibodies (AATF, β-actin, pErk1/2, Erk1/2, pAkt, Akt, pFak, and PEDF) for overnight incubation at 4⁰C. Furthermore, membranes were washed and incubated for secondary antibodies for 2 h at room temperature. The blots were developed using the Western Bright ECL HRP substrate, and images were captured using the Uvitec Alliance Q9 chemiluminescence imaging system. The bands were quantified using ImageJ software. The intensity of each band was normalized to its respective endogenous control, β-actin.

The RNA sequence and clinical information data of HCC tissues (n= 371) and normal liver tissues (n=50) were downloaded from the TCGA-LIHC website of The Genomic Data Commons (GDC: https://portal.gdc.cancer.gov/repository) up to November 15, 2022. TCGA dataset containing only primary tumors was processed. Different stages of HCC clinical samples as follows: Stage1: n=168; Stage2: n=84; Stage3: n=82 and Stage4: n=6. The expression values of AATF in terms of transcripts per million from the whole transcriptome profile were obtained for which a differential expression analysis was performed on normal and different stages of HCC datasets using limma R package at a significance level of FDR<0.05 along with p<0.05. The graphs were plotted using GraphPad Prism software (version 6).

Results were calculated as means ± SEM. Statistical significance was analyzed using Student’s t-test. All statistical analyses were performed using the GraphPad Prism software (version 6), and all experiments were considered significant with p<0.05 [p<0.05 (*, #) or <0.001 (**, ##)].

To investigate the role of AATF in HCC, mRNA and protein levels of AATF were measured in human HCC cell lines and found to be relatively overexpressed in QGY-7703 compared to Hep3B and normal liver tissues ( Figures 1A, B ). Consistent with these observations, the mRNA expression of AATF in human HCC tissues (n=50) was also found to be significantly upregulated compared to adjacent normal liver tissues (n=15) ( Figure 1C ). Histological analysis of human HCC tissues revealed distinctive changes in the cell structure and arrangement of the hepatocytes, confirming different grades of HCC ( Figure 1D ). Furthermore, immunohistochemistry revealed that AATF expression increased gradually from stages I to IV, as well as with the differentiation grades from well differentiated to poorly differentiated HCC (Figure E, Supplementary Figure 1A ). These findings show that AATF expression increases with the HCC stage and loss of differentiation, confirming AATF’s role in the development and progression of HCC. Next, we analyzed the publicly available TCGA database to evaluate the status of AATF in normal (n=50) and HCC (n=371). In support of our results, the data from the TCGA database provided evidence that there are significantly higher AATF differential transcription levels in HCC compared to normal. The levels of AATF were found to be significantly related to the pathological stages and tumor grades of HCC ( Supplemental Figure 2 ).

We investigated the regulatory role of AATF on angiogenesis, a major hallmark of cancer, which is responsible for the rapid recurrence and poor survival rate in HCC patients (35). The effect of AATF on angiogenesis was determined by assessing the impact of its loss of function. The initial step towards achieving this was the establishment of stable QGY-7703 cell lines. The expression of AATF was significantly reduced in the knockdown clones compared to puromycin-resistant control clones ( Figure 2A , Supplementary Figure 3 ).

Cell proliferation, migration and invasion are critical steps for the endothelial cells to form blood vessels in angiogenesis (36). To determine the effect of AATF on angiogenesis, the influence of conditioned media (CM) of control and AATF knockdown QGY-7703 cells on proliferation, migration, and invasion of human umbilical vein endothelial cells (HUVECs) was analyzed. We hypothesized that AATF knockdown would affect the angiogenic properties of HUVECs and vascular development in chick embryos. Consistent with this hypothesis, the CM of the knockdown cells significantly inhibited the proliferation of HUVECs compared to the control ( Figure 2B ). Furthermore, the chemotactic motility of endothelial cells was determined by migration assay and matrigel invasion assay. The results showed that CM of knockdown cells inhibited the migration of HUVECs and significantly reduced cell invasion compared to control, providing strong evidence that AATF knockdown affects the motility and matrix degradation capacity of HUVECs, which are critical for angiogenic sprouting ( Figures 2C, D ).

We further employed the chicken chorioallantoic membrane (CAM) assay, which is a physiological model of embryonic angiogenesis. Similar results were obtained where the CM of AATF knockdown cells showed a remarkable effect on the chicken embryo by significantly decreasing the vascular growth compared to the CM of the control cells ( Figure 2E ).

The proangiogenic and anti-angiogenic genes manifest themselves differently in HCC due to the activation of diverse oncogenic pathways. Our previous study using a human angiogenesis array revealed that PEDF, or SerpinF1, a well-known anti-angiogenic factor, is highly expressed in AATF knockdown cells compared to control cells (32). With this rationale, we investigated the regulatory role of AATF on PEDF levels in control and knockdown QGY-7703 stable cells. The AATF knockdown cells showed upregulated PEDF protein expression compared to controls ( Figure 3A ). Furthermore, PEDF levels were also measured by ELISA using the CM from control and knockdown cells. Similar results were obtained where the secretion of PEDF was significantly higher in the CM of AATF knockdown cells compared to the control ( Figure 3B ).

We next explored the mechanism by which PEDF is upregulated in AATF knockdown cells. PEDF, a member of the serpin superfamily, acts as a natural angiogenesis inhibitor and is found to be significantly downregulated in most cancers, including HCC (37). There is an increasing body of evidence for the involvement of matrix metalloproteinases type 2 (MMP2) and type 9 (MMP9) in the degradation of PEDF. Of note, PEDF acts as a substrate for MMP2 and MMP9 (38). Along the same lines, we examined the expression of MMP2 and MMP9 in CM of control and knockdown QGY7703 cells and interestingly found that MMP2 and MMP9 were down regulated in AATF knockdown cells compared to control ( Figures 3C, D ). Additionally, we also tested the MMP activity by performing gelatin zymography. Consistent with the expression data, there was decreased gelatinolytic activity of MMP2 and MMP9 in knockdown cells compared to control ( Figures 3E, F ). Together, these findings suggested that AATF inhibition downregulates MMP2 and MMP9, which prevents PEDF from being degraded in AATF knockdown cells as opposed to control cells.

Further investigations were carried out to determine whether PEDF played a key role in the suppression of angiogenesis in AATF knockdown cells. To test this hypothesis, we further validated the effect of conditioned media from control and AATF knockdown QGY-7703 cells on the migration and invasion of HUVECs and vascular formation in CAM in the presence of neutralizing anti-PEDF antibody. A non-specific IgG antibody served as an isotype control. The concentration of anti-PEDF antibody was determined by a dose response experiment on the proliferation of HUVECs with control and AATF knockdown CM, and a concentration of 5 μg/ml was chosen for the experiments. Consistent with the previous experiments, the conditioned media of AATF knockdown cells inhibited the migration (30% vs. 100% of cell migration in the control) and invasion (18 vs. 75 cells per field in the control) of HUVECs, and this inhibition was significantly diminished in the presence of PEDF neutralizing antibody [72% of cell migration ( Figures 4A, D ) and invasion assay- 46 cells per field ( Figures 4B, E )]. Similarly, the decrease in the vascular growth caused by the conditioned media of AATF knockdown cells (7 vs. 23 blood vessels in control) was diminished in the presence of PEDF neutralizing antibody (16 blood vessels) ( Figures 4C, F ). Taken together, these data provide solid evidence that knockdown of AATF suppresses angiogenesis in HCC via PEDF.

To examine the mechanisms underlying AATF-mediated angiogenesis in human HCC, we sought to explore several downstream signaling effectors that are responsible for endothelial cell survival and vascular permeability, cell proliferation, and migration. To test this hypothesis, HUVECs were treated with CM from control and AATF knockdown cells and examined for Erk1/2, Akt, and FAK phosphorylation by western blot analysis. Our results showed that the activation of the key components of the angiogenesis signaling pathway, including the phosphorylation of Erk1/2 ( Figure 5A ), Akt ( Figure 5B ), and FAK ( Figure 5C ), was decreased in HUVECs treated with CM from knockdown cells compared to control. To corroborate the involvement of PEDF in the angiogenesis signaling pathway, experiments were carried out in the presence of neutralizing anti-PEDF antibody. Consistent with our previous observations, the activation i.e. phosphorylation of Erk1/2, Akt, and FAK, which was decreased by the AATF knockdown, was diminished in the presence of PEDF neutralizing antibody ( Figures 5A–C ).