Combating pancreatic cancer with ovarian cancer cells

Top agents predicted to reverse gene signature of pancreatic cancer

We screened the candidate agents whose gene signatures are anti-correlated with that of pancreatic cancer using DrugMine. As a result, Figure 1A shows the top ten agents with most significant odds ratio (OR) value by Fisher’s exact test and most significant Spearman’s correlation coefficient (Rho). The p-values of both tests for all agents are almost zero, indicating that the agents- induced gene expression changes are significantly anti-correlated with that of pancreatic cancer, and further suggests that these agents could defend pancreatic cancer. Moreover, strikingly, we noted that ovarian serous carcinoma is among the top ten agents, indicating that it is anti-correlated with pancreatic cancer in gene expression changes. It is well known that cancers mostly share common pathways and mechanisms during their formation and development. However, here clear discrepancy was revealed between ovarian serous carcinoma and pancreatic cancer. This finding hinted that the opposite mechanisms could exist between the two cancers and thus suggested possible novel therapeutics, that is, killing pancreatic cancer with ovarian cancer cells. Based on the above observation, we proposed novel therapeutics for pancreatic cancer treatment as shown in Figure 1B, that is, injection of ovarian cancer cells into pancreatic tumor could inhibit pancreatic cancer.

Injection of ovarian cancer cell SKOV3 inhibited pancreatic tumor development

Firstly, we tested whether ovarian cancer cell SKOV3 could inhibit pancreatic cancer cell SW1990 in vitro, so we sorted and purified SKOV3 cells which were transfected with lentivirus carrying the GFP tag through GFP fluorescence by flow sorting. By co-culturing the SKOV3 cells with SW1990 cells, it was found that SKOV3 could reduce the proliferation rate of SW1990 by about 37% and showed a significant inhibition (Figure 2A, p-value = 0.0036, T-test), suggesting that ovarian cancer cell indeed can inhibit pancreatic cancer cell in vitro.

To confirm whether SKOV3 cells can inhibit pancreatic cancer in vivo, we first established subcutaneous pancreatic cancer xenografts by subcutaneous injection of SW1990 cells, then followed by intervention with low dose (LD, 3×10⁵ cells) or high dose (HD, 6×10⁵ cells) SKOV3 cells twice a week through intratumoral injection. Equivalent normal saline solution (NSCT) was given twice a week by intratumoral injection as blank control. Gemcitabine was used as a positive control in the intervention group (100mg/kg, twice a week by intraperitoneal injection). The mice were euthanized and the tumors were harvested after 11 days of intervention (Figure 2B). When compared with the NSCT group, we found a significant decrease of tumor weight in Gemcitabine group and the SKOV3 HD group after 11 days of treatment (Figure 2C, Gemcitabine vs. NSCT, p-value = 0.0015; SKOV3 HD vs. NSCT, p-value = 0.0278, one-way ANOVA test). When compared with the NSCT group, we found a dramatically decreased tumor volume in the Gemcitabine group, SKOV3 LD group and the SKOV3 HD group after 11 days of treatment (Figure 2D, Gemcitabine vs. NSCT, p-value = 0.0022; SKOV3 HD vs. NSCT, p-value = 0.0098; SKOV3 LD vs. NSCT, p-value = 0.0211, one-way ANOVA test). Not only that, the tumor volume was detected on the 4th and 7th day after treatment, which was significantly reduced compared with the NSCT group (Supplementary File 1). In addition, there was no significant difference in body weight change among the four groups (Figure 2E), but there was a tendency in slowing body weight loss in the SKOV3 HD group. The above results suggest that the ovarian cancer cell indeed can inhibit the growth of pancreatic tumors in vivo.

Potential mechanisms of the proposed CCC therapeutics

As we reported above, the proposed CCC therapeutics indeed inhibited the growth of pancreatic cancer through injection of ovarian cancer cells. As we described above, this therapeutics was proposed based on the observation that gene expression change of ovarian cancer is anti-correlated with that of pancreatic cancer (Rho = -0.31, p-value = 0, Figure 3A). To confirm whether injection of ovarian cancer cell reverses the gene expression change of pancreatic cancer, we identified the gene expression profiles of subcutaneous xenotransplanted tumors with (the high dose group) or without injection of ovarian cancer cells by RNA-sequencing. As expected, the ovarian cancer cell-induced gene expression change (with vs. without ovarian cancer cell treatment) is indeed anti-correlated with that (pancreatic cancer vs. normal pancreas) of pancreatic cancer (Rho = -0.19, p-value = 2.07e-95, Figure 3B). This result means that the ovarian cancer cell treatment indeed statistically reverses the gene expression signature of pancreatic cancer and thus together with the cell and animal results confirm the efficacy of the CCC therapeutics for pancreatic cancer. Given that the proposed CCC therapeutics is based on injection of cancer cells, however, people would regard it as one kind of immunotherapy. To confirm whether it belongs to immunotherapy or not and to explore whether it represents some novel therapeutics, we next performed function and pathway enrichment analysis for the DEGs of pancreatic tumors with and without ovarian cancer cell treatments using tools of DAVID Bioinformatics and WEAT.

Firstly, we investigated the enriched subcellular locations of the DEGs. As a result, both DAVID Bioinformatics and WEAT analysis revealed that the DEGs are mostly enriched in the extracellular space/region and cell membrane/surface (Figure 3C), which is consistent with the enriched functions of signaling, ion transport, extracellular matrix organization, and pancreatic secretion (Figure 3D).

Next, investigating the phenotype and disease these DEGs are involved in may provide critical insights for exploring the mechanisms of the CCC therapeutics. As a result, pancreatic-related diseases are among the top terms (Supplementary File 2). For example, the term of pancreatitis especially hereditary/pediatric pancreatitis is one of the top significant terms in all categories (ClinVar, DisGeNET, GWAS, Human Phenotype Ontology, Jensen DISEASES, and rare disease). In addition, the DEGs are also enriched in a number of other pancreatic-related disease, such as alcoholic pancreatitis, non-diabetic pancreatic disorders, chronic pancreatitis, pancreatic insufficiency, exocrine pancreatic insufficiency, pancreatic pseudocyst, idiopathic chronic pancreatitis, pancreatic ductal adenocarcinoma, acute recurrent pancreatitis, and autoimmune pancreatitis in the DisGeNET category (Figure 3E), and pancreatic steatorrhea in the Jensen DISEASES category. The above results suggest that the ovarian cancer cell treatment strikingly induced changes of genes being involved in pancreas function and phenotype in pancreatic tumors.

Given that the pancreas-related disease including pancreatitis is the top term related with the DEGs, it will be interesting whether the DEGs contribute to these phenotypes through inflammation or immune related functions and pathways. For doing so, we performed functions and pathways analysis (Figure 3D). Strikingly, biological processes such as protein digestion and absorption, proteolysis, response to growth hormone-releasing hormone, pancreas regeneration, positive regulation of cell adhesion, and positive regulation of type B pancreatic cell proliferation were mostly significant, suggesting that the CCC therapeutics could work by enhancing pancreas function, pancreas regeneration and cell adhesion through secreting signaling molecules to trigger proteolysis but not through inflammation and immune related functions and pathways. Moreover, molecular function (MF) analysis showed that the proteolysis could be mediated by serine-type endopeptidase but not other peptidase. Interestingly, the only significant term in MSigDB hallmark category is ‘Epithelial Mesenchymal Transition’ (EMT, FDR = 1.55e-7), suggesting that another possible mechanism of the CCC therapeutics is that the ovarian cancer cell could block the metastasis of pancreatic cancer by inhibiting its EMT process. Specifically, immune-related functional or pathway terms are not identified as significant, suggesting that the CCC therapeutics could not belong to immunotherapy.

We next try to explore genes or targets contributing to or being highly related to these DEGs, including categories of kinase, transcription factor (TF), BioPlex Interactome, and Pfam domains. As a result, 81 kinases and 109 TFs were identified to be mostly co-expressed with the DEGs (Supplementary Files 3, 4). It should be noted that NR5A2 (nuclear receptor subfamily 5 group A member 2) is the top 1 significant TF. Interestingly, it was reported that NR5A2 has critical roles in restraining inflammation in normal mouse pancreas and is considered to be highly relevant to human pancreatitis and pancreatic cancer [23]. In BioPlex analysis, there are 7 significant genes highly connected with these DEGs, including GDF11, GSTT1, FN1, ITGA4, TUBB1, TUBB3, and HSPA5 (Supplementary File 5). All of the 7 genes have critical roles in carcinogenesis, for example, it is known that FN1 (fibronectin 1) is involved in cell adhesion and migration processes including metastasis, and thus is associated with pancreatic cancer prognosis and survival [24]. For the Pfam Domain analysis, the only significant term is ‘Trypsin’ (FDR=4.72e-8), an enzyme produced in pancreas.

African elephant TP53 transgenic SKOV3 cells enhanced the inhibitory effects of pancreatic tumor

In the above section, we confirmed the efficacy of the CCC therapeutics on pancreatic cancer through injection of the SKOV3 ovarian cancer cells. It is then interesting to investigate whether genetically engineered ovarian cancer cell can further enhance the efficacy or not. It is known that TP53 (tumor protein 53) is a crucial tumor suppressor gene, mutated in majority of human cancers [25]. Interestingly, the multiple copies of TP53 and the enhanced p53-mediated cell apoptosis observed in African elephants may have evolved to offer such cancer protection [26]. Therefore, we transferred TP53 gene of African elephant into SKOV3 cells to investigate if it can enhance the inhibition of the growth of pancreatic cancer.

We sorted and purified SKOV3 cells which were transfected with lentivirus containing GFP tag and TP53 gene through GFP fluorescence by flow sorting, and detected the expression of TP53 by q-PCR. The results of q-PCR also showed that the expression of TP53 gene was significantly increased after transfection in SKOV3 cells (Figure 4A).

Similar to the work flow of the previous animal experiments, we upped the dose and frequency of cell injection, and followed by intervention with low dose (LD, 6×10⁵ cells) or high dose (HD, 1.2×10⁶ cells) SKOV3^GFP or SKOV3^TP53+GFP cells every two days through intratumoral injection. As shown in Figure 4B, the mice were euthanized and the tumors were harvested after 18 days of intervention. As a result, the tumor weight in the Gemcitabine group and SKOV3^TP53+GFP LD group had a significant decrease (Figure 4C, Gemcitabine group vs. NSCT, p-value = 0.016; SKOV3^TP53+GFP LD vs. NSCT, p-value = 0.025, one-way ANOVA test). Comparing with the NSCT group, xenograft tumors grown in both the SKOV3^GFP group and the SKOV3^TP53+GFP group of low dose or high dose had a smaller mean size (Figure 4D). The tumor volume was significantly smaller and significantly different from the NSCT group at each measurement after intervention. Furthermore, there was no significant difference in body weight change among the four groups (Figure 4E), but, early in the intervention, treatment with SKOV3^GFP or SKOV3^TP53+GFP cells can slow down the weight loss. In addition, as SKOV3 is also a rapidly proliferating cancer cell, we were concerned that the injected SKOV3 cells might develop into ovarian cancer in the pancreatic tumor. Therefore, we observed the green fluorescence under confocal microscope after frozen section of xenograft tumor. The result showed that no ovarian cancer was formed in the pancreatic tumor after SKOV3^GFP and SKOV3^TP53+GFP injection, but only some scattered GFP fluorescence signals were detected (Figure 4F). The above results indicate that the CCC therapeutics combined with genetic engineering may further enhance the treatment effect.

Animals

BALB/c nude male mice (6-8-week-old, weighing 16-20g) of SPF-class were purchased from Huafukang Animal Experiment Co., Ltd. (Beijing, China). The mice were housed under in a temperature-controlled environment with a 12/12 h light/dark cycle with standard diet and water ad libitum.

Cell lines and lentivirus

Human pancreatic cancer cell line SW1990 was purchased from ATCC Co., Ltd. (USA) and human ovarian cancer cell line SKOV3 was provided by Prof. Lixiang Xue of Peking University Third Hospital. Cell lines were maintained in RPMI 1640 (Gbico, USA) supplemented with 10% fetal bovine serum (Gibco, USA). Cells were cultured at 37° C in a humidified incubator containing 5% CO₂.

Lentivirus vectors containing the GFP tag, GFP tag and TP53 gene of Loxodonta africana were constructed, respectively. Lentivirus production was completed by the Genechem Company (Shanghai, China). SKOV3 cells were infected with the concentrated virus at a multiplicity of infection of 5 in the presence of HitransG A for 12 h. After 12 hours, the supernatant was replaced with fresh medium. GFP-positive SKOV3 cells were sorted by flow cytometry, and expression of TP53 in the infected cells was evaluated by q-PCR.

Cell experiment

10⁵ SKOV3^GFP cell and 10⁵ SW1990 cell / per well were co-cultured in 6-well plate, in which 2×10⁵ SW1990 cells / per well were planted in 6-well plate as a control. After 48 hours, 3 fields were selected from each well to count the total cells in the bright field and GFP cells (SKOV3 cells) in the fluorescence field, then the number of SW1990 cells was equal to the number of total cells minus the number of GFP cells (SKOV3 cells). The relative cell proliferation rate of SW1990 was calculated with a formula: (the cell number of SW1990 × 2 in co-cultured group) / (the cell number of SW1990 in control group) × 100%.

Quantitative real-time PCR assay

Total RNA from SKOV3 cells was isolated by using TriQuick reagent (R1100, Solarbio) and 5 μg RNA was reverse-transcribed for preparing cDNA by a commercial RT kit (Transgene, AH341). Real-time PCR was performed using the AriaMx Real-Time PCR System with Top Green PCR Master Mix (AQ131-01, TransGen Biotech) and the following primer: TP53 (F:TTTCACCCTTCAGATCCGTG; R:GACTGTCCCTTCTTAGACTTCG), GAPDH(F:CTC CTCCACCTTTGACGCTG, R: TCCTCTTGTGCTCTTGCTGG), GAPDH was used as the internal reference.

Establishment of subcutaneous transplantation tumor model

To produce SW1990 donor tumors, SW1990 cell suspension of logarithmic growth phase in the total volume of 0.2 mL was inoculated subcutaneously into the right posterior axillary line of nude mice. When the tumor grows to a certain extent, the tumor was aseptically peeled off and cut into small tissue pieces of 2×2×2 (mm³), and the tumor tissue pieces were inoculated into the right axilla of nude mice subcutaneously through a trocar to complete the subcutaneous pancreatic cancer model. The tumor volume was calculated with a standard formula: width² × length × 0.5. At the end of the experiment, the xenografts were weighted, excised, fixed in formalin or frozen at -80° C for further experiments.

Screening possible agents to defend pancreatic cancer

To explore possible agents for pancreatic cancer defense, we first got the gene-level expression profiles of 28 PDAC specimens and 2 normal pancreatic tissues from donors (GEO accession number: GSE56560) [40]. Genes with fold change (FC) value ≥1.50 or ≤0.67 are taken as the gene signature of PDAC. And then the reverse-transcriptomics based computational tool DrugMine from Co., Ltd. of Beijing JeaMoon Technology was used to explore possible agents (including small molecules, biological molecules, and traditional Chinese medicine etc.) to defend pancreatic cancer. The hypothesis of DrugMine is that if the gene signature of some agent is anti-correlated with that of PDAC, then this agent will be predicted to be able to defend pancreatic cancer. In addition, Spearman’s correlation was also performed in DrugMine based on the whole FC profiles.

RNA- sequencing

RNA sequencing was performed using RNA extracted from tumors in normal saline control (NSCT) and SKOV3 high dose (HD) group by using TriQuick reagent (R1100, Solarbio). Each group consists of 3 biological replicates. RNA degradation and contamination were monitored on 1% agarose gels. RNA purity was checked using the NanoPhotometer® spectrophotometer (IMPLEN, CA, USA). RNA integrity was assessed using the RNA Nano 6000 Assay Kit of the Bioanalyzer 2100 system (Agilent Technologies, CA, USA). RNA-sequencing was performed according to the standard protocol of Novogene Corporation Inc. The RNA-sequencing data are publicly available at GEO (GSE212173). Differential expression analysis was performed using the DESeq2 R package (1.16.1).

Functional and pathway analysis

To explore possible mechanisms of the CCC therapeutics for combating pancreatic cancer with ovarian cancer cells, we applied DAVID Bioinformatics [41] and the weighted functional and pathway analysis tool WEAT [42], through which here we analyzed a number of functional/pathway categories, including ARCHS4 Kinases Coexp, ARCHS4 TFs Coexp, BioPlanet, BioPlex, ClinVar 2019, DisGeNET, DSigDB, GO biological process, GO cellular component, GO molecular function, GWAS Catalog 2019, Human Gene Atlas, Human Phenotype Ontology, Jensen COMPARTMENTS, Jensen DISEASES, KEGG, MSigDB Hallmark 2020, Pfam Domains 2019, Rare Diseases AutoRIF Gene Lists, RNA Seq Disease Gene and Drug Signatures from GEO, and TRANSFAC and JASPAR PWMs. Genes with fold change (FC) >=2.0 or <=0.5 between the pancreatic cancer xenografts with and without SKOV3 treatment as the differentially expressed genes (DEGs).

Quantification and statistical analysis

All data were expressed as the mean ± SD. R (version 4.2) was used for all statistical analyses. For a comparison between two groups, the T-test was used. For comparisons between three or more groups, the one-way ANOVA (one-tailed) was used.