Dihomo-γ-linolenic acid inhibits xenograft tumor growth in mice bearing shRNA-transfected HCA-7 cells targeting delta-5-desaturase

Background We previously demonstrated that knockdown of delta-5-desaturase via siRNA transfection together with dihomo-γ-linolenic acid supplementation inhibited colon cancer cell growth and migration, by promoting the production of the anti-cancer byproduct 8-hydroxyoctanoic acid from Cyclooxygenase-2-catalyzed dihomo-γ-linolenic acid peroxidation. Here, we extend our study to investigate the effects of delta-5-desaturase-knockdown and the resulting intensified dihomo-γ-linolenic acid peroxidation in xenograft tumor mice model. Methods Four-week old nude mice bearing the human colon cancer cell HCA-7/C29 vs. its delta-5-desaturase knockdown analog (via shRNA transfection) were subject to 4-week treatments of: vehicle control, dihomo-γ-linolenic acid supplementation, 5-Fluorouracil, and combination of dihomo-γ-linolenic acid and 5-Fluorouracil. Tumor growth was monitored during the treatment. At the endpoint, the mice were euthanized and the tumor tissues were collected for further mechanism analysis. Results Delta-5-desaturase knockdown (shRNA) together with dihomo-γ-linolenic acid supplementation increased 8-hydroxyoctanoic acid production to a threshold level in xenograft tumors, which consequently induced p53-dependent apoptosis and reduced tumors significantly. The promoted 8-hydroxyoctanoic acid formation was also found to suppress the tumors’ metastatic potential via regulating MMP-2 and E-cadherin expressions. In addition, our in vivo data showed that delta-5-desaturase knockdown along with dihomo-γ-linolenic acid supplementation resulted in anti-tumor effects comparable to those of 5-Fluorouracil. Conclusions We have demonstrated that our paradigm-shifting strategy of knocking down delta-5-desaturase and taking advantage of overexpressed Cyclooxygenase-2 in tumor cells can be used for colon cancer suppression. Our research outcome will lead us to develop a better and safer anti-cancer therapy for patients. Electronic supplementary material The online version of this article (10.1186/s12885-018-5185-9) contains supplementary material, which is available to authorized users.


Background
Cyclooxygenase (COX) is a lipid-peroxidizing enzyme responsible for metabolizing polyunsaturated fatty acids to produce various lipid-derived molecules [1][2][3]. With Cyclooxygenase-1 being the constitutive isoform, Cyclooxygenase-2, the inducible form, can be readily induced in response to various stimuli including cancer promoters [4][5][6]. Overexpression of Cyclooxygenase-2 is a common phenomenon in many types of cancers. For example, it is known to overexpress in 85% of colorectal cancers and to be associated with colon cancer development by catalyzing peroxidation of arachidonic acid (AA, a downstream ω-6 fatty acid) to produce Prostaglandin E2 (PGE2) [7][8][9]. Hence, suppressing Cyclooxygenase-2 via inhibitor molecules has been extensively studied as a complementary therapy for cancer treatment [10,11]. However, Cyclooxygenase-2 inhibitors have normally resulted in limited clinical outcomes for cancer patients as Cyclooxygenase-2 can be readily induced by various stimuli in the cancer environment [4][5][6]12]. In addition, Cyclooxygenase-2 inhibitors have been found to commonly cause gastrointestinal injury and cardiovascular side effects in patients [13][14][15].
The ω-6 s and ω-3 s are two essential classes of dietary fatty acids. The ω-3 s have been shown to possess some anti-cancer activity and used as dietary supplements for cancer prevention and treatment, partially due to their competition against arachidonic acid for Cyclooxygenase-2 [16][17][18][19][20]. However, the more abundant ω-6 s (the ratio of ω-6 s vs. ω-3 s is 10:1 to 30:1 in the western diet [21][22][23]) have not received much research attention in cancer treatment due to the pro-cancer activities derived from Cyclooxygenase-2-catalyzed arachidonic acid peroxidation. Unlike many other research labs focusing on Cyclooxygenase-2 inhibition and ω-3 dietary supplementation in cancer treatment, our lab aims to develop an entirely novel anti-cancer strategy based on two often overlooked aspects: the commonly overexpressed Cyclooxygenase-2 in cancer, and the inevitable and abundant ω-6 s in our daily diet, to be exploited and manipulated to control cancers.
In the present study, we have made the first effort to test our novel anti-cancer concept and strategy using xenograft tumor models in nude mice bearing shRNA-transfected HCA-7/C29 targeting delta-5-desaturase (D5D-KD tumors). We have demonstrated that dihomo-γ-linolenic acid supplementation elevated 8-hydroxyoctanoic acid production in an autocrine manner to a threshold level (> 0.3 μg/g) in delta-5-desaturase-KD tumors and therefore significantly suppressed tumor growth (~40% reduction vs. delta-5-desaturase-WT tumor control). Formation of 8-hydroxyoctanoic acid was also found to induce p53-dependent apoptosis, and inhibited the metastatic potential of delta-5-desaturase-KD tumors.
In addition, dihomo-γ-linolenic acid supplementation along with delta-5-desaturase knockdown was able to greatly promote the efficacy of 5-FU in inhibiting tumor growth (~70% reduction vs. control).
Besides having promising outcomes for treatment of colon cancer, we have also demonstrated that dihomo-γlinolenic acid, along with a genetic delta-5-desaturase knockdown strategy, can suppress the growth, migration, and invasion of many other cancer cells, including pancreatic cancer BxPC-3 [27,28], breast cancer MDA-MB-231 and 4 T1 [29], lung cancer A549, liver cancer HepG2, and their associated xenograft tumors (unpublished research results). Our new strategy of making use of commonly overexpressed Cyclooxygenase-2 for anti-cancer purpose represents a paradigm shifting concept as it challenges the conventional Cyclooxygenase-2 inhibition strategy in cancer treatment. Our on-going research tasks include optimization of dose/duration of dihomo-γ-linolenic acid supplementation, development of a delivering system (e.g., nanoparticles) of delta-5-desaturase-siRNA to tumors, and discovery of effective delta-5-desaturase inhibitors, aiming to translating our new anti-cancer strategy to clinical settings in the near future.
A stable delta-5-desaturase-KD HCA-7/C29 cell line was created via shRNA transfection for the xenograft tumor study. Briefly, two strands of DNA oligonucleotides encoding delta-5-desaturase-targeted shRNA were designed with BLOCK-iT™ RNAi Designer (www.invitrogen.com/rnai) and purchased from Integrated DNA Technologies with the following sequences: target strand, TGCTGTAAT CATCCAGGCCAAGTCCAGTTTTGGCCACTGACTGA CTGGACTTGCTGGATGATTA; and complementary strand, CCTGTAATCATCCAGCAAGTCCAGTCAGTC AGT GGCCAAAACTGGACTTGGCCTGGATGATTAC. The delta-5-desaturase-targeted shRNA was then cloned into pcDNA tm 6.2-GW/miR vector and transformed into E.coli. The plasmid DNA from the expression clone was extracted and transfected into wild type HCA-7/C29 cells for 24 h. For antibiotic selection, the cells were incubated in fresh complete medium containing 10 μg/ml of Blasticidin. The Blasticidin-containing medium was refreshed every 3-4 days until Blasticidin-resistant colonies were identified (1 0-14 days). About 20 Blasticidin-resistant colonies were collected and expanded, followed by western blot analysis to evaluate the knockdown effect. A colony formation assay was conducted in order to determine whether shRNAtransfection affected the growth of HCA-7/C29 cells.

Xenograft tumor model and mouse treatment
Four-week old female nude mice (J:Nu, stock number 007850) were purchased from The Jackson Laboratory (Bar Harbor, ME), and were housed in a pathogen-free Innovive IVC system with water and food ad libitum. After allowing the mice to acclimatize for 1 week, tumor xenografts were established by subcutaneously injecting 2 × 10 6 delta-5-desaturase-WT or delta-5desaturase-KD (shRNA) HCA-7/C29 cells into the hind flank of each mouse. The mice were then fed with a standard diet for two more weeks to allow the tumors to grow, and further divided into four sub-groups for four-week treatments (6 mice per groups): (1) vehicle control; (2) dihomo-γ-linolenic acid ethyl ester at a dose of 8 mg/mouse (in 250 μL 32% ethanol solution), oral gavage, twice a week; (3) 5-FU at 30 mg/kg (in 50 μL PBS), i.v. injection, twice a week; and (4) combination of dihomo-γ-linolenic acid ethyl ester and 5-FU. All animal experiments were approved by the Institutional Animal Care and Use Committee at North Dakota State University.

Tumor size measurement
Tumor size was measured twice a week using a digital caliper during the entire treatment period. Tumor volume was calculated as: V = L × W 2 /2. After four-week treatment, the mice were euthanized to collect tumor tissues for further analysis described in following sections.

Colony formation assay and in vitro apoptosis analysis
Cancer cell survival and apoptosis after various treatments were assessed by a colony formation assay and an Annexin V Apoptosis Detection Kit I, respectively, as described elsewhere [25][26][27][28]. In the colony formation assay in vitro, survival fraction = plating efficiency in treatment group/plating efficiency in control group. An Accuri C6 flow cytometer was used for apoptosis analysis (10,000 cells were counted for each sample). Unstained cells and cells stained with FITC Annexin V only or PI only were used to set up compensation and quadrants. Data were analyzed by FlowJo software.

Quantification of DGLA/AA ratio and PGE2 level
The amount of dihomo-γ-linolenic acid, arachidonic acid, and PGE2 in cells were quantified via LC/MS analysis as described elsewhere [24,25]. For the in vivo study, tumor tissues were frozen in liquid nitrogen, crushed to a fine powder, and mixed with water, methanol, and internal standards (arachidonic acid-d 8 , dihomo-γ-linolenic acid-d 6 , and PGE2-d 9 ). The mixtures were vortexed for 1 min and set on ice for 30 min, followed by the same extraction procedures and LC/MS analysis method as described in the in vitro experiment [24,25].

GC/MS analysis of 8-HOA
GC/MS analysis was employed to determine the amount of 8-hydroxyoctanoic acid (in its derivative of pentafluorobenzyl bromide) formed from cells after various treatments as described elsewhere [26,[28][29][30]. In the in vivo study, tumor tissues were frozen in liquid nitrogen, crushed to a fine powder, and mixed with water, methanol, HCl, an internal standard (hexanoic acid), and dichloromethane, followed by the same extraction and GC/MS analysis procedures as the in vitro experiment [26,[28][29][30].

Immunofluorescence analysis
Immunofluorescence studies were performed to assess the expressions of delta-5-desaturase, Cyclooxygenase-2, cleaved PARP, Ki-67, MMP-2 and E-cadherin in tumor tissues as described elsewhere [30]. Briefly, tumor tissues were fixed with formaldehyde and embedded in paraffin blocks. Tissue sections were deparaffinized with xylene, rinsed, and rehydrated through a graded series of alcohol. For antigen retrieval, the slides were placed in a rack in the retriever (Aptum Biologics Ltd., UK) filled with sodium citrate buffer, the retriever was run for 30 min at preset pressure and temperature. Then the tumor sections were incubated with primary antibodies and secondary antibodies. Cell nuclei were counter-stained with DAPI. The images were acquired with a Zeiss Axio Imager M2 microscope.

Statistics
All the quantification data were presented as mean ± standard deviation (SD) from at least three separate experiments (for in vitro studies), or from six tumor samples per treatment group (for in vivo studies). Statistical differences between groups were evaluated by analysis of variance and post hoc t-test; differences were considered significant with a p-value < 0.05.

Increased levels of 8-HOA lead to suppression of xenograft tumor growth
Tumor size measurements showed that 4 weeks of dihomo-γ-linolenic acid supplementation had no significant effect on the growth of delta-5-desaturase-WT tumors (Fig. 4a). By comparison, dihomo-γ-linolenic acid supplementation significantly decreased the sizes of delta-5-desaturase-KD tumors relative to the control group (Fig. 4b), associated with the elevated 8-hydroxyoctanoic acid production. It was noted that about a 40% size reduction was achieved in delta-5-desaturase-KD tumors in mice treated with dihomo-γ-linolenic acid supplementation vs. delta-5-desaturase-WT tumors in mice treated with vehicle control (Fig. 4a). Data also showed that 5-FU was able to inhibit tumor growth in both the delta-5-desaturase-WT group and the delta-5-desaturase-KD group ( Fig. 4a and b). It is noteworthy that dihomo-γ-linolenic acid supplementation resulted in an average tumor size of~178.2 ± 31.9 mm 3 in delta-5-desaturase-KD tumors (Fig. 4b), leading to a similar effect compared to 5-FU treatment in delta-5-desaturase-WT tumors (~204.3 ± 55.3 mm 3 , Fig.  4a). In addition, concurrent treatment with dihomoγ-linolenic acid along with 5-FU in mice bearing delta-5-desaturase-KD tumors led to an improved tumor size reduction (100.1 ± 24.3 mm 3 , Fig. 4b).
In order to validate shRNA knockdown efficiency during the 4-week treatment, immunofluorescence studies were performed to assess the expression levels of delta-5-desaturase in tumor tissues. Data revealed that delta-5-desaturase-KD tumors had significant lower delta-5-desaturase expressions than delta-5-desaturase-WT tumors for all treatments (Fig. 4c and d).
Western blotting analysis revealed that dihomo-γ-linolenic acid supplementation did not alter the expression of apoptotic proteins in delta-5-desaturase-WT tumors (Fig. 5c). By comparison, in delta-5-desaturase-KD tumors, dihomo-γ-linolenic acid supplementation significantly increased the expression of p53 and decreased the expression of procaspase 9, indicating the activation of the p53-dependant apoptotic pathway (Fig. 5d). We also observed that dihomo-γ-linolenic acid supplementation caused the up-regulation of acetyl histone H3 and Fig. 3 Delta-5-desaturase knockdown leads to accumulation of dihomo-γ-linolenic acid and 8-hydroxyoctanoic acid in xenograft tumor tissues. a LC/MS profile of dihomo-γ-linolenic acid levels in all groups of tumor tissues; b Calculated dihomo-γ-linolenic acid/arachidonic acid ratio from LC/ MS analysis in different tumor tissues. The dashed line represents the basal level of dihomo-γ-linolenic acid/arachidonic acid ratio in tumors from mice receiving no dihomo-γ-linolenic acid supplementation; c GC/MS profile of 8-hydroxyoctanoic acid production from delta-5-desaturase-WT and delta-5-desaturase-KD tumor tissues. The dashed line represents the threshold level of 8-hydroxyoctanoic acid that can be only reached in delta-5-desaturase-KD tumors; d LC/MS profile of PGE2 from delta-5-desaturase-WT and delta-5-desaturase-KD tumor tissue. All of the data represent mean ± standard deviation with n = 6. *: significant difference with p < 0.05 γH2AX in delta-5-desaturase-KD tumors (Fig. 5d), which is consistent with our previous reports in siRNA-transfected delta-5-desaturase cells [25][26][27]. These data together suggested that the elevated endogenous 8-hydroxyoctanoic acid production in delta-5-desaturase-KD tumors could suppress tumor growth, likely via affecting histone acetylation/deacetylation and causing DNA damage.

Elevated 8-HOA production suppresses metastasis potential in D5D-KD tumors
Data from immunofluorescence studies showed that dihomo-γ-linolenic acid supplementation increased the expression of MMP-2 (a marker for tumor metastasis) in delta-5-desaturase-WT tumors, while significantly decreasing MMP-2 expression in delta-5-desaturase-KD tumors ( Fig. 6a and b). Consistently, dihomo-γ-linolenic acid supplementation also increased the expression of E-cadherin (a tumor metastasis inhibitor) in delta-5-desaturase-KD tumors compared to the vehicle control, while no such effect was observed in delta-5-desaturase-WT tumor tissues.

Discussion
We had previously demonstrated that siRNA-delta-5-desaturase knockdown in different types of cancer cells can promote the production of 8-hydroxyoctanoic acid from intensified Cyclooxygenase-2-catalyzed dihomoγ-linolenic acid peroxidation; the 8-hydroxyoctanoic acid served as an HDAC inhibitor to suppress cancer cell growth, migration, and invasion [25][26][27][28][29]. In the present study, we created stable delta-5-desaturase-KD HCA-7/ C29 cells via shRNA transfection and made the first effort to test the anti-tumor effect of our novel strategy in xenograft tumors. Tumor size measurement of delta-5-desaturase-KD tumors during 4-week treatment. Insert: representative tumor photos from each treatment group at the end of the treatment; *: significant difference vs. corresponding control with p < 0.05; c Representative immunofluorescence images for delta-5-desaturase expression in tumor tissues; delta-5-desaturase was stained in pink, cell nuclei were counter stained with DAPI; d Quantification of the mean intensity of delta-5-desaturase staining in each sample. All of the data represent mean ± standard deviation with n = 6. *: significant difference with p < 0.05 Our data showed that shRNA knockdown of delta-5-desaturase in HCA-7/C29 cells promoted 8-hydroxyoctanoic acid production from dihomo-γ-linolenic acid peroxidation, which then significantly suppressed the growth of delta-5-desaturase-KD cells in vitro (Fig. 1). Western blot (data not shown) further confirmed that promoting 8-hydroxyoctanoic acid formation from Cyclooxygenase-2-catalyzed dihomo-γ-linolenic acid peroxidation in delta-5-desaturase-KD cells resulted in a significant increase of acetyl histone H3 and γH2AX. We demonstrated again that the anti-proliferation effect of dihomo-γ-linolenic acid is actually derived from 8-hydroxyoctanoic acid's action to inhibit HDAC and damage DNA in cells [25][26][27]. In addition, our strategy of delta-5-desaturase knockdown and dihomo-γ-linolenic acid treatment also improved the cytotoxicity of 5-FU to cancer cells (Fig. 2).
Consistently, our in vivo data demonstrated that delta-5-desaturase knockdown in xenograft tumors led to elevated levels of 8-hydroxyoctanoic acid in mice with dihomo-γ-linolenic acid supplementation (Fig. 3), which consequently inhibited the tumor growth (Fig. 4, Additional file 2: Table S1 and Additional file 3: Table S2). In addition, while 5-FU treatment alone was able to suppress the growth of both delta-5-desaturase-WT and delta-5-desaturase-KD tumors (Fig. 4, Additional file 2: Table S1 and Additional file 3: Table S2), a two-factor analysis (considering 5-FU and delta-5-desaturase-KD/ dihomo-γ-linolenic acid as the factors, Additional file 4: Table S3) suggested an additive effect on tumor inhibition from the combination of 5-FU and delta-5-desaturase-KD/dihomo-γ-linolenic acid. In our future studies, we plan to test the combinational effects of our strategy with different chemo-drugs, including decitabine and sorafenib, as they are reported to synergize with HDAC inhibitors to exert anti-cancer activities [38,39].
We have also noted that, without dihomo-γ-linolenic acid supplementation, the tumor size of the delta-5-desaturase-KD vehicle control group is slightly smaller than that of the delta-5-desaturase-WT vehicle control group (Fig. 4), consistent with the relatively higher concentration of dihomo-γ-linolenic acid, ratio of dihomo-γ-linolenic acid/arachidonic acid, and concentration of 8-hydroxyoctanoic acid (although 8-hydroxyoctanoic acid never reached the threshold level). In the in vitro system, delta-5-desaturase-KD treatment alone (i.e., without dihomo-γ-linolenic acid treatment) had no influence on HCA-7/C29 cell growth (Additional file 1: Figure S1). Since upstream ω-6 s in the diet may not be able to convert enough dihomo-γ-linolenic acid in the body, certain amount of dihomo-γ-linolenic acid supplementation is necessary to elicit its anti-cancer activities for reaching a threshold level of 8-hydroxyoctanoic acid from Cyclooxygenase-2 peroxidation.
The body weights of the all of experimental mice were monitored throughout the treatment period (Additional file 5: Figure S2), and no significant change was noted among the different treatment groups.
Results also showed that dihomo-γ-linolenic acid supplementation led to increased MMP-2 expression (i.e., higher metastasis potential) and elevated levels of arachidonic acid as well as PGE2 (Fig. 3d) in delta-5-desaturase-WT tumors (Fig. 6). PGE2 has been shown to play a role in cancer migration [40,41]. However, in delta-5-desaturase-KD tumors, dihomo-γ-linolenic acid supplementation suppressed MMP-2 expression, associated with higher levels of 8-hydroxyoctanoic acid and lower levels of PGE2 (Fig. 3). E-cadherin is a cell adhesion molecule; decreased E-cadherin expression in the tumor environment is correlated with a strong invasive potential [42]. Here we observed that dihomoγ-linolenic acid supplementation greatly elevated E-cadherin levels in delta-5-desaturase-KD tumors compared to the vehicle control (Fig. 6), indicating less invasive potential. In this study, no spontaneously metastasizing tumor was observed in the subcutaneous xenograft model, therefore, our on-going research on orthotopic colon tumors will provide more insight into how our strategy would actually perform on metastasizing tumors and in cancer patients, as that model has a tumor microenvironment very similar to the original tumor.
It has been a challenge to deliver therapeutic RNAs to tumors due to various issues and concerns [43]. In our on-going study, we are employing innovative 3-way-junction RNA nanoparticles to specifically deliver All of the data represent mean ± standard deviation with n = 6. *: significant difference with p < 0.05. c and d Western blotting of p53, procaspase 9, acetyl histone H3 and γH2AX in delta-5-desaturase-WT and delta-5-desaturase-KD tumor tissues.The relative expression level of each protein was normalized using β-actin as a loading control; All of the data represent mean ± standard deviation with n = 6. *: significant difference vs. control with p < 0.05 delta-5-desaturase-targeting siRNA into cancer cells/tumors [44][45][46]. The newly developed multi-functional, thermodynamically and chemically stable RNA nanoparticles were designed to harbor cancer targeting ligands as well as delta-5-desaturase-targeted siRNA to inhibit delta-5-desaturase expression specifically in tumor cells.
Our ongoing study has shown that the RNA nanoparticles carrying delta-5-desaturase-siRNA specifically targeting tumors are able to inhibit delta-5-desaturase expression and suppress colon cancer growth when dihomo-γ-linolenic acid is supplemented concurrently. In addition, it has been reported that various small compounds possess Fig. 6 dihomo-γ-linolenic acid supplementation suppresses the metastasis potential in delta-5-desaturase-KD tumor tissues. a Representative images for MMP-2 and E-cadherin expressions in tumor tissues. MMP-2 was stained in red, E-cadherin was stained in green, cell nuclei were counter stained with DAPI; and b Mean intensities of MMP-2 and E-cadherin in each sample. All of the data represent mean ± standard deviation with n = 6. *: significant difference with p < 0.05 potent delta-5-desaturase inhibitory activities which can be potentially applied in our strategy for clinic use [47][48][49]. We are also working on developing new specific and effective delta-5-desaturase inhibitors for use in cancer patients (pending US-provisional patent application).

Conclusion
The present research demonstrated that delta-5-desaturase knockdown and dihomo-γ-linolenic acid supplementation in HCA-7/C29 xenograft tumors results in elevated 8-hydroxyoctanoic acid production, which serves an HDAC inhibitor to induce cell apoptosis pathway and suppress tumor growth. Compared to the more conventional Cyclooxygenase-2 inhibition strategy, our novel strategy of inhibiting delta-5-desaturase and taking advantage of the high Cyclooxygenase-2 expression in cancer cells will lead to better anti-cancer effects in two ways: stimulating an anti-cancer effect from dihomo-γ-linolenic acid while decreasing the pro-cancer effect from arachidonic acid. In addition, considering the fact that cancer cells in general have overexpressed Cyclooxygenase-2 levels and higher fatty acid intake rates than normal cells and tissues [7][8][9]50], we anticipate that our strategy will lead to fewer side effects and safer cancer treatment outcomes.