Cell culture and viability detection

GES1 cells, MGC803 cells, HK2 cells, SW839 cell, NCM460 cells, HT29 cells, HCT116 cells, and SW620 cells were cultured in RPMI1640 medium (containing 2 g·L−1 (11.1 mM) glucose) with 10% FBS, at 37 °C in 5% CO2 and 95% saturated atmospheric humidity. The cells were plated into 96-well plates (1 × 104 cells/100 μL medium) and incubated for 24 h, then the medium was replaced with 100 μL fresh medium containing different concentrations of glucose (0, 1, 2, 3, 4, 5, 6, 7 and 8 g·L−1, also as 0, 5.6, 11.1, 16.7, 22.2, 27.8, 33.3, 38.9 and 44.4 mM, respectively). Then the cells were cultured for another 48 h, and the CCK-8 kit was applied to measure cell viability. The absorbance at 490 nm was determined by a Microplate Reader (Thermo Fisher Scientific). The cell viability = (Value test − Value blank) / (Value control − Value blank) × 100%. The dosages with the viabilities greater than 80% meanwhile significantly different from the control (P < 0.05) were chosen as the appropriate concentrations of glucose used in the following experiments.

Cell apoptosis assay

Cells grew in 6-well plates treated with a growth medium containing increasing concentrations of glucose (1, 2, 3, 4, 5, and 6 g·L−1, respectively). The cells were collected and resuspended in 250 μL binding buffer, then treated with 15 μL FITC-Annexin V buffer and 30 μL propidium iodide (PI) buffer (10 g·L−1, as 15.0 mM), gently vortexed and incubated for 15 min in the dark at 25 °C. Subsequently, 300 μL binding buffer was added into each tube and then analyzed by flow cytometry (BD) within 1 h.

Preparation of lactoferrin with approximate 50% iron saturation

Lactoferrin with 100% iron saturation (Holo-LF) was purchased from Sigma (USA). 11 g Holo-LF was added with 300 mL citrate buffer (pH = 5), and adjusted PH value to 8.2, to promise iron saturation as 49%-51% according to the fitting equation (Y = 0.0643*2.27045X, Y stands for LF iron saturation, X stands for PH value). Then the solution was concentrated and desalted through a 30 kDa ultrafiltration membrane. When the conductivity was less than 1.5 ms·cm−1, the volume was concentrated 5 times and lyophilized (Chinese Invention patent application, Application publication number: CN 110655570 A).

Cell invasion and migration analyses

In the migration test, the upper chambers were seeded with 5 × 103 cells in 200 μL serum-free medium containing normal or high glucose, respectively, and 400 μL of medium (2 g·L−1 or 5 g·L−1 glucose, accordingly) with 15% FBS was added to the lower chambers. Lactoferrin with approximate 50% iron saturation (LF, 0.5 g·L−1, as 6.3 μM) prepared above was added to the upper chamber and cocultured for 24 h, then cell invasion was detected by calculating three random captured pictures [20].

In the invasion test, HT29 cells were plated in a 6-well plate and incubated for 24 h to achieve a cell density greater than 90%. A single lesion with a width of approximately 5.0 mm was scratched across the cell monolayer by mechanical scraping. The cells were then incubated with LF (0.5 g·L−1) dissolved in a normal (2 g·L−1) or high glucose (5 g·L−1) medium. The width of the scratch wound was photographed and scanned again 24 h later, and the recovery rate was measured [20].

mRNA sequencing and DNA methylation profiling

Total RNA (1 μg) extracted from each cell sample was firstly utilized to poly-d(A)-RNA isolation with NEBNext Magnetic Oligo d(T)25 Beads (NEB), and then used for mRNA library preparation with an RNA Library Prep Kit for Illumina (NEB) according to the instruction procedures. All of the libraries were subjected to 150 bp pair-end sequencing on an Illumina HiSeq2000 platform. After sequencing, trimmed and cleansed reads were analyzed by using the Bowtie2 suite to align to the hg19 reference genome and count normalized transcript abundance. Differentially expressed genes (DEGs) were calculated using the DESeq2 package and further analyzed based on GO biological processes, molecular functions, and the KEGG pathway. DNA methylation profiling was studied using the Infinium Human Methylation EPIC Bead Chip. In brief, 6 samples of 500 ng genomic DNA isolated from each treatment (HT29 cells as control group, HT29 cells treated with 0.5 g·L−1 LF in high glucose medium as the LF group, 4 biological repetitions) were treated with EZ DNA methylation kit (Zymo Research), the targets were prepared, labeled and hybridized with the kits and reagents indicated by the Infinium HD Methylation Assay Protocol Guide (15019519B). Processed methylation chip after single base extension and staining was scanned using an iScan reader (Illumina). Generated microarray data were analyzed using Genome Studio software v2011.1 (Illumina), and for quality control, methylation measures with a detection p-value > 0.05 and samples with a CpG coverage < 95% were removed. After initial normalization using internal controls in the Genome Studio software, the methylation levels of CpG sites were calculated as β-values (β = Intensity (methylated)/intensity (methylated + unmethylated)). The data were further normalized and the differential DNA methylation was assessed using the IMA 3.12-R package. P-values were calculated by t-test corrected for multiple hypotheses testing by the Benjamini–Hochberg method in combination with the Illumina custom false discovery rate (FDR) model (www.tandfonline.com). A threshold for differential DNA methylation was set at FDR-corrected P-value lower than 0.05. The functions of the associated genes were further studied based on GO and KEGG pathway analysis using KOBAS.

Animal models

The animal experiments were approved by the Ethics Committee of Chinese Academy of Agriculture Sciences (Beijing, China; permission number: IAS2020-90), conforming to internationally accepted principles in the care and use of experimental animals (NRC, 2011). All surgical procedures were performed under sodium pentobarbital anesthesia, and all efforts were made to minimize the suffering of the mice.

In C57BL/6 mouse model, 80 male mice (18–22 g) were randomly divided into two parts (16 groups in total): groups in the first part were normal control without any treatment, tumor control treated with HT29 cells implantation, tumor control + LF, tumor control + NT5DC3 protein (Novus Biologicals, USA), tumor control + HKDC1 antibody (Abcam, USA), tumor control + LF + NT5DC3 protein, tumor control + LF + HKDC1 antibody, tumor control + LF + NT5DC3 protein + HKDC1 antibody; while groups in the second part included diabetic mice, diabetic mice implanted with HT29 tumor (dia-tumor mice), dia-tumor mice + LF, dia-tumor mice + NT5DC3 protein, dia-tumor mice + HKDC1 antibody, dia-tumor mice + LF + NT5DC3 protein, dia-tumor mice + LF + HKDC1 antibody, dia-tumor mice + LF + NT5DC3 protein + HKDC1 antibody. The 40 diabetic mice were fed with a high-fat diet for 30 days continuously, and the mice were intraperitoneally administered with streptozocin (STZ, 100 mg·kg−1 (0.38 mM·kg−1) b.w.) once on the 31st day. Then the indicators including fasting blood glucose detection (FBG), oral glucose tolerance test (OGTT), glycated serum protein (GSP) and serum insulin (INS) of mice were detected on the 33rd day to confirm the successful model construction (data in Additional file 7: Table S4). 1 × 108 HT29 cells in 150 μL matrigel medium (BD) were subcutaneously injected into the back of each mouse, and when the tumors volume reached 100–120 mm3, the total 80 mice were treated with LF (250 mg·kg−1 b.w.), NT5DC3 protein (50 mg·kg−1 (0.94 mM·kg−1) b.w.), HKDC1 antibody (50 mg·kg−1 (0.42 mM·kg−1) b.w.), or their combinations, respectively. LF was orally administered by gavage, while NT5DC3 protein or HKDC1 antibody was injected through the tail vein every two days at the same time. All the mice were sacrificed on the 28th day, and the tumors were weighed.

In BALB/c nude mouse model, 50 male BALB/c nude mice (18–22 g) were randomly divided into two parts (10 groups in total): groups in the first part were normal control without any treatment, tumor control treated with HT29 cells implantation, tumor control + LF, tumor control + NT5DC3 protein, tumor control + LF + NT5DC3 protein; while groups in the second part included diabetic mice, diabetic mice implanted with HT29 tumor (dia-tumor mice), dia-tumor mice + LF, dia-tumor mice + NT5DC3 protein, dia-tumor mice + LF + NT5DC3 protein. The diabetic mice model was constructed and protein treatments were performed as described in the C57BL/6 mouse model. All the mice were sacrificed on the 28th day, and the tumors were weighed.

In the two mouse models, tumor diameters were detected with a caliper every 4 days, and tumor volume was calculated using the following formula: tumor volume (mm3) = 0.5 × length (mm) × width (mm)2. Individual tumor suppression rate (%) = (the average tumor weight in the control group − the individual tumor weight in the LF treatment groups) / the average tumor weight in the control group × 100%, as the average tumor weight in the control group was calculated by each tumor weight in the control group. Relative tumor volume (RTV, %) = detected volume / volume before dosing × 100% [20].

The 5mC detection and m6A detection

According to the method [41], 1 µg DNA was denatured at 100 °C for 5 min and subsequently chilled at 4 °C for 10 min. One-tenth volume of 0.15 M ammonium acetate (pH 7.5) and 2.5 units of DNase I (TransGen) were added, then the mixture was incubated at 37 °C for 4 h. 2 units of Alkaline Phosphatase (TaKaRa) was added into the solution and incubated for an additional 3 h at 37 °C. Thereafter, the mixture was incubated for 12 h at 37 °C with 40 units of Exonuclease I (TaKaRa). The complete lysis mixture was placed in a refrigerator at 4 °C for LC–MS/MS detection [42]. 1 μg genomic DNA was hydrolyzed by utilizing a DNA degrease Plus™ kit following the manufacturer’s instructions and according to the previous method [43]. In both test samples and standards, the hydrolyzed DNA was analyzed by liquid chromatography-electrospray ionization tandem mass spectrometry with multiple reaction monitoring (LC–ESI–MS/MS-MRM), and the MRM method was applied to monitor three transitions for each analysis, the experiment parameters in the 5mC/m6A detection by MS were demonstrated in Additional file 4: Table S1. Finally, the total amount of 5mC in test samples from the 5mC MRM peak area was calculated by dividing the sum of the 5mC and cytosine peak areas (5mC/C) [43].

Purified total mRNA (200 ng) was digested to its constituent mono-nucleosides according to previous method [42]. MRM mode was applied for the UPLC-MS/MS analysis through monitoring transition pairs in the Additional file 4: Table S1.

Detection of 5mC- and m6A-related genes

100 ng total RNA from cells samples or blood samples was extracted, the total RNA samples were transcribed into cDNA (42 °C for 10 min, 65 °C for 10 s, stored at 4 °C) by PrimeScript™ RT reagent Kit (TaKaRa). Primers of evaluated genes including NT5DC3, HKDC1, Homo sapiens DNA methyltransferase 1 (DNMT), methyltransferase-like 3 (METTL3), methyltransferase-like 14 (METTL14), Wilms' tumor 1-associated protein (WTAP), fat mass, and obesity-associated factor (FTO), AlkB homologue 5 (ALKBH5), YTH N6-methyladenosine RNA binding protein 1 (YTHDF1), YTH N6-methyladenosine RNA binding protein 2 (YTHDF2), YTH N6-methyladenosine RNA binding protein 3 (YTHDF3) and GAPDH, as well as siRNA fragments of these genes, were outlined in Additional file 5: Table S2, and GAPDH was utilized as the internal reference to assure the equal loadings. qRT-PCR was performed using 96-well microwell plates in a total volume of 20 μL, containing 1 μL template cDNA (10 ng·μL−1), 0.5 μL forward primer (10 μM), 0.5 μL reverse primer (10 μM), 10 μL of TB Green® Fast qPCR Mix (TaKaRa). The q-PCR reactions were performed at 95 ℃ for 3 min, followed by 40 cycles of 95 ℃ for 10 s, 60 ℃ for 30 s by using two-step qRT-PCR. All qRT-PCR reactions were performed.

Methylation sites verified by SELECT qPCR

DNA methylation region of NT5DC3 (5mC, CpG island 1) determination and primer design principles: DNA sequence of NT5DC3 (GenBank Reference Sequence: NM_001031701.3) was harvested using NCBI website and MethPrimer (http://www.urogene.org/cgi-bin/methprimer/methprimer.cgi) was utilized for CpG island prediction (CpG island 1: 47–295 bp) and primer design. When DNA is subjected to bisulfite conversion, the bisulfite-sensitive unmodified cytosines (C) are converted to uracils (U) and further replaced by thymidines (T) in PCR amplification, while the methylated cytosines (5mC) could survive the bisulfite conversion and remain unchanged. From the primer candidates provided by MethPrimer, one pair with a high GC content and an annealing temperature close to 60 °C was chosen as the general primers for assessing the methylation status of the predicted CpG island 1 of NT5DC3 (Additional file 5: Table S2).

Protocols of DNA methylation site (5mC CpG island 1) detection: Genomic DNA (gDNA) was extracted using a DNeasy Blood & Tissue Kit (QIAGEN). 1 μg gDNA of each sample, respectively, was subjected to bisulfite conversion using a DNA bisulfite conversion kit (QIAGEN) following the manufacturers' instructions. A thermocycler was used for the conversion with the following procedure: 95 °C for 10 min and 64 °C for 60 min. After bisulfite conversion, gDNA was purified and used for PCR analysis. The specific primers used were listed in Additional file 5: Table S2. The methylation rate was calculated using the ΔΔCt method: methylation rate (%) = 100%/2ΔΔCt.

RNA methylation site of NT5DC3 (m6A, 2309) determination and primer design principles: DNA sequence of NT5DC3 (GenBank Reference Sequence: NM_001031701.3) was firstly harvested using the NCBI website, and three kinds of sequence that could undergo m6A methylation, namely GGACU(T), GAACU(T) and GAACA, were screened. Furthermore, the m6AVar database (http://m6avar.renlab.org/index.html) was utilized for the m6A site prediction, and chr12:101471023( +), namely m6A 2309, was finally chosen for the following investigation (labeled as site X). The nearest adenine (A) on the 5’ upstream of and at least six bases away from the site X was labeled as site N. The RNA methylation-specific primers for both site X and N were designed, respectively. Site X was regarded methylated if the Ct value in site X PCR detection was larger than that in site N detection.

Protocols of RNA methylation site (m6A 2309) detection

NT5DC3-2309 methylation in RNA level was detected mainly through three steps by SELECT qPCR, which were conducted by following Xiao’s protocol and previous references [42, 44, 45]. Firstly, the total RNA (1500 ng) was mixed with 40 nM up primers (Additional file 5: Table S2), 40 nM down primers (Additional file 5: Table S2), and 5 μM dNTP (NEB) in 1.7 μL 10 × CutSmart buffer (NEB), 20 μL total volume. Then the RNA and primers were incubated as the reference [46] introduced, then 20 μL qPCR reaction system was set up and contained 5 μL of the final reaction mixture, 200 nM SELECT primers (see NT5DC3 select in Additional file 5: Table S2), and TB Green® Fast qPCR Mix (TaKaRa). SELECT qPCR was performed with the following program: 95 °C for 5 min; 95 °C for 10 s, 60 °C for 35 s, 40 cycles in total; 95 °C for 15 s; 60 °C for 1 min; 95 °C for 15 s; then hold at 4 °C. Ct values of samples were normalized to their corresponding Ct values of GAPDH.

SAM, SAH detection by MS

SAM and SAH were quantified by LC–MS/MS as described previously [27, 47], with minor modifications to run on UPLC coupled to a XEVO TQ-S micro mass spectrometer. Cell samples were washed three times with ice PBS, followed by bead-beating in 80% methanol: water (LC–MS grade methanol, Thermo Fisher Scientific) at − 20 °C. The extraction mixture was verting for 10 s, and then ultrasonication for 30 min, and centrifuged for 15 min at 12,000 g·min−1. The supernatants were transferred to an autosampler vial, and 5 µL of the mixture was then used for UPLC-MS/MS analysis (Waters). The separation was performed on a BEH Amide column (130 Å, 1.7 µm, 1 mm × 100 mm, 1/pkg, Waters, USA) for nucleosides. Mobile phases consisted of: (A) 100% water, containing 0.1% formic acid, and (B) 100% acetonitrile containing 0.1% formic acid. The following gradient and mass spectrometer operation were applied as reference described [48, 49]. The precursor → product transitions for SAM (m/z 399.3 → 250.3) and SAH (m/z 385.3 → 136.3) were monitored.

Western Blotting analysis

HT29 or NCM460 cells were lysed by RIPA buffer, and the total protein concentration was measured using a BCA kit (Beyotime). The antibodies included: anti-human NT5DC3 (PA5-70919, Invitrogen), anti-human HKDC1(PA5-35894, Invitrogen), and anti-human β-actin (PAB0865, Enzo Life Sciences). For western blotting, the primary antibodies were diluted at 1:1000; the second antibodies were used at 1:3000 dilution. The signals were captured and analyzed by Clinx ChemiCapture software (Clinx).

Clinical samples collection

From July 2020 to October 2020, 30 patients with T2D and 15 T2D/colon cancer coexsitence patients (colon cancer was diagnosed two years after the T2D diagnosis) were enrolled in the present study. Patients' characteristics are outlined in Additional file 6: Table S3. All patients submitted the informed written consent to utilize biological specimens for investigational procedures, according to the Ethics Committee approval of Beijing Friendship Hospital, Capital Medical University (Permission No. 2020-P2-175–02, as Supplementary Ethics Committee approval in Data and Code Availability section). Blood samples of the patients were taken upon admission on the morning, and the patients did not take any food or water 12 h strictly before the blood was sampled. As the normal control group, 30 healthy people were selected and enrolled in the study. Their blood samples were collected upon admission in the morning. The blood samples of each patient were divided into two parts, 4 mL was collected into RNAase-free tubes (BD) containing preservation solution for DNA/RNA isolation, 2 mL was collected into aseptic anticoagulant tubes (BD) for protein detection.

400 μL lysis buffer was added into each 200 μL blood sample and vortexed for 30 s, which was centrifuged (10,000 rpm, 1 min) later, and the supernatant was removed. The nuclear precipitation was treated with several types of buffer and the total genome DNA was extracted, and the DNA level of NT5DC3 was measured by qPCR, as mentioned above. DNA methylation (CpG island 1) and RNA methylation (2309 site) of NT5DC3 were measured by SELECT-qPCR, according to the protocol described above.

100 μL RIPA lysis buffer (with protease inhibitors) was added into each 10 μL blood sample and vortexed for 5 min, then 30 μL 5 × loading buffer was added and boiled for 15 min, and then the protein level of NT5DC3 was detected by western blotting.

Statistical analysis

All data were presented as means ± standard deviation (SD) and analyzed using SPSS 19.0 and GraphPad Prism 6.0 software (GraphPad Inc., San Diego, CA). Statistical analyses were conducted between two groups (control group vs. treatment groups, single treatment group vs. two treatments group) using a Student’s t-test. P values < 0.05 were considered to be statistically significant.