The PC BxPc-3 and Panc-1 cell lines were obtained from the Type Culture Collection of the Chinese Academy of Sciences and cultured in DMEM (Gibco; Thermo Fisher Scientific, Inc.) with 10% fetal bovine serum (FBS; HyClone; Cytiva) and 1% penicillin-streptomycin (Gibco; Thermo Fisher Scientific, Inc.) in a standard incubator with a 5% CO₂ atmosphere at 37°C. To induce normal or high-glucose environments, 0, 5.5 or 25 mM glucose was added to DMEM (mannitol was used to balance the osmotic pressure). After grouping and intervention, relevant assays were performed on the 0-mM glucose group, the euglycemia group (5.5 mM) and the hyperglycemia group (25 mM), and then the total protein was extracted by RIPA buffer (Beyotime Institue of Biotechnology) from the cells as required.

All nude mice and newborn rats were reared in a specific pathogen-free chamber, with the temperature controlled at 24°C and a 12/12 h light/dark cycle, at the Animal Center of Xi'an Jiaotong University (Xi'an, China). The relative humidity in the chamber was 50–80%. An ultraviolet lamp installed in the chamber was used to keep sterile conditions. Food, water and bedding used by the nude mice was provided ad libitum and newborn rats were all sterilized. A total of 18 male nude mice (6 mice per group) and 6 male newborn rats (1 day old, weighing ~1.5 g) were used to perform this study. The average weight of the nude mice at the start of the experiment was ~20 g (age, 6 weeks). The experimental protocols were approved by the Ethical Committee of the First Affiliated Hospital of Xi'an Jiaotong University. This study was conducted in accordance with the Declaration of Helsinki, and all protocols were approved by the Ethical Committee of the First Affiliated Hospital of Xi'an Jiaotong University.

The PNI co-culture model can mimic the peripheral nerve microenvironment of PC tumor tissues (7). Briefly, the construction of this model is separated into three parts: PC cell inoculation, dorsal root ganglion (DRG) extraction and DRG implantation.

The PC cells were pretreated with the aforementioned different glucose levels according to the experimental requirements. Next, 40 µl PC cells (1×10⁵ cells/well) were inoculated into 24-well culture plates to form a round island of cells (no extra medium was added to maintain the morphological integrity of the cell islands). The 24-well culture plates were incubated in a standard incubator with a 5% CO₂ atmosphere at 37°C for ~8 h.

Newborn rats (1–3 days after birth; purchased from the Experimental Animal Center of Xi'an Jiaotong University) were euthanized with CO₂, with a chamber volume displacement rate of 10–30% per minute (experiments performed in 2019). The rats were then sterilized with 75% ethanol (cervical dislocation was not performed to guarantee the spinal nerve structure). Next, dissection was performed and the entire spine, which contained parts of both sides of the ribs, was obtained and immediately placed into 4°C PBS. After placing the spine in cold PBS on ice, the dorsal nerve roots between the ribs and the spine were exposed to access the DRGs. Finally, the residual nerve roots and fibrous tissue were carefully trimmed from around the DRGs.

After 8 h of incubation, the curved boundary of the round cell islands in the 24-well culture plates were clearly delimited. Next, one DRG was implanted behind the cell island, keeping the distance between the edges of the two structures ~2 mm apart. A drop of growth factor-depleted liquid Matrigel (BD Biosciences) was applied to fix and nourish the DRGs. Subsequently, the 24-well culture plates were placed into a standard incubator with a 5% CO₂ atmosphere at 37°C for 30–40 min to induce Matrigel solidification. After solidification, culture media was added carefully to each well, which was replenished every 2 days.

To quantitate the severity of the PNI in this co-culture model, the initial distance (~2 mm) between the edge of the cell island and the DRG was defined as parameter γ, the relative migration distance of the PC cells toward the DRG was defined as parameter α and the relative outgrowth of the DRGs axons toward the PC cells was defined as parameter β. The PC cell invasion index=α/γ, and the DRG outgrowth index=β/γ (Fig. 2A). These parameters were measured by the image processing/analysis software ImageJ (version 1.52a; National Institutes of Health).

As described previously (19,20), the sciatic nerve infiltrating tumor model was adopted to research the influence of hyperglycemia on PNI of PC in vivo. Briefly, the construction of this model is separated into three parts, the induction of hyperglycemia in mice, the transplanting of PC cells into the mouse sciatic nerve and the relative intervention strategy.

Female 4-week-old BALB/c nude mice were purchased from and reared in the Experimental Animal Center of Xi'an Jiaotong University. The mice were separated into three groups (6 mice/group); two groups (control group and 2-ME group) received an intraperitoneal injection of streptozotocin (STZ; MedChem Express; dissolved in sodium citrate buffer, 150 mg/kg) to induce hyperglycemia, while the third group received an intraperitoneal injection of sodium citrate buffer alone as the euglycemic control. After 7 days, the blood glucose levels of each mouse were measured to confirm establishment of the model.

First, the mice were anesthetized with 1% pentobarbital sodium (50 mg/kg, intraperitoneal injection) and the sciatic nerve of the left lower extremity was exposed. Next, 5 µl Panc-1 cell suspension (1×10⁵ cells/well) was slowly (over ~2 min) injected into the perineurium of the sciatic nerve using a 10-µl micro-syringe. Finally, the wound was sutured and the general status of the mice was observed for 2 weeks.

Three strategies were performed in this study to form the following groups: The euglycemia group (saline, 0.1 ml/mice, intraperitoneal injection, daily), the hyperglycemia group (saline, 0.1 ml/mice, intraperitoneal injection, daily) and the hyperglycemia plus 2-mercaptoethanol (2-ME) group (MedChem Express; inhibitor of HIF1α; 50 mg/kg, intraperitoneal injection, daily for 4 weeks).

Considering that the sciatic nerve innervates the hind limb paw muscles, a sciatic nerve score [the nerve function was graded from 4 (normal paw function) to 1 (total paw paralysis) based on the hind limb paw response to manual extension of the body] and the distance of the paw span (the length between the first and fifth toes of the hind limbs) were used as the evaluation indices of sciatic nerve function. Finally, the nude mice were euthanized with CO₂, with a chamber volume displacement rate of 10–30% per minute. Consequently, the generated tumors were extracted to calculate their volume.

HIF1α-specific siRNA (si-HIF1α; sense, 5′-CCACCACUGAUGAAUUAAATT−3′; anti-sense, 5′-UUUAAUUCAUCAGUGGUGGTT−3′) and a negative control scrambled siRNA (si-NC; sense, 5′-UUCUCCGAACGUGUCACGUTT−3′; anti-sense, 5′-ACGUGACACGUUCGGAGAATT−3′) were purchased from Shanghai GenePharma Co., Ltd. Glucose pretreated cells or the non-treated cells seeded in 6-well plates were transfected with 100 nM siRNA (50 pmol/well) using the Lipofectamine^® RNAi MAX Reagent (Invitrogen; Thermo Fisher Scientific, Inc.) for 48 h at 25°C according to the manufacturer's instructions. The cells were used in subsequent experiments 24 h after the transfection.

Transwell Matrigel invasion assays were adopted to measure the invasive ability of the PC cells. First, 8-mm pore Millicell inserts were coated with 25 µl Matrigel at 37°C for 8 h. PC cells (BxPc-3 and Panc-1) were pretreated appropriately according to the design of the study, and cell suspensions (5×10⁴/well) were added to the upper chambers in DMEM containing 1% FBS. Next, 500 µl DMEM containing 20% FBS was placed in the lower chambers. After being incubated for 48 h, the non-invading cells were removed by cotton swab and the invading cells were stained with crystal violet at room temperature for 15 min. The invasion ability was determined by counting the stained cells on the bottom surface of the insert under a light microscope.

Wound-healing assays were adopted to measure the migration ability of the PC cells. PC cells (BxPc-3 and Panc-1) were pretreated appropriately according to the design of the study. The PC cells were serum-starved during the assay. A sterile pipette tip was used to produce a wound line across the plate after the cells had been grown to 90–100% confluence. The cells were allowed to migrate for 48 h. Images were captured at 0 and 48 h post-wounding using a light microscope and the relative migration area was measured by image processing/analysis software ImageJ (version 1.52a; National Institutes of Health).

PC cells (BxPc-3 and Panc-1) were pretreated appropriately according to the design of the experiment and total protein was extracted using RIPA lysis buffer (Beyotime Institute of Biotechnology). Next, western blotting assays were performed as previously described (21). Primary antibodies against HIF1α (1:1,000; catalog no. ab51608; Abcam), NGF (1:1,000; catalog no. ab52918; Abcam) and β-actin (1:5,000; catalog no. 20536-1-AP; ProteinTech Group, Inc.) were incubated at 4°C for 12 h. Next, the goat anti-rabbit HRP-conjugated secondary antibody (catalog no. A0208; Beyotime Insititute of Biotechnology) was diluted at 1:10,000 and incubated for 1 h at room temperature. Subsequently, ECL luminescence reagent (Beyotime Institute of Biotechnology) was applied for ~5 min. Images of the immunopositive bands were captured with a ChemiDoc XRS imaging system (Bio-Rad Laboratories, Inc.) and Image Lab (version 3.0 Beta 3; Bio-Rad Laboratories, Inc.) was used for densitometry analysis. Grayscale analysis was performed using ImageJ (version 1.52a; National Institutes of Health).

Two online bioinformatics analysis websites, GEPIA (http://gepia.cancer-pku.cn/) and Kaplan-Meier Plotter (http://www.kmplot.com) were used to perform the bioinformatics analysis in this passage. Firstly, the Kaplan-Meier Plotter website was adopted to analysis the relapse-free survival and overall survival rates with regard to HIF1α and pancreatic cancer. A PDAC dataset (n=177) in the pan-cancer database (mRNA RNA-Seq sub-database) was divided based on the median value. Next, on the GEPIA website, the PAAD dataset was adopted to analyze the expression level of HIF1α in pancreatic cancer (179 cases) and normal pancreas (171 cases). |Log2FC| and P<0.01 were used as the cut-offs. The correlation between HIF1α and NGF/BDNF in pancreatic cancer was performed via automatic analysis on the GEPIA website.

The data are presented as the mean ± SD. Using the GraphPad Prism software package (version 6.0; Graphpad Software, Inc.), the differences in the in vivo data were assessed with Student's t-test (unpaired) when there were 2 groups. Dunnett's post hoc test in conjunction with one-way ANOVA was performed to compare the differences between the control group and other groups when there were 3 groups. Pearson's correlation test was applied for bioinformatics analysis. All tests were two-sided, and a P<0.05 was considered to indicate a statistically significant difference.

The necessary condition for PNI is a high invasion/migration ability of PC cells. Hence, before the research on the PNI of PC, the influence of hyperglycemia on the invasion/migration ability of the PC cells was first evaluated. Transwell assays showed that 5.5 mM glucose (euglycemia) had little effect on the invasion ability of PC cells compared to the normal media group; however, 25 mM glucose (hyperglycemia) promoted PC cell invasion (Fig. 1A and B). Wound-healing assays also revealed that both euglycemia (5.5 mM glucose) and hyperglycemia (25 mM glucose) accelerated the migration ability of the PC cells compared with the normal media group (Fig. 1C and D). These phenomena revealed that hyperglycemia promoted the invasion/migration of PC cells, which indicated that hyperglycemia may participate in the PNI of PC.

Next, a PC cell-DRG direct co-culture system was used to explore the association between hyperglycemia and PNI. The main focus was placed on the influence of hyperglycemia on PC, so different concentrations of glucose (0, 5.5 and 25 mM) were first used to culture the DRGs alone and the outgrowth of the DRG axons was observed. The results showed that short-term intervention (<3 days) exposure to glucose had no marked influence on the outgrowth of the DRGs (Fig. 1E); however, as the time was extended (>3 days; data not shown), hyperglycemia began to suppress the growth of the axons, which may be due to its destruction of the Schwann cells (16). All of these findings suggested that hyperglycemia promoted the invasion and migration of PC cells and had little effect on the outgrowth of DRGs; thus, the PC cell-DRG direct co-culture system can be adopted for in vitro PNI research.

As shown in Fig. 2A, to evaluate the severity of PNI in the PC cell-DRG co-culture model, the initial distance (~2 mm) between the edge of the cell island and the DRG was defined as parameter γ, the relative migration distance of the PC cells toward the DRG as parameter α (PC invasion index=α/γ) and the relative outgrowth of the DRG axons toward the PC cells as parameter β (DRG outgrowth index=β/γ). Considering that NGF is an essential member of the neurotrophic family that sustains the PNI of PC, the influence of hyperglycemia on the expression of NGF in PC was measured and hyperglycemia was found to result in increased NGF expression in both BxPc-3 and Panc-1 cells (Figs. 2B and S1A). By applying the PC cell-DRG co-culture model, it was also found that hyperglycemia significantly accelerated the invasion of the PC cells toward the DRG axons (Fig. 2C and E); in addition, hyperglycemia also promoted the outgrowth of the DRG axons (Fig. 2D and E). The promotion effect of hyperglycemia on the DRGs was much weaker than that on the PC cells, and this discrepancy may be due to the fact that hyperglycemia not only promotes the secretion of NGF by PC to stimulate the growth of axons, but it also has a certain toxic effect on nerve axons.

A previous study demonstrated that HIF1α mediates PC cell metastasis in hyperglycemia-induced hypoxic conditions (22). Hence, the present study aimed to explore whether HIF1α was participating in PC cell invasion and migration. First, PC cells were cultured with different concentrations of glucose (0, 5.5 and 25 mM) and it was found that hyperglycemia upregulated HIF1α (Figs. 3A and S1B), which is consistent with the previous findings shown in Fig. 2B. Considering that HIF1α is unstable under normal conditions and is stabilized under hypoxia (22), we hypothesize that hyperglycemia aggravated the hypoxic microenvironment of PC.

Next, using two online bioinformatics analysis websites, GEPIA (http://gepia.cancer-pku.cn/) and Kaplan-Meier Plotter (http://www.kmplot.com), HIF1α expression was found to be high in PC and associated with a poor prognosis (Fig. 3B-D). In addition, HIF1α was positively correlated with the expression of neurotrophic factor NGF (Fig. 3E). To measure the influence of HIF1α on PC cell invasion and migration, 2-ME (10 µM) was adopted as an inhibitor of HIF1α (23). The data revealed that hyperglycemia had a similar function with regard to hypoxia as to the expression of NGF and HIF1α, and HIF1α inhibition suppressed the invasion and migration of the PC cells (Figs. 3F-K and S1C). In conclusion, it was demonstrated that hyperglycemia upregulated HIF1α and that HIF1α promoted the invasion and migration of the PC cells.

As aforementioned, HIF1α was demonstrated to promote the invasion and migration of the PC cells. Next, the PC cell-DRG co-culture model was adopted to explore whether HIF1α was essential for the PNI of PC. First, si-HIF1α was used to inhibit the expression of HIF1α, and consistent with previous findings, it was found that the expression of NGF was also decreased in both hypoxic and normoxic conditions (Figs. 4A and S1D). Next, si-HIF1α- or si-NC-treated PC cells were cultured with DRGs, and after 48 h of co-culture, both PC cell invasion and DRG outgrowth were suppressed after HIF1α downregulation (Fig. 4B-D). Combined with the aforementioned results, it can be concluded that HIF1α is essential for PC cell invasion/migration and PNI. However, the present study also found that hyperglycemia promoted the invasion/migration/PNI of PC cells and that hyperglycemia can upregulate HIF1α. Hence, it was next demonstrated that the promoting effect of hyperglycemia on invasion/migration was HIF1α-dependent. Specifically, si-HIF1α or si-NC was used to influence the expression of HIF1α in the PC cells, which were then treated with 25 mM glucose. As shown in Fig. 4E-G, hyperglycemia promoted the invasion/migration of the PC cells, but inhibiting HIF1α abrogated this effect. In conclusion, hyperglycemia facilitated the invasion and migration of PC cells via HIF1α.

The aforementioned results revealed that HIF1α plays a vital role in the hyperglycemia-induced invasion and migration of PC cells. Therefore, the present study next aimed to analyze the role of HIF1α in the PNI of PC in a background of hyperglycemia. si-HIF1α or si-NC were used to influence the expression of HIF1α in the PC cells, which were then cultured with DRGs in normal or high-glucose media. After co-culture for different time points, the PC cell invasion and DRG outgrowth of these groups was measured, and hyperglycemia was found to promote the PNI of PC; however, this effect was abrogated by the inhibition of HIF1α (Fig. 5A-C). In western blot assays, it was found that hyperglycemia upregulated the levels of both HIF1α and NGF; however, consist with the above findings, after inhibition of HIF1α, the expression of NGF was downregulated in both hypoxic and normoxic conditions (Figs. 5D and S1E). Hence, taken together, all of these findings indicated that hyperglycemia promoted the PNI of PC in a HIF1α-dependent manner (Fig. 5E).

The present study demonstrated that hyperglycemia promoted the PNI of PC in a HIF1α-dependent manner in vitro; however, whether this phenomenon occurred in actual organisms was still unclear. Hence, by using STZ injections and sciatic nerve grafts, an in vivo PNI sciatic nerve infiltrating tumor model (Fig. 6A) was established to explore the influence of hyperglycemia on the PNI of PC.

First, the blood glucose levels of the mice were detected on the 7th day after injecting 150 mg/kg STZ or sodium citrate buffer, and STZ treatment was found to effectively induce hyperglycemia in the mice (Fig. 6B). Next, the sciatic nerve function in the three groups of mice was monitored and evaluated using a sciatic nerve score and the distance of the paw span, and it was found that hyperglycemia significantly affected the function of the sciatic nerve; however, HIF1α inhibition improved this phenomenon (Fig. 6C and D).

After the sciatic nerve and the surrounding tumor tissue were dissected out, hyperglycemia was found to promote the infiltration of the PC cells and tumor growth, while HIF1α inhibition partially abrogated this effect (Fig. 6E and F). In conclusion, the present study demonstrated that hyperglycemia promoted the PNI of PC in a HIF1α-dependent manner in vivo.