Nerve Growth Factor (NGF) Encourages the Neuroinvasive Potential of Pancreatic Cancer Cells by Activating the Warburg Effect and Promoting Tumor Derived Exosomal miRNA-21 Expression

Background It has been reported that signaling from the nerve growth factor (NGF) pathway associated with peripheral nerves is able to contribute to perineural invasion (PNI) of pancreatic cancer (PC). Nevertheless, the underlying mechanism by which NGF leads to PNI remained poorly understood. Methods Western blotting was employed to determine NGF level in PC and paracarcinoma tissues and in PC cell lines as well as pancreatic ductal epithelial cells. MiaPaCa-2 and CFPAC-1 cells were treated with 100 ng/ml of NGF or the NGF inhibitor Tanezumab for 24 h, CCK-8 and Transwell assays were employed to test cell proliferation, invasion, and migration, respectively. TrkA expression was knocked down in MiaPaCa-2 and dorsal root ganglion (DRG) cells treated with NGF to determine its effect on the Warburg effect. To reveal that the NGF-TrkA signaling pathway was closely associated with PC PNI, in vitro neuroinvasion model was established by using MiaPaCa-2 cells via coculturing DRG cells in Matrigel. Further, exosomes were extracted from PC cells and identified by examining the levels of specific markers for exosomes. Then RT-qPCR was applied to test miR-21-5p level in tumor derived exosomal (TDE-miR-21-5p). RIP assay was performed to validate NGF and miR-21 binding ability in MiaPaCa-2 cells. Rescue experiments were performed by using coprocessing of Tanezumab and miR-21-5p mimic on MiaPaCa-2 cells, followed by coculture with DRG cells. Subsequently, we used a model of neuroinvasion in nude mice to assess the effect of NGF in vivo on tumor nerve invasion as well as on nociceptive transmission. Results NGF level was preeminently higher in PC tissues and cell lines than in paracarcinoma tissues and normal pancreatic epithelial cell lines. NGF promoted MiaPaCa-2 and CFPAC-1 cell invasion and migration, while Tanezumab treatment showed the opposite results. Besides, NGF binding to TrkA receptors encouraged the intracellular Warburg effect in PC and DRG cells. TrkA blocking-up could restrain NGF induced PC cell migration and neural invasion. Mechanistically, NGF could upregulate TDE-miR-21-5p levels, and DRG cells took up TDE to activate the Warburg effect and stimulate nociceptor gene expression. miR-21-5p inhibitor could abolish the facilitative effect of NGF on PNI in MiaPaCa-2 cells. In vivo tumorigenesis experiments, Tanezumab markedly alleviated nerve invasion of PC cells as well as relieved nociceptive conduction in animal models. Conclusions These findings displayed that NGF/TrkA encouraged the neuroinvasive potential of PC cells by activating the Warburg effect in DRG cells through upregulation of TDE-miR-21-5p expression.


Introduction
Pancreatic cancer (PC) is a relatively common malignancy of the digestive system and one of the human malignancies with the worst prognosis [1]. Approximately 213,000 people worldwide die from PC each year [2]. Its pathogenesis remains unknown, and it is currently only well established that family history, smoking, diabetes mellitus, obesity as well as alcohol abuse are high risk factors for PC [3]. PC has an insidious onset and often has no obvious clinical manifestation in the early stage. However, the disease progresses more rapidly, and until the clinical symptoms appear or the diagnosis is confirmed, most cases lose the opportunity of surgical treatment because of local invasion or hematogenous, lymphatic, and distant metastasis. Surgery is accessible to only less than 20% of patients, with an additional 70% of cases being amenable to only palliative surgery [4]. Even radical surgical cases that achieve clinicopathological cure criteria, curative effects remain unsatisfactory.
Infiltration versus metastasis is one of the most characteristic biological properties of tumors [5]. Generally, it is believed that there are four main pathways between tumor invasion and metastasis, including local invasion, blood, lymphatic, and seeding metastasis [6]. In addition to these four pathways, tumors can undergo infiltration around nerve fibers and metastasis along nerves, that is, perineural invasion (PNI), which refers to the phenomenon of perineural invasion by tumor cells filling the perineurial space, wrapping around nerves in a continuous concentric sheath-like pattern, infiltration, and metastasis of extension along nerves around nerve fibers or into perineurium within the perineurium [7]. Although this more specific biological behavior is seen in biliary tract tumors, PC, parotid gland tumors, prostate cancer, and breast cancer, as well as basal and squamous cell carcinomas of the head and neck, perineural invasion is more common in PC. The fact that pancreatic tissue itself is rich in neural tissue determines the predisposition of PC to perineural invasion, which also constitutes the anatomical basis for the predisposition of pancreatic cancer to perineural tissue invasion. Perineural invasion, which starts within the pancreas, evolves to perineural invasion of the pancreas and subsequent metastasis to distant sites such as retroperitoneum, which is also thought to be an important cause of surgical margin remnant and persistent extra-pancreatic dissemination [8]. In the PNI of PC, a variety of genes or proteins are involved in the process of promoting tumor cell growth and inhibiting their apoptosis. Their interaction creates a favorable microenvironment for the invasiveness and neurotropic growth of tumor cells. Special growth factors produced by nerves or the cells associated with them are important reasons for promoting the growth and survival of cancer cells with a proneural nature [9].
Nerve growth factor (NGF) plays a very important role in this behavior. NGF, an important neuropeptide in the neurotrophin family with a molecular weight of about 140 KD, is composed of three peptide chains, α, β, and γ, bound noncovalently at the ratio of α2βγ2. Subunit β, which consists of an identical skin chain containing 188 amino acid residues each, is the active zone of NGF and the only of the three subunits to have the biological activities of NGF. Under normal circumstances, NGF exerts regulatory effects on neuronal survival, growth and development, differentiation, regeneration, and functional maintenance. Numerous studies have shown that high level of NGF in PC are closely associated with tumor cell migration and perineural invasion. For example, NGF activated the ERK1/2 signaling pathway in PC by regulating CD133 to accelerate cell migration and invasion [10]. Further, tyrosine kinase receptor A (TrkA) is the functional receptor for NGF and when bound to NGF can activate tyrosine kinase signaling systems that initiate cellular activity to produce biological effects. NGF and TrkA are its high affinity receptors that play a crucial role in cancer pathogenesis through cell proliferation, angiogenesis, invasion, and migration. Vera et al. [11] demonstrated that NGF/TrkA could promote the malignant behavior of ovarian cancer cells by upregulating oncogenic proteins such as vascular endothelial growth factor and cmyc expression during ovarian cancer progression. Further, NGF acts by binding to receptors on the surface of target cells, which are divided into a high affinity TrkA and a low affinity receptor p75. NGF/TrkA facilitated PC cell proliferation and invasion in coculture system by activating the PI3K/Akt/GSK signaling cascade [12]. However, the specific mechanism by which the NGF/TrkA pathway leads to PNI in PC remains to be further explored.
Tumor cells consume far more glucose than normal cells. More importantly, even under aerobic conditions glucose in tumor cells is not oxidized exhaustively but is broken down to produce lactate, a phenomenon that was discovered by the German biochemist O.H. Warburg and is therefore called the Warburg effect [13]. Although the capacity rate of glycolysis is lower than that of oxidative phosphorylation, glycolysis can meet the energy demand of rapid tumor cell growth [14]. Moreover, high levels of glycolysis within tumor cells would consume a large amount of glucose in the tumor microenvironment and lead to the accumulation of lactate, which in turn inhibits lymphocytes and tumor infiltration [15]. An increasing number of studies have shown that an excessive level of glycolysis is closely associated with PC progression and invasion [16]. The molecular mechanism by which cancer cells achieve metabolic reorganization is still unclear. Therefore, elucidating the molecular mechanism of metabolic reorganization of tumor cells will provide a new and effective way for tumor targeted therapy.
In this study, an in vitro model generated by coculturing mouse dorsal root ganglion (DRG) neuronal cells and PC cells, and an in vivo PNI model were applied to investigate the function and underlying mechanisms of NGF/TrkA signaling in the progression and pathogenesis of PNI, contributing to the development of therapeutic regimens for PC.

Exosome Isolation and Identification.
Exosomes were isolated by precipitation using exoquick reagent (SBI) according to the manufacturer's instructions. Briefly, conditioned medium was incubated with exoquick reagent (5 : 1) for 12 h or longer, followed by centrifugation at 1500 g for 30 min, and the precipitated exosomes were suspended in 100 mL of PBS, then stored at -80°C. In addition, total exosome isolation (Magen) was performed using serum exosomes. Reagents were added to serum samples and incubated at 4°C for 30 min, followed by centrifugation of 1 ml of serum samples at 2000 g for 30 min to remove cells and debris. Alternatively, exudates observed by transmission electron microscopy were suspended in glutaraldehyde, dripped onto carbon-containing copper mesh and stained with 2% uranyl acetate, then dried and imaged.
2.8. Co-Immunoprecipitation (co-IP). MiaPaCa-2 and DRG cells were lysed with RIPA buffer (Thermo Fisher, USA), and the supernatant was collected and incubated with NGF and TrkA antibodies overnight at 4°C. This mixture was incubated with 100 microliters of A/G agarose beads overnight at 4°C. Then, the agarose sphere-antigen-antibody complexes were collected by transient centrifugation and washed 3 times with PBS. Next, the complex with proteinadded buffer for 5 minutes. Intracellular interacting proteins were analyzed by immunoblotting technique.
2.9. RNA Immunoprecipitation (RIP) Assay. Protease inhibitor EDTA as well as RNase inhibitor (Qiagen, Hilden, Germany) were added to IP lysis buffer to lyse cells. After the addition of magnetic beads preclearing for 30 min, IgG antibody and NGF antibody (Abcam, USA) were added, respectively, and then the added magnetic beads were rotated to mix for 2 h at room temperature. Aspirate supernatant on magnetic stand and wash beads with IP lysis buffer. After removal of proteins by addition of proteinase K, Trizol LS was added to extract RNA and finally subjected to subsequent analysis.
2.10. Assessment of Neural Cancer Cell Interactions. PNI model was used to simulate the tumor and surrounding neural microenvironment. DRGs were placed on ice in DMEM in 24 wells containing 25 μL of Matrigel at a distance of approximately 1 mm from tumor colonies suspended in 25 μL of matrix. To rule out nonspecific metastasis of tumor cells, another 25 microliters of DRG-free male basement membrane was placed on its back. The incubator was placed in an incubator at 37°C and saturated with 5% CO 2 for 20 minutes in a humid environment to perform Matrigel polymerization. After solidification, the DMEM medium was added every 2 days. The cell suspensions on adjacent sides were photographed by using an inverted microscope technique. To quantitatively study the growth of tumor cells, we determined the tumor boundary and tumor boundary as the γ parameter, the tumor cell metastasis distance as the α parameter, and the direction of the DRG as the tumor growth length. Invasion index was α/γ, and DRG growth index was β/γ. The moving distance is performed by the image analysis software of the microscope imaging system (NIS-Elements, Nikon Instruments Inc., China).

Cell Migration and Invasion
Assay. The ability of cell invasion and migration was detected by treatment with 8.0-μm chamber plates. The upper surface of the Transwell filter we used was coated with or without Matrigel (BD, New Jersey, USA). Firstly, cells were planted into the 8.0 μm chamber plates, then 300 μL of serum-free DMEM medium was added to the upper compartment of the chamber, and then 500 μL of DMEM medium supplemented with 10% FBS was added to the lower chamber for 48 h incubations. Then, the noninvasive cells on the upper side of the chamber were suspended with a cotton swab, and then the invasive cells were fixed in 4% paraformaldehyde and stained with a crystal violet solution. We stained infiltrating cells by using an Olympus IX70 inverted microscope (Olym-pus Corp, Tokyo, Japan) and randomly selected the best six fields of view, and each experiment was repeated three times.
2.12. Cell Proliferation Assay. The treated cells from each group were seeded in 96 well culture plates at a cell number per well of 3 × 103, and 5 replicate wells were set in each group. Cell culture supernatants were removed at 0 h, 24 h, 48 h, and 72 h after seeding, and 10 μl of CCK-8 solution was added to each well according to the CCK-8 cell proliferation activity assay kit instructions (Sigma-Aldrich, USA). After incubation at 37°C in 5% CO2 for 1 h, the OD value at 450 nm was measured by using a microplate reader (Bio-Rad, USA), and the cell growth curve was plotted.
2.14. In Vivo PNI Model. Animal experiments were approved and supervised by the Animal Ethics Committee of Huazhong University of Science and Technology. 5 week old male athymic BALB/C nude mice were obtained from Experimental animal center of Huazhong University of Science and Technology, and subsequently randomly divided into 2 groups, including PNI + PBS and PNI + Tanezumab. MiaPaCa-2 cell suspension was injected subcutaneously into the midline of the back of the mice. Tanezumab was then administered at 10 mg/kg/kg/d, and tumors were excised within 8 weeks and histologically examined to determine neural PNI. In order to evaluate the role of NGF in tumors that invade along nerves, the nerve invasion model is basically performed in this way. Under the microscope, MiaPaCa-2 cells were injected into the sciatic nerve of paralyzed rats separated from the tibial and peroneal nerve bifurcations. 3 ml of normal saline was injected into the right sciatic nerve as a sham-operated control group. 3 microliters of cytoplasm were microinjected with a 5 microliter microsyringe at a concentration of 1 x 10 5 cells per microliter. In order to evaluate the function of the sciatic nerve innervating the paw muscles of the hindlimbs, we measured the following parameters: rough movements, 10 minutes per week, and repeated bite marks on the hindlimbs; limb function, according to the degree of response of the hindlimb paws to manual stretching, divided into 4-1 (total paw paralysis), sciatic nerve function index, the unit of elongation (mm) between the 1st and 5th toes of the hind legs of mice, 7 weeks per week.
2.15. Statistical Analysis. SPSS 22.0 and GraphPad Prism 7.0 were used for data analysis and mapping. The pairwise comparisons were analyzed using the chi-square test. The measurement data were represented as mean ± SEM with normal distribution and homogeneity of variance. Student's t-test was performed for the comparison between two groups. The means of the different groups were compared 4 Oxidative Medicine and Cellular Longevity using one-way or two-way analysis of variance (ANOVA) following Tukey's post hoc test. P < 0:05 was considered as statistically significant difference. All experiments were repeated 3 times (n = 3).

NGF Indicated the Unfavorable Prognosis of PC Patients.
To explore the potential involvement of NGF in PC, we analyzed the differential level of NGF in PC samples with the aid of a bioinformatics. Figure 1(a) was shown as differential genes on chromosomes in the PC samples, and the differential expression genes on chromosome 1, where NGF was located, were found to be most enriched. Figure 1(b) revealed that NGF was prominently higher in pancreatic cancer than in normal samples. Subsequently, we performed RT-qPCR to determine the level of NGF in 80 paired PC tissues and matched with normal tissues, and three groups of samples were randomly selected to detect the NGF protein expression level by using western blotting. As shown in Figure 1(c)-(d), the NGF level was higher in tumor tissues of PC patients than that in the corresponding normal tissues. Similarly, NGF was overexpressed in human PC cell lines compared with that of normal pancreatic epithelial cell lines, especially in MiaPaCa-2 cells (Figure 1(e)-1(f)).

NGF Promoted PC Cell Migration and Invasion In Vitro.
Next, we sought to analyze the effect of NGF on the proliferation and invasive capacity of PC cells. MiaPaCa-2 and CFPAC-1 cells were treated with 100 ng/ml of NGF or the NGF inhibitor Tanezumab for 24 h. The Western blotting results revealed that NGF level was effectively modulated in MiaPaCa-2 and CFPAC-1 cells (Figure 2

NGF Binding to TrkA Receptors Encouraged the Intracellular Warburg Effect in PC Cells and DRG Cells.
It has been reported that Warburg effect is able to participate in regulating neural function [17]. First, we detected and validated NGF and TrkA receptor binding in PC and DRG cells by using protein co-immunoprecipitation (Figure 3(a)). Subsequently, to understand the role of NGF-TrkA signaling in PC and DRG cells, we knocked down TrkA expression in MiaPaCa-2 and DRG cells treated with NGF to determine its effect on the Warburg effect. The efficiency of TrkA siRNA on its level in PC and DRG cells was confirmed by Western blotting (Figure 3(b), p < 0:01). Glucose uptake analysis, lactate production analysis, and ATP analysis discovered that NGF overexpression increased the glucose uptake, lactate production, and ATP accumulation. Further, TrkA knockdown partially reversed the facilitative effects of NGF on glucose uptake, lactate production, and ATP accumulation (Figure 3(c)-3(e), p < 0:05). ECAR analysis discov-ered that NGF promoted the glycolytic capacity, which was reversed by TrkA inhibition (Figure 3(f), p < 0:01). In short, these findings revealed that NGF could encourage the Warburg effect of PC and DRG cells.
We knocked down TrkA expression in MiaPaCa-2 and DRG cells treated with 100 ng/ml of NGF for 48 h to determine its effect on the Warburg effect. (a) Co-IP was employed to test the interaction between NGF and TrkA in MiaPaCa-2 and DRG cells. (b) Western blotting was employed to test the level of TrkA in MiaPaCa-2 DRG cells transfected with TrkA siRNA. (c-e) Glucose uptake, lactate production, and ATP assays were employed to test glucose uptake, lactate production, and ATP accumulation in MiaPaCa-2 DRG cells treated with 100 ng/ml of NGF or/ and transfected with TrkA siRNA for 48 h. (f) Extracellular acidification rate (ECAR) was employed to test and detects the glycolytic capacity of DGR cells treated with 100 ng/ml of NGF or/and transfected with TrkA siRNA for 48 h. N = 3, * P < 0:05, * * P < 0:01.
3.4. NGF/TrkA Activated the PI3K/AKT/mTOR Signaling Pathway and Enhanced the Interaction between PC Cells and DRG Cells. To investigate potential regulators of NGF/ TrkA signaling that induce cell invasion and migration, we focused our attention on downstream pathways. We used western blot assay to detect PI3K/AKT/mTOR pathwayrelated proteins. The results demonstrated that PI3K, p-AKT, and p-mTOR were upregulated in NGF overexpression PC cells, whereas transfection of TrkA siRNA reversed the results (Figure 4(a), p < 0:01). Moreover, PC and DRG cell migration and invasion numbers were increased after NGF treatment, and this result was partially counteracted by TrkA downregulated (Figure 4(b)-4(e), p < 0:01). Signaling through the NGF-TrkA pathway was involved not only in PC cell growth but also in the process of PNI in PC cells [18]. To confirm whether the NGF/TrkA axis is involved in the PNI of PC cells, a model of neuroinvasion was constructed in vitro by culturing rat DRG cells and MiaPaCa-2 cells using a Matrigel coculture system. The invasion schematic is shown in Figure 4(f). The results revealed that overexpression of NGF memorably increased the MiaPaCa-2 invasion index and DRG growth index, whereas TrkA knockdown impaired the MiaPaCa-2 invasion index and DRG growth ability. Furthermore, TrkA knockdown was also able to partially reverse the promoting effect of NGF overexpression on the MiaPaCa-2 invasion index and DRG growth index (Figure 4(g)-4(h), p < 0:01). In a word, these findings reveal that TrkA blockade attenuates NGF induced PC cell migration and neuroinvasion as well as PI3K/Akt/ mTOR pathway activation.

miR-21-5p Inhibitor Suppressed PNI in PC Cells to
Relieve Nociceptive Transmission. To further investigate whether miR-21-5p has an effect on PNI in PC cells, MiaPaCa-2 cells were transfected with miR-21-5p inhibitor cocultured with DRG cells. Figure 6(a) is shown as a  7 Oxidative Medicine and Cellular Longevity coculture system schematic. We discovered that miR-21-5p level, cell migration number, ATP accumulation, glucose absorption, lactate production, and glycolytic capacity, as well as pain state dependent protein TRPV1 expression, were observably increased in DRG cells under coculture conditions, which could be partially reversed by inhibition of    9 Oxidative Medicine and Cellular Longevity miR-21-5p ( Figure 6(b)-6(i), P < 0:05). Therefore, we hypothesized that TDE-miR-21-5p is an important regulator of PNI in PC cells.

NGF Encouraged PC Cell Neuroinvasive Potential by
Activating the Warburg Effect and Upregulating TDE-miR-21-5p Expression. To further explore whether NGF and miR-21-5p are functionally relevant in regulating PC, rescue experiments were performed by using cotreatment of Tanezumab and miR-21-5p mimic on MiaPaCa-2 cells, followed by coculture with DRG cells. Figure 7(a) is presented as the overexpression efficiency of mir-21-5p mimic. Subsequently, intracellular miR-21-5p level also decreased in DRG cells of the Tanezumab-treated MiaPaCa-2 cell group after coculture, and the miR-21-5p level was reversed after transfection of miR-21-5p mimic (Figure 7(a) , p < 0:01). Furthermore, Tanezumab treatment significantly decreased the invasion index and DRG growth index of MiaPaCa-2 cells, and the invasion number, cell migration number, ATP accumulation, glucose absorption, lactate production, and glycolytic capacity of MiaPaCa-2 and DRG cells, and TRPV1 protein level in MiaPaCa-2 cells, which were reversed by miR-21-5p upregulation (Figure 7(b)-7(m) , p < 0:05). In conclusion, inhibition of miR-21-5p expression in MiaPaCa-2 cells was able to abolish the promoting effect of NGF on PNI in PC cells.

NGF Knockout Markedly Alleviated Nerve Invasion by PC Cells as Well as Relieved Nociceptive Conduction in
Animal Models. Subsequently, we used a model of neuroinvasion in nude mice. A total of 15 nude mice were randomly divided into 3 groups: Sham, PNI + PBS, and PNI + Tanezumab. The cell suspension was injected into the middle of the back of the mouse. Eight weeks after injection, tumors were obtained histologically to observe whether the nerves had PNI. The relationship between NGF and PNI was studied by western blot analysis. Using immunohistochemical methods, nerve tissue marker (S100), and tumor marker (CK19) was evaluated in the neural invasion model (Figure 8(a)). Next, mice in the Tanezumab group developed hindlimb dysfunction after 15 days and left hindlimb paralysis by week 7, which was significantly prolonged compared with the PBS group. Seven of the ten mice in the PNI + PBS group became completely paralyzed at week 7, whereas only two mice in the Tanezumab group became paralyzed. All mice in the sham group had normal hindlimb function (Figure 8(b)). Sciatic nerves were excised for HE staining after mice were sacrificed at 8 weeks. HE staining displayed   compared to the sham group, and were back increased after treatment with Tanezumab (Figure 8(f)-8(g)). Further, the results revealed that NGF inhibition could reduce tumor volume and invasive nerve diameter. Additionally, compared with the sham group, PNI + PBS substantially promoted NGF and miR-21-5p expression in mouse tumor tissues, which were dramatically reversed by Tanezumab treatment (Figure 8(h)-8(i)). These data suggest that elimination of NGF can inhibit tumor growth and sciatic nerve invasion into the spinal cord.

Discussion
PNI is one of the most characteristic biological behaviors of pancreatic cancer and is also one of the leading causes of local recurrence and poor prognosis after curative resection [20]. At present, it is generally accepted that neuroinvasion refers to the existence of reciprocal chemotactic interactions between tumor cells and nerve cells, resulting in the infiltrative growth of nerve bundles with tumor cell extension [21]. In this experiment, Matrigel/DRG and mouse sciatic nerve infiltration models were used to investigate the effect of NGF on the formation of PNI in PC cells and preliminarily explore the paracrine loop of NGF/TrkA/miR-21-5p. Normally, NGF regulates neuronal survival, growth and development, differentiation, and regeneration and functional maintenance [22]. First, we determined NGF expression in PC. NGF level in PC tissues and cell lines was prominently upregulated than in normal tissues and normal pancreatic epithelial cell lines. Besides, NGF promoted MiaPaCa-2 and CFPAC-1 cell invasion and migration, while Tanezumab treatment inhibited cell invasion and migration. Zhu et al. revealed that the mRNA levels of NGF and TrkA in PC were increased by 2.7-and 5.6-fold compared with normal pancreatic tissue, respectively. Moreover, mRNA levels of NGF and TrkA were higher in tumors with perineural invasion than in those without perineural invasion, and perineural invasion was more likely to occur in tumors with high expression of NGF and TrkA, suggesting that NGF released by cancer cells does not act on itself through paracrine or autocrine actions, but rather acts on TrkA presented at the perineurium to bring nerves into contact with cancer cells, which in turn leads to TrkA-mediated perineural invasion [23]. Thus, it illustrates that the NGF/TrkA pathway plays a role in the dissemination and invasion of neural processes along nerves in PC.
In 1920, the German biochemist Warburg discovered that liver cancer cells, in contrast to normal hepatocytes, were more active in their glycolytic activity, leading to the discovery that the programmed metabolism of cancer cells differed from that of other normal cells [24]. Even under oxygen replete conditions, cancer cells metabolize glucose through anaerobic glycolysis to generate the energy needed for cell survival, a metabolic feature of aerobic glycolysis known as the Warburg effect, which is manifested by a high rate of glucose uptake, active glycolysis, and high levels of metabolite lactate [25]. A research found that circ_03955 through miR-3662/HIF-1α Axis activation of the Warburg effect functions as a tumor promoter, which may provide a new perspective for PC treatment [26]. ARF6 may contribute to pancreatic cancer development by promoting the Warburg effect [27]. Li et al. [28] displayed that silencing of PKM2 exhibited suppressive effects on pancreatic cancer tumor growth and invasion by altering the Warburg effect. Ye et al. [29] confirmed that lncRNA FEZF1-AS1 made an

16
Oxidative Medicine and Cellular Longevity indispensable contribution to the invasion of pancreatic cancer cells by maintaining the Warburg effect. Additionally, the Warburg effect has been reported to participate in tumor development and also accelerate the neurological diseases. Formaldehyde upregulates the Warburg effect in hippocampal tissue, as evidenced by increased ATP, glucose, and lactate production [30]. Therefore, we speculate that NGF/ TrkA signaling may contribute to PC and DRG cell migration and PNI by activating the Warburg effect. TrkA knockdown partially reversed the facilitative effects of NGF on glucose uptake, lactate production, and ATP accumulation. Accumulating evidence has established that the PI3K/AKT/ mTOR pathway is one of the key intracellular signaling pathways, and its activation has recently been found to be frequently aberrantly activated in many tumors and plays a critical role in regulating key physiological and pathological cellular processes, including cell proliferation, invasion, cancer progression, and chemoresistance as well as angiogenesis [31]. Currently, many studies have shown that Akt/mTOR signaling promotes the Warburg effect and tumor development by increasing GLUT1 trafficking and activation of glycolytic enzymes, as well as by inducing glycolysis [32]. In this study, we further tested the level of the PI3K/AKT/ mTOR pathway proteins under NGF/TrkA pathway stimulation and blockade, and the results showed that PI3K, p-AKT, and p-mTOR protein expression were upregulated in NGF overexpression PC cells, whereas transfection of TrkA siRNA reversed the results. Exosomes, small extracellular vesicles with a diameter of 30-150 nm that are secreted by almost all living cells in the body and have a phospholipid bilayer resembling the structure of the cell membrane, have been closely followed by researchers in recent years [33]. Exosomes carry a variety of biologically active substances such as DNA, RNA, proteins, liposomes, and metabolic small molecules associated with their cell of origin, which reflect the body's physiopathological state and suggest disease-related biological information, including tumors [34]. Tumor cell or other cells in the tumor microenvironment create a microenvironment suitable for tumor survival and progression by secreting exosomes and transporting tumor related active substances such as DNA, RNA, or proteins into the target cells via exosomes, which in turn regulate the biological functions of the target cells, thereby promoting the occurrence and development of tumors [35]. High baseline miR-21 plasma concentrations are associated with clinical outcomes in cancer patients after induction chemotherapy [36]. Exosomal miR-21-5p could increase the tumor volume, size, and weight of ovarian cancer in vivo and promote the development of ovarian cancer [37]. Tang et al. [38] showed that miR-21-5p may encourage cell proliferation, migration, and invasion by disrupting Smad7 expression in lung cancer cells. Besides, high expression of miR-21-5p was demonstrated to be associated with rectal tumor TNM stage and lymph node metastasis [39]. A study using nanomedicines to codeliver miR-21-5p antisense oligonucleotides and gemcitabine for PC treatment showed that the integration of miR-21-5p gene inhibition and gemcitabine treatment by using single-chain variable fragment functionalized nano-particle carriers exerted synergistic antitumor effects on PC cells [40]. miR-21-5p was obviously increased in PC cells and tissues [41]. Besides, the effect of miR-21-5p on the Warburg effect is positively correlated with the activity of the Wnt/β-catenin pathway [42,43]. The results of this study demonstrate that miR-21-5p is highly enriched in MiaPaCa-2-derived exosomes. We discovered that TDE-miR-21-5p is a considerable regulator of PNI in PC cells. Our results confirmed that NGF was able to upregulate miR-21-5p expression. Furthermore, knocking down the expression of miR-21-5p in MiaPaCa-2 cells was able to abolish the promoting effect of NGF on PNI in PC cells. Because, we speculated that NGF/TrkA promoted neuroinvasion in pancreatic cancer may be associated with upregulation of miR-21-5p by NGF.
As mentioned above, NGF and miR-21-5p levels were prominently overexpressed in PC tissues and cell lines. NGF/TrkA promotes the neuroinvasive potential of PC cells by activating the Warburg effect in DRG cells through upregulation of TDE-miR-21-5p expression. Collectively, this study demonstrates that a paracrine loop between PC cells and the DRG mediated by NGF/TrkA/miR-21-5p contributes to PNI.

Data Availability
The datasets used during the present study are available from the corresponding author on reasonable request.