1,4-dinitrosopiperazine specifically induces malignant nasopharyngeal transformation through the cytochrome P450 enzyme

Abstract

Background

1,4-Dinitrosopiperazine (DNP), a nitrite-derived compound commonly present in preserved salted foods, has been confirmed to induce malignant transformation of nasopharyngeal epithelial cells in animal experiments, but its tissue specificity and carcinogenic mechanisms remain unclear. In this study, tree shrews were used to investigate the tissue specificity and carcinogenic mechanisms of DNP.

Methods

A pathological atlas of tree shrew nasopharyngeal tissue was constructed via anatomical and pathological examination. Pharmacokinetics and tissue affinity were assessed through single-dose administration, quantifying blood/tissue distribution via liquid chromatography-tandem mass spectrometry‌ (LC‒MS/MS). Chronic intervention was established via 12-month DNP exposure, with micro positron emission tomography/computed tomography (PET-CT) imaging and pathology confirming malignant transformation. Network toxicology analyzed DNP-associated head/neck carcinomas, followed by qPCR validation of key mechanisms.

Results

A three-dimensional reconstruction model and tissue atlas of the tree shrew nasopharynx were developed. LC-MS/MS analysis revealed peak DNP accumulation in the nasopharynx at 24 h post-injection (with significantly higher retention compared to other tissues), followed by complete systemic clearance within 72 h. After 12-month chronic DNP intervention, microPET/CT imaging and pathological examination confirmed precancerous lesions or malignant transformation in nasopharyngeal tissues of all tree shrews. Network toxicology highlighted cytochrome P450 (CYP450) enzyme activation as closely associated with malignant progression, with CYP19A1 (HR = 1.52) and CYP2D6 (HR = 0.745) expression showing significant prognostic relevance. qPCR validation demonstrated marked CYP19A1 upregulation and CYP2D6 downregulation in animal nasopharyngeal tissues, indicating their role as key pathogenic biomarkers.

Conclusion

These findings confirm DNP’s selective nasopharyngeal accumulation drives carcinogenesis via CYP450 activation, with CYP19A1 overexpression and CYP2D6 suppression may serve as key pathogenic biomarkers and potential regulatory factors in DNP-induced malignant nasopharyngeal transformation.

Graphical abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12935-026-04269-4.

Introduction

Nasopharyngeal carcinoma (NPC) is a frequently occurring malignant tumor in southern China [1]. Owing to its significant geographical distribution, Cantonese-style salted fish has been identified in epidemiological studies as a risk factor for NPC [2]. Previous studies have shown the occurrence of nasopharyngeal malignancies (NPM) in rats fed Cantonese-style salted fish [3, 4]. Additionally, through compositional analysis of Cantonese-style salted fish, high levels of nitrosamines have been detected both before and after cooking [5], further corroborating the relationship between Cantonese-style salted fish and the occurrence of malignant nasopharyngeal transformation.

To further investigate the carcinogenic mechanism of Cantonese-style salted fish, Yao et al. [6, 7] attempted to establish a nasopharyngeal malignancy animal model using rats. Various nitrosamines extracted from Cantonese-style salted fish (such as N-nitrosodiethylamine, 4-nitrosomorpholine, and 1,4-dinitrosopiperazine [DNP]) were tested. After prolonged exploration, subcutaneous injection of DNP was shown to induce NPM in rats [8]. However, there are anatomical and immunological differences in the nasopharynx between rats and humans: (1) NPM mainly occurs in the crypts adjacent to the nasopharyngeal tonsils, which is a transition zone between the ciliated and squamous epithelia in humans, whereas rats lack this transition zone; and (2) rats lack tonsils and circumjacent lymphatic tissues, which results in a lack of chronic inflammatory stimulation and an immune microenvironment favorable for NPM development. These differences limit the development of this model [9]. Therefore, a more advantageous species is needed for establishing an NPM animal model.

The tree shrew, as a close relative of primates, shares similar characteristics, physiological functions, and biochemical metabolism with humans [10] and has promising applications as an animal model for human diseases [11–15]. In previous studies, we reported that the abundant lymphatic tissues throughout the body of tree shrews can provide a foundation for the formation of an immune microenvironment specific to NPM [16]. Therefore, we speculate that it may be possible to use tree shrews to establish NPM animal models to investigate the initial mechanisms of NPM development. This research provides a potential animal model for studying the pathogenesis of NPM.

Materials and methods

Experimental animals

A total of 64 adult male tree shrews (average age, 1 year ± 2 months; average weight, 114.6 ± 7.2 g) were utilized in this study. 2 animals were used for the characterization of nasopharyngeal anatomy, 30 for the investigation of DNP tissue affinity, and 32 for the chronic DNP intervention. For the sample size of the chronic DNP intervention, it was determined a priori using G*Power 3.1 software (Heinrich-Heine-Universität Düsseldorf, Germany) for the independent samples t-test. Based on preliminary data from our pilot study, the minimum required sample size per group was calculated to be 6 animals. Considering potential attrition in the long-term intervention, we initially allocated 16 animals to the DNP group, 8 to the DMSO group, and 8 to the control group to ensure sufficient statistical power for key analyses. The animals were procured from the Experimental Animal Center of Kunming Medical University. Throughout the experimental procedures, humane care was provided to the animals following the 3R principle.

DNP

DNP was synthesized by Toronto Research Chemicals (CAS number: 140-79-4), and its molecular structure is illustrated in Fig. 2A. The purity of the synthesized product, confirmed by high-performance liquid chromatography (HPLC), exceeded 98%. The DNP powder was fully dissolved in dimethyl sulfoxide (DMSO) to create a 25% DNP stock solution, which was stored at 4 °C and used on the same day.

Fig. 2.

Dynamic changes in DNP pharmacokinetics, tissue accumulation, and immunotoxicity. (A) Molecular structure of DNP. (B) Drug concentration‒time curve of the peripheral blood of tree shrews. (C) Drug concentration‒time curves for different organs. After 12 h, DNP was detected in all the tissues. After 24 h, DNP could not be detected in any tissues except for the nasopharyngeal tissue and liver. At this time, the DNP content in the nasopharyngeal tissue remained high. By 48 h, only low levels of DNP could be detected in the liver, and by 72 h, DNP was completely cleared. (D) Changes in the proportion of WBCs in the intervention group. After the injection of DNP, the neutrophil ratio significantly increased at 2 h and began to decrease at 6 h, whereas the lymphocyte ratio significantly decreased, gradually returning to normal by 72 h. (E) Many toxic granules appear in the cytoplasm of some GRANs and can be easily confused with eosinophilic granules in eosinophils. Error bars represent mean ± SD; n = 6 for each group in (B, C, D)

Nasopharyngeal dissection in tree shrews

Two tree shrews were euthanized via an intraperitoneal injection of 1.5 mL of 1% pentobarbital sodium solution. Immediately after euthanasia, a small animal multirow micro-CT imaging system (Aloka LaTheta LCT 200, Japan) was used to obtain a series of ultrathin coronal images with a thickness of 40 μm. After the limbs were fixed, the skin was incised along the midline from the mandibular angle to the xiphoid process. The skin and excess muscles were removed, and a coronal cut was made along the white solid line indicated in Supplementary Fig. 1A. The brain and eyes were removed, and the nasal and nasopharyngeal tissues were completely immersed in formalin fixative for 24 h, and then decalcified in 10% EDTA at room temperature. After reaching the decalcification endpoint, dehydrated in an automatic dehydrator, and embedded in paraffin.

Liquid chromatography-tandem mass spectrometry‌ (LC‒MS/MS) detection of DNP content in plasma and tissues

Thirty tree shrews were randomly divided into 5 groups (blank control, 12-, 24-, 48-, and 72-hour groups), with 6 animals in each group. A 1% DNP working solution was prepared by diluting the 25% DNP stock solution with saline. At a dose of 50 mg/kg, the DNP working solution was subcutaneously injected into the shoulder blade area. Blood samples were collected 18 times via an indwelling needle within 72 h after injection, with 80–100µL harvested each time [17]. After each blood collection, an equal volume of normal saline was administered through the intraperitoneal indwelling needle to compensate for the blood volume removed. One group was euthanized before injection, and the other groups were euthanized at 12, 24, 48, and 72 h post-injection. Nasopharyngeal tissues (including soft palate, nasopharyngeal roof, lateral wall, and posterior wall tissues) and the liver, kidneys, spleen, lungs, stomach and colon were collected after blood was carefully removed. After cleaning with cold PBS, 50 mg of tissue was weighed, homogenized in 200 µL of PBS using an ultrasonic homogenizer, and centrifuged, after which the supernatant was collected for detection.

The DNP powder was dissolved in the preinjection supernatant of different tissues to prepare the DNP intermediate solution, which was diluted into a standard solution with concentrations ranging from 0 to 8000 ng/mL. After filtration through a microporous membrane, the standard solution was stored at 4 °C until use. LC‒MS/MS was performed on an AB SCIEX Triple QuadTM 5500 + mass spectrometer (AB SCIEX, USA) with an ACQUITY UPLC® HSS T3 1.8 μm column (Waters, USA). Mobile phase A was acetonitrile (containing 0.1% formic acid), and mobile phase B was water. A gradient elution procedure was used: 0–5 min, mobile phase A: B = 2.0:98.0. The gradient concentration standard solution was first detected, and a standard curve of the quantitative ion peak area to DNP concentration was drawn. Method precision, recovery, and matrix effects were calculated. The DNP concentration in the tissue supernatant was subsequently detected.

Construction of the tree shrew NPM animal model

Thirty-two tree shrews were randomly divided into the DNP (n = 16), DMSO (n = 8), and control (n = 8) groups. Those in the DNP group were subcutaneously injected with DNP working solution at a dose of 50 mg/kg every 3 days for 12 months. The dosage and administration route of DNP were determined based on previous carcinogenic studies [18] and toxicological pilot experiments of DNP in rodent models. Those in the DMSO group were injected with a 4% DMSO-saline solution, and those in the control group were injected with saline. The body weight of each tree shrew was measured twice a month, and the injected dose of the DNP working solution was adjusted promptly.

Immunohistochemistry

The paraffin was removed from the slices using xylene and gradient alcohol. Antigen retrieval was performed in citrate buffer, and the tissues were stained for CK5 (Abcam 52635, 1:200), p63 (Abcam 124762, 1:5000), and Ki67 (Abcam 15580, 1:500). A DAB chromogenic detection system (Zsbio, China) was used for staining, followed by counterstaining with Mayer’s hematoxylin.

Pathological diagnostic criteria and blinded assessment

Pathological classification of nasopharyngeal mucosal lesions was performed in accordance with the Head and neck tumors: WHO classification of tumors series [19], with the following criteria: (a) Atypical hyperplasia: Mild to moderate cellular atypia, structural disorder limited to the lower 1/3–2/3 of the epithelial layer, and intact basement membrane; (b) High-grade atypical hyperplasia: Severe cellular atypia, structural disorder involving ≥ 2/3 of the epithelial layer, and intact basement membrane; (c) Carcinoma in situ: Atypical cells involving the entire epithelial layer, intact basement membrane, and no stromal invasion; (d) Invasive carcinoma: Atypical cells breaking through the basement membrane, invading the submucosa or deeper stroma, accompanied by stromal reaction. All paraffin sections were independently evaluated by 2 pathologists with more than 10 years of experience in head and neck pathological diagnosis under single-blind conditions (unaware of animal grouping and intervention measures). Disagreements on lesion classification were resolved through joint review and consensus. The consistency of diagnostic results between the two pathologists was assessed using the Kappa test, with Kappa > 0.8 indicating excellent consistency.

¹⁸F-FDG micro positron emission tomography/computed tomography (PET-CT)

MicroPET/CT (MadicLab PSA, China) was used for 18F-FDG imaging. The tube voltage was set to 80 kV, the tube current was set to 0.7 mA, and the scanning layer thickness was set to 100 μm. After intravenous injection of the 18F-FDG imaging agent through the femoral vein, the tree shrews were connected to a gas anesthesia machine, which induced and maintained anesthesia with isoflurane until the end of imaging. CT scanning was immediately performed after PET scanning.

Analysis of the human head and neck squamous cell carcinoma (hHNSCC) dataset

The STAR count data and corresponding clinical information of hHNSCC were downloaded from the TCGA database (https://portal.gdc.cancer.gov). The data in TPM format were subsequently extracted and normalized using log2(TPM + 1). Finally, after the samples with both RNA-seq data and clinical information were retained, a total of 504 samples were used for subsequent analyses. The log-rank test was used to compare the survival differences between the above two groups via Kaplan‒Meier (KM) survival analysis. For the KM curve, the p value and hazard ratio (HR) with a 95% confidence interval (CI) were calculated by the log-rank test and univariate Cox regression. The raw data of the hHNSCC single-cell dataset were sourced from GSE103322. Immune infiltration analysis was performed via CIBERSORT, and statistical analysis was performed using R software version 4.0.3.

Total RNA extraction from tissues and quantitative real-time PCR (qPCR)

Total RNA from tissues was extracted using the Total RNA Isolation Kit according to the manufacturer’s instructions (Tiangen DP419, China), and PrimeScript™ RT Master Mix (Takara RR036Q, China) was used for cDNA synthesis. All the qPCRs were performed using the SYBR^® Green Real-time PCR Master Mix (Takara CN830S, China) and the Bio-Rad CFX96 Real-time PCR System (Bio-Rad, USA). The primers used are listed in Supplementary Table 4.

Data analysis

3D Slicer (Brigham and Women’s Hospital, USA) was used to perform three-dimensional reconstruction of the CT images of the tree shrew nasopharynx, with different epithelia labeled according to the HE-stained tissue sections. PMod (PMod Technologies LLC, Switzerland) was used for reconstruction and fusion of the microPET/CT images and measurement of parameters such as FDG uptake in the target region. GraphPad Prism (GraphPad Software, USA) was used to plot the data. SPSS (IBM, USA) was used for data analysis. The data are expressed as the mean ± standard deviation (x ± s), and the relative standard deviation (RSD) and relative error (RE) were calculated. The significance level was set at p < 0.05.

Results

Anatomy and 3D reconstruction of the nasopharynx in tree shrews

A clear anatomical and histological atlas of the tree shrew nasopharynx is essential for constructing an animal model of NPM. We conducted gross dissection and micro-CT examinations of the nasopharynx, and the anatomical and imaging data are described in Supplementary Text 1. To visualize the anatomical structures and epithelial distribution in the tree shrew nasopharynx, 3D reconstruction was performed on micro-CT images, in which different colors were used to denote various types of epithelial tissues. Distinct structures, such as the posterior nostril, nasopharyngeal roof, posterior wall, and ostium pharyngeum tubae auditivae, were clearly identifiable. The epithelial distribution in the tree shrew nasopharynx from the posterior nostril to the pharynx was as follows: ciliated columnar epithelium with short tufts (Fig. 1a), ciliated columnar epithelium with long cilia (Fig. 1b), pseudostratified ciliated columnar epithelium (Fig. 1d), and stratified squamous epithelium (Fig. 1e). The pharyngeal tonsil was covered by ciliated columnar epithelium with long cilia, as indicated in light blue (Fig. 1c), with sparse and short cilia.

Fig. 1.

3D reconstruction and epithelial distribution of the nasopharynx of a tree shrew (HE sections magnification 400×). (A) and (B) show the 3D reconstruction of the nasopharynx from two different perspectives. As shown in Fig. 1A, the bottom wall of the nasopharynx was removed to better display the structures of the nasopharyngeal roof and posterior wall. Different colors and numbers are used to label the epithelium of different tissue types in the model. The short ciliary epithelium is labeled in pink (a); the long ciliary epithelium is labeled in yellow (b); the pseudostratified ciliated epithelium is labeled in red (c); the tonsil region is labeled in light blue (d); and the stratified squamous epithelium is labeled in dark blue (e). The corresponding HE images are on the right side of Fig. 1, and the same colors and lowercase letters are used to label the small squares in the upper left corner of each image

The end of the pharyngeal tonsil was located in the transition zone from the ciliated columnar epithelium to the pseudostratified ciliated columnar epithelium. The pharyngeal ostium of the auditory tube was also in the transition zone of the epithelium, as shown in Fig. 1A. Near the free edge of the soft palate, there was a transition zone from the pseudostratified ciliated columnar epithelium to the stratified squamous epithelium (PESE). Figure 1B shows a frontal view of the left side of the soft palate, which visually reveals the epithelial distribution on the nasopharyngeal side (nasopharyngeal floor). On the oral side of the soft palate, there was stratified squamous epithelium, and the number of epithelial layers increased toward the vicinity of the pharynx.

Affinity of DNP for major tree shrew tissues

The specific induction of NPM by DNP has been confirmed in previous studies [6, 7], prompting speculation about its affinity for specific tissues. By testing gradient DNP-plasma and tissue homogenate standard solutions, we confirmed that the LC‒MS/MS method is suitable for measuring DNP concentrations in the range of 5-8000 ng/mL in tissues (specific data, charts, and descriptions are recorded in Supplementary Text 2). A single subcutaneous injection of 1% DNP working solution at a dose of 50 mg/kg was administered to the tree shrews in the scapular region. The drug concentration‒time curve is presented in Fig. 2B, where the peak drug concentration is 6626.67 ng/mL, the time to reach peak concentration is 8 min, the terminal elimination half-life is 1.72 h, and the clearance rate is 2.345 L/h/kg. DNP concentrations in the liver, spleen, kidneys, lungs, stomach, colon, and nasopharyngeal tissues were measured at 12, 24, 48, and 72 h. Concentration‒time curves of drug in different organs were plotted, with the tissue DNP concentration as the ordinate and the injection time as the abscissa (Fig. 2C). After 12 h, DNP was detectable in all tissues, with the highest average DNP concentration in the nasopharyngeal tissue and the lowest in the lungs. By 24 h, DNP was undetectable in all tissues except for the nasopharyngeal tissue, which had a significantly greater average DNP content than did the liver tissue (1567.9 ng/g vs. 849.6 ng/g, p = 0.002). At 48 h, only the liver still presented low levels of DNP, and complete clearance was achieved by 72 h. It is inferred that DNP exhibits a selective retention property in nasopharyngeal tissue, which may be attributed to multiple potential mechanisms including specific tissue affinity, slower clearance rate, tissue-specific sequestration, or local metabolic differences in the nasopharynx. In general, this prolonged local exposure affords sufficient time for DNP-induced epithelial cell injury.

Blood smears were used to classify white blood cells in the intervention group, which revealed a significant increase in the proportion of neutrophilic granulocytes (GRAN) at 2 h after DNP injection, which began to decrease after 6 h. The lymphocyte (LYM) proportion significantly decreased (p < 0.05) at 6 h. The different cell proportions gradually returned to normal by 72 h (Fig. 2D). Additionally, blood smears revealed a substantial number of toxic granules in the cytoplasm of some GRANs (Fig. 2E).

Construction of the tree shrew NPM animal model

On the basis of previous studies yielding anatomical and histological atlases of the tree shrew nasopharynx, confirming the affinity of tree shrew nasopharyngeal tissue for DNP, and exploring the damage caused by DNP to nasopharyngeal epithelial cells, we began constructing a long-term intervention to establish a tree shrew NPM animal model. During the model construction process, a total of 19 tree shrews reached the scheduled endpoint (12 months), including 8 in the DNP group (initially 16; survival rate, 50%), 6 in the DMSO group (initially 8; survival rate, 75%), and 5 in the control group (initially 8; survival rate, 62.5%). Among the thirteen tree shrews that exited the experiment, eight died due to accidental factors such as intestinal torsion, intestinal telescoping, hepatic abscess, and systemic infection; two died due to urinary system diseases such as renal surface hemorrhage and pyuria; one sustained a hind limb fracture during cage movement and required treatment, preventing continuation of the experiment; one exhibited severe depressive behavior at the 8th month of the intervention, characterized by self-harm, reduced physical activity, prolonged corner dwelling time, and weight loss, leading to withdrawal from the experiment; and one experienced spontaneous seizures at the 6th month of the intervention, resulting in withdrawal from the experiment. No malignant lesions were found in any of the deceased animals after autopsy or pathological examination. The weight changes in each group were recorded (Fig. 3A). The weights of the animals in the DNP group slowly decreased, whereas the weights of those in the DMSO and control groups did not significantly change. At the scheduled endpoint, the activity, food and water intake, fur, body size, and posture of the tree shrews in the DMSO and control groups were not significantly different (Fig. 3B). However, the tree shrews in the DNP group appeared skinny, with sparse fur and a lack of luster, and some exhibited symptoms such as unilateral or bilateral nasal discharge, hoarseness, and drooling (Fig. 3C).

Fig. 3.

Long-term DNP intervention effects on body weight, physiological manifestations, and nasopharyngeal ¹⁸F-FDG uptake. (A) Weight changes in each group after long-term intervention. n = 16 (DNP), n = 8 (DMSO), n = 8 (Control) at baseline, n = 8 (DNP), n = 6 (DMSO), n = 5 (Control) at the endpoint. (B) The body shape of the animals in the control group at the end of the experiment was not significantly different from that at the beginning of the experiment. (C) The animals in the DNP group at the scheduled endpoint were skinny, with sparse and lackluster hair, and exhibited nasal discharge. (D) ¹⁸F-FDG microPET/CT imaging results revealed almost no nasopharyngeal uptake of ¹⁸F-FDG in the DMSO group or the control group but significant nasopharyngeal uptake in the DNP group

¹⁸F-FDG PET/CT is the most commonly used method for diagnosing tumors [20]. Here, we used ¹⁸F-FDG microPET/CT to identify the site of tumor occurrence in a tree shrew NPM animal model. To determine the optimal imaging time window, dynamic scans were initially performed on animals in the control group, which revealed that the 40- to 50-minute period after intravenous injection of ¹⁸F-FDG had low background interference and did not significantly affect the physiological uptake of ¹⁸F-FDG in the tissue structures surrounding the nasopharyngeal tissue (detailed images and descriptions are recorded in Supplementary File 3). Subsequently, static imaging was performed on the tree shrews via microPET/CT. The imaging results, shown in Fig. 3D, indicated significantly increased nasopharyngeal ¹⁸F-FDG uptake in the DNP group but almost no nasopharyngeal uptake in the DMSO or control groups. The minimum uptake, maximum uptake, average uptake, and selected volume of ¹⁸F-FDG in the selected nasopharyngeal tissue region were measured on the midline sagittal plane, and the results are listed in Table 1. Compared with the control group, the DNP group presented significantly greater maximum and average ¹⁸F-FDG uptake, with no significant difference between the DMSO group and the control groups. Additionally, no abnormal uptake of ¹⁸F-FDG was observed in other organs of the tree shrews.

Table 1.

¹⁸F-FDG uptake in the nasopharynx

Group	Min (%ID/mL)	Max (%ID/mL)	Average (%ID/mL)	Volume (ccm)
Control	0.33 ± 0.03	1.09 ± 0.26	0.67 ± 0.06	0.0053 ± 0.00047
DMSO	0.29 ± 0.03	1.18 ± 0.31	0.71 ± 0.08	0.0055 ± 0.00050
DNP	0.38 ± 0.09	2.77 ± 1.15**	1.06 ± 0.29*	0.0043 ± 0.00047

^* indicates p < 0.05; ^** indicates p < 0.01. Final sample sizes for PET-CT analysis: Control group (n = 5), DMSO group (n = 6), DNP group (n = 8). All surviving animals completed PET-CT scanning at the 12-month endpoint. Statistical analysis was performed by one-way ANOVA followed by LSD post-hoc test

Initially, we conducted autopsies and histopathological examinations on animals that reached the study endpoint, verifying that no remarkable dysplastic changes were evident in organs other than the nasopharyngeal tissue in the DNP group. Histopathological evaluations were performed on all surviving animals: Control group (n = 5), DMSO group (n = 6), DNP group (n = 8). The consistency of pathological diagnosis was verified using Kappa test (Kappa = 0.83, 95%CI:0.71–0.95, p < 0.001) among two independent pathologists and excluding observer bias. Lymphoid tissues, including palatine tonsils, the thymus, cervical lymph nodes and the spleen, presented varying degrees of follicular hyperplasia but without notable structural disruption or dysplastic alterations (Supplementary Fig. 5). The nasopharyngeal tissue differed notably. The nasopharyngeal roof and posterior wall surfaces in the control and DMSO groups were smooth, with no abnormal proliferative tissue. HE staining revealed no significant lymphoid follicle hyperplasia in the pharyngeal tonsil area, and the ciliated columnar epithelium was arranged neatly or regularly, with a uniform cell size and no atypical cells (Fig. 4A1, C1). In the stratified squamous epithelium and epithelial transition zone, no regions with atypical proliferation were observed, and the structure was orderly, with no lymphocyte aggregation in the basal layer (Fig. 4B1, D1). Immunohistochemistry revealed that the nasopharyngeal epithelial cells in various regions were neatly arranged, with intact basal membranes and a normal cell proliferation rate. In the DNP group, 7 tree shrews presented rough nasopharyngeal roof and posterior wall surfaces, with nodular elevations and accompanying mucosal tissue congestion (Fig. 4 and Supplementary Fig. 6). HE staining revealed slightly dilated surrounding blood vessels, reactive hyperplasia of pharyngeal tonsil lymphoid follicles, thickening of the ciliated columnar epithelium, and local atypical hyperplasia (Fig. 4E1, G1), with active proliferation of nasopharyngeal epithelial cells and lymphocytes in the region (Fig. 4E4, G4). In some regions of the PESE, severe atypical hyperplasia occurred, with an irregular arrangement of tissue structures, cell dysplasia, and lymphocytic infiltration (Fig. 4F1). In one case, significant cell dysplasia was observed in the region of stratified squamous epithelium, with variable cell sizes, deep staining of cell nuclei, and involvement of the entire epithelial layer, with a tendency to grow toward the basal membrane, accompanied by lymphocytic infiltration (Fig. 4H1). Immunohistochemistry revealed active proliferation of nasopharyngeal epithelial cells, increased layering, disordered arrangement of basal cells, and poor basal membrane integrity. In summary, among the 8 surviving animals in the DNP group, 1 case was diagnosed with carcinoma in situ, 6 with high-grade atypical hyperplasia, and 1 with atypical hyperplasia.

Fig. 4.

Microscopic view of the nasopharynx in the control, DMSO, and DNP groups, with lesion areas (indicated by blue solid line boxes) and pathological results, and HE staining revealed the general morphology of the primary lesion areas. CK5 staining was used to visualize the epithelial cell cytoplasm, p63 staining was used to identify basal cell nuclei, and Ki67 staining was used to evaluate cell proliferation. In the control group, the surface of the nasopharyngeal roof and posterior wall appeared very smooth. Pathological findings revealed a normal transition from pseudostratified ciliated columnar epithelium to stratified squamous epithelium (PESE) at the epithelial transition zone (A1-A4, B1-B4). The conditions of the DMSO group were similar to those of the control group (C1-C4, D1-D4). In the DNP group, the nasopharyngeal roof and posterior wall surfaces appeared rough, with nodular elevations and accompanying mucosal congestion. HE staining revealed mildly dilated surrounding blood vessels and reactive hyperplasia of pharyngeal tonsil lymphoid follicles. Focal atypical hyperplasia, characterized by epithelial ingrowth with a subbasement membrane growth tendency, is observed, resulting in irregular arrangement and compromised integrity of basal cells. Cell proliferation is highly active within lesion areas

Mechanistic analysis of DNP-induced NPM in tree shrews

To explore the possible mechanisms by which DNP induces NPM in tree shrews, we sought to identify potential target genes from public datasets. Notably, NPM is classified as a distinct subtype of HNSCC, sharing core molecular pathways with other HNSCC subtypes [21, 22]. Due to the limited sample size and lack of comprehensive clinical follow-up data in public NPM-specific datasets, we used the hHNSCC dataset (n = 504), which provides complete gene expression profiles and survival information for prognostic analysis. We find commonalities with the occurrence of hHNSCC, we obtained 120 potential target genes of DNP from ChEMBL21 [23]. These genes were then combined and analyzed with the differentially expressed genes from the hHNSCC dataset. Ultimately, 18 overlapping genes were identified (Fig. 5A), 17 of which were cytochrome P450 family genes. KEGG pathway enrichment analysis was performed on these 18 genes (Fig. 5B), revealing that the top three significantly enriched pathways were chemical carcinogenesis-DNA adducts, metabolism of xenobiotics by cytochrome P450, and drug metabolism-cytochrome P450. To identify the key genes among them, the degree algorithm was used to rank the contributions of gene interactions, and a PPI network diagram was constructed. CYP3A4, CYP1A1, and CYP19A1 were located at the core of the network and might play key roles (Fig. 5C). Furthermore, a forest plot was drawn, and survival analysis was conducted. CYP19A1 is a significant risk factor for NPC, whereas CYP2D6 is a significant protective factor, and there may be an antagonistic effect between them (Fig. 5D-F).

Fig. 5.

Integrative genomic analysis of DNP in hHNSCC: pathway enrichment, PPI networks, and prognostic CYP19A1/CYP2D6 modulation with intervention validation. (A) There are 18 genes that overlap between DNP target genes and significantly differentially expressed genes in hHNSCC; (B) KEGG pathway enrichment analysis of the 18 genes; (C) PPI network diagram based on the degree algorithm; (D) Forest plot of the 18 genes, showing that only CYP19A1 is a significant risk factor for the occurrence of hHNSCC, whereas CYP2D6 is a significant protective factor; (E) and (F) Survival analysis of CYP19A1 and CYP2D6 in hHNSCC; (G) PCR validation was performed in the nasopharyngeal tissues of the DNP and control groups. CYP19A1 was significantly increased after DNP intervention, whereas CYP2D6 was significantly decreased. (H) The expression levels of CYP19A1 and CYP2D6 were validated in several major tissues in the control group. Error bars represent mean ± SD; n = 5 (Control), n = 8 (DNP)

qPCR validation of these 18 overlapping genes was performed in the nasopharyngeal tissues of the DNP and control groups. Ultimately, 3 genes were excluded because of low amplification efficiency. Among the remaining 15 overlapping genes, 9 were significantly increased in the DNP group, 3 were significantly decreased, and 3 were not significantly different (Fig. 5G). Among these genes, CYP19A1 was significantly increased after DNP intervention, whereas CYP2D6 was significantly decreased. This trend is consistent with the analysis results for hHNSCC. It is known that cytochrome P450 family genes are expressed mainly in the liver, but their expression in other tissues is not clear. To verify the expression abundance of these genes among different tissues and avoid false positives due to subtle changes in gene expression levels, PCR was performed on several major tissues in the control group (Supplementary Fig. 7). Both CYP19A1 and CYP2D6 had high expression levels in the nasopharynx (Fig. 5H).

We explored the main target cells of CYP19A1 and CYP2D6 within tumor tissues. The expression levels of CYP19A1 and CYP2D6 in different cell types were determined in the hHNSCC single-cell dataset, and both were significantly expressed in the malignant and CD8⁺ Tex cell clusters (Fig. 6A). Subsequently, immunoinfiltration analysis was performed (Fig. 6B), revealing that CYP19A1 was significantly highly expressed in CD8⁺ T cells and follicular helper T cells, whereas CYP2D6 was significantly expressed at low levels (p < 0.01). This is consistent with the results of previous experiments.

Fig. 6.

CYP19A1/CYP2D6-driven immune dysregulation and carcinogenic mechanism of DNP via T-cell reprogramming. (A) The expression abundance of CYP19A1 and CYP2D6 in different cell types was analyzed in the hHNSCC single-cell dataset. Both genes were significantly expressed in the malignant and CD8⁺ Tex cell populations. (B) Immune infiltration analysis revealed that CYP19A1 was significantly highly expressed, whereas CYP2D6 was significantly expressed at low levels in T-cell CD8⁺ and T-cell follicular helper subsets (p < 0.01), which was consistent with previous experimental results. (C) Mechanistic inference diagram: After DNP enters the body, it resides in the nasopharyngeal region and is activated by the CYP enzyme system in the form of dealkylation. On the one hand, it covalently binds with the amino groups on nucleotides in DNA, causing DNA mutations or chromosomal aberrations during cell division, thereby accelerating the carcinogenesis process. On the other hand, it promotes the expression of CYP19A1 and inhibits the expression of CYP2D6, driving the continuous differentiation of stem-like PD-1^lo CD8⁺ T cells into dysfunctional PD-1^hi CD8⁺ Tex cells, which eventually undergo apoptosis. This process reduces the immune system’s ability to surveil and eliminate tumor cells

Discussion

In humans, the depression between the tympanic tonsil and the posterior wall of the pharynx is called the pharyngeal recess, which is covered mainly by PESE; this area is the most common and earliest site of NPC [24]. There is relatively rich lymphatic tissue around the pharyngeal recess, which provides a foundation for the formation of a special immune microenvironment in NPM [25–27]. In the ostium pharyngeum tubae auditivae of the tree shrew, there is also lymphatic tissue and PESE. Therefore, there are many similarities between the anatomical and histological characteristics of the nasopharynx of tree shrews and humans, suggesting that tree shrews are potentially more suitable for establishing NPC animal models.

Histopathological examination of the nasopharyngeal tissues from the DNP group revealed lesions typically affecting two sites: the eustachian tonsil and the PESE. The lesions presented two typical signs: damage to nasopharyngeal epithelial cells and infiltration of immune cells. These findings are similar to the pathological features of human NPC, where changes in the immune microenvironment play crucial roles in NPC development and progression [28, 29]. Previous studies have reported an association between CYPs and NPC [30, 31], but the specific mechanisms involved remain unclear. Through this study, we hypothesize that DNP may interact with CYPs at multiple levels and induce the development of NPC. First, DNP enters the body, accumulates in the nasopharyngeal region, and is activated by the CYP enzyme system in the form of dealkylation [32]. The activated DNP can subsequently (1) covalently bind with amino groups on nucleotides in DNA, forming nitroso-nucleotide compounds. These compounds can easily lead to DNA mutations or chromosomal aberrations during cell division [33], thereby accelerating the carcinogenesis process. (2) These compounds promote the expression of CYP19A1 and inhibit the expression of CYP2D6 locally, driving the continuous differentiation of stem-like PD-1^lo CD8⁺ T cells into dysfunctional PD-1^hi CD8⁺ Tex cells, which eventually undergo apoptosis [34]. This reduces the immune system’s ability to surveil and eliminate tumor cells, thereby promoting the initiation and progression of tumors (Fig. 6C). It is important to emphasize that the current study establishes correlative relationships rather than direct causal links between CYP19A1/CYP2D6 expression and DNP-induced malignant nasopharyngeal transformation. The upregulation of CYP19A1 and downregulation of CYP2D6 are closely associated with DNP exposure and poor prognostic outcomes in hHNSCC, but we cannot definitively conclude they are “drivers” of carcinogenesis without functional manipulation experiments. Instead, these findings suggest they may act as key biomarkers and potential regulatory factors involved in CYP450-mediated metabolic activation of DNP and subsequent immune dysregulation.

The animal model we constructed exhibited a small range of lesions, and most pathological observations involved atypical hyperplasia. We speculate that there are two reasons limiting the range and degree of the lesions: (1) NPC occurs in humans mostly between 40 and 60 years of age [35, 36], and long-term intervention involving environmental factors has been proven to play an important role [20]. Therefore, the intervention and observation times may not be long enough to be major limiting factors. (2) EBV infection is a key co-factor for NPC development, as persistent EBV infection and progressive genomic changes promote the clonal evolution of NPC. The expression of latent EBV gene products alters multiple cell pathways by regulating the host microenvironment, while EBV-induced methylation inactivates multiple cancer-related genes [37–39]. However, successful construction of an animal model of NPC with EBV infection alone is still challenging, and the interaction between EBV and other environmental factors may be necessary for the occurrence of NPC. We have already verified that tree shrews can serve as an animal model for EBV infection [40]; therefore, we will use both EBV and DNP to improve the model in our future work.

This study has several limitations. First, animal attrition during the experiment exceeded expectations. It is important to note that all deaths were attributed to accidental factors or spontaneous diseases, which were unrelated to DNP intervention or nasopharyngeal malignant transformation. Thus, the attrition is unlikely to introduce selection bias. Additionally, post-hoc power analysis using G*Power showed that the final sample sizes (n = 8 for the DNP group, n = 5–6 for the control/DMSO groups) retained a statistical power of 82% for detecting differences in key outcomes, which meets the conventional threshold for robust statistical inference. Nevertheless, the reduced sample size may limit the generalizability of our findings, and future studies with larger sample sizes and longer follow-up are warranted to confirm the conclusions. Another limitation is the lack of functional validation experiments. The current mechanistic inference is based on correlative evidence from animal models, hHNSCC dataset analysis, and immune infiltration patterns, rather than direct functional evidence. Additionally, in this research, we did not conduct an in-depth exploration of the specific biochemical process by which DNP is activated by the CYP enzyme system in the body and its metabolic products, nor did we determine which metabolic product can accelerate the carcinogenesis process of cells. Although this is highly important for elucidating the specific mechanism of DNP carcinogenesis, the overall extremely small dosage of DNP makes this process very challenging.

Conclusion

Our research provides a 3D reconstructed model and tissue atlas of the tree shrew nasopharynx. LC‒MS/MS analysis validated the affinity of DNP for tree shrew nasopharyngeal tissue. The successful construction of a DNP-induced tree shrew model of precancerous lesions or malignant transformation in nasopharyngeal tissues, verified by microPET/CT and pathology. Network pharmacology analysis highlights the involvement of DNP’s selective nasopharyngeal accumulation is closely associated with nasopharyngeal carcinogenesis via CYP450 pathway activation, with CYP19A1 overexpression and CYP2D6 suppression serving as key pathogenic biomarkers and potential regulatory factors.

Abbreviations

CT

Computed Tomography

CYP450

Cytochrome P450

DNP

1,4-Dinitrosopiperazine

GRAN

Granulocytes

HPLC

High-performance liquid chromatography

hHNSCC

Human head and neck squamous cell carcinoma

LC‒MS/MS

Liquid chromatography-tandem mass spectrometry‌

NPC

Nasopharyngeal carcinoma

NPM

Nasopharyngeal malignancy

NMR

Nuclear magnetic resonance

PESE

Pseudostratified ciliated columnar epithelium to the stratified squamous epithelium

PET-CT

Positron emission tomography/computed tomography

qPCR

quantitative real-time PCR

RSD

Relative standard deviation

RE

Relative error

Author contributions

**Yiwei Feng: ** Conceptualization, Methodology, Writing - Original Draft. **Wei Xia: ** Methodology, Software, Writing - Original Draft. **Pengcheng Zhao: ** Validation, Formal analysis, Writing - Original Draft. **Li Gong: ** Visualization, Validation. **Xunwei Feng: ** Validation. **Nan Shi: ** Investigation. **Jiawei Wang: ** Data Curation. **Shenghui Ke: ** Methodology, Software. **Zhen Yin: ** Resources. **Ketong Ji: ** Validation. **Yongjing Lai: ** Project administration. **Jingyu Li: ** Writing - Review & Editing. **Xiang Bin: ** Supervision. **Xiang Yi: ** Supervision. **Guangyao He: ** Funding acquisition, Project administration. **Anzhou Tang: ** **Writing - Review & Editing, Project administration, Funding acquisition.**.

Funding

This study was supported by the Natural Science Foundation of China (U21A20371, 82160217 and 82260519), the Guangxi Natural Science Fund (2020GXNSFAA297235 and 2020JJA140398), Guangxi Scholarship Fund of the Guangxi Education Department of China and Key Laboratory of Early Prevention and Treatment for Regional High-Frequency Tumours (Guangxi Medical University), the Ministry of Education (GKE-ZZ202412, GKE-ZZ202303 and GKE-ZZ202231), Guangxi Science and Technology Program (AD25069077).

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The use of experimental animals in this study was approved by the Animal Ethics Review Committee of Guangxi Medical University, with approval numbers 201911060 and 202301001.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.