Abstract

   During ex situ conservation, the adaptability of giant pandas to
   environmental changes is greatly challenged. The issue of natural
   reproduction in captive giant pandas remains unresolved both
   domestically and internationally. It hypothesized that the restricted
   natural reproductive capacity may be linked to abnormal mating behavior
   expression due to physiological stress resulting from incompatible
   pairings in confined environments. To test this hypothesis, we utilized
   ultra-high performance liquid chromatographytandem
   quadrupole-time-of-flight mass spectrometry (UPLC-Q-TOF/MS) to analyse
   urine metabolites in captive adult giant pandas during their breeding
   period. Simultaneously, enzyme-linked immunosorbent assay was employed
   to measure the levels of cortisol and epinephrine in urine, providing
   insight into the psychological state of captive giant pandas during
   mate selection by examining all metabolites and related biochemical
   pathways. This comprehensive approach aims to fully elucidate the
   physiological mechanisms underlying the decline in natural reproductive
   capacity. The metabolomics findings indicate that the aberrant
   expression of natural mating behaviour in captive adult male and female
   giant pandas may be associated with dysfunction in amino acid metabolic
   pathways. The activation of these metabolic pathways is linked to
   psychological stress, such as the tryptophan metabolic pathway and
   GABAergic synapse pathway. The results of physiological indicators
   indicate a significant correlation between the expression of natural
   mating behaviour in captive adult pandas and the hormone urine
   cortisol, which is associated with physiological stress. These findings
   indicate that the atypical manifestation of natural mating behaviour in
   captive adult giant pandas may be associated with physiological stress
   induced by incompatible pairings within confined environments.

Introduction

   The giant panda (Ailuropoda melanoleuca) is a distinctive species in
   China and serves as a flagship species for biodiversity conservation
   ([36]Wei et al., 2015). In order to ensure the preservation of this
   particular species, China has implemented an ex situ conservation and
   breeding programme. While the implementation of the ex situ
   conservation breeding programme has yielded phased results in giant
   panda protection, persistent challenges such as the decline of
   instinctive behaviour within the captive population have significantly
   impeded the smooth execution of the programme ([37]Xie, 2016;
   [38]Swaisgood et al., 2018). During ex situ conservation, significant
   alterations have occurred in the habitat of giant pandas (including
   early weaning in the nursery, maternal separation, non-parental care
   and exposure to unnatural stressors such as tourist noise), posing
   considerable challenges to their environmental adaptability. These
   environmental changes can impact the behaviour of giant pandas, leading
   to observable distinctions in behavioural patterns between wild and
   captive pandas, including the manifestation of abnormal behaviours
   (stereotyped behaviours) and a decline in instinctive behaviours
   (mating behaviours) ([39]Martin-Wintle et al., 2020; [40]Zhang et al.,
   2021). Captive giant pandas generally exhibit a low natural
   reproduction rate. The factors influencing the natural reproductive
   efficiency of captive giant pandas are complex and variable. The
   hypothesis suggests that the atypical manifestation of natural mating
   (NM) behaviour in captive giant pandas may be associated with
   psychological stress resulting from pair incompatibility within captive
   environments, given their reputation as a wild species known for strong
   mate selection ([41]Zhang et al., 2004; [42]Martin-Wintle et al.,
   2015). However, direct evidence or markers supporting this hypothesis
   are currently lacking. Our ability to explain the adaptive state of
   captive giant pandas is currently limited when relying solely on
   behavioural and hormonal manifestations. Metabolomics is the systematic
   investigation of the biochemical pathways and interactions of
   metabolites within biological specimens, providing insight into
   cellular, tissue, and organismal metabolic activities ([43]Cao et al.,
   2020). Stress, as a pervasive systemic response, intricately
   intertwines with the regulation of stress hormones, often heralding
   substantial alterations in the body’s metabolic landscape. To test our
   hypothesis, we aim to utilize metabolomics technology to discern subtle
   differences in metabolite profiles between experimental and control
   groups. This effort will contribute to a deeper understanding of the
   underlying biological processes that govern the complex interplay
   between stress, behaviour and metabolism ([44]Khamis et al., 2017).

   In scientific research, stress is commonly categorized into two types
   based on the intensity and duration of the stressor: acute stress,
   which typically involves exposure to a single, brief (<30 min)
   stressor; and chronic stress, which generally entails exposure to a
   complex, persistent (~1 month) stressor ([45]Chrousos and Gold 1992).
   Metabolic regulation under stressful conditions serves as a protective
   mechanism in life processes, with moderate levels of stress aiding the
   body in adapting to environmental changes and thereby enhancing chances
   of survival and reproduction. However, excessive or prolonged states of
   stress, particularly chronic stress, can lead not only to metabolic
   disorders but also to the onset of severe mental illnesses such as
   depression ([46]Olff, 1999; [47]Ehlert et al., 2001). The alteration of
   material metabolism represents a key characteristic of the stress
   process ([48]Kivimäki et al., 2022). The physiological response
   elicited by stress typically encompasses a sequence of neuroendocrine
   activities ([49]Carrasco and Van de Kar 2003). In the context of acute
   stress, activation of the human body’s sympathetic–adrenal–medullary
   system and hypothalamic–pituitary–adrenal (HPA) cortex system results
   in peripheral effects characterized by elevated concentrations of
   epinephrine and cortisol in the bloodstream. These changes are now
   relatively well established ([50]McEwen, 2008). Cortisol serves as the
   principal effector hormone within the HPA axis (HPAA) stress response
   system and represents a crucial marker for evaluating the intensity of
   acute stress ([51]O'Connor et al., 2021). When animals experience
   sudden fright, are relocated to a new environment or undergo
   long-distance transportation, they exhibit a stress response
   characterized by a rapid increase in cortisol concentration in the body
   ([52]Chmelíková et al., 2020; [53]Kovács et al., 2021). Epinephrine, a
   stress hormone released by the adrenal medulla, plays a pivotal role in
   acute stress response through its stimulation of the sympathetic
   nervous system ([54]Gerra et al., 2000). If the stressor persists, it
   triggers activation of the sympathetic–adrenal medulla system, this
   results in the release of epinephrine from the adrenal medulla, which
   is rapidly distributed through the bloodstream to further prepare the
   body for a response, leading to a variety of physical, psychological
   and behavioural changes ([55]O'Connor et al., 2021). In addition to
   alterations in individual biomarkers such as cortisol and epinephrine,
   metabolomics has also unveiled numerous novel biomarkers that provide
   insight into the metabolic pathways associated with the pathogenesis of
   stress-related diseases. For instance, psychosocial stress triggers the
   activation of certain metabolic pathways that utilize amino acids
   ([56]Obled et al., 2002), leading to a decrease in the relative
   concentrations of essential amino acids such as lysine and tryptophan,
   as well as a reduction in the relative concentration of conditional
   essential amino acid arginine and vitamin B group nicotinamide
   ([57]Morgan et al., 2022; [58]Reeds and Jahoor 2001). The impact of
   stress on the body is more comprehensively reflected through changes in
   endogenous metabolites of small molecules within organisms.

   Hence, in order to investigate the hypothesis that the decline in NM
   behaviour of captive giant pandas is linked to physiological stress
   induced by incompatible pairings in confined environments, this study
   used urine ultra-high-performance liquid chromatography-tandem
   time-of-flight mass spectrometry (UHPLC-TOF/MS) technique to compare
   metabolite levels in the urine of captive giant pandas exhibiting
   different NM behaviours during peak estrus. Additionally, enzyme-linked
   immunosorbent assay (ELISA) was employed to determine cortisol and
   epinephrine levels in the urine of male and female captive giant
   pandas, aiming to elucidate the physiological mechanism underlying the
   decline in NM behaviour through the analysis of metabolites, related
   biochemical pathways and hormone levels.

Materials and Methods

Animals and ethics statement

   In this study, 12 captive giant pandas from the Chengdu Research Base
   of Giant Panda Breeding (Panda Base) were selected as the subjects for
   research ([59]Table 1). The essential information and breeding history
   of the experimental giant pandas are presented in [60]Tables 2 and
   [61]3. The pandas were divided into two groups: the NM group,
   consisting of three female and three male adult giant pandas with
   successful NM experience (able to produce offspring through NM after
   reaching adulthood), denoted as F-NM (female-natural mating) and M-NM
   (male-natural mating). The artificial insemination (AI) group comprised
   three female and three male adult giant pandas without successful NM
   experience (unable to produce offspring through NM after reaching
   adulthood), represented as F-AI (female-artificial insemination) and
   M-AI (male-artificial insemination). The study protocol was approved by
   the Institutional Animal Care and Use Committee of Chengdu Research
   Base of Giant Panda Breeding (approval number: 2020013).

Table 1.

   Information on the giant pandas in this study
   Name Studbook Groups Gender The number of samples collected for
   metabolomics tests The number of samples collected for ELISA
   Gong Zai 711 NM Male 2 3
   Ying Ying 724 AI Male 2 3
   Cheng Shuang 857 AI Male 2 3
   Mei Lan 649 NM Male 2 3
   Lou Abao 703 NM Male 2 3
   Xi Lan 731 AI Male 2 3
   Er Qiao 823 AI Female 2 3
   Ya Yun 796 AI Female 2 3
   Jing Jing 598 AI Female 2 3
   Mei Lun 870 NM Female 2 3
   Mei Bao 801 NM Female 2 3
   Zhao Mei 990 NM Female 2 3
   [62]Open in a new tab

   Notes: All adult male giant pandas that are unable to mate naturally
   have attempted NM without success, resulting in the need for semen
   collection and AI to obtain offspring. In rare instances, adult male
   giant pandas capable of natural mating may encounter sexual
   incompatibility, resulting in the need for semen collection and
   artificial insemination for females. The three selected giant pandas
   successfully produced offspring through both NM and semen collection,
   demonstrating their reproductive capabilities.

Table 2.

   Reproductive history information of the captive male giant pandas (The
   reproductive histories of male giant pandas are informed by statistical
   data derived from our previously published article in AABS) ([63]Zhang
   et al., 2021)
   Name Birth year The number of successful NM (Reproductive history) The
   number of offspring produced by NM (Reproductive history) The number of
   offspring produced by sperm collection and AI (Reproductive history)
   The number of successful NM (2021) Number of sperm collection (2021)
   Sampling time
   Ying Ying 2008 0 0 6 0 0 3-24-2021
   Cheng Shuang 2012 0 0 1 0 1 2-12-2021
   Xi Lan 2008 0 0 2 0 1 3-15-2021
   Mei Lan 2006 12 20 4 2 0 3-28-2021
   Lou Abao 2007 4 4 2 2 5 3-21-2021
   Gong Zai 2008 5 5 2 3 0 3-27-2021
   [64]Open in a new tab

   Notes: In the 2021 mating season that we sampled, amongst the male NM
   group, Lou Abao not only successfully completed 2 NM sessions and sired
   offspring, but also provided semen samples for AI of experimental
   female giant pandas on five occasions, albeit without successful
   reproduction. In the male AI group, aside from Cheng Shuang and Xi Lan,
   who contributed semen once for AI of experimental female giant pandas
   without producing offspring, Ying Ying did not provide semen samples
   during that year.

Table 3.

   Reproductive history information of the female giant pandas (The
   reproductive history of female giant pandas based on historical
   statistical data from Chengdu Research Base of Giant Panda Breeding for
   2000–21)
   Name Birth year The number of successful NM (Reproductive history) The
   number of offspring produced by NM (Reproductive history) The number of
   offspring produced by AI (Reproductive history) Time of the first birth
   Sampling time
   Er Qiao 2011 0 0 1 2017 3-22-2021
   Ya Yun 2010 0 0 1 2018 3-23-2021
   Jing Jing 2005 0 0 2 2015 4-15-2021
   Mei Lun 2013 1 1 0 2021 3-27-2021
   Mei Bao 2010 1 1 1 2019 3-20-2021
   Zhao Mei 2010 1 1 0 2021 4-19-2021
   [65]Open in a new tab

   Notes: In the 2021 mating season, all adult female giant pandas in the
   female NM group (Mei Lun, Mei Bao and Zhao Mei) successfully engaged in
   NM for the first time and subsequently gave birth to offspring.
   Conversely, within the female AI group, none of the adult female giant
   pandas (Er Qiao, Ya Yun and Jing Jing) successfully completed NM;
   however, they all successfully produced offspring through AI.

Feeding and management

   All the giant pandas included in this study resided at the Panda Base
   in Sichuan, China (104°E, 30°N). Each panda was housed individually and
   provided with consistent indoor and outdoor activity areas,
   environmental enrichment and surroundings. The feeding, management
   protocols and food supply were standardized across all individuals.

Urine sampling

   During this period, all adult male giant pandas were sampled and housed
   separately in the Moon or Sun delivery room of the Panda Base for the
   entire breeding season. Adjacent animal enclosures on both sides
   contained adult female giant pandas in estrus, including those involved
   in the experiment. Sampling took place after mating attempts during the
   2021 mating season (March–April), with urine collected within 3 h
   following successful NM for Gong Zai, Mei Lan and Lou Abao and within
   3 h following unsuccessful NM attempts for Ying Ying, Cheng Shuang and
   Xi Lan. The non-invasive method was utilized to collect urine samples
   from male giant pandas during estrus. The Panda Base demonstrated that
   the appearance of the urinary luteinizing hormone peak indicated the
   timing of ovulation and facilitated the determination of the optimal
   time for NM ([66]Cai et al., 2017). We collected urine samples from all
   female giant pandas during their estrus period. Based on hormone
   fluctuations, we were able to determine the peak time range of
   oestrogen (~6 h), and the urine samples required for testing were
   collected during this specific time frame as illustrated in [67]Fig. 1.
   After the panda has urinated, it should be lured into a nearby cage and
   the urine transferred into a 2-ml frozen storage tube using a
   disposable syringe. Care should be taken to select uncontaminated and
   undiluted urine, with the date, time and individual name clearly
   marked. The urine should then be immediately sealed with a sealing film
   without any chemical treatment, refrigerated, transported back to the
   laboratory and stored in a −80°C refrigerator for testing.

Figure 1.

   Figure 1
   [68]Open in a new tab

   Sampling protocol for adult giant pandas in captivity during the mating
   season

Liquid chromatography-mass spectrometry/mass spectrometry analysis

   Urine samples were collected in 5-ml vacutainer tubes containing
   ethylene diamine tetra acetic acid (EDTA) as the chelating agent. The
   samples were then centrifuged at 1500 × g for 15 min at 4°C.
   Subsequently, 150-μl aliquots of urine samples were stored at −80°C
   until UHPLC-Q-TOF/MS analysis. Prior to analysis, the samples were
   slowly thawed at 4°C and an appropriate quantity was added to a
   pre-cooled solution of methanol/acetonitrile/water (2:2:1, v/v). The
   mixture was then vortexed, sonicated at a low temperature for 30 min
   and allowed to stand at −20°C for 10 min before being centrifuged at
   14000 × g for 20 min at 4°C. Following this step, the supernatant was
   removed and dried under vacuum. Reconstitution of the sample involved
   adding 100 μl of an aqueous solution of acetonitrile (acetonitrile:
   water = 1:1, v/v), which was then re-mixed by vortex agitation and
   centrifuged again at 14 000 × g for 15 min at 4°C prior to supernatant
   analysis.

   The samples were submitted to Shanghai Applied Protein Technology Co.,
   Ltd for untargeted metabolomics analysis, following the detailed
   detection procedures outlined in our previously published paper
   ([69]Zhang et al., 2022). The general procedure involves the
   utilization of the Agilent 1290 Infinity LC ultra-high-performance
   liquid chromatography system with a HILIC column for sample separation,
   followed by analysis using the ABTriple TOF 6600 mass spectrometer to
   collect primary and secondary spectra. The samples were subjected to
   separation via the Agilent 1290 Infinity LC ultra-high-performance
   liquid chromatography system and subsequently analysed using the Triple
   TOF 6600 mass spectrometer (ABSCIEX). Detection was carried out in both
   positive and negative ion modes of electrospray ionization (ESI).
   Following data extraction by XCMS software, metabolite structure
   identification and data preprocessing were conducted prior to
   experimental data quality assessment, culminating in data analysis.

ELISA analysis

Determination of hormone in urine samples

   The concentration of cortisol in urine was quantified using the Cayman
   Cortisol ELISA Kit (No. 500360) with a sensitivity of 35 pg·ml^−1, an
   intra-batch coefficient of variation <13.4% and an inter-batch
   coefficient of variation <25.8%. The general procedure involved
   preparing the standard and diluting the sample, followed by adding
   ELISA buffer, standard, sample, cortisol tracer and cortisol antibody
   to the enzyme-labelled plate coated with goat anti-mouse
   immunoglobulin. The plate was then incubated at 4°C overnight.
   Subsequently, after washing the plates on the following day, Ellman’s
   reagent and tracer were added and incubated in darkness for 1.5–2 h
   before measuring absorbance at 405 nm using a Thermo Scientific
   Multiskan MK3 spectrophotometer.

   The concentration of epinephrine in urine was quantified using the IBL
   Epinephrine ELISA Kit (No. RE59251) with a sensitivity of 0.2 ng·ml^−1,
   intra-batch coefficient of variation <8.7%, and inter-batch coefficient
   of variation <12.1%. Briefly, standard, control and urine samples were
   added to the boric acid affinity gel-coated macro titration plate.
   Subsequently, incubation and washing steps were performed with double
   steaming water, buffer and acylation reagent as required followed by
   overnight incubation at 4°C. On the following day, the enzyme plate
   coated with goat anti-rabbit antibodies was utilized for addition of
   enzyme solution along with standard, control and urine samples. This
   was followed by sequential addition of epinephrine antiserum, enzyme
   coupler and PNPP chromogenic solution according to specified incubation
   and washing steps. The reaction was terminated by adding PNPP
   termination solution before measuring absorbance at 405 nm using a
   Thermo Scientific Multiskan MK3 spectrophotometer.

Creatinine assays

   The urine was diluted 20 times and added to a 96-well plate with
   0.05 ml of 0.04 mol·l^−1 picric acid and 0.05 ml of 0.75 mol·l^−1 NaOH.
   After incubation at room temperature (25°C) for 15 min, the optical
   density (OD) value at 492 nm was measured using a Thermo Scientific
   Multiskan MK3 spectrophotometer. Samples with a creatinine (Cr) value
   <0.1 mg·ml^−1 were considered contaminated with water and excluded from
   further analysis based on the OD value not being used in these cases.

Statistical analyses

Liquid chromatography-mass spectrometry/mass spectrometry

   The data analysis encompassed univariate and multidimensional
   statistical analysis, differential metabolite screening, correlation
   analysis of differential metabolite, as well as Kyoto Encyclopedia of
   Genes and Genomes (KEGG) pathway analysis. Detailed descriptions of the
   statistical analyses can be found in our previous publications
   ([70]Zhang et al., 2022).

ELISA

   Creatinine values were utilized to standardize for variations in water
   content within each urine sample. The hormone concentration in each
   urine sample was divided by the creatinine concentration using the
   following formula: A = creatinine content (mg·ml^−1) = [creatinine
   (μg·ml^−1) × dilution ratio]/1000; B = cortisol content
   (ng·ml^−1) = cortisol content per well (ng/well) × dilution release
   ratio; C = creatinine to correct cortisol (cortisol ng/mg Cr) = B/A.
   Similar procedures were applied to determine epinephrine levels.

   The experimental data were processed using Excel and analysed by
   SPSS24.0. Group comparisons were conducted using the t-test, and the
   data were presented as x ± SEM. The significance level was set as
   α = 0.05.

Results

Urine metabolomics analysis of giant pandas with different mating behaviours

Analysis of intergroup differences in metabolomics experimental data

   Based on univariate statistical analysis, we conducted a comprehensive
   examination of the urine metabolomics data from captive adult male and
   female giant pandas exhibiting different NM behaviours (normal and
   abnormal). We analysed all detected metabolites (FC > 1.5 or FC < 0.67,
   P < 0.05) in both positive and negative ion modes, visualizing the
   results through volcanic maps. Subsequently, we identified the top 10
   upregulated and downregulated metabolites with qualitative names for
   labelling ([71]Fig. 2). Furthermore, we utilized multidimensional
   statistical analysis and Partial Least Squares Discrimination Analysis
   (OPLS-DA) to establish a model expressing the relationship between
   sample categories based on metabolite expression levels. The OPLS-DA
   model parameters of captive adult male giant pandas were obtained
   through 7-fold cross-validation as R^2X = 0.313, R^2Y = 0.991,
   Q^2 = 0.79 in the positive ion mode and R^2X = 0.372, R^2Y = 0.986,
   Q^2 = 0.727 in the negative ion mode. The model parameters for captive
   adult female giant pandas in negative ion mode were R^2X = 0.478,
   R^2Y = 0.991, Q^2 = 0.883, whilst those in positive ion mode were
   R^2X = 0.562, R^2Y = 0.996, Q^2 = 0.95. The results indicate good
   stability of the model ([72]Fig. 3). To prevent overfitting during
   modelling, a permutation test was conducted to validate the model’s
   robustness and validity. The gradual decrease in permutation retention
   led to a corresponding decrease in R2 and Q2 of the random model,
   demonstrating that the original model did not suffer from overfitting
   and remained robust throughout the analysis process ([73]Fig. 4).

Figure 2.

   [74]Figure 2
   [75]Open in a new tab

   Volcano plot comparing urine metabolites between the NM and AI groups
   in captive male and female giant pandas; A: captive male giant pandas
   in positive ion mode; B: captive male giant pandas in negative ion
   mode; C: captive female giant pandas in positive ion mode; D: captive
   female giant pandas in negative ion mode. Note: The x-axis in the
   figure represents the logarithm of log2 Fold Change, whilst the y-axis
   represents the negative logarithm of -log10 significance P-value.

Figure 3.

   [76]Figure 3
   [77]Open in a new tab

   OPLS-DA of urine metabolites between NM group and AI group in captive
   male and female giant pandas; A: captive male giant pandas in positive
   ion mode; B: captive male giant pandas in negative ion mode; C: captive
   male giant pandas in positive ion mode; D: captive male giant pandas in
   negative ion mode.

Figure 4.

   [78]Figure 4
   [79]Open in a new tab

   Permutation test (base on 200 permutations) of urine metabolites
   between NM group and AI group in captive male and female giant pandas.
   R^2Y indicated the model’s explanation for the Y variable rate; Q^2
   indicated the predictive ability of the model.

Differential metabolites and bioinformatics analysis

   In this experiment, strict OPLS-DA VIP > 1 and P-value <0.05 were
   utilized as the criteria for significant difference metabolite
   screening. A total of 94 different metabolites was identified in the
   urine of adult male giant pandas in captivity, with 46 metabolites
   detected in the positive mode ([80]Fig. 5) and 48 in the negative mode
   ([81]Fig. 6). When compared to the M-NM group, a decrease was observed
   in 74 urine metabolites such as malic acid, citrate, phenylalanine,
   glutamine, tryptophan and tyrosine in giant pandas from the M-AI group;
   conversely, an increase was noted in 20 urine metabolites including
   serotonin, 5-aminovalerate and betaine. A total of 177 different
   metabolites was identified in the urine of adult female giant pandas in
   captivity, with 86 metabolites detected in positive mode ([82]Fig. 7)
   and 91 metabolites in negative mode ([83]Fig. 8). In comparison to the
   F-NM group, the F-AI group exhibited a decrease in 35 urine metabolites
   such as malic acid, succinic acid, DL-glutamic acid and aspartic acid,
   whilst an increase was observed in 144 urine metabolites such as
   kynurenic acid and 5-aminovaleric acid betaine.

Figure 5.

   Figure 5
   [84]Open in a new tab

   Bioinformatics analysis of differential metabolites between NM group
   and AI group in captive male giant pandas: significant differential
   metabolite hierarchical clustering results of male giant pandas in
   captivity under positive ion mode. Note: Each row in the figure
   represents a differential metabolite, with the ordinate indicating
   significantly differentially expressed metabolites, and each column
   representing a group of samples, with the abscissa providing sample
   information. Metabolites exhibiting similar expression patterns are
   clustered together on the left.

Figure 6.

   [85]Figure 6
   [86]Open in a new tab

   Bioinformatics analysis of differential metabolites between NM group
   and AI group in captive male giant pandas: significant differential
   metabolite hierarchical clustering results of male giant pandas in
   captivity under negative ion model.

Figure 7.

   [87]Figure 7
   [88]Open in a new tab

   Bioinformatics analysis of differential metabolites between NM group
   and AI group in captive female giant pandas: significant differential
   metabolite hierarchical clustering results of female giant pandas in
   captivity under positive ion model.

Figure 8.

   [89]Figure 8
   [90]Open in a new tab

   Bioinformatics analysis of differential metabolites between NM group
   and AI group in captive female giant pandas: Significant differential
   metabolite hierarchical clustering results of female giant pandas in
   captivity under negative ion model.

   Metabolic pathway enrichment analysis using KEGG was conducted on urine
   metabolites from captive adult male and female giant pandas. The
   results indicated that the expression of NM behaviour in captive adult
   male giant pandas is associated with changes in the metabolic pathways
   of phenylalanine, tyrosine and tryptophan biosynthesis, biosynthesis
   amino acid, propionate metabolism, protein digestion and absorption, as
   well as tryptophan metabolism ([91]Fig. 9A). Additionally, the
   expression of NM behaviour in captive adult female giant pandas was
   associated with changes of the metabolic pathways primarily
   encompassing the citrate cycle (TCA cycle), biosynthesis amino acid,
   alanine, aspartate and glutamate metabolism, and γ-aminobutyric acid
   synaptic formation pathway ([92]Fig. 9B).

Figure 9.

   [93]Figure 9
   [94]Open in a new tab

   KEGG pathway enrichment analysis of differentially expressed
   metabolites between NM group and AI group in captive male and female
   giant pandas; A: male giant pandas in captivity; B: female giant pandas
   in captivity. Note: The vertical axis of the bar graph represents each
   KEGG metabolic pathway, whilst the horizontal axis denotes the number
   of differentially expressed metabolites within each pathway. The colour
   gradient reflects the P-value of the enrichment analysis, with darker
   shades indicating smaller P-values and greater significance in
   enrichment. Additionally, the numerical values on each column represent
   the rich factor.

Differences of cortisol and epinephrine in urine of giant pandas with
different mating behaviours

   There were no significant differences in cortisol and epinephrine
   levels between the two groups (cortisol: P = 0.746; epinephrine:
   P = 0.179) ([95]Table 4).

Table 4.

   Urine cortisol and epinephrine levels of captive adult pandas
      Cortisol/(ng·mg^−1) Epinephrine/(ng·mg^−1)
   NM 24.716 ± 12.177     4.560 ± 2.310
   AI 26.273 ± 12.871     5.934 ± 1.769
   P  0.746               0.179
   [96]Open in a new tab

Differences in cortisol and epinephrine in the urine of giant pandas of
different sexes with different mating behaviours

   The results of urine cortisol revealed that female giant pandas in the
   NM group exhibited significantly lower levels of urine cortisol
   compared to those in the AI group (P = 0.006). Conversely, male giant
   pandas in the NM group showed higher levels of urine cortisol compared
   to those in the AI group, with a trend towards approaching significance
   (P = 0.088). Additionally, among all giant pandas displaying NM
   behaviour, male individuals demonstrated significantly higher urine
   cortisol content than their female counterparts (P = 0.033), whereas
   male giant pandas unable to exhibit NM behaviours displayed
   significantly lower urine cortisol content than females (P = 0.026)
   ([97]Table 5).

Table 5.

   Urine cortisol and epinephrine levels of captive adult pandas
   Cortisol/(ng·mg^−1) P-value Epinephrine/(ng·mg^−1) P-value
   Female Male Female Male
   NM 18.592 ± 4.159 33.844 ± 16.167 0.033 4.584 ± 1.437 4.540 ± 3.189
   0.978
   AI 37.474 ± 16.788 18.764 ± 6.207 0.026 6.693 ± 2.022 4.669 ± 0.713
   0.155
   P-value 0.006 0.088 0.094 0.948
   [98]Open in a new tab

   The results of urine epinephrine analysis indicated that the levels in
   the female NM group were lower than those in the female AI group, with
   a trend towards significance (P = 0.094). However, no significant
   differences were observed between the male NM group and the male AI
   group (P = 0.948).

Discussion

   The factors affecting the natural breeding efficiency of captive giant
   pandas are complex and dynamic ([99]Swaisgood et al., 2004; [100]Zhang
   et al., 2004; [101]Li et al., 2017). Given the conspicuous mate
   selection behaviour documented in free-ranging giant pandas, the
   academic consensus suggests that the diminished reproductive success in
   captivity may be attributed to the absence of autonomous mate choice,
   which could lead to enforced copulation with mismatched partners and
   consequent psychological stress (frustration) ([102]Zhang et al., 2004;
   [103]Martin-Wintle et al., 2015). However, this speculation lacks
   direct evidence or markers. Currently, our understanding of the
   adaptive state of captive giant pandas remains limited when relying
   solely on behavioural and hormonal manifestations. This metabolic
   response to stress represents a physiological reaction of the organism
   to external stimuli, leading to perturbations in internal homeostasis
   and subsequent modifications in metabolic pathways. In our
   investigation, we employed UPLC-Q-TOF/MS metabolomics technology to
   analyse urine samples from captive adult giant pandas to explore the
   hypothesis that the manifestation of NM behaviour in captivity is
   associated with psychological distress arising from incompatible
   pairings.

   The results indicated significant differences in urine metabolite
   abundance between the NM and AI groups in both adult male and female
   giant pandas. The discrepancies observed may be correlated with
   substantial alterations in amino acid metabolic pathways, encompassing
   the activation of pathways tied to psychological status, such as those
   involving tryptophan, alanine, aspartate and glutamate metabolism.
   Previous research has shown that disruptions in tryptophan metabolic
   pathways could potentially result in imbalances of active metabolites,
   which are closely associated with psychiatric disorders such as
   depression and schizophrenia ([104]Alkhalaf and Ryan 2015). The study
   also revealed a significant positive correlation between pathways
   related to amino acids such as alanine, aspartic acid and glutamate
   metabolism with outcomes of successful mating attempts. Consistent with
   these findings, our study observed a significant upregulation of
   metabolites (aspartic acid, DL-glutamic acid and citrate) associated
   with the metabolic pathways of alanine, aspartic acid and glutamate in
   the urine of captive adult giant pandas exhibiting normal NM behaviour.
   These results suggested that the expression of NM behaviour in captive
   giant pandas is linked to their emotional state.

   Our study revealed that the urinary serotonin levels in male pandas
   from the AI group were lower compared to those in the NM group, and
   there were significant alterations in the metabolites associated with
   tryptophan metabolism. Additionally, in the AI group, there was a
   notable enrichment in metabolic pathways related to the nervous system
   (GABAergic synapses), along with significant increases or decreases in
   inhibitory neurotransmitter amino acids and metabolites linked to brain
   development (DL-glutamic acid, succinic acid and 5-aminovalerate
   betaine). These changes in metabolites are considered potential risk
   markers for stress-related disorders ([105]Deakin, 1996; [106]Leppik
   et al., 2018; [107]Haikonen et al., 2022), supporting the scientific
   hypothesis that psychological frustration due to pair incompatibility
   is a key factor contributing to the decline of NM behaviour among
   captive giant pandas.

   By employing metabolomics techniques, we have initiated an
   investigation into the complex physiological changes occurring across
   all metabolites and metabolic pathways within the body. Our initial
   analysis has indicated that the abnormal expression of NM behaviours in
   captive giant pandas may be attributed to psychological stress
   resulting from incompatible pairings. To further support our hypothesis
   and gain deeper insights, we have also utilized ELISA methodologies to
   precisely quantify the levels of stress-related hormones (cortisol and
   epinephrine) during the mating season of these captive pandas. Our
   findings suggest that female giant pandas in the AI group exhibit
   significantly elevated levels of urinary cortisol compared to those in
   the NM group, indicating potential hindrance of NM behaviour due to
   frustration from pair incompatibility among adult female giant pandas.
   Additionally, previous research has demonstrated a positive association
   between urinary cortisol levels and stereotyped behaviour exhibited by
   giant pandas ([108]Pan et al., 2022). The most recent study has
   revealed that the stereotyped behaviour of giant pandas is associated
   with the success rate of NM. Specifically, male giant pandas exhibiting
   more stereotyped behaviours demonstrate a higher likelihood of
   successful NM during the mating season, potentially attributed to
   heightened sexual desire and fertility ([109]Martin-Wintle et al.,
   2020). Conversely, the stereotyped behaviour of female captive pandas
   exhibits a negative correlation with the expression of NM behaviour
   ([110]Martin-Wintle et al., 2020). This may be due to highly
   stereotyped female pandas inadvertently signalling a lack of sexual
   receptivity to male pandas, thereby diminishing their attractiveness
   ([111]Mason, 2010). Our results on urinary cortisol levels in the urine
   of giant pandas also support the significance of the behavioural
   observations reported by [112]Martin-Wintle et al. (2020).
   Specifically, our results indicate that male pandas exhibit more
   frequent stereotyped behaviours and are more likely to mate naturally
   when their urine cortisol levels are higher, whereas female pandas show
   the opposite pattern, with lower urine cortisol levels being associated
   with greater ease of NM. While there were no statistically significant
   differences in the epinephrine indicators pointing to chronic stress, a
   trend towards chronic stress was observed. Numerous human clinical
   studies have established a link between chronic psychological stress
   and dysfunction of the HPAA. Modulating the activity of the HPAA
   impacts the function of the sympathetic–adrenal–medullary system,
   leading to the release of epinephrine ([113]Schmalbach et al., 2020),
   influencing the human autonomic nervous system and eliciting
   physiological and behavioural manifestations of psychological disorders
   ([114]Olff, 1999). Due to the lack of clear indicators for assessing
   chronic stress in animals and psychological stress resulting from
   chronic stress, particularly in studies involving wild animals, we must
   rely on relevant research findings from human studies for guidance.
   Additionally, based on preliminary urine metabolomics results, there is
   evidence suggesting that neurotransmitter metabolites and pathways
   associated with psychological stress may be linked to the reproductive
   success of captive giant pandas ([115]Zhang et al., 2022). For the
   subsequent investigation, it is imperative to integrate the behavioural
   science of giant pandas. The correlation between the aberrant
   manifestation of NM behaviour and chronic stress in captive giant
   pandas was examined through the utilization of brain tissue physiology,
   immunology and neurobiology.

   In summary, the findings of this study suggest that the unconventional
   display of NM behaviour in captive adult giant pandas may be linked to
   physiological distress caused by incompatible pairings within
   restricted habitats. These results offer a theoretical foundation for
   the scientific conservation of captive giant panda populations and the
   enhancement of their welfare.

Acknowledgements