Abstract Background Gram-negative bacteria are the main bacterial pathogens infecting Chinese giant salamanders (Andrias davidianus; CGS) in captivity and the wild, causing substantial economic losses in the CGS industry. However, the molecular mechanisms underlying pathogenesis following infection remain unclear. Results Spleen-derived macrophages from healthy CGS were isolated, cultured, and identified using density gradient centrifugation and immunofluorescence. A macrophage transcriptome database was established 0, 6, and 12 h post lipopolysaccharide stimulation using RNA-sequencing. In the final database 76,743 unigenes and 4,698 differentially expressed genes (DEGs) were functionally annotated. Gene Ontology and Kyoto Encyclopedia of Genes and Genomes enrichment results showed that DEGs were concentrated in toll-like receptor–nuclear factor kappa B-related immune pathways. Ten DEGs were validated 12 h after lipopolysaccharide (LPS) stimulation. Although the common LPS recognition receptor toll-like receptor 4 was not activated and the key adaptor protein MyD88 showed no significant response, we observed significant up-regulation of the following adaptors: toll/interleukin-1 receptor domain-containing adaptor inducing interferon-β, tumour necrosis factor receptor-associated factor 6, and transforming growth factor-β activated kinase 1, which are located downstream of the non-classical MyD88 pathway. Conclusions In contrast to that in other species, macrophage activation in CGS could depend on the non-classical MyD88 pathway in response to bacterial infection. Our study provides insights into the molecular mechanisms regulating CGS antibacterial responses, with implications for disease prevention and understanding immune evolution in amphibians. Supplementary Information The online version contains supplementary material available at 10.1186/s12864-024-10888-w. Keywords: Chinese giant salamander, Lipopolysaccharide, Macrophages, MyD88, Non-classical pathway, RNA-sequencing Background Immunity is an important physiological function in animals that facilitates recognition and resistance to exogenous pathogenic microorganisms, as well as the removal of ageing, dead, damaged, or mutated cells to maintain homeostasis. Immune defence systems depend on both innate and acquired responses. Innate immunity, which represents an organism’s first line of defence against pathogen infection, can quickly sense and destroy pathogens to prevent infection [[40]1]. Innate immunity relies mainly on natural immune cells (phagocytic and non-phagocytic cells), which include mononuclear macrophages, neutrophils, eosinophils, and basophils [[41]2]. Macrophages are phagocytes and antigen-presenting cells that are widely involved in pathogen clearance, inflammatory response mediation, and maintenance and repair of tissue microenvironment homeostasis [[42]3]. Therefore, studies on the pathogenesis and regulatory mechanisms of disease-targeting macrophages have significant therapeutic potential. Upon invasion by pathogenic microorganisms, macrophages first rapidly identify the invading microorganisms at the infection site through pattern-recognition receptors (PRRs), then activate downstream signalling pathways and induce the secretion of cytokines such as interleukin (IL) and tumour necrosis factor (TNF) [[43]4]. Toll-like receptors (TLRs), a type of PRR on the surface of macrophages that mediate clearance and self-protection against pathogenic microorganisms, are type I transmembrane proteins comprising a leucine-rich extracellular region, a cysteine-rich transmembrane region, and an intracellular (toll/interleukin-1 receptor, TIR) domain [[44]5]. To date, 13 TLR genes have been identified in mammals (TLR1–TLR13) [[45]6], whereas fish TLRs include TLR1–TLR5, TLR13, TLR22, and TLR20a–TLR20e [[46]7, [47]8]. In contrast, TLR family members in the giant salamander are not well understood. TLR family members mediate multiple distinct anti-infective signalling pathways based on whether the TIR domain of a TLR molecule binds to the adaptor protein myeloid differentiation factor 88 (MyD88) [[48]9]. In mammals, TLR1, TLR4, TLR5, and TLR7 signalling is MyD88-dependent; participation of TRAF6 and IL-1 receptor-associated kinase in the signal transduction process then activates nuclear factor kappa B (NF-кB) to produce an immune response. In contrast, the MyD88-independent sig. nalling pathway involves receptor binding to TRIF or TRIF-related activator molecule (TRAM) However, in poultry, only the classical MyD88 pathway exerts antimicrobial immune responses [[49]10], indicating that the existence and activation modes of the two TLR signal transduction pathways differ among species. Lipopolysaccharide (LPS) is an important cell-wall component of gram-negative bacteria that supports their ability to infect hosts. After bacterial infection, the body activates corresponding PRRs, with TLR4 offering an important line of defence against early stages of bacterial infection [[50]11]. In mammals, LPS binds to TLR4 receptors on the cell membrane with the assistance of three proteins, LBP, MD2, and CD14, and recruits Mal proteins to further activate MyD88. This ultimately triggers IκBα protein phosphorylation and translocation of the p50/p65 complex into the nucleus, which enhances the transcription of downstream factors [[51]12]. Although the TLR family is conserved among vertebrates, TLR4 is not present in all species, including some species of fish [[52]13, [53]14]. For example, TLR4 expression has not been confirmed in the Chinese giant salamander (Andrias davidianus; CGS) of the clade Caudata, and the genes and signalling pathways activated after LPS-induced macrophage stimulation remain unclear. CGS is the largest living fossil species among extant amphibians, occupying an important position in the evolutionary history of aquatic and terrestrial organisms. CGS populations have declined drastically because of a reduction in suitable habitats, over-capture by humans, and infectious diseases [[54]15]. To limit CGS decline, China has implemented artificial breeding programs, with several artificial farms in Shaanxi, Sichuan, Hubei, and Zhejiang Provinces [[55]16]. However, an increasing number of disease outbreaks threaten the survival of CGS and the viability of CGS artificial breeding programs, with bacterial infections representing a particular threat. In this study, we developed an experimental method for studying the interactions between pathogenic microorganisms and CGS immune systems to analyse the genes and signalling pathways activated by LPS in macrophages. Specifically, we isolated spleen-derived macrophages from CGSs, which were then cultured and characterised. We then utilised RNA-sequencing to establish transcriptome databases of macrophage responses at 0, 6, and 12 h after LPS stimulation. This study provides insights into the molecular mechanisms underlying antibacterial responses in CGS, and has implications for disease prevention and understanding immune evolution in amphibians. Methods Animal care and use Two-year-old healthy CGSs were obtained from Hanzhong Ancient Giant Salamander Co., Ltd (Hanzhong, China). All animal procedures were conducted according to methods approved by the Animal Research and Ethics Committee of the Shaanxi Institute of Zoology to minimise animal suffering. Macrophage isolation Undiluted Percoll separation solution (GE Pharmacia, London, UK) was prepared with 1.5 M NaCl and 100% Percoll separation solution at a ratio of 9:1. The following Percoll separation solutions were prepared for subsequent experiments: 40% Percoll solution (6.14 mL 0.15 M NaCl + 3.86 mL 100% Percoll separation solution, 40% Percoll buoyancy density of approximately 1.056 g/mL); 30% Percoll solution (7.14 mL 0.15 M NaCl + 2.86 mL 100% Percoll separation solution, 30% Percoll buoyance density of approximately 1.043 g/mL). CGSs were wiped with 75% (v/v) ethanol, anaesthetised by immersing in 0.6 g/L MS-222 (tricaine methanesulfonate) and 0.6 g/L soda (sodium bicarbonate, Church & Dwight, Co, USA) [[56]17], then euthanised while unconscious. Spleens were collected and washed three times with sterile phosphate-buffered saline and twice with serum-free medium (Sigma, St. Louis, Missouri, USA). Spleens were then ground and passed through a 70-µm cell sieve, filtered, and centrifuged at 1,000 rpm for 5 min. The cells were suspended in 40% and 30% Percoll solution, and the cell layer above the 30% Percoll was carefully collected. The cells were then suspended in M199 medium containing 1,000 units/mL penicillin (Sigma, St. Louis, Missouri, USA), 1,000 µg/mL streptomycin (Sigma, St. Louis, Missouri, USA), and 20% foetal bovine serum, cultured at 25 °C with 2% CO[2], and stained with Trypan Blue (Sigma, St. Louis, Missouri, USA). To detect the distribution and phenotype of spleen-derived macrophages from CGSs, immunofluorescence was used with a macrophage surface-specific Anti-F4/80 antibody (Abcam, Cambridge, UK) and a goat anti-rat IgG (H + L) cross-adsorbing secondary antibody Alexa Fluor™555 (Thermo Fisher Scientific Inc., Mississauga, ON). Alexa Fluor™ 555 antibody-labelled cells fluoresce red in response to excitation wavelengths of 555 nm, and all nuclei stained using DAPI fluoresce blue in response to excitation wavelengths of 350–360 nm. Macrophage purification Macrophages purified using gradient centrifuge separation were cultured at a density of 5 × 10^6 cells per plate for 12 h before LPS stimulation. Non-adherent cells were removed, and the medium was replaced with M199 medium (Thermo Fisher Scientific Inc., Mississauga, ON) with 100 ng/mL LPS (Sigma, St. Louis, Missouri, USA). After incubation for 0, 6, and 12 h (AD_0h, AD_6h, and AD_12h, respectively), the culture flasks were washed vigorously to remove dead cells before harvesting the remaining live cells. RNA extraction, purification, and sequencing Total RNA was extracted using TRIzol reagent (Invitrogen, Waltham, MA, USA) according to the manufacturer’s instructions. RNA quality and completeness were examined using a NanoDrop instrument (Thermo Scientific, Waltham, MA, USA) and Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). Only high-quality RNA (OD[260/280] = 1.8–2.2, OD[260/230] ≥ 2.0, RNA integrity number > 8.0, 28 S:18 S ≥ 1.0, > 1 µg) was used for next -generation sequencing. Oligo (dT) magnetic beads were used to concentrate the mRNA, which was ligated with random hexamers and used to generate cDNA. The synthesised cDNA was subsequently purified, end-repaired, A-tailed, and ligated to sequencing adapters. RNA purification, reverse transcription, library construction, and sequencing were performed at Shanghai Majorbio Bio-Pharm Biotechnology Co., Ltd. (Shanghai, China), according to the assay manufacturers’ instructions (Illumina, San Diego, CA). Finally, a cDNA library was obtained via PCR amplification and purified using AMPure XP beads. The libraries, generated on a NovaSeq 6000 platform (Biomarker Technologies, Beijing, China) with paired-end reads 150 bp in length, were named AD_0h (LPS stimulation for 0 h, n = 3), AD_6h (LPS stimulation for 6 h, n = 3), and AD_12h (LPS stimulation for 12 h, n = 3). Functional annotation of unigenes Quality control of raw reads was performed using FastQC (version 0.11.3). Reads with adapters, repeated sequences, missing nucleotides > 10%, and/or of low quality (with over 50% of Q-values ≤ 20% bases) were discarded. De novo transcriptome assembly was conducted using Trinity, a short-read assembly programme, with default parameters. The reads were combined and overlapped using a k-mer length of 25 bp to form longer fragments (termed contigs). Functional annotation of unigenes was conducted by comparing sequences with those in the following public databases: NCBI non-redundant protein (Nr, e-value ≤ 10^− 5), SwissProt (e-value ≤ 10^− 5), evolutionary genealogy of genes: non-supervised orthologous groups (eggNOG, e-value ≤ 10^− 5), and protein family (Pfam, e-value ≤ 0.01). GO functional classification was performed using Blast2GO v2.5 software (e-value < 10^− 6) and the WEGO web tool based on Nr and Pfam annotations. Pathways were assigned using the KEGG database (e-value ≤ 10^− 10). For unigenes that failed to align to any of the databases, sequence directions and gene-coding regions were predicted using ESTScan (3.0.3) software. DEG enrichment analyses DEGs at different times after LPS stimulation were identified using the DESeq2 R package. The resulting p-values were adjusted using the Benjamini–Hochberg method to limit the false discovery rate to less than 0.01. Genes with an adjusted p-value < 0.01 and fold change ≥ 2.0 were considered DEGs. To gain insight into the potential functions and metabolic pathways associated with the DEGs, further GO and KEGG enrichment analyses were conducted using TBtools software v1.0697. Significant pathways were defined based on a corrected p-value ≤ 0.01. Gene expression analyses Candidate genes involved in signalling pathways activated in response to LPS infection were selected and identified using quantitative PCR (qPCR). The SYBR Green method and LineGene 9600 System (Bioer Technology, Hangzhou, China) were used with the following cycling conditions: denaturation for 10 min at 95 °C, followed by 40 cycles at 95 °C for 10 s, 58 °C for 30 s, and 72 °C for 25 s. β-actin was used as the internal reference control, and no template was used as a negative control. Primers were designed using Primer Premier 5 and are listed in Table [57]1. All assays were performed in triplicate to reduce errors. Table 1. Primer pairs designed for PCR validation of differentially expressed genes Genes Primer sequences(5’—3’) Tm (ºC) MyD88 F: GAAAAAAGGTGTAGGAAGATGG 53.05 R: GCAGACAGTAATGAACCGCAAG 58.82 TRIF F: GGGTTTCTCCATTCTTCCATC 53.45 R: CCCAGTCTTGTCTTCGCATT 56.45 TRAF6 F: CTGACGGTGAAATGCCCGAATGAA 61.48 R: CAACGGACAGCTCTGGTCGTGATC 63.29 TAK1 F: CTTCGGCAGTTGTCTCGTGT 58.60 R: CGCCTTGGGAGCACTGTAA 57.85 TAB3 F: ACATCTCACGGTTCTTCTCCTTT 57.70 R: GGGATGGTTGGCTATGTATTGGT 56.76 p65 F: CAGCAACAGACCCATCCAA 54.74 R: CGTTCTCCTCATCGGATCTGCTAC 62.14 IL-8 F: CATGTCATTGGCAAGAACGGGAA 59.25 R: TCCAAACACACCTCCTGCCCTGATT 62.12 TNF-α F: CTCCAGCAGCACCACCTAA 57.32 R: GAAAGTTGGTGTTCGGTGC 55.16 IFN-α F: TGCTGGTCCTGTTCTCCTC 57.32 R: GGAATCGGTGGACTGAAGC 57.32 IL-1β F: TGAACGCAGACCTGATGGA 55.16 R: GCCGCTGCTTCATCTTCTC 57.32 β-actin F: TGAACCCAAAAGCCAACCGAGAAAAGAT 62.63 R: TACGACCAGAGGCATACAGGGACAGGAC 65.26 [58]Open in a new tab F: forward primer, R: reverse primer Results Macrophage isolation Percoll density gradients of 30–60% were used for separation; the results showed that a 30–40% Percoll separation solution was optimal. Optical microscopy showed that CGS macrophages were round or oval, half attached to the plate wall, and aggregated in clusters with good growth after 1–2 days of culture. A macrophage surface-specific antibody F4/80 immunofluorescence assay was performed on the isolated spleen cells after density gradient centrifugation (Fig. [59]1). Macrophages accounted for 96% of cells after density gradient centrifugation, but only approximately 6% of cells without density gradient centrifugation (Fig. [60]S1). Fig. 1. [61]Fig. 1 [62]Open in a new tab Immunofluorescence results of spleen-derived macrophages in giant salamanders before centrifugation (first and second rows) and after centrifugation (third and fourth rows) of splenocytes. Primary anti-F4/80 antibody was added during immunofluorescence for samples shown in the first and third rows, whereas phosphate-buffered saline was added as a negative control for samples shown in the second and fourth rows Sequencing characterisation and de novo assembly After quality assessment and data filtering, 19.5 Gb of clean data was obtained. All Q30 scores of the clean data were above 98.3%, and the Q20 scores were more than 95.01% for each sample. In addition, the average GC content was 50.36%, with no samples containing overly high (> 80%) or low (< 20%) GC content, suggesting that the de novo sequencing was high quality (Table [63]S1). From the clean reads, the Trinity Program produced 92,220 transcripts with an average length of 721.25 bp and N50 size of 1,302 bp, and assembled 76,743 unigenes with a mean length of 670.49 bp and N50 size of 1,167 bp. Among these unigenes, the sequence lengths ranged from 200 to > 2,000 bp (Table [64]S2). A total of 52,095 unigenes (68%) were ≤ 500 bp, 12,186 (16%) were 501–1,000 bp, and 15,462 (16%) were > 1,000 bp (Fig. [65]2). Finally, alignment was performed between the sequences of assembled unigenes; over 82% of reads showed correspondence, indicating their suitability for further analyses. Fig. 2. Fig. 2 [66]Open in a new tab Length distribution map of unigenes. x-axis represents unigenes at different length intervals and y-axis represents the number of unigenes in a certain length interval Functional annotation of unigenes Functional annotation was completed for a total of 74,199 unigenes; 15,240 (20.54%), 18,680 (25.18%), 24,685 (33.27%), 27,551 (37.13%), 19,923 (26.85%), and 16,728 (22.54%) unigenes were annotated according to the GO, KEGG, eggNOG, NR, Swiss-Prot, and Pfam databases, respectively (Fig. [67]3). Unigenes successfully assigned to the GO database were classified into 40 subcategories of three main ontologies based on their putative functions: biological process (BP) (35.62%), cellular component (CC) (39.26%), and molecular function (MF) (25.11%) (Fig. [68]4A). Fig. 3. [69]Fig. 3 [70]Open in a new tab Functional annotation of unigenes in seven public databases. Different colours represent different databases and numbers on top of each column represent the number of annotated unigenes in each database Fig. 4. [71]Fig. 4 [72]Open in a new tab GO classification statistical analysis. A: Histogram of GO annotation analysis; vertical axis represents the second-level classification of GO, horizontal axis represents the number of unigenes/transcripts included in this classification, and the three colours on the right represent the three branches of GO (BP, CC, MF). B: Subcategories of the CC group; C: subcategories of the BP group; and D: subcategories of the MF group After refining our analysis of the three main ontologies, 51.16% of the BP unigenes were functionally annotated to cellular processes, 35.1% to metabolic processes, and 29.59% to biological regulation (Fig. [73]4C). For CC, 57.02% of the unigenes were annotated to the cell, 31.16% to organelles, and 24.78% to the membrane (Fig. [74]4B). For MF, 55.85% of the unigenes were annotated to binding and 38.1% to catalytic activity (Fig. [75]4D). Functional annotation of DEGs Pairwise comparison of data from the three RNA libraries was performed. Compared to the baseline, cells stimulated by LPS for 6 h (AD_6h compared to AD_0h) had 1,763 DEGs, of which 1,503 were up-regulated and 260 were down-regulated, whereas cells stimulated by LPS for 12 h (AD_12h compared to AD_0h) had 4,698 DEGs, of which 4,413 were up-regulated and 285 were down-regulated. Compared to 12-h LPS exposure, 6-h LPS exposure (AD_6h compared to AD_12h) resulted in 2,100 DEGs, of which 1,724 were up-regulated and 376 were down-regulated (Fig. [76]5). Intracellular gene expression changed when the spleen-derived macrophages of CGS were stimulated by LPS, and the changes were more obvious as exposure time increased. Fig. 5. Fig. 5 [77]Open in a new tab Differential expression analysis. Horizontal axis represents the different comparison groups; vertical axis represents the corresponding up/down-regulated gene/transcript number; red indicates up-regulation; blue indicates down-regulation KEGG annotation analysis was performed to gain insights into the metabolic pathways associated with these DEGs, as well as their functions. The results showed that these DEGs were mainly annotated into the following six categories: metabolism, genetic information processing, environmental information processing, cellular processes, organismal systems, and human diseases. Among the different stimulation time comparisons, human diseases accounted for the most annotated genes (976, 1113 and 1817 DEGs), indicating that the CGS spleen-derived macrophages showed a disease-infection type of response (e.g., immune activation) after LPS stimulation. In the subclass comparisons, signal transduction accounted for the most annotated genes (339, 383 and 637 DEGs), revealing that macrophages were activated and signal transduction was enhanced after 6 h of LPS stimulation. Interestingly, the number of DEGs annotated to infectious diseases (bacterial) and immune system were decreased slightly after a small peak at 6 h, and then continued to increase (Fig. [78]6). These results indicated that macrophages were activated in the immune response to LPS and that the increase in the degree of activation has a rising pattern of multiple peaks with stimulation time. Fig. 6. [79]Fig. 6 [80]Open in a new tab Comparison of KEGG classification statistical analysis. Vertical axis represents the name of the KEGG metabolic pathway, horizontal axis represents the number of unigenes/transcripts annotated to the pathway; and different colours represent different categories into which the KEGG metabolic pathway can be divided. A: Comparison of 0 h and 6 h of LPS stimulation; B: Comparison of 6 h and 12 h of LPS stimulation; C: Comparison of 0 h and 12 h of LPS stimulation KEGG enrichment analyses were performed to understand the biological relationships of the DEGs. The results showed that the up-regulated genes after LPS stimulation were enriched in 244 signalling pathways. The top 20 results with significant enrichment effects were determined according to the criterion padjust < 0.05. Among them, the NF-κB signalling pathway had the largest enrichment factor, containing 77 DEGs (Fig. [81]7). The results indicated that the 12-h LPS stimulation of spleen-derived macrophages activated the NF-кB signalling pathway to induce the transcription of downstream cytokines. The signalling pathways were mapped according to gene activation in Fig. [82]8. Fig. 7. [83]Fig. 7 [84]Open in a new tab KEGG pathway enrichment analysis. Horizontal axis shows the enrichment rate; vertical axis is the − log10 adjusted p-value; each bubble in the picture represents a KEGG pathway; bubble size is proportional to the number of unigenes/transcripts enriched by the KEGG pathway; and red line and black circle indicates the most significant enrichment pathway Fig. 8. [85]Fig. 8 [86]Open in a new tab NF-kappaB signalling pathway. Genes with a green background were detected in this sequencing; genes with red boxes were significantly up-regulated; genes with black boxes showed no significant changes in expression after LPS stimulation; and genes with a white background were not detected in this sequencing Gene expression analyses We speculated that the NF-κB signalling pathway was activated after 12 h of macrophage stimulation by LPS and relied on TRIF/TRAM for downstream signalling, which ultimately led to the translocation of p50/p65 into the nucleus to promote the transcription of IL-8, IL-6, MIP-1α, and MIP-1β. To test this hypothesis, 10 genes potentially involved in NF-κB signalling (MyD88, TRIF, TRAF6, TAK1, TAB3, p56, IL-8, TNF-α, IFN-α, and IL-1β) were selected for qPCR validation. The results revealed no significant change in MyD88 expression. However, the expression of TRIF increased significantly after 12 h of LPS stimulation, showing an approximately four-fold increase from that in the control group. Together, these results suggest that the NF-κB signalling pathway was activated through TRIF rather than through MyD88. Expression of the downstream factors TRAF6, TAK1, TAB3, and p65 was significantly increased (394-, 16-, 6.7-, and 28.7-fold, respectively) (Fig. [87]9). The expression of pro-inflammatory factors IL-1β and TNF-α was not increased by 12 h of LPS stimulation, whereas the expression of IL-8 and IFN-α was significantly up-regulated (2.6- and 12.2-fold, respectively), suggesting that pro-inflammatory responses in CGS macrophages were predominantly mediated by IL-8 and IFN-α. Fig. 9. [88]Fig. 9 [89]Open in a new tab qPCR verification of 10 genes related to immunology that were dependent on the non-classical MyD88 pathway. Ctrl: control group without LPS stimulation; LPS: experimental group treated with 100 ng/mL LPS for 12 h (n = 3; * p < 0.05, ** 0.01 < p < 0.05, *** p < 0.01) Discussion Macrophages are an important part of the innate immune system, with functions of antigen presentation, phagocytosis, foreign substance-killing, autoimmune initiation, and anti-tumour activity [[90]18]. Currently, the Percoll density gradient centrifugation method is primarily used to obtain macrophages from aquatic animals, although neutrophils and lymphocytes may also be present. Macrophage isolation and culture methods have previously been applied to rainbow trout [[91]19], spotted wolffish [[92]20], carp [[93]21], shark [[94]22], and zebrafish [[95]23]. Adhesion is an important marker to distinguish macrophages from other cells; thus, the obtained cells should be purified by differential adhesion after isolation and culture [[96]24]. At present, no researchers have reported methods for separating CGS splenic macrophages. Therefore, in this study, we used multiple density gradients of 30–60% Percoll for pre-experiments, and found that splenic macrophages from CGS were more easily obtained with 30–40% Percoll isolation solution. The macrophages were round or oval in shape, and half of them were attached to the plate wall. After 1–2 days of culture, cluster aggregation and good growth were observed. In addition, we stained the cells with Trypan Blue and macrophage surface-specific antibody F4/80 [[97]25], and the rate of immunofluorescence positivity was more than 96%. Therefore, we successfully developed an operable and pure culture system to obtain spleen macrophages from CGS, providing an experimental method for studying the interactions between pathogenic microorganisms and the immune systems of amphibians, especially caudates. Innate immunity is a complex biological process [[98]26]. Macrophages express various PRRs when pathogenic microorganisms invade the body, including TLRs, NOD-like receptors [[99]27, [100]28], C-type lectin receptors [[101]29], RIG-I-like receptors [[102]30, [103]31], and AIM2-like receptors [[104]32], with the TLR family being the most widely studied. In response to different pathogenic microorganisms, TLRs activate their corresponding downstream receptors to initiate an immune response. In this study, we compared the transcriptomes of splenic macrophages from CGS exposed to LPS for 12 h and identified 4,698 DEGs, including 4,413 up-regulated genes and 285 down-regulated genes. According to GO enrichment analysis, the DEGs were enriched in cellular processes, metabolic processes, cellular components, the response to stimuli, the developmental process, and the immune system process. According to KEGG enrichment analysis, DEGs were mainly annotated to signal transduction, the NF-кB signalling pathway, Fc gamma R-mediated phagocytosis, Th17 cell differentiation, and other pathways. Further analysis showed that the TLR–NF-кB signalling pathway was highly enriched. However, TLR4, the traditional recognition receptor for LPS in mammals, was not annotated, indicating differences in TLR ligand recognition between species. TLR4 in zebrafish is also unresponsive to LPS stimulation, which has been attributed to a lack of LPS recognition ability by TLR4 because of the absence of a corresponding TLR4 extracellular region, rather than a change in TIR domain signalling ability [[105]13]. Other studies have suggested that zebrafish are tolerant to LPS, which may be because the N-terminal end of the adaptor protein, an molecule upstream of MyD88 in zebrafish TLR4 signalling, is 105 amino acids larger than that in mammals [[106]33]. This indicates that lower vertebrates, especially fish and amphibians, may be tolerant of LPS. Compared to mammals, extremely high concentrations of LPS are required to stimulate immune cells in fish for the desired effect in vitro. In addition, scavenger receptors can recognise LPS instead of TLR4 and participate in the negative regulation of NF-κB in puffer fish [[107]34]. Whether such a phenomenon exists in amphibians remains to be studied. The signal transduction patterns of TLRs differ among species. One major signalling pathway is dependent on the adaptor protein MyD88; that is, this pathway relies on MyD88 and TIRAP to induce expression of the inflammatory response factors IL-1 and TNFα [[108]35]. The other signalling pathway operates independently of MyD88 and depends on another class of adaptor proteins, TRIF and TRAM, to activate NF-кB and antiviral type I IFN responses. Among the TLR family members, most are involved in MyD88-dependent signalling pathways, whereas only TLR3 signals through the MyD88-independent pathway, and only TLR4 uses both signalling pathways [[109]36]. MyD88, TIRAP/Mal, and TRIF adaptor proteins have been found in the zebrafish genome. Zebrafish TLR3 mediates the inflammatory response by inducing IFN-β expression and NF-κB activation through the adaptor protein TRIF. However, Mal does not perform this function. Fluorescence confocal microscopy previously revealed that the subcellular structure of zebrafish TRIF protein localises to the Golgi membrane and retains the ability to induce IFN expression and NF-κB activation [[110]37]. Overexpression of TRIF in grass carp also leads to the up-regulation of IRF7 and IFN-I gene expression during antiviral immune response [[111]38]. In addition, IRF9 negatively regulates TRIF-mediated activation of the NF-κB transcription factor in fish [[112]39]. These results suggest that, in lower vertebrates, TLR family members operate through two signal transduction pathways, similar to those in mammals; however, differences in the main functional regions of TLRs may lead to differences in the selection of adaptor proteins or method of signal activation. In our study, unlike the MyD88-dependent classical pathway and MyD88-independent pathway in mammals, or the MyD88 classical pathway-only antibacterial immune response in poultry, LPS stimulation led to no significant difference in the expression of MyD88 classic pathway-related factors, whereas MyD88-independent pathway-related factors such as TRIF, TRAF6, and TAK1 were significantly up-regulated. This indicates that the TLR signalling pathways that activate inflammatory signals after bacterial infection in amphibians may differ from those in other vertebrate species. Conclusions By isolating splenic macrophages from CGS and applying transcriptome sequencing and qPCR, we revealed activation of the NF-κB signalling pathway and significant up-regulation of the downstream immune-related genes IL-8 and IFN-α after LPS stimulation. We also found that LPS recognition in CGS was dependent on the non-canonical TRIF/TRAF6 pathway rather than on the traditional TLR4-MyD88 pathway. Elucidating these molecular mechanisms will aid research into the recognition receptors on the membrane surfaces of amphibian cells after bacterial stimulation and the key genes and signal transduction pathways involved in innate immune responses. In addition, this study highlights different immune response patterns, providing a reference for understanding the immune evolution of amphibians and other aquatic animals, and has implications for the development of CGS protection measures. Electronic supplementary material Below is the link to the electronic supplementary material. [113]Supplementary Material 1^ (319KB, pdf) [114]12864_2024_10888_MOESM2_ESM.docx^ (70.2KB, docx) Supplementary Material 2: Table S1. Summary of statistics from Illumina sequencing. Table S2. Statistical data of assembly results. Figure S1. Histogram of the macrophage percentage in the spleen of the Chinese giant salamander before and after cell centrifugation. Acknowledgements