AM seeds used in this experiment were purchased from a professional traditional Chinese medicine market in Anguo (Heibei, China). Soilless cultivation was performed in a light incubator by using an improved Hoagland Modified nutrient Salts (Coolaber, China) under a light intensity of 2,000~2,500 lux, a diurnal temperature of 25 °C (16 h), a night temperature of 20 °C (8h). Healthy plants with consistent growth were selected when AM seedlings were 40 d old, and part of them were sampled for leaves, stems, and roots, which were rapidly frozen with liquid nitrogen after collection and then kept at -80°C for qRT-PCR experiments

The AM genomic files used for the study were downloaded from the GPGD website (http://www.gpgenome.com/species/109). TBtools software was used to extract and convert all coding sequences in the AM genomic to protein sequences (Chen et al., 2020). Firstly, CHIs in AM were identified using a Simple Hidden Markov Model (HMM) Search function in TBtools software using HMM mapping of CHI structural domains downloaded from the Pfam database (PF16035, PF16036, and PF02431) (E-value <1 × 10⁻¹⁰). Second, the CHI proteins identified in A. thaliana and G. max were used for BLASTP (E-value <1 × 10⁻⁵) with AM protein sequences to further screen for CHIs. Finally, the final AmCHIs was obtained after removal of redundant sequences. The physicochemical properties of AmCHIs were performed using the ExPASy database (https://www.expasy.org/). Prediction of subcellular localization of AmCHI proteins using the online site Euk-mPLoc (http://www.csbio.sjtu.edu.cn/bioinf/plant/#).

The ClustalX (Thompson et al., 2002) software was employed to conduct sequence alignment analyses of protein from AM, A. thaliana, G. max, L. corniculatus, Solanum lycopersicum, and Oryza sativa derived from the Phytozome (Goodstein et al., 2012). The neighbor-joining approach was employed for phylogenetic analysis using the MEGA (Kumar et al., 2018), and the bootstrap repetition number was 1,000. Landscaping the phylogenetic tree using the EvolView website (https://www.evolgenius.info/evolview-v3/#login).

Genome sequences of Cannabis sativa (GCA_029168945.1), S. lycopersicum (GCA_000188115.4), and Malus domestica (GCF_002114115.1) were obtained from the NCBI database (https://www.ncbi.nlm.nih.gov/). The MCScanX Wrapper function in TBtools was used for gene collinearity analysis, and Advanced Circos was used for inter-and intraspecific collinearity visualization analysis and to label the chromosomal position of the AmCHI genes.

The conserved motifs of AmCHI proteins were identified through the use of MEME Suite (version 5.5.0), with 10 motifs and the rest set as default parameters. Information on phylogenetic trees, motifs, introns, and exon regions was integrated and visualized using the TBtools. A DNA sequence 2000 bp upstream of the AmCHIs coding gene was extracted as promoter region for cis-element prediction using PlantCare database and visualized using TBtools software.

Raw transcriptome data of AM seedlings at different tissues were obtained through sequencing. The number of reads mapped to each gene was calculated using FeatureCounts (v2.0.3) (Liao et al., 2014). The FPKM value of each gene was then calculated based on the gene length, and heatmaps were plotted using TBtools software.

Total RNA extraction from samples using RNA extraction kit. Synthesize the first strand cDNA using the NovoScript^®Plus All-in-one 1st Strand cDNA Synthesis SuperMix (gDNA Purge) (Novoprotein, Shanghai, China) kit. The qRT-PCR system was configured using the StarLighter HP SYBR Green qPCR Mix (Universal) (Beijing Foreverstar Biotech), and upsampling assays were performed using the AriaMx System platform (Agilent Technologies, Hercules, CA, USA). The 18s rRNA gene was employed as an internal reference. The samples were analyzed in triplicate, and the final expression of each AmCHI was determined using the 2^−ΔΔCT method (Livak and Schmittgen, 2001).

AmCHI3, AmCHI4, and AmCHI5 were seamlessly cloned into the pET28a vector carrying the MS Fastcloning MultiS kit plus (Msunflowers Biotech). The constructed recombinant plasmid and the empty vector were transformed into BL21(DE3) Chemically Competent Cell (Weidi Biotechnology). A selection of positive clones was cultivated in liquid medium of LB composition, containing kanamycin (50mg/L) to OD₆₀₀ = 0.6, and 0.5 mmol/L isopropyl-betaD-thiogalactopyranoside (IPTG) was added. Expression was induced at 37 °C, 30 °C, 25 °C, 20 °C, and 16 °C for 8h, respectively. The induced bacterial solution was collected, and the proteins were extracted and purified by His tag protein purification kit. The protein samples were electrophoresed by SDS-PAGE. In order to further identify the protein of AmCHIs, the purified protein was analyzed by Western Blot (WB). The primary antibody utilized was Rabbit Anti-His tag antibody. The secondary antibody employed was Goat Anti-Rabbit IgG H&L/HRP, and the PVDF membrane was treated with BiossECL Plus WB Substrate kit (Bioss, Beijing, China). Finally, the Amersham Imager 600 (Cytiva, USA) was used for photo observation. The reagents used in the WB experiments were brought grom Boaosen Biotechnology Co., LTD (Beijing, China).

The activity of the AmCHI enzyme was quantified using a purified recombinant AmCHI protein (10ug/15ug), tris-HCl buffer (1M, pH 7.6), and 40 mmol naringenin chalone or isoliquiritigenin (Sichuan Vicky Biotechnology Co., Ltd., Sichuan, China) as substrate in 200µl system. The reaction conditions of the system with naringenin chalcone as substrate were vortexed for 10s at room temperature, subsequently, the sample was extracted twice with ethyl acetate, after which the solvent was evaporated using a nitrogen blower. Residues were dissolved in methanol and analyzed by UPLC. The reaction time of the system with Isoliquiritigenin as substrate was 37°C for 1h, and the rest was the same as above. Denatured proteins was used in the control group. The mobile phase was composed of acetonitrile (A) and water (B). Naringenin chalcone and naringenin were detected using the following gradient procedure was employed: 0~30 min, 45% A. Liquiritigenin and isoliquiritigenin were detected using the following gradient procedure was employed: 0~20 min, 30~80% A; 20~25 min, 100% A. The detection wavelength was 270 nm.

The sense oligonucleotides (sODN) sequence aligned to the AmCHI3 gene fragment and the reverse complementary antisense oligonucleotides (AsODN) sequence were designed using the online software Soligo (https://sfold.wadsworth.org/cgi-bin/index.pl). In order to prevent the primer from being degraded by nucleases, it is necessary to perform thiol modification on three bases at each end of the primer. The sODN primer and AsODN primer for the AmCHI3 gene were diluted to 5 µM using dd H2O, and then Astragalus seedlings grown for 14 d were clipped at the roots and inserted into the diluted primers, and incubated for 48 h. The samples treated with the sODN primer served as the control group. For each treatment, ten plants with uniform growth were selected, and the experiment was replicated three times. The leaves of the treated plants were collected, rapidly frozen in liquid nitrogen, and then stored at -80 °C for subsequent molecular experiments and metabolic assays.

The leaves of AM that had been treated with the antisense oligodeoxynucleotide (AsODN) and sense oligodeoxynucleotide (sODN) primers were completely ground into a fine powder in liquid nitrogen. Subsequently, the powdered sample was lyophilized using a freeze dryer. Precisely 100 mg of the powder was weighed out and then 1200 microliters of 70% methanol was added to it. The mixture was sonicated in an ice-water bath for one hour. The sample was centrifuged at 12,000 rpm for 10 minutes. Subsequently, the supernatant was carefully extracted and filtered through a 0.22-μm syringe-driven membrane filter for LC-MS analysis. The samples were detected by LCMS-9030 (SHIMADZU, Japan) equipped with a Ultimate^® UHPLC XB-C18 column (1.8μm, 2.1 by 100mm). The mobile phase consisted of water containing 0.1% formic acid (designated as A) and acetonitrile (designated as B). The following gradient procedure was employed: 0~2 min, 95~60% A; 2~8 min, 60~30% A; 9~10 min, 5% A; 10~11 min, 5~95% A; 11~14 min, 95% A. The flow rate was set at 0.3 ml/min, and the column temperature was maintained at 40°C. The isoflavone metabolites including calycosin (CA), calycosin-7-glucoside (CAG) and formononetin (FO) (Shanghai yuanye Bio-Technology).

Eight AmCHI genes were found in the AM genomic sequence, and the detailed information and physicochemical properties of all encoded proteins were presented in Supplementary Table 1 . They were named AmCHI1–AmCHI8 based on their chromosomal locations. The proteins encoded by AmCHIs ranged from 177 (AmCHI1) to 498 (AmCHI2) amino acids and corresponding MWs ranged from 20020.76 (AmCHI1) to 54944.30 Da (AmCHI2), and their theoretical pIs ranged from 4.94 (AmCHI4) to 9.36 (AmCHI8). The subcellular localization results showed that AmCHI1, AmCHI2, AmCHI3, AmCHI5, AmCHI6, AmCHI7, AmCHI8 proteins were localized in the chloroplast, and AmCHI4 protein was localized in the cytoplasm.

To more comprehensively investigate the evolutionary origin and functional diversity of CHI genes in plants, this study constructed a phylogenetic tree using the Neighbor-Joining (NJ) method. ( Supplementary Table 2 ). The results were consistent with the expectation that the AmCHIs could be categorized into four types ( Figure 1 ). Types I and II had one each, type III had the most AmCHIs with four, and type IV had two. The CHI genes in AM were highly identical to those from G. max and L. corniculatus, which are also members of the Fabaceae family. Notably, AmCHI3 was a unique type II gene in the Fabaceae family that may play a critical role in flavonoid biosynthesis in AM.

Chromosomal localization maps were created based on information about the location of AmCHI genes on chromosomes of AM ( Figure 2 ). The results showed that these eight AmCHI genes were distributed unevenly on 5 chromosomes. AmCHIs had the highest density of three on chromosome 2, followed by two on chromosome 6, and one on chromosomes 1, 3, and 9, respectively. No CHI gene was produced via segmental duplication in the AmCHI gene family ( Figure 2A ).

In this study, we investigated the genetic relationships and evolutionary trends of AmCHI family using collinearity analysis of AM, M. domestica, C. sativa, and S. lycopersicum. M. domestica had a closer kinship with AM, with five AmCHIs mapped to nine MdCHI genes ( Figure 2B ). In contrast, C. sativa and S. lycopersicum had only three AmCHIs mapped to the corresponding three CsCHIs and three SlCHIs ( Figure 2C ), which were identical (AmCHI2, AmCHI5, and AmCHI7) ( Figure 2D ). Therefore, it is possible that these genes have similar functions.

Conserved motifs are of paramount significance in the processes of gene family identification and classification. Ten motifs were found among the eight AmCHIs and named motif 1–motif 10. The results showed that each AmCHI contained 4–9 motifs with a width range of 8–30 amino acids ( Figure 3B ). Motif 1 and motif 3 were present in all heat shock factor proteins in C. sativa. The same types may have the same conserved motifs, and the composition of the conserved motifs in AmCHIs further supported the results of the phylogenetic classification ( Figure 3A ).

The structural diversity of the CHIs was revealed by analyzing the exon–intron structure of the AmCHI genes. The amount of introns in the AmCHIs varies from 2 to 12, with most AmCHI genes consisting of 2–4 introns ( Figure 3C ). Within the same type, most members had similar exon–intron numbers and arrangements, suggesting that closely related AmCHI genes may have similar structures.

For investigating the potential functions of the eight CHIs in AM, we characterized and classified the cis-elements in promoter regions of these genes. The analysis revealed that a total of 209 cis-elements were screened from the promoter regions of all AmCHI genes. The identified cis-elements were classified into three functional categories, including nine growth-related, five hormone-related, and three stress-related responsive elements ( Figure 4 ) ( Supplementary Table 3 ). All eight AmmCHI genes contained at least one cis-element from these three categories, suggesting a critical role of the CHI gene family in all stages of AMM growth, development, and response to various stresses. We observed that the promoter regions contained binding sites for MYB transcription factors and were involved in the light and drought responses, suggesting that under certain circumstances, the AmCHI gene may be regulated by the AmMYB gene.

Gene expression pattern analysis can provide supporting evidence for mining gene functions. To elucidate the functions of AmCHIs during the seedling growth and secondary metabolism, we performed the transcriptome data from the seedling leaves, stems, and roots and extracted the FPKM values of all CHI genes ( Figure 5A ). Based on the heat map, AmCHI2, AmCHI3, AmCHI4, AmCHI5, and AmCHI8 expressed in all tissues (FPKM>0.5) ( Supplementary Table 4 ), with AmCHI4 being the most highly expressed. Conversely, the AmCHI7 transcript was not detected in any of the tissues. The AmCHI6 gene showed an expression specific to leaf tissue. AmCHI4 of type I and AmCHI3 of type II, which may have catalytic functions, and AmCHI5, which is specifically expressed in roots, were selected for qRT-PCR to verify the reliability of the RNA-seq ( Figure 5B ) ( Supplementary Table 5 ). The analysis indicated that the expression trends of the AmCHI3, AmCHI4, and AmCHI5 in the leaves, stems, and roots of AM plants were consistent with that of the RNA-seq.

To verify whether the AmCHI3, AmCHI4 and AmCHI5 genes encode functional CHI enzymes, recombinant AmCHI proteins were cloned from AM and inserted into a pET28a expression vectors with a His tag. Heterologous expression of recombinant proteins of AmCHI was attempted in BL21 (DE3) Chemically Competent Cell induced by 0.5 mM IPTG for 8 h at different temperatures (37, 30, 25, 20, and 16°C). The crude proteins were extracted and purified using a protein extraction kit. SDS-PAGE analysis showed that the crude protein extracted from E. coli had a clear, highly expressed protein at 23 kDa ( Supplementary Figure 1A ). After the purification of the crude protein, a clear single band at 23 kDa was observed ( Supplementary Figure 1B ), judged to be the recombinant AmCHI protein, and the size of the protein was in agreement with the predicted value. The recombinant AmCHI3 protein showed the highest expression at 30°C, and recombinant AmCHI4 and AmCHI5 showed the highest expression at 37°C. The recombinant AmCHI3, AmCHI4, and AmCHI5 proteins were expressed in soluble form in E. coli, and they were constitutively expressed at 37 °C with a small amount without IPTG induction. The analysis of the WB indicated that the product induced by the recombinant AmCHI expression strain exhibited His-tagged bands, detected using a His-specific antibody, at approximately 23 kDa, which is the band size of the expected the recombinant protein. ( Figure 6A ). The results showed that the AmCHI recombinant proteins had adequate reactogenicity and, thus, enzymatic activity compliant proteins were obtained.

In order to verify that AmCHI encodes functional CHI enzymes, purified AmCHI3, AmCHI4, and AmCHI5 proteins were used for in vitro enzymatic reactions. The findings indicated that Type I AmCHI4 solely catalyzes the conversion from naringenin chalcone to naringenin. Additionally, increasing the dosage of AmCHI4 can enhance the catalysis of a larger quantity of naringenin chalcone into naringenin. ( Figure 6B ). The type II AmCHI3 protein can not only catalyze the reaction of naringin chalcone to naringin, but also the reaction of isoliquiritigenin to Liquiritigenin. Increasing the protein content of AmCHI3 can promote the conversion of more isoliquiritigenin to Liquiritigenin. However, increasing the protein content of AmCHI3 cannot catalyze the conversion of naringin chalcone to more naringin ( Figure 6C ). The typeIII AmCHI5 protein does not have any catalytic function, which is consistent with previous reports.

AsODN is a technology to inhibit gene expression by specifically binding to the mRNA sequence of the target gene ( Figure 7A ). In this study, AmCHI3 gene was tested ( Supplementary Table 6 ). The secondary structure of the AmCHI3 gene was analyzed by the online analysis software Soligo, and the optimal specific primers were designed based on the oligo binding energy (-6.4 kcal/mol) and nucleotide composition (50% GC). The primers were diluted to 5µM using ddH₂O, and the root-cut AM seedlings were inserted into the primers, and samples were collected and tested after 2 d of incubation in the dark ( Figure 7B ). The results indicated that, when compared with the control (sODN), treatment with the AsODN primer led to a significant reduction in the expression level of the AmCHI3 gene ( Figure 7C ). Concurrently, the contents of CA, CAG and FO also decreased significantly ( Figure 7D ). This is contrary to the results of overexpression of AmCHI3 gene, indicating that AmCHI3 gene is a positively related enzyme gene for the synthesis of isoflavone compounds in AM.