To identify full-length UDPGP genes in different species, we searched the UDPGP genes using HMMER software (Potter et al., 2018). A total of 454 full-length primary protein sequences were identified in 76 organisms, including plants (58 species: 404 sequences), chlorophyte (seven species: 21 sequences), animals (six species: 16 sequences), fungi (three species: 11 sequences), kinetoplastid (one species: two sequences), and bacteria (one species: 0 sequence). The detailed information is in Supplementary Table 2.

To study the origin and evolutionary history of the UDPGP genes in different species, we constructed a phylogenetic tree with the Maximum Likelihood (ML) method. The phylogenetic tree with gene names and bootstrap values (UFBoot/SH-aLRT/aBayes) is shown in Supplementary Figure 1. The topology of the ML tree showed that UDPGP genes were clustered into three clades, which were defined as UAP, UGP, and USP. Besides, The UGP clade could be divided into two subgroups, UGP-A and UGP-B. These results were similar to the OrthoFinder analysis results (Supplementary Table 3). The orthogroup analysis divided the gene family into seven orthogroups, and the first four largest orthogroups, OG1, OG2, OG3, and OG4, were corresponding with the subgroup UGP-A, UAP, USP, and UGP-B, respectively, while three minor orthogroups (OG5, OG6, and OG7) contained UAP and UGP-A members. Notably, no homologous UGP-B and USP members were identified in animals, fungi, and kinetoplastid. Furthermore, Escherichia coli does not contain any homologous genes belonging to UDPGP in this study, which is corresponding to that of previous studies (Kleczkowski et al., 2004; Führing et al., 2013). Then, we analyzed the distribution of each subgroup in the 58 plants (Figure 1 and Supplementary Table 2). The result showed that the numbers of UDPGP genes significantly varied among the major lineages of plants, ranging from the two members detected in chlorophytes including Micromonas sp. RCC299 and Micromonas pusilla to the 19 sequences identified in Gossypium raimondii. In general, the numbers of UAP and USP genes were relatively stable among the plants, ranging from 0 to 5 (UAP) and 0 to 4 (USP), while the members in UGP (UGP-A and UGP-B) varied greatly, spanning from 0 to 14 (G. raimondii). Specifically, we did not find any full-length UAP genes in Kalanchoe fedtschenkoi or USP genes in Volvox carteri. In addition, Micromonas sp. RCC299, Micromonas pusilla, and Ostreococcus sp. Lucimarinus had no UGP (UGP-A or UGP-B) genes (Figure 1 and Supplementary Table 2).

The calculated isoelectric point (pI) and molecular weight (MW) of each UDPGP sequence are shown in Supplementary Table 2. The average MWs of subgroups UAP, UGP-A, UGP-B, and USP were 56,234.91, 47,351.92, 93,678.22, and 66,044.22 Da, respectively. The results showed remarkable differences among the four subgroups, with subgroup UGP-A containing the smallest MW and subgroup UGP-B containing the heaviest MW. The pI for the UDPGP genes ranged from 4.33 to 9.99, implying a wide range of activity in microcellular environments. The subgroup UAP had more acidified pI with an average of 6.00 compared with the other subgroups, suggesting possible functional divergence of UAP (Supplementary Table 2). All these showed the conserved physicochemical features in UDPGP genes.

The loss or gain of introns leads to different gene structures, makes genes more complex, and acts as the foundation of gene evolution (Fedorova and Fedorov, 2003). Previous studies have shown that introns play an important biological role in regulating gene expression (Castillo-Davis et al., 2002; Le Hir et al., 2003). In the current study, we constructed a phylogenetic tree using protein sequences from 16 representative plant species (Figure 2) and analyzed the exon–intron structures (Supplementary Figure 2). The exon–intron organization of UDPGP genes was significantly different among the four subgroups, while the structure within each subgroup was conserved, indicating the conservative characteristics in the UDPGP gene family (Supplementary Figure 2). Besides, the numbers of exons in each subgroup are similar, ranging from 15 in UAP to 21 in UGP-A typically.

In the evolutionary history of the UDPGP gene family, the number of members was relatively stable, resulting in an elementary gene family. As shown in Figure 3A, six of 12 chromosomes contain UDPGP genes in Oryza sativa. Three OsUGP genes (OsUGP1, OsUGP2, OsUGP3) posited on chromosome 9, chromosome 2, chromosome 3, respectively. Synteny analysis of the UDPGP family in rice showed that collinearity blocks between UDPGP members only existed in clade UAP (OsUAP1 and OsUAP2) and located on chromosome 8 and chromosome 4. Furthermore, the genome of O. sativa only contains a single OsUSP gene located on chromosome 6. The similarity of OsUAP1 and OsUAP2 indicated that these two genes originated from duplication. Among the five chromosomes in Arabidopsis, only chromosome 4 carried no UDPGP genes, and others had one or two UDPGP genes (Figure 3B). Chromosome 3 contained two ATUGP genes (ATUGP1 and ATUGP3), and chromosome 5 contained ATUSP and ATUGP2 genes at both arms. Besides, chromosome 1 and chromosome 2 separately hold GlcNA.UT1 (ATUAP1) and GlcNA.UT2 (ATUAP2) genes. In addition, we also detected duplication events, which were similar to O. sativa, in A. thaliana. The ATUAP1 and ATUAP2, ATUGP1 and ATUGP2 were two paired collinearity genes detected by McScanX, while no tandem gene pairs were identified in both O. sativa and A. thaliana. All these results showed that the members of UDPGP genes were stable and the expansions of the UDPGP gene family were only caused by segmental duplication.

The transcriptomic profile reflects the tissue-specific function. RNA sequencing data from Rice Expression Database (RED) were downloaded to analyze the UDPGP genes in O. sativa expression pattern in eight (anther, callus, leaf, panicle, pistil, root, seed, shoot) different tissues (Figure 4). The results showed that the expression level of LOC_Os09g38030 (OsUGP1) was significantly higher than that of other gene members in the UDPGP family. Anther during flowering is the highest expressing tissue detected with OsUGP1. The LOC_Os02g02560 (OsUGP2) and LOC_Os01g15910 (OsUGP3) also belonged to the UGP clade in rice, whose expression levels were greatly lower than that in OsUGP1. For OsUGP2, anther (before and during flowering) and panicle (7 days before heading and 7 days after flowering tissues) in specific development stages exhibited higher expression levels than other tissues. Notably, OsUGP3 showed almost no expression in the seven tissues. Besides, for the UAP clade LOC_Os04g52370 (OsUAP2) and LOC_Os08g10600 (OsUAP1), the expression patterns were similar to each other and the expression level of OsUAP2 gene was slightly higher than OsUAP1 in some specific tissues, such as root and shoot from 7-day seedlings. In addition, for the OsUSP gene, a single member of the rice USP clade, the expression profile was ubiquitous in all tissues with a relatively low level. These results suggest that rice UDPGP genes can be expressed in various tissues to perform their roles with different expression levels as needed.

Plant hormones, abscisic acid (ABA) and jasmonic acid (JA), are important regulatory factors involved in various biological processes, including cell death (Sreenivasulu et al., 2006; Reinbothe et al., 2009). The expression patterns of rice UDPGP genes under ABA and JA treatments are similar in both shoot (Figure 5A) and root (Figure 5B) tissues. Importantly, OsUAP1, OsUGP3, and OsUSP were generally upregulated in shoot tissue under the ABA and JA stress, and OsUSP was also upregulated in root tissue under the ABA and JA stress.

Cadmium (Cd) contamination has become a big issue in food safety, especially in rice (Liu et al., 2020). Under the very low Cd and low Cd treatments, the expression level of all six rice UDPGP genes were uniformly upregulated in root tissue except for OsUGP3 (Figure 5D). The three genes (OsUAP1, OsUGP1, and OsUSP) also showed increased expression trend in shoot tissue (Figure 5C) under the same treatments. When stressed by cold, the four rice UDPGP genes (OsUGP1, OsUGP2, OsUGP3, and OsUSP) were raised both in shoot (Figure 5E) and root (Figure 5F).

Moreover, the expressions of UDPGP genes in O. sativa under abiotic (salinity, dry, flood, and osmotic) and non-metal [phosphate (P)] (Supplementary Figure 3) treatments were also investigated. In general, OsUAP1 and OsUSP were found to be upregulated under dry, flood, and osmotic stresses of shoot, and OsUSP was also upregulated under flood and osmotic stresses of root. Moreover, OsUAP2 and OsUGP2 separately showed upregulated expression under dry stress of shoot and osmotic stress of root. No significant expression changes were identified for UDPGP genes under salinity and P treatment in both shoot and root tissues.

The expression data above together implied that rice UDPGP genes may be involved in environmental endurance.

Phylogenetic analysis of the UDPGP homologs yielded three big different clades and four distinct subgroups (Figure 6A). In previous reports, UGP-A clade contained a nucleotide-binding loop (NB-loop) and substrate binding loop (SB-loop) at the active center and an insertion loop (I-loop) at the C-terminus site (Steiner et al., 2007; Führing et al., 2015; Chi et al., 2016; Figure 6B). For the UAP clade, only NB-loop and I-loop were reported (Peneff et al., 2001; Figure 6B). In the USP clade, NB-loop and SB-loop were described (Dickmanns et al., 2011; Figure 6B). Based on the NB-loop reported by Geisler et al. (2004), we identified the differences of the NB-loop in the three clades through multiple sequence alignment (MSA) (Figure 6C). The results showed that NB-loop was diverse in different subgroups. In general, all the NB-loops contained a start with “GGxG.”

Then, the motifs from 16 represented species were analyzed based on the primary protein sequences using MEME software. As shown in Supplementary Figure 4, motifs 1, 4, 13, and 14 existed in all subgroups, while motifs 20, 18, and 15 were unique sequences at the N-terminus in UAP, USP, and UGP-A subgroups, respectively. Besides, motif 14 was a common motif at the C-terminus in all UDPGP clades except the UAP clade, which carried extra motifs 5, 9, and 17. Furthermore, motifs 6, 16, and 19 were only located in the USP clade, while the UGP-A subgroup contained distinctive motifs 7, 8, 10, 12, and 14. In our study, we defined a motif 8 with “NPSIELGPEFKKVGNFLSRFKSIPSIVELDSLKVSGDVWFG” sequence, which overlapped with a previously reported motif “RFKSIPSI,” and this motif was proven to be an essential element for the phosphorylation and binding with 14-3-3 protein (Toroser et al., 1998; Cotrim et al., 2018). Moreover, motif 2 “KLAVLLLAGGLGTRLGCTGPK” displayed in all clades except UGP-B; this motif was corresponding to that of a previous study that reported a high essential signature motif “LX2GXGTX6PK” (Mio et al., 1998) and three amino acids (G, R, and K) were proven vital for the activity of the enzyme. All the above contributed to the form of the UDPGP gene family and diverged it into three large clades (UAP, USP, and UGP). Thus, the conservation of these additional motifs in their respective clades may play a key role in their functional specificity.

In OsUAP1 mutant osuap1, guanine (G) was replaced by thymine (T) at the position of 712 bp in the CDS, resulting in the 238th amino acid changed from glycine (Gly) to cysteine (Cys) and thus lost function of UAP enzymatic activities for the OsUAP1 protein (Wang et al., 2015; Supplementary Figure 5). This key amino acid site is located in motif 2, which exhibited in all clades except UGP-B, and this site might be involved in the uridine recognition region (Wang-Gillam et al., 2000). Some single amino acid and fragment deletion mutations were performed to study the key amino acid of UDPGPs. In the UAP clade, 14 site-directed mutants were used to study the key amino acid affecting the catalytic activity, including Giardia intestinalis (Mok and Edwards, 2005), Homo sapiens (Wang-Gillam et al., 2000; Peneff et al., 2001), Saccharomyces cerevisiae (Mio et al., 1998), and Aspergillus fumigatus (Raimi et al., 2020; Supplementary Figure 6 and Supplementary Table 4). In G. intestinalis (Mok and Edwards, 2005), when G108 (corresponding to G125 in OsUAP1) and G210 (corresponding to G236 in OsUAP1) were substituted by alanine, the enzymatic activity of UAP was significantly reduced. In H. sapiens (Wang-Gillam et al., 2000; Peneff et al., 2001), R115, P220, G222, G224, Y227, and G111 were replaced by alanine to study the catalytic property. The results showed that R115, G222, G224, and G111 were key amino acids to maintain the activity of the enzyme, corresponding to R129, G236, G238, and G125 in OsUAP1. Besides, the G111 and R115 were located in the NB-loop as well as motif 2, indicating the importance of the sequence. In Saccharomyces cerevisiae (Mio et al., 1998), three amino acids were studied, including G112, R116, and K123, which equated to G125, R129, and K136 in OsUAP1. The enzymatic activity was severely diminished when these three sites were replaced by alanine. In A. fumigatus, five site mutations were performed to study the key amino acids in AfUAP gene (Raimi et al., 2020). Three (K148, Y330, and K437) of them showed a key role in impacting Km value of UTP or GlcNAc-1P, corresponding respectively to K136, Y320, and K416 in OsUAP1 in rice. Several mutants in Hordeum vulgare (Martz et al., 2002; Meng et al., 2009a), Solanum tuberosum (Katsube et al., 1991), Cricetulus griseus (Flores-Díaz et al., 1997), and H. sapiens (Chang et al., 1996), including single amino acid and fragment deletion, were used to study the key amino acids of UGP (Supplementary Figure 7). In H. vulgare (Martz et al., 2002; Meng et al., 2009a), G91, C99, L117, I118, V119, K127, K128, L135, L136, L137, Y192, and K260 were key amino acids to provide the activity of proteins, which corresponded to G88, C96, L114, I115, V116, K124, K125, L132, L133, L134, Y189, and K257 in OsUGP1. And when K183, K332, K405 were replaced by alanine, the activity was not greatly impacted in HvUGP (HORVU5Hr1G087810.2). These three sites corresponded to the K180, K329, and K402 in OsUGP1. In addition, the authors trimmed different lengths of N and C terminals to study the key role in regulating the activity. Results showed that the majority (Ncut-21, Ncut-27, Ccut-8, Ccut-67, and Ccut-101) largely reduced the activity except for the cut 32 amino acids in C terminal, indicating the importance of N and C terminals for catalytic activity. In S. tuberosum (Katsube et al., 1991), five site mutants were studied through substituting with glutamine, including K263, K329, K367, K409, and K410. Among them, K263 and K367 were two important sites to keep the activity of UGP in S. tuberosum, which corresponded to K257 and K361 in OsUGP1. When G115 was replaced by aspartic acid (D) in Cricetulus griseus (Flores-Díaz et al., 1997), the activity of UGP was largely affected, which corresponded to G88 in OsUGP1. In H. sapiens (Chang et al., 1996), eight site mutants were developed including C123, W218, H266, W333, R389, R391, R422, and R445. Among them, C123, W333, R389, R391, and R422, corresponding to C96, W299, R354, R356, and R387 in OsUGP, were key amino acids for sustaining the activity. Among these, G115 in CgUAP was mapped to NB-loop, which corresponded to motif 2 assigned by MEME. And K236 in StUGP was a key amino acid located in SB-loop, which was in agreement with motif 3. The rest of the amino acids were located in motifs 4, 7, 8, and 11. Furthermore, no visible differences of activity were described when K183, K332, and K405 in HvUGP (Martz et al., 2002; Meng et al., 2009a) and W218, H266, and R445 in HsUGP (Chang et al., 1996) were substituted by other amino acids, indicating that these sites were not key amino acids in UGP protein. In addition, these four amino acids were not located in any motif assigned by MEME. The only one study reported on the key amino acid of USP was from Leishmania major (Prakash et al., 2019). V330, F383, and V199 were key sites for the activity of USP in LsUSP, corresponding to Q353, F405, and V239 in OsUSP (Supplementary Table 4 and Supplementary Figure 8). These sites were located in motifs 13, 16, and 19. Moreover, we compared the key amino acids, important loops (NB-loop, SB-loop, and I-loop), and motifs from MEME. The results showed that the amino acids in the loops or the motifs more likely contributed to the catalytic activity of UDPGPs.

Previous studies showed that the mutants of UDPGP genes lead to aborting of enzymatic activity, resulting in cell death (Tresse et al., 2008). Besides, the family numbers of UAP and USP clades are conserved, and many species contained only two UAP genes or a single USP gene. To address the question if the amino acid sequences are conserved in UDPGP genes, we scanned the protein sequences of UDPGPs in 4,726 rice accessions in RiceVarMap¹. A total of 4, 2, 5, 2, 23, and 7 gene variants for OsUAP1, OsUAP2, OsUGP1, OsUGP2, OsUGP3, and OsUSP were identified, respectively (Table 1). Furthermore, 32 of 43 variation sites located in the non-motif region, and only 11 posited on the motif sites. Among the 11 variants, only two of them showed a relatively higher proportion in nature. The 502nd amino acid (in Motif 8) changed from Ser to Asn in OsUGP3 with 40.40% and the 462nd amino acid changed (in Motif 6) from Ser to Ala in OsUSP with 28.30% (Table 1). In the previous study, one amino acid in Motif 8 was mutant in wheat, and the result showed it only slightly lowered the enzymatic activity (Meng et al., 2009a). In addition, the number of OsUGP3 variants was greater than other members (Table 1), but the expression levels of OsUGP3 were much lower than other members (Figure 4), which may indicate that this gene is under evolution. All above suggest that most UDPGP genes are conserved in the evolution.

Alternative splicing events are important posttranscriptional regulatory mechanisms, which could reduce the enzyme activities or completely abolish the activity (Kelemen et al., 2013). Here we studied the alternative splicing events of the UDPGP gene family in O. sativa. Four (OsUAP1, OsUAP2, OsUSP, and OsUGP1) of six UDPGP genes carried more than one transcript sequence from the rice annotation file. The OsUAP1 gene had three different types of mRNA sequences, OsUAP1.1, OsUAP1.2, and OsUAP1.3. The OsUAP1.1 transcript was the longest form with 489 amino acids, while the OsUAP1.2 transcript lacked exon sequence near 5’ untranslated region (5’ UTR), and the OsUAP1.3 transcript lost exons at 3’ UTR (Figure 7A). The OsUAP2 gene had two similar transcripts with a difference of six nucleotides at the end of the fifth exon (Figure 7A and Supplementary Figure 9). The OsUSP gene and the OsUGP1 gene both had two transcripts with the exon differences closing to the 3’ UTR.

To compare the structure differences for these gene isoforms, we predicted the 3D structures of proteins using I-TASSER. The results showed that the spatial structures among the three OsUAP1 isoforms were similar, while OsUAP1.2 (Figure 7B) and OsUAP1.3 (Figure 7C) lacked some folding due to the deficiency of exons at 5’ UTR and 3’ UTR. And for OsUAP2 isoforms, the predicted tertiary structures of OsUAP2.1 and OsUAP2.2 were almost the same (Figure 7D) due to the tiny difference between OsUAP2.1 and OsUAP2.2 variants, with OsUAP2.2 discarding two amino acids compared to OsUAP2.1 (Supplementary Figure 9). In addition, the structures of OsUGP1.1, OsUGP1.2, OsUSP.1, and OsUSP.2 were also predicted to compare differences in the spatial structures (Supplementary Figure 10). The results showed that the N and center domains were similar, while the C terminal was diverse because of the lack of sequences in OsUGP1.2 and OsUSP.2.

To understand if the alternative splicing events in UDPGP genes affect the catalytical characteristics, we performed the enzymatic activity analysis in the UAP clade. ¹H-nuclear magnetic resonance (¹H-NMR) spectroscopy was used to record the enzymatic reaction of OsUAP1.1, OsUAP1.2, and OsUAP1.3 in situ. In the time-gradient enzymatic progression at 60 min, forward conversion of GlcNAc-1-P (5.36 ppm) to UDP-GlcNAc (5.52 ppm) was observed with OsUAP1.1 (Figure 8A, line 2), but not with the glutathione S-transferase (GST) control (Figure 8A, line 1), OsUAP1.2 (Figure 8A, line 3), and OsUAP1.3 (Figure 8A, line 4). Besides, the reverse conversion of UDP-GlcNAc (5.52 ppm) to GlcNAc-1-P (5.36 ppm) was identified with OsUAP1.1 (Figure 8B, line 2), but not with the GST control (Figure 8B, line 1), OsUAP1.2 (Figure 8B, line 3), and OsUAP1.3 (Figure 8B, line 4). Furthermore, OsUAP1.1 could also catalyze the reverse conversion of UDP-GalNAc (5.55 ppm) to GalNAc-1-P (5.39 ppm) (Figure 8C, line 2), whereas GST (Figure 8C, line 1), OsUAP1.2 (Figure 8C, line 3), and OsUAP1.3 (Figure 8C, line 4) could not perform this reaction. Furthermore, we also conducted the same reaction in the OsUAP2 gene. The OsUAP2 gene was another member of the UAP clade in rice, which contained two isoforms (OsUAP2.1 and OsUAP2.2). The results showed that both OsUAP2.1 and OsUAP2.2 could catalyze the reaction from GlcNAc-1-P to UDP-GlcNAc, UDP-GlcNAc to GlcNAc-1-P, and UDP-GalNAc to GalNAc-1-P (Supplementary Figures 11A–C).

Losing UAP enzymatic activity for OsUAP1 induces early leaf senescence (Wang et al., 2015). Actually, all three alternatives (OsUAP1.1, OsUAP1.2, and OsUAP1.3) were mutated, corresponding to the 238, 203, and 238 in amino acid sequences (Gly to Cys), respectively (Supplementary Figure 5). Accordingly, this mutant osuap1 was used to identify the function of OsUAP1.1, OsUAP1.2, OsUAP1.3, OsUAP2.1, and OsUAP2.2. Results showed that the osuap1 exhibited early leaf senescence, and overexpressing OsUAP1.1 in osuap1 could restore the healthy leaf (C1, C2, and C3 in Figure 8D) but not with OsUAP1.2 (C4, C5, C6 in Figure 8E) and OsUAP1.3 (C7, C8, C9 in Figure 8F). All the above indicated the key role of the N and C terminals in maintaining the activity of UAP. The UAP protein would lose its function without N or C terminal, which corresponded to a previous study on UGP protein (Supplementary Figure 7; Meng et al., 2009a). Then, we test whether the OsUAP2.1 and OsUAP2.2 could restore the osuap1 mutant. The results showed that both the OsUAP2.1 (Supplementary Figure 11D) and OsUAP2.2 (Supplementary Figure 11E) also complemented the function of OsUAP1.1, making the leaves of osuap1 grow normally without early leaf senescence.

The above results illustrated that the loss of N and C terminals in UDPGP gene isoforms did not affect the overall 3D structures, but these N- and C-terminal sequences may be important for the UDPGP gene isoforms to maintain or change their enzymatic activity. The molecular mechanisms for the existence of inactive UDPGP gene isoforms through alternative splicing are not clear. We proposed the following speculations: (1) the inactive UDPGP gene isoforms are mistake alternative splicing; (2) the inactive UDPGP gene isoforms are under evolution for new functions; (3) the inactive UDPGP gene isoforms have effective but unknown functions, for example, competing with active isoforms to bind the enzymatic substrates to regulate the enzymatic activities of active isoforms; (4) lack of the C terminal could reduce the spatial block of the UDPGP, resulting in forming an inactive dimer in the plant (Decker et al., 2012). All above represent potential regulating mechanisms of alternative splicing isoforms of UDPGP in the plant, and further experiments are needed to validate the hypothesis.

Protein sequences, transcript sequences, genomic sequences, and GFF annotation files of 58 green plants, seven chlorophytes, six animals, three fungi, one kinetoplastid, and one bacterium were downloaded from Phytozome² and Ensembl website³ (Supplementary Table 1). UDPGP homologs were identified by the following steps: (1) A hidden Markov model (HMM) of UDPGP (ACC: PF01704.19) was downloaded from the Pfam website. (2) The UDPGP HMM was used to direct HMMSEARCH with the parameters E-value < 0.1. (3) Pfam-A.hmm was a manually corrected database and was downloaded from the Pfam website (April 18, 2020). Hmmscan software was used to identify the UDPGP domain with default parameters, and a total of 454 non-redundant primary protein sequences were retrieved for further analysis.

To explore the phylogenetic relationships of the UDPGP genes in the plant, animal, and microbial lineages, 454 full-length non-redundant primary protein sequences were used to perform MSA analysis using MAFFT V7.271 (Katoh and Standley, 2013) with default parameters. The MSA was submitted to IQ-TREE v2.1.1 (Minh et al., 2020) to be tested for the best substitution model, and the model with the lowest Bayesian Information Criterion (BIC) was selected as the best model (LG + I + G). Then, the phylogenetic tree was inferred by the ML method. We measured branch supports using the Ultrafast Bootstrap (UFBoot) algorithm with 1,000 replicates, the SH-aLRT, and approximate transformation Bayes test (aBayes). The tree was visualized using Interactive Tree of Life (iTOL)⁴ (Letunic and Bork, 2019) and FigTree v1.4.4⁵.

The coding sequence (CDS) information of the 105 full-length primary protein sequences from 16 representative plants was retrieved from the GFF annotation files and submitted to the TBtools (Chen et al., 2018) to visualize the exon–intron organization of UDPGP genes in representative species. Motif analysis was performed using MEME suite v5.3.0 (Bailey et al., 2009), which scans for motifs recurring in a set of sequences. Motif analysis was carried out using the MEME server⁶, keeping the width of the motif at 6–50 amino acids, the number of motifs was 20, and the other parameters set to default. The gene structure and conserved motif patterns were visualized by the TBtools (Chen et al., 2018). The pI/MW tool from ExPASy (Gasteiger et al., 2005) was used to compute the theoretical pI and MW of each sequence.

The UDPGP gene expression data of O. sativa were downloaded from the Rice Expression Database⁷ and TENOR database⁸. The expression abundance with different treatments was transformed by log₂RPK and visualized using TBtools.

All full-length sequences from the 16 species (Amborella trichopoda, A. thaliana, Brachypodium stace, Citrus sinensis, Glycine max, Marchantia polymorpha, O. sativa, Panicum hallii, Physcomitrella patens, Prunus persica, Ricinus communis, Selaginella moellendorffii, Setaria italica, Solanum lycopersicum, Sorghum bicolor, and Triticum aestivum), including basal angiosperm, spikemoss, mosses, liverwort, monocots, and eudicots were selected to study the evolution of the UDPGP gene family (Supplementary Table 1). Orthologous genes were generated by OrthoFinder v2.2.7 (Emms and Kelly, 2015) with default parameters. In total, seven orthogroups were identified across the 16 species, and 105 (100%) of the input genes were assigned to orthogroups.

The collinearity of O. sativa and A. thaliana was detected with MCScanX (Wang et al., 2012). The result was visualized using TBtools (Chen et al., 2020).

To understand and infer the evolutionary expansion history of the UDPGP family, we used a total of 105 full-length homologs from the 16 species.

The I-TASSER (Yang et al., 2014) was used to model the structures of all O. sativa UDPGP genes, including OsUAP1, OsUAP2, OsUGP1, OsUGP2, OsUGP3, and OsUSP. I-TASSER generates simulated protein structures depending on the pairwise structure similarity. Then, we selected the top models based on the C-score as representative structures. The 3D structures of proteins were visualized in PyMOL software.

The GST gene fusion constructs of UAP isoforms were generated. The full-length CDS of the OsUAP1.1, OsUAP1.2, and OsUAP1.3 isoforms were specifically amplified using primers GST-OsUAP1.1 (F: cgGGATCCatggcggagatcgtggtggc, R: cgGAATTCctaaaatgaaatctcactcggtgc), GST-OsUAP1.2 (F: cgGG ATCCatggatgtacacagcc, R: cgGAATTCctaaaatgaaatctcactcggtgc), and GST-OsUAP1.3 (F: cgGGATCCatggcggagatcgtggtggc, R: cgGAATTCtcaagctttaagcctgccgtg). The full-length CDS of the OsUAP2.1 and OsUAP2.2 isoforms were amplified using primers GST-OsUAP2 (F: cgGGATCCatgaaggagatagtggttgggtcg, R: cgGAATTCctagaaggaaatctcactcggcg). PCR products were inserted into pGEX-6P-1 using the restriction enzyme sites BamHI and EcoRI. Then, the recombinant vectors were transferred into E. coli DH5α and sequenced to check if the constructions were correct. Expression and purification of the fused GST-UAP isoforms were performed using the same method as described (Wang et al., 2015). The reaction of enzymatic activities of UAPs were performed as described previously (Wang et al., 2015), with 0.5 μg purified GST-UAP recombinant proteins used in the reaction mixture. Examination of GlcNAc-1-P/GalNAc-1-P and UDP-GlcNAc/UDP-GalNAc was performed by ¹H-NMR as described (Zhang et al., 2011; Wang et al., 2015). Data acquisition started at 60 min respectively after the addition of an enzyme to the reaction mixture.

The overexpression vectors of UAP gene isoforms were constructed. The full-length CDSs of the OsUAP1.1, OsUAP1.2, and OsUAP1.3 isoforms were specifically amplified using primers OsUAP1.1-OE (F: cagtGGTCTCa gttgatggcggagatcgtggtggc, R: cagtGGTCTCaagagctaaaatgaaatctca ctcg), OsUAP1.2-OE (F: cagtGGTCTCagttgatggatgtacacagcccact, R: cagtGGTCTCaagagctaaaatgaaatctcactcg), and OsUAP1.3-OE (F: cagtGGTCTCagttgatggcggagatcgtggtggc, R: cagtGGTCTC aagagctaaaatgaaatctcactcg). The full-length CDSs of the OsUAP2.1 and OsUAP2.2 isoforms were amplified using primer OsUAP2-OE (F: cagtGGTCTCagttgatgaaggagatagtggttgg, R: cagtGGTCTCaagagctagaaggaaatctcactcg). PCR products were inserted into the binary vector pBWA(V)BU (reconstructed from pCAMBIA3300) using BsaI sites for the digesting-link one-step reaction. The recombinant vectors were transferred into E. coli DH5α and sequenced to check if the constructions were correct. Correct vectors were introduced into Agrobacterium tumefaciens EHA105 and then transformed into the osuap1 calli. Positive transgenic plants were confirmed using the phosphinothricin solution (20 mg/L) screening and then cultivated under natural conditions.