Besides abiotic stress, miRNAs are also regulators of biotic stress tolerance in soybean. Soybean cyst nematode (SCN) is a major soybean pest causing significant yield loss (Shaibu et al., 2020). The miR159 family was found to be responsive to cyst nematodes in genome-wide miRNA identification studies (Li et al., 2012; Tian et al., 2017). Among the members of the miR159 family, the mature miR159-3p was found to be abundant in plants (Millar et al., 2019). In a study of miR159-3p using transgenic soybean hairy roots, it was found that the overexpression of pre-miR159a could improve the resistance to cyst nematodes (Lei et al., 2021). Using psRNATarget, GAMYB genes were predicted to be the targets of miR159-3p (Lei et al., 2021). By 5’RACE, six GAMYB genes, Glyma.04G125700, Glyma.06G312900, Glyma.13G073400, Glyma.13G187500, Glyma.15G225300, and Glyma.20G047600, which encode GmMYB33c, GmMYB33d, GmMYB33b, GmMYB33e, GmMYB33f, GmMYB33a, respectively, were shown to be the cleavage targets of miR159-3p (Lei et al., 2021). As expected with the targeting of these transcripts by miR159-3p, the expression of miR159-3p was repressed during SCN infection while these target genes were upregulated (Lei et al., 2021). Compared to the wild-type, without SCN infection, these target genes were down-regulated in pre-miR159a/b/c/d/e/f over-expressers while up-regulated in the short tandem target mimic (STTM) lines of miR159a-3p/e-3p and miR159b-3p/f-3p (Lei et al., 2021). In addition, it was found that the abundance of miR159 was significantly induced by gibberellin (GA) at 10 μM, which also led to the down-regulation of the expressions of the MYB genes (Lei et al., 2021). GA treatment enhanced the resistance to SCN by soybean plants. This study shows the regulation of cyst nematode resistance through the interactions among a phytohormone, an miRNA and transcription factors, although the target genes of these MYB transcription factors are not clear. These interactions should also be taken into consideration when designing molecular breeding programs for desirable traits.

Besides regulating stress tolerance, the overexpression of gma-miR1508a also conferred dwarfism and increased cell wall thickness (Sun et al., 2020). It was proposed that the dwarf plants may have a higher transpiration rate and water loss under drought compared to the wild-type, thus resulting in drought sensitivity (Sun et al., 2020). Another example of growth-regulatory miRNA is Gma-miR156b which delays the flowering time of soybean but improves the yield (Cao et al., 2015; Sun et al., 2019). The overexpression of Gma-miR156b conferred enhanced stem thickness, branching, and yield compared to the wild-type, without affecting the height of the plant (Sun et al., 2019). The number of pods, number of seeds, seed length, seed width, seed thickness, seed weight, and yield were all increased in the over-expresser compared to the wild-type (Sun et al., 2019). Such a combination of traits is favorable for breeders. Gma-miR156b overexpression also enhanced the meristematic activity at the vegetative growth stage (Sun et al., 2019). By 5’RACE, 15 GmSPL (Glycine max SQUAMOSA PROMOTER BINDING PROTEIN-LIKE) genes were found to have their transcripts cleaved by Gma-miR156b (Sun et al., 2019). Among the GmSPL proteins, GmSPL9d was shown to interact with GmWUSa/b, which are the transcription factors defining the shoot stem cell niche and are involved in axillary meristem initiation (Sun et al., 2019).

Nodulation is an important strategy by soybean to acquire organic nitrogen and is a trait selected for during breeding (Kueneman et al., 1984). In soybean, miR482, miR1512, and miR1515 were induced by Bradyrhizobium japonicum inoculation in root (Li et al., 2010). Using transgenic soybean hairy roots, the nodule number was enhanced by either the overexpression of miR482 or miR1515 under a constitutive promoter or the expression of miR482 or miR1515 under a nodulation-inducible promoter (Li et al., 2010). By using 5’RACE, Glyma.12G28730, which encodes a GSK3 (glycogen synthase kinase 3)-like kinase, and Glyma.09G02920, which encodes a putative Dicer-like protein, were found to be the cleavage targets of miR482 and miR1515, respectively (Li et al., 2010). In soybean root hair, gma-miR2606b and gma-miR4416 were found to be repressed upon B. japonicum inoculation (Yan et al., 2016). However, the overexpression of gma-miR2606b and gma-miR4416 led to increased nodule number and decreased nodule number, respectively (Yan et al., 2016). This could be due to the fact that the overexpression of gma-miR2606b led to the repression of Glyma.07G02290, which encodes a mannosyl-oligosaccharide 1, 2-alpha-mannosidase (MNS), while the overexpression of gma-miR4416 led to the repression of Glyma.11G29920 (GmRIP1) which encodes a putative rhizobium-induced peroxidase (RIP; Yan et al., 2016). These same gene targets of miR4416 and miR2606b were also predicted by the Parallel analysis of RNA Ends (PARE) library (Yan et al., 2016). As a result of the repression of gma-miR2606b upon B. japonicum inoculation, the expression of GmRIP1 was induced in the root hair (Yan et al., 2016).

The miRNAs gma-miR171o and gma-miR171q were found to have differential expressions between uninoculated roots and nodules resulting from B. japonicum inoculation (Hossain et al., 2019). Gma-miR171o has reduced expression in nodules compared to uninoculated roots while the opposite trend was observed for gma-miR171q (Hossain et al., 2019). However, the overexpression of either gma-miR171o or gma-miR171q in transgenic soybean hairy roots could both inhibit nodulation upon B. japonicum inoculation (Hossain et al., 2019). PARE analysis revealed that GmSCL6-1 (Glycine max Scarecrow like 6–1 ) and GmNSP2 (Glycine max Nodulation-signaling pathway 2 ), which encode GRAS-family transcription factors, were the possible targets of gma-miR171o and gma-miR171q, respectively (Hossain et al., 2019). In transgenic soybean hairy roots, the overexpression of gma-miR171o and gma-miR171q led to the repression of GmSCL6-1 and GmNSP2.1, respectively (Hossain et al., 2019). Using GFP as the cleavage reporter, transient expressions of gma-miR171o or gma-miR171q with GmSCL6-1 or GmNSP2.1, respectively, in Nicotiana benthamiana leaves demonstrated the cleavage of GmSCL6-1 transcripts by gma-miR171o and those of GmNSP2.1 by gma-miR171q (Hossain et al., 2019). When the construct of GmSCL6-1 or GmNSP2.1 with a mutated cleavage site was transformed into soybean hairy roots, nodulation induced by B. japonicum inoculation was enhanced (Hossain et al., 2019). Altogether, these results indicated that miR171o and gma-miR171q negatively regulate nodulation by cleaving the transcripts of GmSCL6-1 and GmNSP2.1, respectively. In another study, miR399b was found to be induced in soybean roots inoculated with Sinorhizobium fredii, with successful nodulation (Fan et al., 2021). Using transgenic soybean hairy roots, it was shown that the overexpression of miR399b improved the growth and nutrient acquisition of S. fredii-inoculated plants, as reflected in the enhanced whole-plant inorganic phosphate (Pi) concentration, whole plant ureide concentration, nodule number and leaf node number (Fan et al., 2021). The function of miR399b was further confirmed by the miR399b knock-down mutant in which opposite trends of the above parameters were observed (Fan et al., 2021). In uninoculated roots, inoculated roots and nodules, miR399b had opposite expression trends to those of GmPHO2a and GmPHO2b, which inhibit high-affinity Pi transporters (Fan et al., 2021). The cleavage of GmPHO2a and GmPHO2b transcripts by miR399b was validated by 5’RACE (Fan et al., 2021). It was then suggested that miR399b promotes Pi uptake by the plant by negatively regulating GmPHO2a and GmPHO2b. This enhanced Pi uptake probably in turn promoted nitrogen fixation in nodules.

The above examples show the capacity of a single miRNA to regulate multiple transcripts and multiple traits simultaneously, such as both Pi and nitrogen acquisitions, to improve the nutritional status of soybean. The interaction of soybean with rhizobia to form nitrogen-fixing nodules is an important trait for improving the nutritional status of the crop. Accumulating evidence shows that such an interaction is regulated by ncRNAs. It should be noted that usually one single ncRNA could regulate multiple transcripts, which could be useful for improving the efficiency of molecular breeding.

Seed coat color is also an agronomic trait of interest to breeders (Heiser, 1988). GmDCL2 (Glycine max DICER-LIKE2 ) encodes a dicer-like protein which mediates the generation of 22-nucloetide siRNAs from long inverted repeats-derived transcripts (Jia et al., 2020). When both copies of GmDCL2 (GmDCL2a and GmDCL2b) were mutated by CRISPR/Cas9, the seed turned from yellow to brown (Jia et al., 2020). The levels of 22-nucleotide siRNAs in the Gmdc2a/2b mutant were greatly reduced (Jia et al., 2020). Specifically, a series of 22-nucleotide siRNAs from the antisense CHS1 (chalcone synthase 1) region and the sense CHS3 region were found in the wild-type but not in the Gmdc2a/2b mutant (Jia et al., 2020). Besides, other 21-nucleotide long secondary siRNAs from both antisense and sense strands of other CHS genes such as CHS2, CHS7, and CHS8 were also absent in the Gmdc2a/2b mutant (Jia et al., 2020). It was reasoned that the disruption of siRNA generation led to the increase in CHS mRNA levels in the seed coat and resulted in the darker seed coat color (Jia et al., 2020). These observations suggest the negative role of siRNAs in the stability of their target mRNAs. This example also shows the power of transcriptional regulation of multiple genes by ncRNAs.

Examples of agronomic traits regulated by ncRNAs in soybean are listed in Table 1.

Ribosomal proteins have been reported to regulate abiotic stress responses in soybean, such as cold stress responses. Several cold-sensitive yeast mutants have been shown to be defective in the assembly of ribosomal subunits (Bayliss and Ingraham, 1974). A ribosomal protein L34-like protein in soybean, GmSOL34, was induced in root tips and embryonic axes by low temperatures. The expression level of GmSOL34 was positively correlated to the increased duration of imbibition (Cheng et al., 2010). By overexpressing the sense and antisense in Arabidopsis, the transgenic SOL34ox plants showed a negative correlation between the increased SOL34 level and the survival rate of Arabidopsis under low-temperature conditions, while the transgenic Arabidopsis overexpressing the antisense SOL34 exhibited a better adaptation to freezing temperatures than the wild type plants during short cold imbibition (Cheng et al., 2010). Therefore, it was suggested that SOL34 might play a negative role in plant response to temperature stress.

Similarly, the soybean ribosomal proteins GmRPS13, GmRPS6, and GmRPS37 were found to be involved in cold stress adaptation (Kim et al., 2004). Unlike cold-stress proteins such as the soybean cold-stress response protein SRC1, the genes encoding these ribosomal proteins were not induced at the early stage of the cold stress treatment until three days after treatment started. RPS6 is in the mRNA-binding site of the 40S subunit of the cytosolic ribosome (Nygård and Nilsson, 1990). It is proposed that the late induction of GmRPS13, GmRPS6, and GmRPS37 might assist in the proper ribosome assembly and protein translation under cold stress (Kim et al., 2004).

Components in the translational machinery provide a structure for ribosomes or their associated factors and play a crucial role in translation. As mentioned above, the translational machinery might play a role in stress-coping, as some translation-related genes are stress-responsive. For example in soybean, the ribosomal protein L2 showed a transient down-regulated mRNA and protein levels upon encountering the pathogen Phytophthora sojae and its elicitors, as well as exposure to heavy metals (Ludwig and Tenhaken, 2001). As demonstrated using the yeast ribosome model, ribosomal protein L2 is essential for the peptidyl-transferase activity of ribosomes (Diedrich et al., 2000). By measuring the autoradiography of the newly synthesized proteins incorporating ³⁵S-labeled amino acids, it is remarkable that the rpL2 mRNA down-regulation was well-correlated with the transient loss of newly synthesized proteins. This observation persisted with the programmed cell death-induced expression of DD1-51 (EMBL: AJ289152), suggesting the presence of a block on protein synthesis at the translation stage but not on the transcription of mRNAs after the Phytophthora sojae infection (Ludwig and Tenhaken, 2000). It was hypothesized that the long lag time between mRNA accumulation and the appearance of newly synthesized proteins was caused by the down-regulated rpL2. Such down-regulation of the ribosomal protein L2 might directly lead to less efficient translation of cellular mRNAs, which favors the re-modeling of stress-coping mechanisms in cells (Ludwig and Tenhaken, 2001).

Besides ribosomal proteins, translation elongation factors are also reported to be involved in stress adaptation. For example, the Arabidopsis los1 mutant, which is deficient in producing eukaryotic translation elongation factor 2 (eEF2), was shown to have impaired cold sensing (Guo et al., 2002). It was further reported that the cold-induced expression of the early response transcriptional activators, C-repeat/dehydration responsive element-binding factor1 (CBF/DREB1), was enhanced by the los1-1 mutation. Protein synthesis is normal in los1–1 mutant plant at warm temperatures but is blocked in the cold as shown with the autoradiography of newly synthesized proteins incorporating ³⁵S-labeled amino acids. The altered translation efficiency might be involved in stress adaptation in the plant (Guo et al., 2002).

The comparison of proteomes among different salt-tolerant genotypes of soybean revealed a possible involvement of the translational machinery in salt stress adaptability (Ma et al., 2012). Multiple translation-related factors involved in the initiation and elongation of the peptide chain showed altered abundance when under salt stress (Ma et al., 2012). A similar inhibition of translation elongation by eEF2 under cold stress was observed in the human cells when compared to that in the Arabidopsis los1-1 mutant (Guo et al., 2002; Knight et al., 2015). It was reported that the phosphorylation of eEF2 by the eEF2 kinase reduced the physical interference between mRNA-bound ribosomes around the start codon and thus slowed down the translation activity by inhibiting ribosome translocation under cold stress. Besides eEF2, elongation factor 1α (EF1α) was also reported to have unstable expression under stress conditions (Chung et al., 2009; Saraiva et al., 2014). For example, SLTI100, which encodes EF1α in soybean, was reported to be induced under stresses including low temperature, salinity, drought, and ABA treatment (Chung et al., 2009). It was reported that the mRNA levels of the soybean EF1α family were responsive to drought stress or ABA treatment, by being either up- or down-regulated (Gao et al., 2019). Under salt stress, the genes were shown to be up-regulated (Gao et al., 2019). Among the genes of the soybean EF1α family, GmEF4 was found to be induced under salt stress (Gao et al., 2019). Composite soybean plants having GmEF4-overexpressing hairy roots survived better under drought and high salinity compared to the empty vector control as revealed by the biomass, proline contents, and H₂O₂ and O⁻ contents (Gao et al., 2019). Other than its involvement in abiotic stress, GmEF1A is also the host factor of the soybean mosaic virus (SMV) during viral pathogenesis (Luan et al., 2016).

Translation elongation factors have also been demonstrated to regulate the responses to biotic stress in soybean. It was shown that the virus-induced gene silencing (VIGS) of GmEF1A in soybean did not alter the morphology of the soybean plant but alleviated soybean mosaic virus (SMV) accumulation, ER stress accumulation, and the SMV-induced cell death, probably the result of interrupting the interaction between the potyviral P3 protein and GmEF1A (Luan et al., 2016). Similarly, knocking down GmEF1B, which encodes a guanine nucleotide exchange factor that restores the GTP moiety to reactivate EF1A, could also enhance the soybean resistance to SMV (Luan et al., 2016).

Besides translation elongation, translation initiation factors could also regulate stress responses in soybean (Alam et al., 2010; Yin and Komatsu, 2015; Gallino et al., 2018; Cho et al., 2019; Wang et al., 2021). eIF4G was reported to act as a hub in translation initiation and mediate the recruitment of additional initiation factors (Sonenberg et al., 1978). The eukaryotic translation initiation factor iso4G (GmeIFiso4G-1a) exhibited a specific drought induction profile in the slow-wilting soybean cultivar N7001 compared to the drought-sensitive TJS2049 soybean cultivar (Gallino et al., 2018). The heterologous expression of soybean GmeIFiso4G-1a in Arabidopsis improved the tolerance to osmotic, salt, drought, and low-temperature stresses (Gallino et al., 2018). In another report, it was shown that the Arabidopsis mutant eifiso4G1, a homolog of the soybean GmeIFiso4G-1a, had increased sensitivity to submergence due to the disruption of the interaction between eIFiso4G1 and Snf1-related protein kinase 1 (SnRK1), which phosphorylates eIFiso4G1 for translational regulation (Cho et al., 2019).

Flooding affected the mRNA levels of some genes that encode ribosomal proteins, and altered the abundances of nuclear proteins and phosphoproteins in the soybean root tip (Yin and Komatsu, 2015). For instance, under flooding, the mRNA levels of the ribosomal protein S24/S35 family and eukaryotic translation initiation factor 4G (eIF4G) were lowered although the protein levels were not affected (Yin and Komatsu, 2015).

Besides stress tolerance, translation elongation factors also regulate the growth and development of plants. For example, the knockout mutation of Arabidopsis eIFiso4G, which is an isoform of elF4G, reduced the plant’s adaptability to dehydration, as well as causing slower growth and development (Lellis et al., 2010). In another report, using proteomic analyses of the soybean root, it was shown that the level of eukaryotic translation initiation factor 5A (eIF5A) was decreased after flooding (Alam et al., 2010; Wang et al., 2021). Salt and heavy metal stress induced the mRNA levels of OseIF5A-1 and OseIF5A-2 in rice (Chou et al., 2004). In Tamarix androssowii, the overexpression of TaeIF5A1 also enhanced the tolerance of the transgenic poplar plants to various abiotic stresses, including salt stress and cadmium stress (Wang et al., 2012).

Transgenic Arabidopsis with constitutively suppressed AteIF5A-2 showed a reduction in pathogen growth and therefore stronger resistance against Pseudomonas syringae pv tomato (PstDC3000), while the overexpression of AteIF5A-2 showed similar susceptibility to the PstDC3000 as the wild type (Hopkins et al., 2008). It was suggested that the eIF5A level might determine the rate of translation of the mRNA species required for specific stress responses. This example hints at the importance of the proper abundance of the translational regulators. Compared to the model plant Arabidopsis, the involvement of soybean translation initiation factors in regulating the responses to biotic stress is not well-studied. However, the example in Arabidopsis may indicate the direction for similar research in soybean.