The amino acid sequences of Arabidopsis CLE proteins were downloaded from The Arabidopsis Information Resource (TAIR) (https://www.Arabidopsis.org/). The 12-amino acid CLE motifs were used as queries in TBLASTN search of the P. virgatum v1.1 genome with a threshold e-value of 500 in the Phytozome v12.1 database (https://phytozome.jgi.doe.gov/pz/portal.html) (Goodstein et al., 2012). Repeated hits were removed and only one hit at each chromosomal location was kept. Genes at each chromosomal location were identified as potential CLE genes. If there was no annotated gene at the target location, a 5-kb genome fragment with the target location in the center was then retrieved and subjected to an analysis with the online software, FGENESH, for gene predictions on the softberry website (http://linux1.softberry.com/) (Solovyev et al., 2006). Amino acid sequences encoded by each of the potential CLE genes were checked for the C-terminal conserved CLE motifs. All of the newly identified motifs were used as queries in TBLASTN searches, as described in the preceding steps. The analysis was iterated until no more CLE candidate could be identified. All of the CLE candidates were compared with the previously reported CLE genes in P. virgatum (Zhang et al., 2020). The final TDIF/TDIFL gene sequences were obtained using cloning based on the nucleotide sequences from Phytozome (Supplementary Figure 1).

To classify PvCLE genes, all 12-aa CLE motifs encoded by AtCLE and PvCLE genes were extracted, including the PvCLE genes encoding multiple CLE motifs. Phylogenetic trees containing the 12-aa CLE motif sequences were constructed as previously described (Zhang et al., 2020). Based on the clustering of CLE motifs, PvCLE genes in the same group as AtTDIF/TDIFL genes were predicted as PvTDIFL genes. To further analyze the evolutionary relationship between PvTDIFL and AtTDIF/TDIFL genes, phylogenetic trees containing the full-length amino acid sequences for TDIF/TDIFL from both P. virgatum and Arabidopsis were constructed using the MEGA 5.05 software (Hall, 2013) with the following parameters: alignment, Muscle, phylogeny construct or test, Maximum Likelihood Tree, and number of bootstrap replication = 1000. Signal peptides that target the TDIF/TDIFL proteins to the secretary pathway were predicted with the SMART Server using a normal model (http://smart.embl-heidelberg.de/smart/set_mode.cgi?NORMAL=1) (Letunic and Bork, 2018). Gene structure analysis was performed using the Gene Structure Display Server 2.0 (http://gsds.cbi.pku.edu.cn/) (Hu et al., 2015). The Format of Gene Features was set as fast-all (FASTA) Sequence. Other features containing signal peptides and motifs were uploaded in Browser Extensible Data (BED) format. Phylogenetic Tree was uploaded in Newick format based on full-length amino acid sequences (Hu et al., 2015). Multiple alignments were performed using the DNAMAN v6.0 software (Lynnon Biosoft, Quebec, Canada). Weblogo-Create Sequence Logos (http://weblogo.berkeley.edu/logo.cgi) was used for comparative analysis of motif conservation and for conservation analysis of each amino acid site of TDIF/TDIFL motifs in Arabidopsis and switchgrass (Crooks et al., 2004). Isoelectric point (pI) and molecular weight (MW) were calculated using the ExPASy-ProtParam tool (http://web.expasy.org/protparam/) (Bjellqvist et al., 1993).

Arabidopsis thaliana ecotype Columbia-0 (Col-0) and P. virgatum were used in this study. Arabidopsis seeds were surface sterilized in 75% (v/v) ethanol for 1 min and 5% (v/v) sodium hypochlorite for 15 min, with occasional gentle shaking. The seeds were then washed five times with sterilized distilled H₂O (dH₂O). After stratification at 4°C for 2 days, the seeds were placed in liquid Murashige and Skoog (MS) medium (pH 5.8) containing 1.5% sucrose (w/v) or on solid MS medium (pH 5.8) containing 1.5% sucrose (w/v) and 0.7% plant agar (w/v). The conical flasks containing liquid medium and the plates containing solid medium were placed in a tissue culture room maintained at 23°C and grown in a photoperiod containing 16 h of light at a fluence rate of 100 μmol photons m⁻² s⁻¹ followed by 8 h of dark. The conical flasks were placed on an orbital shaker at a rotational speed of 80 rpm. The plates were positioned vertically. Arabidopsis were grown in soil under the same temperature and light conditions, except that light intensity was set as 150 μmol photons m⁻² s⁻¹. The P. virgatum plants used in this study were grown at Huazhong Agricultural University, Wuhan, China.

The TDIF/TDIFL peptides derived from Arabidopsis TDIF and PvTDIFL motifs (Supplementary Table 1) were chemically synthesized by GenScript Biotech Corporation (Nanjing, China). Twenty milligrams of each peptide was provided at a purity ≥ 90% (w/w). Peptides were dissolved in double distilled water (ddH₂O) at a stock concentration of 1 mg/ml and stored at −80°C for future use. Synthetic TDIF and PvTDIFL peptides were added to liquid medium at a final concentration of 10 μM as previously described (Whitford et al., 2008). Arabidopsis seeds were treated with 10 ml of medium in conical flasks on an orbital shaker at a rotational speed of 80 rpm for 10 days. Each treatment required six flasks, with six seeds per flask.

Five different tissues of P. virgatum, namely, seed (mature seeds), rachis (bearing mature seeds), leaf (fully expanded leaf), stem (middle internode of seedling), and root were sampled in triplicates and immediately frozen in liquid nitrogen. Total RNA was extracted using the 2 × CTAB method (Li et al., 2008). The pellet was dissolved in 30 μl of RNase free ddH₂O. The concentration and quality of RNA were measured with a NanoDrop® 2000 spectrophotometer (Thermo Scientific, Wilmington, Delaware, USA). The RNA samples with A₂₆₀/A₂₈₀ ratios that ranged from 1.8 to 2.1 and A₂₆₀/A₂₃₀ ratios ≥2 were stored at −80°C for future use.

To quantify the relative expression levels of PvTDIFL genes in different samples, 1 μg of total RNA was used as a template to synthesize cDNA using the PrimeScript™ RT reagent Kit with gDNA Eraser (Perfect Real Time) (TaKaRa, Dalian, China). The qRT-PCR reactions were then prepared using 2 × HSYBR qPCR Mix without ROX (ZOMANBIO, Beijing, China). A standard 2-step amplification protocol was run in a LightCycler® 96 Real-Time PCR System (Roche, USA). The comparative C_T method (ΔΔCt method) was used for relative quantification of real-time PCR (Pfaffl, 2001). PvUBQ6 (Pavir.5NG345900) was used as the reference gene (Gimeno et al., 2014). All gene specific primers (Supplementary Table 2) were designed using the Primer Premier 5.0 software (www.PremierBiosoft.com). All real-time PCR reactions were run in three biological replicates and two technical replicates.

The coding sequences (CDS) from PvTDIFL3^MR3 and PvTDIFL3^MR2 were amplified using PCR with gene specific primers (Supplementary Table 2). The CDS of PvTDIFL1 was artificially synthesized (Sunny Technology, Shanghai, China). The products were cloned into a Gateway™ entry vector pDONR201 using the BP recombination reaction (Invitrogen). The clones were analyzed using DNA sequencing. The correct fragments were subsequently recombined into the destination vector, pK2GW7 (Karimi et al., 2002). Transformation of A. thaliana was performed using the Agrobacterium tumefaciens-mediated floral dip method (Clough and Bent, 1998).

Arabidopsis seedlings were photographed with a Canon EOS 7D digital camera to obtain high-resolution images. Root length was measured by using the ImageJ software (http://imagej.nih.gov/ij). The images presented in the figures were representative images from images of 24 individual seedlings for each line for the plate-grown plants and 10 individual seedlings for each line for the soil-grown plants, respectively.

Histological analysis of the vasculature was performed by using semi-thin sections. The upper part of the hypocotyl, 2–3 mm in length, was cut with a scalpel. A small amount of tissue from the stem and petiole of the rosette was kept to mark the upper end of the hypocotyl segment. The hypocotyl segments were fixed in formalin-acetic acid-alcohol (FAA) fixative (50% ethanol, 10% glacial acetic acid, 5% formaldehyde, v/v/v) at 4°C for 24 h. Dehydration was carried out by immersing the specimen in an ascending series of ethanol. Lastly, the specimens were infiltrated and embedded with Technovit® 7100 resin (www.kulzer-technik.com). The hypocotyl specimens were transversely sectioned at a thickness of 2.5 μm using a RM2265 Microtome (Leica BIOSYSTEMS, Nussloch, Germany) beginning from the upper end. The rosette tissues were first trimmed and monitored by sectioning and microscopy. The 10th to 20th sections of the hypocotyls were stained with 0.05% (w/v) aqueous toluidine blue and visualized with a BX53 light microscope (OLYMPUS, Tokyo, Japan). The cells in the stele were marked by using the ImageJ software and the number of cells was counted manually.

Statistical analysis was performed with six biological replicates and a t-test was used with double sample variance assumption. The letters a, b, and c and asterisks indicate statistically significant differences relative to the control.

Sequence data from this article are from the P. virgatum v1.1 genome in the Phytozome v12.1 database and TAIR10 associated with the following accession numbers: AtEF1-α (At5G60390), PvUBQ6 (Pavir.5NG345900), AtCLE41 (AT3G24770), AtCLE42 (AT2G34925), AtCLE44 (AT4G13195), AtCLE46 (AT5G59305), PvTDIFL1 (Pavir.Ab03264), PvTDIFL2 (Pavir.Aa00134), and PvTDIFL4 (Pavir.J06189). The sequence data for PvTDIFL3^MR3 and PvTDIFL3^MR2 has been submitted to GenBank and are associated with accession numbers MZ463198 and MZ463199, respectively.

To identify CLE genes in P. virgatum (PvCLE), sequences similar to the Arabidopsis CLE motifs were identified in the P. virgatum v1.1 genome (https://phytozome.jgi.doe.gov/pz/portal.html). The PvCLE genes were predicted and clustered using the recently reported method (Zhang et al., 2020). In total, 93 PvCLE genes were identified, including 91 annotated PvCLEs in the P. virgatum v1.1 genome and two novel PvCLEs identified in this study (Figure 1, Supplementary Table 3). Since there were three PvCLEs that encoded multiple CLE motifs, all PvCLEs encoded a total of 97 CLE motifs. The total number of PvCLEs was nearly three times of the number of AtCLEs. Based on the clustering, PvCLEs were divided into six groups (Figure 1A, Supplementary Figure 2). The Weblogo of CLE motifs showed that Group 1, including eight AtCLEs (AtCLE1-7, and AtCLV3) and 21 PvCLEs, had two conserved amino acids, namely, Asp (D) at the 8th residue and His (H) at 12th residue. These two residues were replaced with Asn (N) in Group 3, which contained 13 AtCLEs (AtCLE8-14, AtCLE16-17, and AtCLE19-22) and 38 PvCLEs. Residue number 12 was Asn (N) in Group 2, which contained five AtCLEs (AtCLE18, AtCLE25-27, and AtCLE45) and 20 PvCLEs (Supplementary Figure 2). The Pavir.Fa00904.1 from Group 2 encoded two repeats of RRVRRGSDPIHN, therefore the Group 2 PvCLEs encoded a total of 21 CLE motifs. Group 4 was comprised of nine TDIF/TDIFL genes, with four members from Arabidopsis (AtCLE41, 42, 44, and 46) and five genes from P. virgatum. In Group 4, there were two PvCLEs encoding multiple TDIFL motifs (see the next paragraph for details), and the total number of TDIFL motifs encoded in P. virgatum was as many as eight. Group 5 had two Arabidopsis members (AtCLE40 and AtCLE43) and three PvCLEs. In general, the motif conservation in Group 5 was lower relative to groups 1 through 4. Six PvCLE genes were found upon encoding CLE motifs that are very different from the AtCLE motifs and thus, fell into “Group others” (Figure 1A, Supplementary Table 3). The CLE motifs are rather conserved between the AtCLEs and PvCLEs (Figures 1B,C).

In total, five TDIF homologous genes were identified in the P. virgatum genome (Figure 2, Supplementary Figure 1, Table 1). The three TDIF homologs that were annotated are Pavir.Ab03264, Pavir.Aa00134, and Pavir.J06189, namely PvTDIF-like1 (PvTDIFL1), PvTDIFL2, and PvTDIFL4 in this study, respectively. A novel TDIFL gene, PvTDIFL3, was identified from TBLASTN searches of the P. virgatum v1.1 genome with a threshold e-value of 500. The predicted PvTDIFL3 protein product contained three potential TDIFL motifs and hence named PvTDIFL3^MR3. The MR3 is referring to that it has three motif repeats (Figure 2, Table 1). When gene-specific primers were used to amplify the CDS of PvTDIFLs, a shorter fragment derived from PvTDIFL3^MR3 was also cloned that was named PvTDIFL3^MR2 (Figures 2A–C). The amino acid sequence of PvTDIFL3^MR2 is identical to that of PvTDIFL3^MR3, except that 47 amino acid residues are missing from the middle. The missing 47 amino acid residues constitute the second TDIFL motif and its flanking sequences of PvTDIFL3^MR3 (Figure 2C). The PvTDIFL3^MR2 matched to the same chromosomal location as PvTDIFL3^MR3 in TBLASTN search of the P. virgatum v1.1 genome. In a phylogenetic analysis, amino acid sequences from the five PvTDIFL genes clustered in the same clade as the amino acid sequences from Arabidopsis TDIF/TDIFL genes (Figures 1, 2A). The nucleotide sequences were subjected to the online gene prediction program FGENESH on the Softberry website (http://linux1.softberry.com/), to evaluate the genomic sequence and gene structure of these five genes.

All the TDIF/TDIFL proteins from Arabidopsis and switchgrass contained a signal peptide for the secretory pathway at the N-terminus (Figures 2B,C). Similar to the Arabidopsis TDIF proteins, PvTDIFL1, 2, and 4 had a single TDIFL motif at the C-terminus. However, PvTDIFL3^MR3 and PvTDIFL3^MR2 had three and two TDIFL motifs, respectively (Figures 2B,C). The PvTDIFL proteins ranged from 92 to 175 amino acid residues in length. Their theoretical molecular weights (MW) and isoelectric points (pI) ranged from 10.01 to 18.50 kDa and from 5.81 to 11.94, respectively (Table 1).

A multiple sequence alignment analysis revealed that PvTDIFL1 and 2 proteins shared the same motif, HEVPSGPNPDSN, which is different from the Arabidopsis TDIF motif at the 10th residue in that an Ile (I) residue is changed to an Asp (D) residue. Motif 1 and motif 3 of PvTDIFL3^MR3/PvTDIFL3^MR2 were different at the 2nd and 10th residues relative to the TDIF motif. The 12th residue of Motif 2 from PvTDIFL3^MR3 had an Asn (N) to His (H) substitution. The 2nd, 7th, and 10th residues of the motif from PvTDIFL4 were different from the TDIF motif (Figure 2C). The conserved TDIF/TDIFL dodecapeptides in Arabidopsis and switchgrass were compared by using a WebLogo analysis. Similar to the Arabidopsis TDIF, the 4th and 7th residues of the TDIFL motifs in PvTDIFL proteins were conserved Pro (P) residues. However, their 2nd, 10th and 12th residues were less conserved (Figures 2D,E).

In Arabidopsis, CLE41 is expressed in the phloem and the neighboring pericycle cells in the root and hypocotyl. CLE44 is expressed in the phloem, pericycle and endodermal cells (Hirakawa et al., 2008). To analyze the expression pattern of PvTDIFL genes in the different tissues of switchgrass, in silico expression analysis and qRT-PCR were both performed.

The in silico expression data for PvTDIFL1, 2, and 4 were downloaded from Phytozome. PvTDIFL1 and 2 were both expressed at the highest levels in the inflorescence panicle, rachis and shoot. The PvTDIFL3^MR3 and PvTDIFL3^MR2 were not annotated in Phytozome and thus, no in silico expression data was available. The expression of PvTDIFL4 was barely detectable in the tissues studied (Figure 3A).

For qRT-PCR, five switchgrass tissue samples were collected in triplicates. These tissues included seed, rachis, leaf, stem, and root. The PveEF-1α, PvACT12, and PvUBQ6 were evaluated as internal reference genes (Gimeno et al., 2014). The PvUBQ6 was selected because of its uniform expression levels in different tissues. The qRT-PCR results showed that the expression levels of PvTDIFL1 were in the rachis, which is consistent with the in silico data (Figure 3B). The expression of PvTDIFL3^MR3 was relatively high in the rachis and leaf. In contrast, the expression of PvTDIFL3^MR2 was more specific to the rachis (Figures 3C,D). However, the expression of PvTDIFL2 and PvTDIFL4 was undetectable, probably due to low levels of expression. The relatively high levels of expression in the rachis provides evidence for PvTDIFL genes contributing to vascular development, similar to the Arabidopsis TDIF genes.

Exogenous TDIF peptide promotes procambial cell divisions, which leads to a remarkable increase in the vascular development of the hypocotyl in Arabidopsis (Hirakawa et al., 2010a). To study the activity and functional conservation of PvTDIFL peptides, four types of PvTDIFL peptides corresponding to the TDIFL motifs from PvTDIFL1, PvTDIFL3^MR3 and PvTDIFL3^MR2 were chemically synthesized and exogenously applied to Arabidopsis seedlings, at a concentration of 10 μM. The Arabidopsis TDIF peptide was used as the positive control (Whitford et al., 2008).

Arabidopsis seedlings were grown in liquid MS media either with or without TDIF/TDIFL peptides for 10 d. To test whether these peptide treatments affected the vasculature of the hypocotyl, semi-thin transverse sections were prepared. The results showed that the application of Arabidopsis TDIF peptides induced increases in the size of the stele with significant increases in cell numbers, as previously reported (Figures 4A,B,G) (Whitford et al., 2008). Similarly, the number of cells in the stele increased significantly in the seedlings treated with PvTDIFL_1p, 2p, and 3p, but not as much as in the TDIF-treated seedlings (Figures 4A–E,G). In contrast, the PvTDIFL_4p treatment caused an unexpected decrease in the number of cells in the stele, due to the His (H) substitution for Asn (N) at the 12th residue of PvTDIFL_4p (Figures 4A,F,G). These data indicate that the activities of the three peptides (PvTDIFL_2p, 3p, and 4p) encoded by PvTDIFL3 diverged.

As described above, PvTDIFL1, 3^MR3 and 3^MR2 differ in their motif numbers, motif sequences and peptide activities. To further investigate their functions in plant development, the 35S promoter was used to drive the expression of PvTDIFL1, 3^MR3 and 3^MR2 in Arabidopsis. The relative expression levels of the PvTDIFL transgenes in Arabidopsis were analyzed by RT-PCR (Figure 5A). The transgene expression levels and morphologies of two representative lines for each transgene were compared to Col-0. Four-week-old soil-grown plants were photographed (Figures 5B–H). In comparison with the wild-type plants (Figure 5B). Both lines harboring the 35S:PvTDIFL1 transgene developed smaller rosettes with small, round, and bushy leaves relative to Col-0 (Figures 5B–D). Similar results were obtained when AtCLE42 and 44 were overexpressed in Arabidopsis (Strabala et al., 2006). In contrast, the morphological changes of 35S:PvTDIFL3^MR3 and 35S:PvTDIFL3^MR2 plants were less severe and less consistent (Figures 5E–H).

To further quantify the phenotypic changes of the 35S:PvTDIFL plants, the above-mentioned lines were grown on vertical plates for 2 weeks to observe the morphological changes in roots. A significant decrease in the length of the primary root was observed in all heterologous expression lines, except for line 35S:PvTDIFL3^MR2-1 (Figures 6A–G,O). Similar short-root phenotypes were observed in rice and pine after treatments with synthetic TDIF peptides (Kinoshita et al., 2007; Strabala et al., 2014). The heights of the inflorescence of 35S:PvTDIFL plants were measured after 6 weeks of growth in soil, 2 weeks after the initiation of flowering. Extreme dwarfism was observed in the lines harboring the 35S:PvTDIFL1 transgene. The heights of their inflorescences were 40% of the wild-type plants (p ≤ 0.001, n = 10). The height of the inflorescence was also decreased in the lines harboring the 35S:PvTDIFL3^MR3 transgene. The heights of the inflorescence from the line 35S:PvTDIFL3^MR3-1 and -2 were 54% (p ≤ 0.001, n = 10) and 34% (p ≤ 0.001, n = 10) of the wild type plants, respectively. In contrast, no significant decrease in the height of the inflorescence was observed in the lines harboring the 35S:PvTDIFL3^MR2 transgene (Figures 6H–N,P).

To investigate the functions of PvTDIFL genes in vascular development, the hypocotyls of 6-week-old Arabidopsis were sectioned and analyzed using light microscopy. In comparison with Col-0 (Figures 7A,E), heterologous expression of PvTDIFL1 induced a drastic increase in the number of cells in the stele, suppressed the differentiation of xylem and disrupted the organization of the vascular tissue (Figures 7B,F). Heterologous expression of PvTDIFL3^MR3 induced a reduction in the size of xylem and an increase in the size of phloem without disrupting the organization of the vascular tissue and did not influence the diameter of the hypocotyl (Figures 7C,G). Heterologous expression of PvTDIFL3^MR2 led to a similar albeit somewhat attenuated phenotype relative to the heterologous expression of PvTDIFL1. For instance, the size of treachery elements was not reduced in the plants expressing PvTDIFL3^MR2 (Figures 7D,H). The above results are consistent with the previous reports on the TDIF genes in Arabidopsis and Populus (Hirakawa et al., 2008; Whitford et al., 2008; Etchells and Turner, 2010; Etchells et al., 2015; Li et al., 2018). Although PvTDIFL1 contains only one single TDIFL motif, it had the greatest influence on vascular development. On the contrary, multiple TDIFL motifs did not increase the activity of either PvTDIFL3^MR3 or PvTDIFL3^MR2 relative to PvTDIFL1.

In this study, 93 putative CLE genes were identified in the genome of P. virgatum, by using a novel method that was recently developed (Zhang et al., 2020) followed by gene cloning. In Arabidopsis, all of the 32 CLE genes encode proteins with a single CLE motif, except for CLE18, which contains a second CLEL motif (Meng et al., 2012). Surprisingly, five CLE proteins containing multiple CLE motifs were identified in O. sativa, T. aestivum, and Medicago truncatula (Oelkers et al., 2008). A total of 59 CLE proteins in 27 plant species contain multiple CLE motifs (Goad et al., 2017). Genes encoding proteins containing multiple CLE motifs have been recently identified in various plants, despite that the CLE motifs from the same protein might or might not be identical to each other (Goad et al., 2017). Most functional studies on CLE genes have been conducted with genes encoding a single CLE motif. The knowledge of the CLE genes encoding multiple motifs remains limited. In this study of CLE genes in P. virgatum, it was found that three PvCLE genes—Pavir.Fa00904.1, PvTDIFL3^MR3, and PvTDIFL3^MR2 encode proteins containing multiple CLE motifs, which allowed the functions of the CLE motif-containing proteins encoded by these genes to be studied.

Among the three PvCLEs that encode multiple CLE motifs, PvTDIFL3^MR3 and 3^MR2 appear to encode CLE peptides that are homologous to the TDIF peptide. The function of the TDIF peptide in vascular development has been well-studied in several species, such as Arabidopsis, Populus, Marchantia polymorpha (Ito et al., 2006; Hirakawa et al., 2008, 2010a,b; Etchells and Turner, 2010; Etchells et al., 2015; Kondo and Fukuda, 2015). However, these previously studied TDIF/TIDFL genes encode a single TDIF/TDIFL motif. Therefore, functional analysis of the TDIFL peptides from the PvTDIFL3^MR3 and 3^MR2 proteins could provide a better understanding of the regulatory activities of TDIF/TDIFL peptides.

In total, five PvTDIFL-encoding genes were identified in P. virgatum. The motif sequences of PvTDIFL1/2 and TDIF are very similar. Indeed, only the 10th amino acid residue of the motif is different between PvTDIFL1/2 and TDIF. The TDIFL peptide from PvTDIFL1/2 represents the largest group of TDIF/TIDFL peptides in monocots (Zhang et al., 2020). The number of different residues in the TDIFL motifs of PvTDIFL3^MR3, 3^MR2, and 4 ranges from two to three (Figures 2B,C). The PvTDIFL3^MR3 and 3^MR2 contain three and two TDIFL motifs, respectively. Sequence alignments of PvTDIFL proteins show that PvTDIFL3^MR3 had one more repeat of the TDIFL motif within its flanking sequence relative to PvTDIFL3^MR2. The PvTDIFL3^MR2 was a “byproduct” of cloning PvTDIFL3^MR3 by using gene specific primers for PvTDIFL3^MR3. Meanwhile, PvTDIFL3^MR3 and 3^MR2 were matched to the same chromosomal locus, which provides evidence that these two genes could be alleles. An alternative explanation is that PvTDIFL3^MR2 is simply missing from the v1.1 genome sequence of P. virgatum.

Alanine scanning mutagenesis indicated that the 2nd, 5th, 7th, 10th, and 11th residues of the TDIF motif do not contribute to its activity. However, the H¹, V³, G⁶, N⁸, P⁹, and N¹² substitutions caused severe losses of TDIF activity (Ito et al., 2006). In order to investigate the activities of PvTDIFL peptides based on the predicted motifs, four PvTDIFL peptides were synthesized and applied to Arabidopsis seedlings. The peptide activities were evaluated based on their ability to influence cell numbers and cell arrangement in the vasculature of the hypocotyl.

The results showed that PvTDIFL_1p, 2p, and 3p had similar and significant activities in promoting cell division, disrupting vascular cell patterns and inhibiting xylem development, although their activities were weaker than the Arabidopsis TDIF peptide (Figures 4A–E,G). On the contrary, PvTDIFL_4p (corresponding to the 2nd motif of PvTDIFL3^MR3) had no TDIF activity. Instead, PvTDIFL_4p induced a 50% reduction in the number of cells in the hypocotyl, which is consistent with PvTDIFL_4p serving as an antagonist of peptides that possess TDIF activity (Figures 4F,G). The sequence alignment of the PvTDIFL peptides that included the TDIF peptide showed that PvTDIFL_1p, 2p, and 3p have different residues at the 2nd and/or 10th residues, which are not critical positions for TDIF activity (Ito et al., 2006). The PvTDIFL_4p has an extra substitution, with a His (H) instead of an Asn (N) at the 12th residue (Figure 4H), which is one of the critical positions (Ito et al., 2006). This H-to-N substitution is consistent with its inhibitory activity in the exogenous peptide treatment experiments.

It has been hypothesized that a full CLE protein precursor carrying multiple motifs can release several active peptides after processing, which could play an amplification effect (Oelkers et al., 2008). To better understand the function of PvTDIFL genes, PvTDIFL1 (one motif), PvTDIFL3^MR3 (three motifs), and PvTDIFL3^MR2 (two motifs) were expressed in Arabidopsis. Heterologous expression of PvTDIFL1 mimicked the phenotypes of the Arabidopsis plants that overexpressed the endogenous TDIF genes, such as CLE41 and CLE44 (Strabala et al., 2006; Etchells and Turner, 2010). However, heterologous expression of PvTDIFL3^MR3 and 3^MR2 produced more subtle phenotypes than the heterologous expression of PvTDIFL1 (Figures 5–7). An amplification of TDIFL activity in Arabidopsis plants expressing PvTDIFL3^MR3 or 3^MR2 was not observed although these genes encode peptides with multiple motifs. The lack of the amplification effect has several possible explanations. Firstly, PvTDIFL3^MR3 and 3^MR2 were heterologously expressed in Arabidopsis. Because Arabidopsis does not have any CLE proteins with multiple CLE motifs, it might not be able to process PvTDIFL3^MR3 or 3^MR2 protein precursors efficiently. Secondly, the motif sequences of PvTDIFL peptides are not the same as the endogenous TDIF/TDIFL peptides of Arabidopsis. Therefore, the affinity of the PXY/TDR receptor in Arabidopsis for the PvTDIFL peptides is difficult to predict and requires more experimentation to understand. Thirdly, although the typical sequence of the TDIF peptide is HEVPSGPNPISN (Ito et al., 2006), the conserved sequences flanking the TDIFL motifs from PvTDIFL3^MR3 and 3^MR2 provide evidence that possible variants of the mature motifs are encoded by these two genes. Finally, the experiments in this study have multiple variables, therefore it is hard to know whether all the TDIFL motifs from PvTDIFL3^MR3 and 3^MR2 have been successfully processed. The proteomics analysis of small proteins based on liquid chromatography with tandem mass spectrometry (LC-MS/MS) technique (Wang et al., 2020) can be applied to the analysis of transgenic lines harboring the transgenes that encode multiple PvTDIFL motifs with various alanine substitutions. This may help to analyze the processing of multiple PvTDIFL motifs.

The PvTDIFL3^MR3 contains a 47-amino acids insertion between the amino acid residues 77 and 123. This insertion introduces an extra TDIFL motif in PvTDIFL3^MR3 relative to PvTDIFL3^MR2 (Figure 2C). The extra TDIFL motif is corresponding to PvTDIFL_4p (Figure 4H). A moderate TDIF-overexpression phenotype was apparent in the hypocotyl sections of the 6-week-old 35S:PvTDIFL3^MR2 plants, which provides evidence that at least one of the PvTDIFL peptides (motif 1 and motif 3) was successfully processed. On the contrary, although vascular development in 35S:PvTDIFL3^MR3 plants was reduced, the vascular organization of the hypocotyl was well maintained. These data suggest a processing of motif 2 from PvTDIFL3^MR3 and indicate that PvTDIFL_4P may serve as an antagonist of PvTDIFL_2P (PvTDIFL3^MR3_motif 1), PvTDIFL_3P (PvTDIFL3^MR3_motif 3), and the endogenous TDIF peptide. The inhibitory effect derived from motif 2 of PvTDIFL3^MR3 in the PvTDIFL3^MR3 heterologous expression experiments is consistent with the inhibition of stele development by PvTDIFL_4p that was observed during the exogenous peptide treatments in 10-day-old Arabidopsis seedlings (Figures 4F,G).

Previous reports showed that although TDIF has no inhibitory effect on root elongation in Arabidopsis (Ito et al., 2006; Whitford et al., 2008), TDIF mildly inhibits root elongation in rice and pine (Kinoshita et al., 2007; Strabala et al., 2014). In this study, heterologous expression of PvTDIFLs significantly shortened the roots of Arabidopsis, especially in the 35S:PvTDIFL1 lines (Figures 6A–G). These data are consistent with the diversification of TDIF/TDIFL gene function in different species. However, little is known about the mechanism responsible for these differences in peptide activity.

As a PBC, P. virgatum is a commonly used material to study the synthesis of biomass. In this study, heterologous expression of PvTDIFLs in Arabidopsis caused dwarf seedlings and a disordered vasculature, indicating that constitutive heterologous expression of PvTDIFLs reduced biomass. It is consistent with the previous studies on overexpression of the endogenous TDIF genes in Arabidopsis (Hirakawa et al., 2008; Whitford et al., 2008; Etchells and Turner, 2010). However, the phloem-specific expression of PttCLE41 leads to increased woody biomass in Populus and thus, demonstrates that it is possible to increase the biomass by manipulating the TDIF–PXY/TDR signaling module in plants (Etchells et al., 2015). In hybrid poplar, the PttWOX4 genes act downstream of PXY/TDR to control cell division activity in the vascular cambium and hence, to increase stem girth (Kucukoglu et al., 2017). Overexpression of the WOX gene STF (STENOFOLIA) improves biomass yields in grasses (Wang et al., 2017). In this study, five TDIF/TDIFL genes in P. virgatum were identified. Three of these genes in Arabidopsis were cloned and heterologous expressed. Overexpression technologies and exogenous peptide treatment experiments gave a better understanding of the functions and activities of various TDIFL peptides, including PvTDIFL peptides, and also provide clues that will drive future applied research on biomass improvement by manipulating the TDIF–PXY/TDR–WOX4 signaling pathway in plants.