CIA2 has NLS at aa 62-65 and 291-308
Previous study has shown that CCT motifs serve as an NLS for many transcription factors (Kurup et al., 2000; Strayer et al., 2000; Robson et al., 2001). We used WoLF PSORT software (Horton et al., 2007) to predict the locations of NLS in CIA2, which identified NLS at aa 56-59 (RKPR), 62-65 (RKRP), 291-308 (K-rich) and 383-426 (CCT motif).
To establish if these NLS were functional, we fused full-length (FL) and various deletion fragments of CIA2 to the C terminus of the GUS coding sequence (Fig. 1a). Potyvirus nuclear inclusion protein (GUS-NIa) represented the positive control (Carrington et al., 1991). These chimeric constructs were driven by the CaMV 35S promoter and were transiently expressed in onion epidermal cells by microparticle bombardment. GUS-CIA2FL, GUS-CIA2D1-61 and GUS-CIA2D309-435 (Fig. 1b C1-3, D1-3 and G1-3) were all localized in the nucleus, demonstrating that aa 1-61 and 309-435 of CIA2 are not essential for protein entry into the nucleus. However, signal for GUS-CIA2D1-65, GUS-CIA2D291-435, GUS-CIA2D62-65, GUS-CIA2D291-308 (Fig. 1b E1-3, F1-3, H1-3 and I1-3) was evenly distributed throughout the cells, indicating that these deleted fragments might include functional NLS. Thus, CIA2 has two functional NLS at aa 62-65 and 291-308, so the CCT motif in CIA2 is not a functional NLS because that motif is localized at aa 383-426.
CIL is a nuclear protein and its NLS is located at aa 47-50
We also employed WoLF PSORT to predict the subcellular localization of CIL and the sequence position of its NLS. The software predicted four NLS at aa 41-44 (RRPR), 47-50 (RKRP), 255-262 (K-rich) and 342-385 (CCT motif). Applying a similar strategy to that used for CIA2, we found that chimeric GUS-CIL FL, GUS-CILD1-46, GUS-CILD254-394 and GUS-CILD245-262 (Fig. 1c C1-3, D1-3, F1-3 and H1-3) all localized in the nucleus, suggesting that CIL aa 1-46 and 245-394 are not essential for nuclear entry. However, signal of both GUS-CILD1-50 and GUS-CIA2D47-50 (Fig. 1c E1-3 and G1-3) was evenly distributed in the cells, indicating that CIL has an NLS at aa 47-50. Thus, as for CIA2, the CCT motif in CIL (aa 341-384) is not a functional NLS.
CIA2 interacts with other nuclear proteins
CIA2 may form complexes with other nuclear proteins to modulate physiological or developmental processes. To explore the biological functions of CIA2, we used a bait plasmid pGBKT7 containing fused GAL4 BD and 435-aa CIA2 fragments to screen an Y2H Arabidopsis cDNA library (Clontech Mate and Plate Library - Universal Arabidopsis) for putative CIA2-interacting proteins.
We identified 12 CIA2-interacting nuclear proteins from this assay, four of which represent nuclear proteins with unknown functions (Table 1). Notably, CIA2 is able to interact both with itself or CIL, indicating that CIA2 might form a homodimer or heterodimer to interact with other proteins. Three of the 12 identified proteins are nuclear factor Y subunit subdomains (i.e., NF-YB1, NF-YC1 and NF-YC9). Also known as CCAAT-box factors (CBF) or HEME activator proteins (HAP), NF-Y proteins have been independently identified from yeast, mammals, and plants, and are transcription factors and histone-folding proteins that are conserved across species (Mantovani, 1999). NF-Y proteins comprise three major subunits—NF-YA (CBF-B or HAP2), NF-YB (CBF-A or HAP3), and NF-YC (CBF-C or HAP5)—that together can form heterotrimeric complexes and interact with other proteins to influence many aspects of plant life, such as seed development, flowering time regulation, primary root elongation, abscisic acid signaling, and drought resistance (Mantovani, 1999; Petroni et al., 2012; Zhao et al., 2016). Thus, through its interactions with NF-YB1, NF-YC1 or NF-YC9, CIA2 could participate in some of those biological processes.
CO, another of the 12 interacting partners, is a 373-aa transcription factor that induces transcription of the flowering locus T (FT) gene, which controls the floral transition of shoot apical meristem (Abe et al., 2005; Corbesier et al., 2007; Jaeger et al., 2007). In Arabidopsis, CO forms a protein complex with NF-Y proteins. CO/NF-Y complexes bind to CO-response element (CORE) and the CCAAT-box of the FT promoter to increase FT expression and induce flowering (Wenkel et al., 2006; Tiwari et al., 2010). ABSCISIC ACID INSENSITIVE 3 (ABI3) also interacts with CO (Kurup et al., 2000), with the maize ABI3 orthologue VIVIPAROUS 1 (VP1, Finkelstein et al., 2002) playing essential roles in abscisic acid-mediated regulation of seed maturation, sensitivity to desiccation, and precocious germination (Ooms et al., 1993; Nambara et al., 1994; Parcy et al., 1994 and 1997). However, manifestation of a late-flowering phenotype upon ectopic ABI3 expression and the early-flowering phenotype of the abi3-4 mutant suggest that this protein is also involved in regulating flowering time (Kurup et al., 2000; Hong et al., 2019).
Moreover, ARABIDOPSIS RESPONSE REGULATOR 3 (ARR3) and ARR4 exhibit redundant functions in regulating signaling cascades in response to cytokinin and rhythmic expression of oscillator genes such as circadian clock associated 1 (CCA1) and TOC1 (Kakimoto, 2003; Salomé et al., 2006). Taken together, the CIA2-interacting proteins ABI3, ARR3, CO, NF-YB1, NF-YC1 and NF-YC9 all seem to be at least partially involved in controlling flowering, strongly indicating a role for CIA2 in flower development.
Confirmation of protein interactions among CIA2-interacting candidates
To confirm the interactions between CIA2, CIL and these six flowering regulators, we ligated different CIA2 and CIL fragments to pAS2-1 vector harboring GAL4 BD or the full-length (FL) proteins to pACT2 vector harboring GAL4 AD for Y2H assay. Plasmids were co-transformed into AH109 yeast cells and plated on medium with various concentrations of 3-AT. We further validated our findings by colony-lift filter assay to determine expression levels of b-galactosidase (LacZ) activity.
We found that only full-length CIA2 (CIA2FL) and the CIA2 1-189aa fragment interacted with all eight CIA2-interacting candidates (Fig. 2), showing that it is the N-terminal 1-189 fragment of CIA2 that interacts with other proteins. Notably, CIA2 could also interact with CIL and itself through the C-terminal 330-435 aa. However, the CCT motif (CIA2 379-435aa) alone was not sufficient for interaction with any candidate proteins. Similar to CIA2, CIL also interacted with the six flowering regulators via its N-terminal 1-156aa fragment. Interestingly, CIL only interacted with CIA2 and itself via the C-terminal 290-394aa fragment. Again, the CCT motif (CIL 337-394aa) alone was not sufficient to interact with any candidate proteins.
To exclude bias from a plasmid effect in the divergent self- and cross-interaction results for CIA2 and CIL, we performed a domain swap analysis. We found that both the N-terminal CIA2 1-189aa and C-terminal CIA2 330-435aa fragments were mutually interactive (Fig. 3). However, the N-terminal CIA2 1-189aa fragment no longer interacted with full-length CIL when it was placed in an AD vector, indicating that CIL indeed specifically interacts with the C termini of both CIA2 and itself via its C-terminal 290-394 fragment.
Next, we used BiFC to establish if CIA2 and CIL interact in planta. Full-length CIA2 and CIL protein were separately fused behind the N or C terminus of enhanced yellow fluorescent protein (nEYFP or cEYFP), and transiently expressed in onion epidermal cells by particle bombardment (Waadt et al. 2008), and then observed by fluorescence microscopy. The results demonstrate that both CIA2 and CIL mutually interact in the nucleus of onion epidermal cells (Fig. 4).
CIA2 and CIL interact with the CCT motif of CO
CO utilizes its C-terminal CCT motif to interact with the NF-YB and NF-YC subunits, with those interactions allowing CO/NF-Y complex to bind to the FT promoter to increase transcript yield of FT, thereby promoting flowering (Wenkel et al., 2006; Tiwari et al., 2010). However, alternative splicing of CO transcript generates a shorter 274-aa CO protein lacking the CCT motif at the C terminus. This truncated CO protein interacts with full-length 373-aa CO at the N-terminus, causing degradation of the 373-aa CO spliced variant to delay flowering time (Gil et al., 2017). Determining how CIA2/CIL and CO interact would help establish if and how CIA2/CIL is involved in flowering regulation. We used Y2H assay to investigate this topic. We found that CIA2FL and CIA2 1-189aa interacted with the COFL, CO 181-373aa and CO 298-373aa fragments (each of which harbors the CCT motif), but the CIA 190-379aa, CIA2 330-435aa and CIA2 379-435aa did not (Fig. 5). We observed similar results for CIL fragments (Fig. 5). Together, these data show that it is the N-terminal regions of CIA2/CIL that interact with the C-terminal CCT motif of CO.
Evolutionary relationships of CIA2 and CIL homologues
To investigate the evolutionary relationships between CIA2 and CIL and further understand their physiological mechanisms, we conducted bioinformatics analysis on homologous proteins. We anticipated that proteins playing an important regulatory role in plant cells would exhibit conserved motifs across species. Since we had found that it is the N-termini of CIA2 and CIL that interact with the six flowering regulators (Table 1 and Fig. 2), we conducted a BLAST analysis of the N-terminal 200-aa sequences of Arabidopsis CIA2 and CIL to extract homologous sequences from the Gramene (www.gramene.org) and NCBI (National Center for Biotechnology Information; www.ncbi.nlm.nih.gov) databases (Altschul et al., 1997). We obtained 280 protein sequences using default settings, revealing that CIA2 and CIL homologues are widely present across the plant kingdom. In Table S1, we present summary data on 70 angiosperms (such as plants of the Cruciferae, Solanaceae, Fabaceae and Poaceae families) and 2 gymnosperms (Ginkgo biloba and the early flowering Amborella trichopoda) for which full-length sequences were available.
Phylogenetic analyses of N-terminal 200-aa sequences from these 72 plants using MEGA-X software (Kumar et al., 2018) revealed that CIA2 and CIL evolved from the same ancestral protein (Fig. 6). For each of the Brassicaceae, Fabaceae and Poaceae families, we observed two clusters of generally intermixed CIA2 and CIL homologues, but only one cluster for solanaceous plants. The resulting phylogeny demonstrates that the ancestral gene of CIA2/CIL has convergently replicated into two functional sequences and subsequently diverged during the evolution of true flowering plants. Such an evolutionary process has occurred independently in various plant families (Fitter et al., 2002).
Significance of the N-terminal CC1 motif of CIA2/CIL for interactions with flowering regulators
Apart from the CCT motif, previous bioinformatics analysis using the MEME software (Bailey et al., 2009) identified five additional conserved motifs in CIA2 and CIL (named CC1~5, Fig. S1). These motifs in CIA2 are located at aa 45-86, 162-188, 252-262, 334-359 and 364-381 for CC1 to CC5, respectively, whereas in CIL they are located at aa 30-71, 137-163, 205-215, 290-315 and 322-339.
Interestingly, the CC1 motif covers the NLS we identified in CIA2 and CIL (Fig. 1). Using the MUSCLE software in MEGA-X (Edgar, 2004a and 2004b; Kumar et al., 2018), we found that the amino acid sequence of the CC1 motif is similar to highly-conserved subdomain YA1 of NF-YA subunits (Fig. 7). The YA1 subdomain interacts with NF-YB/NF-YC dimer (Romier et al., 2003), and three of its arginine residues (R24, R25 and R26) modulate NF-Y complex assembly. R24K or R24-26A mutation of Arabidopsis NF-YA proteins (also named HAP2) resulted in lack of HAP complex formation and impeded DNA binding (Mantovani, 1999). Moreover, R25G and R26P mutation of HAP2 also abolished HAP complex formation and resulted in loss of DNA binding, but the R25K mutation did not affect HAP2 function (Xing et al., 1994). Since residues 24-26 of the YA1 subdomain are almost identical to the NLS of the CC1 motif of CIA2/CIL (Fig. 7), we speculate that CIA2 and CIL interact with CO and NF-Y proteins via their N-terminal CC1 motifs.