The A subunits of NF-Y generally localize to the nucleus and most NF-YA proteins can bind to the CCAAT cis-element in the promoter regions of target genes, but with different affinities (Calvenzani et al., 2012; Petroni et al., 2012; Laloum et al., 2013; Nardini et al., 2013). Protein structure analysis indicated that two conserved α-helix domains (A1 and A2) are present in the core regions of NF-YA subunits (Figure 1). The 20-amino-acid α helix A1 is in the N-terminal of the core region and functions in the interaction with NF-YB and NF-YC subunits, whereas the 21-amino-acid α helix A2 is in the C-terminal and provides sequence-specificity for recognition and binding of the CCAAT cis-element (Petroni et al., 2012; Laloum et al., 2013).

The NF-YB and NF-YC subunits contain the highly conserved Histone Fold Domain (HFD, also termed as Histone Fold Motif, HFM) (Figure 1), which functions in protein-DNA and protein-protein interactions (Frontini et al., 2004; Kahle et al., 2005; Laloum et al., 2013). The HFD domains of NF-YB and NF-YC subunits are more closely related to the core histone H2B and H2A, respectively (Dolfini et al., 2012; Petroni et al., 2012; Laloum et al., 2013). Typically, the HFD domain of NF-YB/YC is formed by a minimum of three α-helices (α1, α2, and α3) separated by two loops, similar to the HFD domains of histones H2B and H2A, and even similar to the HFD domain of the β and α subunits of NC2 (Arents and Moudrianakis, 1995; Mantovani, 1999). In NF-YB, the α1 helices contain the putative DNA-binding domain. In NF-YC, the α1 helices contain a few conserved amino acids which supposed to be the putative DNA-binding domains (Laloum et al., 2013). Indeed, the α1 helix makes essential stabilizing contacts with the DNA backbone flanking the central CCAAT pentamer (Nardini et al., 2013). NF-YB and NF-YC form a heterodimer via their HFDs, specifically via α2 and α3 (Kim et al., 1996; Zemzoumi et al., 1999; Frontini et al., 2004). The fourth helix of the conserved domain of NF-YB and NF-YC (called αC) has been proposed to function in interactions with other protein in mammals (Romier et al., 2003; Laloum et al., 2013). Formation of NF-YB/YC heterodimers is necessary for the translocation of NF-YB from the cytoplasm to the nucleus since NF-YB family members lack a nuclear localization signal (NLS), in contrast to the NF-YA and NF-YC subunits (Liu and Howell, 2010; Hackenberg et al., 2012). The HFD domains of NF-YB and NF-YC subunits assemble on a head-to-tail fashion to form a dimer, thus producing a structural scaffold for DNA-binding and interaction with NF-YA subunits (Nardini et al., 2013).

Crystal structure analysis of the NF-Y/CCAAT complex in mammals indicated that NF-YA binds to NF-YB/YC and inserts an α-helix into the minor groove of DNA, thus providing a sequence-specific contact to the CCAAT box. By contrast, NF-YB/NF-YC subunits form heterodimers that bind to the DNA sugar-phosphate backbone, mimicking the nucleosome H2A/H2B-DNA assembly (Romier et al., 2003; Nardini et al., 2013; Nardone et al., 2016). NF-YB/NF-YC interacts with DNA through non-specific HFD-DNA contact, which is similar to those of the core histone H2A/H2B. NF-YB/NF-YC also interact with NF-YA to produce the NF-Y complex with its sequence-specific recognition properties and nucleosome-like capabilities of non-specific DNA-binding. Thus combinations of subunits lead to stable DNA-binding activity of the NF-Y complex (Oldfield et al., 2014). Recently, the first crystal structure of the Arabidopsis NF-YB6/NF-YC3 dimer has been solved, which revealed that AtNF-YB6 and AtNF-YC3 subunits interact in a head-to-tail fashion and form a classical histone-like pair (Gnesutta et al., 2016; Nardone et al., 2016). The structural information on the NF-Y/CCAAT complex has given crucial insight into the molecular mechanism responsible for the architecture of the NF-YB/NF-YC heterodimer and the NF-Y heterotrimer, and also for the capacity of recognition and binding to DNA of NF-Y complexes.

There are 30 predicted NF-Y members in the Arabidopsis genome; in theory, this could result in about 1000 heterotrimeric combinations (Petroni et al., 2012). To define the molecular mechanism of NF-Y in plants, two fundamental questions about NF-Y complexes should be considered. First, how many heterotrimeric combinations actually exist in vivo and, second, what determines the specificity of those interactions and of DNA-binding. To define the unique NF-Y complexes, tissue- and development-specific expression patterns for all the subunits of Arabidopsis NF-Y have been investigated using stable promoter:beta-glucuronidase (GUS) fusion reporter lines (Siefers et al., 2009). The different subunits of NF-Y are specifically expressed in different tissues and organs during special developmental stages (Gusmaroli et al., 2001, 2002; Siefers et al., 2009; Cao et al., 2011; Sorin et al., 2014), or in response to environmental changes (Pant et al., 2009; Des Marais et al., 2012), suggesting that only some combinations of NF-Y subunits can be assembled and act in different developmental stages or under certain stimuli or conditions. In addition, yeast two-hybrid and three-hybrid systems have also been used to detect protein-protein interactions and formation of heterotrimers between Arabidopsis NF-Y subunits (Calvenzani et al., 2012; Hackenberg et al., 2012; Sato et al., 2014). Yeast two-hybrid assays indicated that most of the NF-Y subunits of the same type (A-A, B-B, or C-C) could not form homodimers or heterodimers, and NF-YA and NF-YB show little or no interaction. However, NF-YB and NF-YC subunits could interact in many combinations and form different heterodimers (Calvenzani et al., 2012; Hackenberg et al., 2012). These systematic studies of the protein interactions between NF-Y subunits have facilitated the discovery of complete NF-Y complexes, but tissue, development, and even time-dependent specific combinations should also be considered.

Research based on yeast three-hybrid and co-immunoprecipitation assays showed that NF-Y could function as a NF-YA-YB-YC heterotrimeric complex in vivo (Table 1) (Hou et al., 2014; Sato et al., 2014). In theory, Arabidopsis NF-Y subunits can form 1000 different heterotrimeric combinations (Petroni et al., 2012); however, only a few NF-Y complexes have been verified through molecular and biochemical studies (Table 1). One possible reason is that some or many heterotrimeric of NF-Y may be transient and highly dynamic in vivo, and thus hard to detect. Nevertheless, with the emergence of new experimental methods and in-depth studies of NF-Y family members, more and more NF-Y complexes are being identified and explored.

In the past few years, a growing number of studies have reported that subunits of NF-Y not only form heterodimeric and heterotrimeric complexes, but also interact with other proteins in various kinds of complexes. For example, multiple NF-YC subunits (C1, C3, C4, and C9) and NF-YB subunits (B2 and B3) can interact with CONSTANS (CO) to regulate flowering time (Wenkel et al., 2006; Kumimoto et al., 2010; Cao et al., 2014; Xu F. et al., 2016). In addition, basic region/leucine zipper motif (bZIP) type transcription factors bZIP28 and bZIP67 interact with NF-Y subunits to regulate endoplasmic reticulum (ER) stress and seed development, respectively (Yamamoto et al., 2009; Liu and Howell, 2010). Phytochrome interacting factor 4 (PIF4) interacts with LEAFY COTYLEDON1 (LEC1/NF-YB9) to regulate hypocotyl elongation (Huang et al., 2015b). REPRESSOR OF ga1-3 (RGA) and RGA-LIKE2 (RGL2) interact with NF-YA2, NF-YB2/B3, or NF-YC3/C4/C9 to regulate gibberellin (GA) and photoperiod-mediated flowering time, or abscisic acid (ABA) and GA signaling pathways during seed germination, respectively (Hou et al., 2014; Liu et al., 2016). Studies of the interaction of NF-Y with other proteins regulating target gene expression have largely improved our understanding of the physiological roles of NF-Y in many biological activities (Table 1).

Since NF-Y regulates gene expression mainly as a protein complex, it is important to understand its specific regulatory mechanisms. Previous reports have proposed that NF-Y modulates expression of downstream target genes mainly by two mechanisms (Figure 2). In the first mechanism, the NF-YB-YC heterodimer assembles in the cytoplasm, then translocates into the nucleus, where it interacts with NF-YA to form an active heterotrimer (Figure 2A) (Hackenberg et al., 2012; Laloum et al., 2013). The NF-YA-YB-YC complex mainly binds to the CCAAT box in the promoter regions of the downstream target genes through NF-YA, and regulates the expression of target genes, which is a highly conserved transcriptional regulation mechanism for NF-Y complexes in yeast, mammals, and plants (Mantovani, 1999; Frontini et al., 2004; Liu and Howell, 2010; Dolfini et al., 2012; Petroni et al., 2012; Hou et al., 2014; Sato et al., 2014). For example, the AtNF-YA4-YC2-YB3 complex binds to the CCAAT cis-element in the promoter region of BINDING PROTEIN 3 (BiP3) through AtNF-YA4 and up-regulates the expression of ER stress-induced genes (Liu and Howell, 2010). In addition, the AtNF-YA6 subunit interacts with the AtNF-YB6/YC3 dimer and the resulting trimer can directly bind CCAAT cis-elements, even a probe containing the CCAAT box derived from the human Heat Shock Protein70 (HSP70) promoter (Nardone et al., 2016). Most NF-Y complexes bind to the CCAAT box mainly through NF-YA; however, in plants, a recent study in rice reported that OsNF-YB1 might bind to the CCAAT box of Sucrose Transporter1 (SUT1), SUT3, and SUT4 (Bai et al., 2016). However, the in vitro and in vivo data are not sufficient to prove OsNF-YB1 binding to CCAAT box directly, since some sequences flanking the CCAAT box could bind via NF-YB (Nardini et al., 2013).

In the second molecular mechanism, the NF-YB/NF-YC heterodimer associates with specific transcription factors to form a complex (NF-YB-YC-transcription factor), and regulates the expression of their target genes through the binding of transcription factors to specific cis-elements in the promoters of various target genes (Wenkel et al., 2006; Yamamoto et al., 2009; Kumimoto et al., 2010) (Figure 2B). In this mechanism, NF-YA probably could suppress the formation of the NF-YB-YC-transcription factor trimeric complex by preventing the combination of transcription factors with NF-YB-YC. Although this kind of transcriptional regulation has been proposed by many studies, while the direct molecular evidences are still required since the opinion of compete of YA with TF derived from supposes upon some opposite observations. For example, AtB9-C2-bZIP67 directly binds to the ABA-response elements (ABREs) through bZIP67 in the promoter regions of SUCROSE SYNTHASE 2 (SUS2) and CRUCIFERIN C (CRC), then activates the expression of SUS2 and CRC and promotes seed development (Yamamoto et al., 2009). However, NF-YA subunits strongly inhibited the expression of CRC by competing with bZIP67 to form an NF-YA-YB9-YC2 complex, which indicated different members of the NF-Y subunits play distinct roles in plant development and growth (Adrian et al., 2010). In mammals, genome-wide association with selected transcription factors indicated that the NF-Y complex directly binds to the CCAAT box and associates with other cis-elements by interacting with different transcription factors (Dolfini et al., 2016; Zambelli and Pavesi, 2016). Recently, a ChIP-sequencing approach in rice identified genome-wide downstream targets of OsNF-YB1, indicating that OsNF-YB1 directly binds to the CCAAT box and associates with the GCC box by interacting with an ERF-type transcription factor (Xu J.J. et al., 2016).

Some special transcriptional regulatory mechanisms do not fit these two major regulatory models in Arabidopsis. For instance, AtNF-YA2 can directly bind to the NFYBE cis-element (not the CCAAT box) of the SUPPRESOR OF OVEREXPRESSION OF CONSTANS1 (SOC1) promoter to regulate the expression of SOC1 (Hou et al., 2014). In addition, a recent study indicated that NF-YC subunits (C1, C3, C4, and C9) directly interact with HISTONE DEACETYLASE 15 (HDA15), recognize the promoter of hypocotyl elongation-related genes, and repress their expression through the deacetylation of H4 in a light-dependent manner (Tang et al., 2016). A previous study also showed that NF-Y directly activates the expression of SOC1 partly through the H3K27 demethylase RELATIVE OF EARLY FLOWERING 6 (REF6; Hou et al., 2014). Therefore, the NF-Y complex may function together with various epigenetic factors to regulate the transcription of downstream target genes. Certainly, deeper research will discover more and more novel mechanisms.

In plant, the different individual subunits of NF-Y have attracted extensive attention in recent years, but the many functions of the NF-Y complex remain only partially defined. In contrast to yeast and mammals, plants have a large NF-Y family, generally having multi-gene families encoding each subunit. This provides more combinations of NF-YA-YB-YC complexes in different developmental stages or under certain conditions, and increases functional complexity. This indicates that NF-Y is widely involved in the intricate regulatory processes in plants, compared with its more narrow roles in other organisms.

The downstream target genes and upstream regulators of NF-Y, and its functional redundancy and specificity are largely unknown. The diversity of NF-Y subunits and their many potential combinations, as well as likely redundant and divergent functions, provide a substantial challenge for work aiming to tease apart the functions of the different subunits. In mammals, NF-Y is a pioneer factor that binds to the CCAAT box in the core promoter and enhancer region (Nardini et al., 2013; Oldfield et al., 2014), but its targets remain elusive in plants. In addition, other fundamental questions about NF-Y, such its transcriptional regulation, and post-translational modification, need to be investigated. NF-Y complexes definitely function as essential regulatory hubs for many processes in plant, but its functional redundancy remains a problem for further investigation. Therefore, future studies on NF-Y will play an indispensable part in plant science.