The Arabidopsis thaliana pifq quadruple mutant (pif1, pif3, pif4, pif5) used in this study was described by Shin et al. (2009), and the wild-type control plants used in this study were Arabidopsis ecotype Columbia-0 (Col-0). Arabidopsis seeds were surface-sterilized with 20% bleach for 20 min and then washed four times with sterile distilled-deionized H₂O. After vernalization for 2 days at 4°C, seeds were germinated on GM plates (4.74 g/L Murashige and Skoog [MS] salts, 10 g/L 1% sucrose, 0.5 g/L MES, 8 g/L agar, pH 5.8). Generally, the Arabidopsis plants were grown under long-day conditions (16 h light/8 h dark, light intensity 100 μmol m^-2 s^-1, 22°C).

Seedlings of maize inbred line B73 and Nicotiana benthamiana were grown in growth chambers under a 12-h light/12-h dark cycle at 210 μmol m^-2 s^-1 of light at 28°C. Two weeks after planting, seedlings of maize inbred line B73 were transferred to constant light conditions and then harvested at different Zeitgeber times to measure the diurnal expression of ZmPIF4 and ZmPIF5. Three weeks after planting, seedlings of the maize inbred line B73 were harvested and separated into roots, coleoptiles, stems, and leaves for detection of the tissue expression patterns of ZmPIFs.

Plant total RNA was extracted with an Ultrapure RNA Kit (CWBIO, China) following the manufacturer’s instructions. The first-strand cDNA was synthesized using EasyScript One-Step gDNA Removal and cDNA Synthesis SuperMix (TransGEN Biotech, China). The cDNA was then diluted to 60 ng/μL and 2 μL was used for qPCR. Quantitative PCR was performed using UltraSYBR Mixture (CWBIO, China) in an ABI QuantStudio 6 Flex Real-Time PCR System (ABI). The RT-qPCR procedure was described in Ma et al. (2016). The expression level of UBIQUITIN1 and 18S rRNA were used as internal controls for RT-qPCR in Arabidopsis and maize, respectively. Plant material for RT-qPCR was collected from three biological replicates, and three technical replicates were performed for each experiment.

Seedlings of Col-0, pifq mutants, and transgenic lines of ZmPIF4 and ZmPIF5 were grown on medium in the dark for 4 days, and then the direction of gravity was changed by 90° and the bending angle of the hypocotyl elongation zone was measured every hour using ImageJ program. The contents of amyloplasts in the endodermis of the hypocotyl elongation zone and the columella cells in the root cap of these seedlings were detected by I₂–KI staining following the method described by Ma et al. (2017).

To generate transgenic ZmPIF4-OE and ZmPIF5-OE lines in the pifq or wild-type Col-0 background, the coding regions of ZmPIF4 and ZmPIF5 were PCR-amplified from cDNA of the maize inbred line B73 with the primer pairs ZmPIF4-F and ZmPIF4-R, and ZmPIF5-F and ZmPIF5-R (see Supplementary Table S2 for more information on all of the primers used in this study). The fragments of ZmPIF4 and ZmPIF5 were then inserted into a BamHI and SpeI digested pPZP211-35Spro::3FLAG empty binary vector (Ma et al., 2016) to produce 35Spro::ZmPIF4-3FLAG and 35Spro:ZmPIF5-3FLAG. Finally, these two plasmids were transformed into the pifq mutant plants and the Col-0 plants, to generate ZmPIF4-OE/pifq, ZmPIF5-OE/pifq, ZmPIF4-OE, and 35Spro:ZmPIF5-OE lines. More than 20 independent lines of each transformation were selected with kanamycin and verified by RT-qPCR and western blot assays. The procedure for the western blot was described previously (Ma et al., 2017). All immunoblots were repeated at least twice and a representative experiment is shown (Supplementary Figure S5 ).

Plasmids 35Spro::ZmPIF4-GFP and 35Spro:ZmPIF5-GFP were generated by PCR-amplifying the coding region of ZmPIF4 and ZmPIF5 from 35Spro::ZmPIF4-3FLAG or 35Spro:ZmPIF5-3FLAG using primer pairs GFP-ZmPIF4-F and GFP-ZmPIF4-R, and GFP-ZmPIF5-F and GFP-ZmPIF5-R. The fragments containing the coding region of ZmPIF4 and ZmPIF5 were then inserted into pPZP211-35Spro::GFP digested with BamHI and XbaI. The various plasmids were then transformed into the Agrobacterium strain GV3101 and infiltrated into leaves of N. benthamiana together with the 35Spro::H2B-Cherry (Howe et al., 2012). Three days after infiltration, the N. benthamiana plants were transferred into dark conditions for 12 h, the lower epidermis of the infiltrated leaf was peeled, and the fluorescence signals were observed with a two-photon laser confocal microscope (Zeiss).

For subcellular localization analysis of ZmPIF4 and ZmPIF5 in protoplasts, 30-day-old Arabidopsis wild-type plants (Col-0) grown under short day (8-h light/16-h dark) conditions were used to isolate the protoplasts. Plasmids 35Spro::ZmPIF4-GFP and 35Spro:ZmPIF5-GFP were transformed into the protoplasts using PEG method (Yoo et al., 2007). Incubated for 18 h in darkness to allow the constructs to be expressed, and then the fluorescent signals were observed with a laser confocal microscope (Zeiss).

The plasmids pGBKT7-ZmPIF4 and pGBKT7-ZmPIF5 were generated by PCR-amplifying the coding region of ZmPIF4 and ZmPIF5 from 35Spro::ZmPIF4-3FLAG and 35Spro::ZmPIF5-3FLAG using the primer pairs ADBD-ZmPIF4-F and ADBD- ZmPIF4-R, and ADBD-ZmPIF5-F and ADBD-ZmPIF5-R. The fragments containing the coding region of ZmPIF4 and ZmPIF5 were inserted into pGBKT7 (Clontech) digested with BamHI and NdeI to produce pGBKT7-ZmPIF4 and pGBKT7-ZmPIF5.

The plasmid pGADT7-RGA was generated by PCR-amplifying the coding region of RGA from Arabidopsis Col-0 with the primer pair ADBD-RGA-F and ADBD-RGA-R. Then, the fragment containing the RGA coding region was inserted into pGADT7 (Clontech) digested with BamHI and SacI to produce pGADT7-RGA. The yeast two-hybrid procedure was performed following the manufacturer’s instructions (Clontech).

The plasmids nLUC-ZmPIF4, cLUC-ZmPIF4, nLUC-ZmPIF5, and cLUC-ZmPIF5 were generated by PCR-amplifying the coding regions of ZmPIF4 and ZmPIF5 from 35Spro::ZmPIF4-3FLAG and 35Spro::ZmPIF5-3FLAG using the primer pairs LCI- ZmPIF4-F and LCI-ZmPIF4-R, and LCI-ZmPIF5-F and LCI-ZmPIF5-R. The fragments containing the coding regions of ZmPIF4 and ZmPIF5 were inserted into KpnI and BamHI digested pCAMBIA1300-nLUC and pCAMBIA1300-cLUC vectors (Chen et al., 2008) to produce nLUC-ZmPIF4, cLUC-ZmPIF4, nLUC-ZmPIF5, and cLUC-ZmPIF5. The coding region of RGA was PCR-amplified from pGADT7-RGA with the primer pair LCI-RGA-F and LCI-RGA-R, and inserted into KpnI and SalI digested pCAMBIA1300-nLUC and pCAMBIA1300-cLUC vectors to produce nLUC-RGA and cLUC-RGA. Each of the nLUC- or cLUC- fused plasmids was transformed into Agrobacterium strain GV3101 and then infiltrated into leaves of N. benthamiana. Three days after infiltration the infiltrated plants were transferred into darkness for 12 h and the luciferase activity was measured using an in vivo imaging system (XENOGEN) with potassium luciferin as the substrate.

To generate plasmids ZmPIF4-YFP^N and RGA-YFP^C , ZmPIF4 and RGA were amplified with the primer pairs YFP-ZmPIF4-F and YFP-ZmPIF4-R, and YFP-RGA-F and YFP-RGA-R. Full-length cDNAs of ZmPIF4 and ZmPIF5 were subcloned into the TOPO vectors and then recombined into pSITE-nEYFP and pSITE-cEYFP (Martin et al., 2009), respectively. The plasmids ZmPIF4-YFP^N and RGA-YFP^C were then transformed into the Agrobacterium strain GV3101 and infiltrated into leaves of N. benthamiana. Three days after infiltration, the fluorescent signals were observed with a laser confocal microscope (Zeiss).

To identify the putative PIFs family proteins in maize, we performed BLASTP analysis using the PIF protein sequences of Arabidopsis as query sequences. This identified 15 putative ZmPIF family proteins in maize (Supplementary Table S1 ). A phylogenetic tree with the protein sequences of PIF family proteins from Arabidopsis and maize showed that the putative ZmPIFs were closely related to AtPIFs (Figure 1A ). Analysis of the conservation of the protein sequence of each putative PIF showed that seven proteins (ZmbHLH16, ZmbHLH27, ZmbHLH36, ZmbHLH76, ZmbHLH115, ZmbHLH165, and ZmbHLH198) have a highly-conserved APB domain, and three (ZmbHLH76, ZmbHLH165, and ZmbHLH198) have a highly-conserved APA domain (Figure 1B ). Both APA and APB domains were also identified in the ZmPIF3.1 and ZmPIF3.2, two previously reported PIF family proteins in maize encoded by ZmbHLH76 and ZmbHLH165, respectively (Figure 1B , left lower panel). Multiple alignments with the full-length protein sequences of AtPIF4, AtPIF5, ZmbHLH16, ZmbHLH27, and the putative PIF4 and PIF5 proteins in other land plants showed that ZmbHLH16 (GRMZM5G865967) and ZmbHLH27 (GRMZM2G165042) are very similar to AtPIF4 and AtPIF5 (Figure 1C and Supplementary Figure S1 ). ZmbHLH16 and ZmbHLH27 were therefore renamed ZmPIF4 and ZmPIF5, respectively.

RT-qPCR analysis showed that both ZmPIF4 and ZmPIF5 are expressed in the roots, stems, coleoptiles, and leaves at the six-leaf stage (Figure 2A ). Under 12 h light/12 h dark diurnal conditions, RT-qPCR analysis also demonstrated that ZmPIF4 and ZmPIF5 have similar temporal expression patterns, with their transcript levels increasing overnight, peaking at dawn, and decreasing during the day (Figure 2B ). In addition, to verify whether the transcript level of ZmPIF4 and ZmPIF5 is regulated by R or FR light, three-leaf stage seedlings of maize inbred line B73 grown under white light conditions were transferred to FR light for 1 h and then transferred to R light for various times. RT-qPCR analyses revealed that the ZmPIF4 and ZmPIF5 transcript levels were rapidly induced by FR light, but repressed by R light (Supplementary Figure S2A ).

Further, we investigated the protein subcellular localization of ZmPIF4 and ZmPIF5 in Arabidopsis protoplasts and N. benthamiana epidermal cells. As shown in Figure 2C and Supplementary Figure S2B , the florescent signals of ZmPIF4-GFP and ZmPIF5-GFP fusion proteins were only observed in the nucleus. This suggests that both ZmPIF4 and ZmPIF5 localize in the nucleus, consistent with a potential function as transcription factors.

In Arabidopsis, the pifq quadruple mutant displays a constitutive photomorphogenic phenotype under continuous dark conditions, including short hypocotyls, open cotyledons, and the loss of negative gravitropism (Shin et al., 2009). To examine whether ZmPIF4 and ZmPIF5 could have PIF function, we tested whether they could complement the pifq phenotype by generating ZmPIF4-OE/pifq and ZmPIF5-OE/pifq plants. RT-qPCR analysis showed that the transcripts of ZmPIF4 and ZmPIF5 were present at high levels in the transgenic ZmPIF4-OE/pifq and ZmPIF5-OE/pifq plants (Figure 3E ). The hypocotyl length of pifq was significantly shorter than that of Col-0, consistent with previous reports (Leivar et al., 2008). When ZmPIF4 and ZmPIF5 were over-expressed in the pifq background, their hypocotyls were significantly elongated relative to pifq (Figures 3A–D ). The cotyledons were completely closed in ZmPIF4-OE4/pifq and ZmPIF4-OE5/pifq, and only some of the plants in the ZmPIF5 transgenic line had open cotyledons. This may be related to the level of protein expression. The partial complementation supports the hypothesis that both ZmPIF4 and ZmPIF5 have physiological functions similar to those of AtPIF4 and AtPIF5.

Next, we investigated whether the ZmPIF4 and ZmPIF5 proteins are light labile using the Arabidopsis transgenic lines expressing ZmPIF4 and ZmPIF5. As shown in Figure 3F , ZmPIF4 and ZmPIF5 proteins accumulate to high levels at night, peak at ZT20, and are present at low levels during the day. Furthermore, we checked whether the levels of ZmPIF4 and ZmPIF5 proteins are more stable under dark conditions. The ZmPIF4 and ZmPIF5 transgenic seedlings were transferred from light (at ZT4) to darkness for various times and then used to perform western blots. As shown in Figure 3G , the accumulation of ZmPIF4 and ZmPIF5 proteins increased in the darkness. All these results indicate that ZmPIF4 and ZmPIF5 might be light labile and subject to degradation during the day, similar to AtPIF4 and AtPIF5.

In the Arabidopsis pifq mutant, the negative gravitropism of hypocotyls is completely disrupted in dark conditions (Kim et al., 2011). Overexpression of ZmPIF4 and ZmPIF5 partially restored the negative gravitropism of hypocotyls in darkness, especially in ZmPIF4-OE/pifq (Figure 4A ). This was further supported by growing seedlings of Col-0, pifq mutants, and transgenic lines of ZmPIF4 and ZmPIF5 on medium for 4 days in the dark, then changing the direction of gravity by 90° and measuring the bending angle of the hypocotyl elongation zone every hour (Figures 4B,C ). The ZmPIF4 transgenic lines were able to respond quickly to gravity, but the pifq mutant and the ZmPIF5-OE transgenic lines responded more slowly (Figures 4B,C ).

In darkness, accumulation of PIF proteins suppresses the conversion of starch-filled endodermal amyloplasts to plastids and thus plays an important role in the plant’s response to gravity (Kim et al., 2011). To test whether ZmPIF4 or ZmPIF5 could complement the amyloplast defect in pifq mutants, we detected amyloplasts by I₂–KI staining. Amyloplasts were detected in both the hypocotyl and the root cap in Col-0 plants, but only in the root caps of the pifq mutants, consistent with previous studies (Kim et al., 2011). In the seedling plants of ZmPIF4-OE4/pifq and ZmPIF4-OE5/pifq, amyloplasts were detected in both the endodermis of the hypocotyl elongation zone and in the columella cells of the root cap. Indeed, the transgenic lines of ZmPIF4 stained more strongly, compared to the ZmPIF5 transgenic lines (Figures 4D,E ). This indicates that both ZmPIF4 and ZmPIF5 might be involved in the plant response to gravity, with ZmPIF4 playing a primary role.

To test whether ZmPIFs could regulate skotomorphogenesis and the shade avoidance response, we next overexpressed FLAG-tagged versions of the ZmPIFs (ZmPIF4-3FLAG and ZmPIF5-3FLAG) under the control of the constitutive 35S promoter in the Arabidopsis wild-type control Col-0 plants. Three independent transgenic ZmPIF4 over-expression lines (OE8, OE10, and OE11), and ZmPIF5 over-expression lines (OE2, OE3, and OE12) were selected and used to perform further analysis (Supplementary Figure S3A ). The hypocotyls of all the transgenic lines were significantly longer than those of Col-0 in continuous dark conditions, which indicates that both ZmPIF4 and ZmPIF5 can participate in skotomorphogenesis in Arabidopsis (Figures 5A–D ).

Further, we checked the hypocotyl elongation phenotype of all the ZmPIF4 and ZmPIF5 transgenic lines in the pifq mutant and Col-0 backgrounds under long-day conditions with high (Supplementary Figure S3B ) or low R:FR (Figures 5E,F ), respectively. Under long-day conditions with high R:FR (white light, R:FR = 5), the ZmPIF4 and ZmPIF5 transgenic lines displayed shorten hypocotyls, without significant differences between ZmPIF4 and ZmPIF5 (Supplementary Figure S3B ). By contrast, under long-day conditions with low R:FR (F:FR = 0.35), the transgenic lines of ZmPIF4, but not ZmPIF5, in both pifq mutant and Col-0 backgrounds displayed longer hypocotyls, compared with the control plants (Figures 5E,F ). This suggests that ZmPIF4 is involved in shade avoidance responses. Furthermore, RT-qPCR analyses revealed that the transcript levels of PIL1, PIL2, HFR1, IAA19, and TAA1 in 7-day-old seedlings grown under low R:FR conditions were significantly increased in ZmPIF4-OE4/pifq and ZmPIF4-OE5/pifq, and ZmPIF4-OE8 and ZmPIF4-OE10, compared with pifq mutant or Col-0 wild-type control plants, respectively (Figure 5F ). These observations suggested that ZmPIF4 regulates cell elongation of the hypocotyl under low R:FR conditions, probably by promoting the expression of genes related to the shade avoidance response.

To further verify the physiological function of ZmPIF4 in the shade avoidance response, we examined the phenotype of adult plants of the ZmPIF4 transgenic line grown under long-day with high R:FR conditions. As shown in Figures 6A,B , three independent transgenic overexpression lines of ZmPIF4 showed earlier flowering times and had fewer rosette leaves at flowering, compared with wild-type control plant Col-0. In addition, the phenotypes of continuous shade avoidance, including elongated petioles, reduced leaf area, accelerated leaf senescence, slender inflorescences, and plant lodging were observed in the ZmPIF4 over-expression lines (Figures 6C–F and Supplementary Figure S4 ), but not the ZmPIF5 overexpression lines (data not shown). The chlorophyll and carotenoid contents were lower in the ZmPIF4 overexpression lines (Figure 6E ), and RT-qPCR analysis revealed that the transcript levels of the chlorophyll biosynthesis genes GUN4, HEMA1, and CHLH were significantly decreased in adult plants of the ZmPIF4 overexpression lines (OE8 and OE10), compared with wild-type control plant Col-0 (Figure 6G ). All these results suggested that overexpression of ZmPIF4 resulted in a constitutive shade avoidance response, even under high R:FR conditions.

To examine whether ZmPIF4 can interact with the GA signaling pathway to coordinately regulate hypocotyl elongation and the shade avoidance response, we first tested whether it can interact with the DELLA protein REPRESSOR OF GA1-3 (RGA), a negative regulator of GA signaling that interacts with Arabidopsis PIFs. Yeast two-hybrid assays indicated that ZmPIF4 can interact with RGA directly in vitro (Figure 7A ). Further, luciferase complementation imaging (LCI) assays and bimolecular fluorescence complementation (BiFC) assays showed that ZmPIF4 can interact with RGA in planta (Figures 7B,C ). In addition, the hypocotyl elongation of pifq mutant is less responsive to GA treatment, compared with wild-type control plant Col-0 under darkness conditions. Overexpression of ZmPIF4 in the pifq mutant background completely rescued the lack of response to GA treatment (Figures 7D,E ), which suggests that ZmPIF4 might affect the GA signaling pathway, possibly by interacting with RGA to influence hypocotyl elongation.