Directed dimerization: an in vivo expression system for functional studies of type II phytochromes



Type II phytochromes (phy) in Arabidopsis form homodimers and heterodimers, resulting in a diverse collection of light-stable red/far-red (R/FR) sensing photoreceptors. We describe an in vivo protein engineering system and its use in characterizing the activities of these molecules. Using a phyB null mutant background, singly and doubly transgenic plants were generated that express fusion proteins containing the phyB–phyE N–terminal photosensory regions (NB–NE PSRs), a nuclear localization sequence, and small yeast protein domains that mediate either homodimerization or heterodimerization. Activity of NB/NB homodimers but not monomeric NB subunits in control of seedling and adult plant responses to R light is demonstrated. Heterodimers of the NB sequence with the chromophoreless NBC357S sequence, which mimic phyB Pfr/Pr photo-heterodimers, mediate R sensitivity in leaves and petioles but not hypocotyls. Homodimerization of the NC, ND and NE sequences and directed heterodimerization of these photosensory regions with the NB region reveal form-specific R-induced activities for different type II phy dimers. The experimental approach developed here of directed assembly of defined protein dimer combinations in vivo may be applicable to other systems.


Phytochromes (phy) are dimeric plant photoreceptors for red (R) and far-red (FR) light that undergo a reversible photo-induced conformational change between an inactive R-absorbing Pr form and an active FR-absorbing Pfr form (Quail, 2002). Phy-related R/FR light-sensing receptors have been identified in eubacteria, cyanobacteria, fungi and plants (Bhoo et al., 2001; Sharrock, 2008). Phy apoproteins are modular: they contain a covalently attached linear tetrapyrrole chromophore embedded in a 500–700 amino acid N–terminal globular photosensing region (PSR) (Quail et al., 1995; Rockwell et al., 2006). This conserved PSR comprises three major protein domains: PAS (Per/ARNT/SIM), GAF (cGMP phosphodiesterase/adenylcyclase/FhlA) and PHY (Phy-associated; Sharrock, 2008). The N–terminal PSR contains all of the amino acid sequences required to mediate reversible Pr–Pfr conformational photo-transformation. In bacteria, fungi and algae, the C–terminal modules of phy apoproteins are light-regulated two-component histidine kinases (Yeh et al., 1997; Davis et al., 1999; Froehlich et al., 2005). In higher plants, phy C–terminal modules have a more complex domain structure, containing two adjacent PAS domains and a histidine kinase-related sequence that lacks some amino acid residues essential to histidine kinase activity but has an ATP-binding domain-like sequence. The functions of the two phy ends are independent to a great extent. The N–terminal PSR mediates R light sensing/signaling (Matsushita et al., 2003), and the C–terminal region mediates dimerization and nuclear translocation (Chen et al., 2003). When expressed in plants, the PSR of phyB (NB), without the C–terminal module, confers R light sensitivity but only when it is fused to a heterologous protein that causes the fusion protein to dimerize (β–glucuronidase) and a nuclear-localization signal (NLS) that promotes import into the nucleus (Matsushita et al., 2003).

Multiple PHY genes, such as the Arabidopsis PHYA–PHYE genes, are expressed in plant cells. The phy apoproteins encoded by these genes have distinct regulatory functions, as determined by analysis of their respective null mutants (Reed et al., 1994; Aukerman et al., 1997; Devlin et al., 1998, 1999; Monte et al., 2003). The most striking difference is between the function of phyA, which mediates very low fluence responses to a broad range of light colors and high-irradiation responses to continuous FR, and the functions of phyB–phyE, which are involved in R/FR reversible responses and shade sensing (Mathews, 2006). Because of this functional distinction, phyA is referred to as a type I phy and phyB–phyE are referred to as type II phytochromes. An additional factor that complicates the plant phy photoreceptor array is that the type II phytochromes associate in many heterodimer forms (Sharrock and Clack, 2004; Clack et al., 2009). It is not clear whether each of these phy forms has a distinct regulatory activity.

We have used a synthetic biology approach to characterize the functions of individual homodimeric and heteromeric type II Arabidopsis phytochromes. Small protein–protein interaction domains from the yeast GAL4, Bem1 and Cdc24 proteins were substituted for the phy C–termini to produce either obligate homodimers or obligate heterodimers of the N–terminal phy PSRs (NX). The engineered NB/NB homodimers and ND/NB heterodimers are highly active as photoreceptors, complementing phyB mutant seedling de-etiolation and flowering phenotypes. Dimers of NB with chromophoreless NBC357S subunits and the NC/NB and NE/NB heterodimers are active in petioles and cotyledons but not in hypocotyls. In addition, hyperactivity to low R levels is induced by several of these constitutively nuclear fusion constructs, indicating that nuclear translocation of phy may be limiting for plant photo-responses when the Pfr/Ptotal ratio is low. These findings confirm and extend the model for modular organization of plant phy proteins, and illustrate the utility of a directed protein dimerization system, here used in plants but potentially applicable in many in vivo systems, to study the roles of protein quaternary interactions.


The PB1 domains of the yeast Bem1 and Cdc24 proteins form obligate heterodimers

Analysis of the establishment of cell polarity in budding yeast (Saccharomyces cerevisiae) has shown that the C–terminal PB1 domains of the Bem1 and Cdc24 proteins interact directly with each other to form a Bem1/Cdc24 heteromeric complex (Ito et al., 2001). To test whether the Bem1 and Cdc24 dimerization domains bind to themselves in addition to binding to each other, yeast two-hybrid assays were performed with varying sizes of PB1 domain protein sequences from Bem1 (Bem–317, Bem–128 and Bem–82) and Cdc24 (Cdc–182 and Cdc–95). As shown in Figure S1, no interaction of any of the Bem1 or Cdc24 regions with themselves was detected, whereas binding interactions between all combinations of Bem1 PB1 domains with Cdc24 PB1 domains were observed. Strong interaction was also observed in yeast two-hybrid assays when the Bem–82 and Cdc–95 sequences were fused to the N–terminal 651 amino acid phyB sequence (NB–Bem–82 and NB–Cdc–95). These results suggest that these small yeast protein domains may be used to assemble client protein domains, such as the phy N–terminal PSR modules, into dimers in a directed manner in vivo.

Bem/Cdc domain-directed dimerization of the NB PSR in vivo

As illustrated in Figure 1(a), transgenes referred to as NB–Bem and NB–Cdc were constructed, in which epitope-tagged nuclear-localized phyB fusion proteins are expressed from the CaMV 35S promoter. The NB–Bem transgene product consists of the phyB PSR (amino acids 1–651, NB) fused to Bem–82, followed by a Myc6 epitope tag and the SV40 NLS. The NB–Cdc transgene product comprises NB fused to the Cdc–95 region, followed by the His6 epitope tag and the SV40 NLS. The NB PSR contains the native phyB N–terminus, the PAS, GAF and PHY domains, and the approximately 20 amino acid protease-sensitive ‘hinge’ region, but none of the downstream C–terminal domains. A fluorescent protein tag was not included in these constructs because of the possibility that it may influence dimerization (Smith and Ward, 1998). The NB chimeric coding sequences were inserted into vectors that confer either kanamycin or gentamicin resistance in plants. Singly and doubly transgenic Arabidopsis plants were generated in which the 35S:NB–Bem and 35S:NB–Cdc transgenes are expressed individually or co-expressed in a phyB null mutant background. Single-locus phyB(NB–Bem) T3 lines and phyB(NB–Cdc) T3 lines were identified and crossed with each other to produce doubly transgenic true-breeding phyB(NB–Cdc/NB–Bem) F3 progeny lines.

Figure 1.

Structure and expression of phy PSR fusion transgenes. (a) Structures of the four groups of chimeric phytochrome transgene protein products: NB–Bem, NX–Cdc, NX–GALMyc6 and NX-GALHis6 (X indicates multiple phy N–terminal PSR coding sequences). A representation of the predicted Cdc/Bem-directed NX/NB heterodimer is shown. Bem and Cdc are the Bem–82 and Cdc–95 regions, respectively (see Figure S1); GAL is the yeast GAL4 homodimerization domain. (b) Immunoblot analysis of the levels of the NB–Bem and NB–Cdc fusion proteins in singly and doubly transgenic lines. Two full-length phyB-expressing lines, phyB(PHYB–Myc6) line 210 and phyB(His6PHYB) line 336, were used as internal standards for quantification (Figure S2). Extracts from 5-day-old white light-grown seedlings of the indicated lines were fractionated by SDS–PAGE. Blots of the gels were probed with the anti-His6 or anti-Myc antibody as indicated. The numbers above the lanes indicate independent lines, and those below the lanes indicate the expression levels in the transgenic lines relative to the normal wild-type phyB level. The asterisk indicates a non-specific band detected by the anti-His6 antibody.

Extracts of seedlings of the various lines were analyzed by immunoblotting with anti-Myc and anti-His6 antibodies (Figure 1b). In order to compare the expression levels of these proteins with the normal wild-type phyB level, extracts of phyB(PHYB–Myc6) line 210 and phyB(His6PHYB) line 336, in which epitope-tagged full-length (FL) phyB coding sequences are expressed from the PHYB promoter, were included on the blots. Both of these full-length phyB proteins complement the phyB mutant phenotype, and their expression levels were quantified relative to the native WT phyB level on immunoblots using anti-phyB monoclonal antibody B6B3 (Figure S2). The expression levels of the NB chimeric proteins, which lack the anti-phyB monoclonal antibody B6B3 epitope, were then assessed by comparison with these proteins on immunoblots that detect the epitope tags (Figure 1b). Despite being transcribed from the 35S promoter, the NB–Bem and NB–Cdc proteins are present at steady-state levels that are similar (0.8–1.2-fold) to native phyB in NB–Bem and NB–Cdc plant lines. The phyB(NB–Cdc/NB–Bem) lines contain 2.5–4-fold higher levels of both the NB–Cdc and NB–Bem proteins than the progenitor singly transformed lines, indicating that dimerization stabilizes these chimeric proteins. A 35S:NBC357S–Cdc transgene, with the same sequence as NB–Cdc but containing a mutation changing the chromophore attachment cysteine residue at position 357 to serine, was constructed and transformed into the phyB host and phyB(NB–Bem) line 7. The phyB(NBC357S–Cdc) monomer lines and the phyB(NBC357S–Cdc/NB–Bem) directed dimer lines contain similar transgene product levels to the corresponding phyB(NB–Cdc) and phyB(NB–Cdc/NB–Bem) lines, and the fusion proteins are stabilized when they are co-expressed (Figure 1b).

To demonstrate Bem/Cdc domain-mediated dimerization of the transgene products, extracts of phyB(NB–Bem), phyB(NB–Cdc) and phyB(NB–Cdc/NB–Bem) seedlings were fractionated by size-exclusion chromatography (SEC). Figure 2(a) shows that the NB–Bem (93.2 kDa) and NB–Cdc (85.2 kDa) proteins are recovered in fractions consistent with their monomer molecular mass when expressed by themselves, but are recovered in fractions consistent with a dimer molecular mass (178.4 kDa) when co-expressed. To confirm this, the NB–Bem protein was immunoprecipitated with anti-Myc antibody from extracts of the phyB(NB–Cdc/NB–Bem) and phyB(NBC357S–Cdc/NB–Bem) lines. Figure 2(b) shows that the His6-tagged NB–Cdc and NBC357S–Cdc proteins co-precipitate with Myc-tagged NB–Bem.

Figure 2.

Demonstration of Cdc/Bem domain and GAL domain-directed dimerization in vivo. (a) Size-exclusion chromatography (SEC) analysis. Extracts of white light-grown seedlings were fractionated on a Superose 6 SEC column. Column fractions were separated by SDS–PAGE gels, blotted and probed with anti-Myc or anti-His6 antibodies. Profiles of scanned immunoblots for the NB–Bem, NB–Cdc, NB–Cdc/NB–Bem, ND–Cdc/NB–Bem and NB–GALH transgene products from the indicated transgenic lines are shown. (b) Immunoblots of extracts and anti-Myc antibody immunoprecipitates of lines expressing the indicated transgene proteins. Seedlings were grown for 5 days under continuous white light, extracts were prepared and immunoprecipitated using anti-Myc antibody, and immunoblotting was performed using the anti-His6 antibody.

Dimerization of the NB PSR with itself or with chromophoreless NB PSR induces different sets of R responses

Figure 3(a) shows that, while the singly transformed phyB(NB–Cdc) and phyB(NB–Bem) lines have elongated hypocotyls and small cotyledons under R like the phyB mutant parent, co-expression and dimerization of the fusion proteins in phyB(NB–Cdc/NB–Bem) lines significantly complements these phenotypes. The hypocotyl lengths and cotyledon angles for these lines are shown in Figure 3(b,c). These results demonstrate that monomeric NB fusion proteins are non-functional for these seedling responses and that dimerization activates them. It is notable that, while the R-induced cotyledon angle phenotype is fully complemented by the NB/NB dimers, the hypocotyl response to R is only partially restored. For plant type I phyA, distinct signaling activities have been demonstrated for cytoplasmic and nuclear-localized molecules (Rosler et al., 2007), but the role of cytoplasmic type II phytochromes in plants remains unclear (Huq et al., 2003; Hughes, 2013). As all of the phy PSR fusion constructs used in our experiments contain the SV40 NLS, it is possible that the differences observed between the activities of NX PSR fusion proteins compared with native full-length phytochromes, such as those seen in Figure 3(a,b), reflect the lack of a cytosolic pool of the engineered molecules.

Figure 3.

R light responses of phyB lines containing Cdc/Bem domain-mediated NB/NB or NBC357S/NB dimers. (a) Morphology of phyB seedlings expressing monomeric NB–Cdc or NB–Bem, the NB–Cdc/NB–Bem dimer, full-length phyBC357S, the NBC357S–Cdc monomer or the NBC357S–Cdc/NB–Bem dimer grown for 5 days under continuous R (25 μmol m−2 sec−1) at 20°C. Scale bar = 3 mm. (b) Hypocotyl lengths of seedlings of the indicated lines grown under continuous R (25 μmol m−2 sec−1) or in the dark for 5 days at 20°C (values are means ± SE;= 25–30). (c) Angles between the two cotyledons of seedlings of the indicated lines grown under continuous R (25 μmol m−2 sec−1) for 5 days at 20°C (values are means ± SE;= 25–30).

The chromophoreless (achromo-) mutant phyB–C357S has been shown to be inactive in R control of hypocotyl length (Clack et al., 2009), and, as expected, the phyB(FL PHYBC357S) and achromo-NB monomer phyB(NBC357S–Cdc) lines are very similar in phenotype to the phyB mutant parent (Figure 3a,b). Interestingly, phyB(NBC357S–Cdc/NB–Bem) lines also show almost no R-induced suppression of hypocotyl elongation. It is predicted that, in these lines, NBC357S–Cdc/NB–Bem dimers assemble and form Pfr/PrC357S PSR photo-conformational heterodimers when exposed to R light, but these appear to be inactive in mediating R sensitivity in hypocotyl cells. However, the R-induced increases in cotyledon expansion and cotyledon angle are both strongly restored in phyB(NBC357S–Cdc/NB–Bem) lines, demonstrating that Pfr/PrC357S phyB PSR photo-heterodimers are functional and that they are differentially active in promoting R responses in petiole and cotyledon cells (Figure 3a,b). Time courses of changes in hypocotyl length and cotyledon area under continuous R confirm that cotyledon area is much more sensitive to the presence of nuclear NBC357S–Cdc/NB–Bem dimers than is hypocotyl length throughout seedling development (Figure S3). Activity of NBC357S–Cdc/NB–Bem dimers in controlling cotyledon angle was observed both when the R stimulus is delivered continuously (Figure 3c) and when it is delivered in 1 min hourly pulses of low or high fluence (Figure S4). This shows that intermittent brief production of NB Pfr/PrC357S activates this response, and that rapid dark reversion of phyB Pfr/Pr to Pr/Pr (Hennig and Schäfer, 2001) either does not occur in NBC357S/NB dimers or that signaling is rapid and only partially mitigated by this reversion.

Yeast GAL4 domain-directed homodimerization of the NB or ND PSRs, but not NC or NE, partially complements phyB mutant phenotypes

As a second method to dimerize phy PSR regions in a directed way in vivo, the homo-dimerization domain of the yeast GAL4 protein (Hidalgo et al., 2001) was fused to the C–terminus of the NB PSR region, followed by an epitope tag and the NLS (Figure 1a). Figure 4(a) shows that the phyB(35S:NB–GAL–His6–NLS) and phyB(35S:NB–GAL–Myc–NLS) lines, which differ only in their epitope tags and are abbreviated as NB–GALH and NB–GALM, express their transgene products at levels three and ten times that of endogenous phyB, respectively. SEC of an extract of a line expressing NB–GALH shows that it migrates as a discrete dimer (Figure 2a). Homologous N–terminal regions of phyC (amino acids 1-602), phyD (amino acids 1-655) and phyE (amino acids1-592) were also incorporated into GAL4-based chimeric transgenes with the NLS, and transformed into phyB host plants. These chimeric proteins are present in the phyB(ND–GAL), phyB(NC–GAL) and phyB(NE–GAL) lines at five, 0.25 and four times the normal phyB level present in WT Arabidopsis, respectively (Figure 4a).

Figure 4.

Expression levels and activities of GAL domain-mediated NX homodimers. (a) Immunoblot analysis of transgene-encoded protein levels in phyB(NX–GAL) lines (X = B, C, D and E), and two control lines: phyB(PHYB–Myc6) line 210 and phyB(His6PHYB) line 336. Total protein extracts from 5-day-old white light-grown seedlings were fractionated by SDS–PAGE, blotted, and probed with anti-His6 or anti-Myc antibodies. Numbers below the lanes indicate the expression levels in the transgenic lines relative to the normal wild-type phyB level. The asterisk indicates a non-specific band detected by the anti-His6 antibody. (b) Morphology of phyB seedlings expressing the indicated NX–GAL transgenes grown for 5 days under continuous R (25 μmol m−2 sec−1) at 20°C. Scale bar = 3 mm.

Seedlings of lines expressing the NX–GAL fusion proteins grown under continuous R light are shown in Figure 4(b). The NB–GALH and NB–GALM homodimers are both highly active in complementing the phyB hypocotyl elongation and cotyledon expansion phenotypes. Lines 336 and 210, which over-express His6- or Myc-tagged full-length phyB by two- to fourfold, show the expected mild R hypersensitivity relative to wild-type (Wester et al., 1994). In Figure 4, it is clear that dimerization of the NB PSR, whether by the Bem/Cdc interaction or by homodimerization via the GAL domain, fails to restore high-fluence R control of hypocotyl elongation in the phyB null mutant to the same extent as expression of full-length native phyB in lines 336 or 210. This suggests that either the absence of the normal phyB C–terminal sequences or alteration of the interaction of the NB regions in the engineered dimers influences their signaling function for this response. As these constructs only form dimers between NB subunits, it is also possible that this reduced signaling reflects a lack of phyB/phyC, phyB/phyD or phyB/phyE heterodimers. Compared to NB–GALH, the ND–GAL fusion protein is more highly expressed but less active in R sensing/signaling, particularly with regard to hypocotyl elongation (Figure 4a,b). This suggests that native phyD/phyD homodimers, which do form in Arabidopsis (Sharrock and Clack, 2004), signal less effectively through hypocotyl cell light-transduction pathways than phyB/phyB homodimers. The NC–GAL and NE–GAL fusion transgene products are inactive in phyB-mediated seedling R responses (Figure 4b). Native full-length phyC and phyE do not form homodimers in Arabidopsis (Clack et al., 2009), so this latter finding is not unexpected.

Constitutive nuclear localization of NB PSR dimers confers hypersensitivity to low-fluence R light

Figure 5(a) shows R fluence response curves for hypocotyl elongation in representative phyB(NB–GAL), phyB(ND–GAL) and phyB(NB–Cdc/NB–Bem) lines. Seedlings expressing full-length phyB, including the WT and transgenic lines 210 and 336, show the expected negative slope of hypocotyl length when exposed to increasing R fluence rates, with the two- to fourfold phyB over-expressing lines 210 and 336 exhibiting hypersensitivity relative to WT at fluences >0.1 μmol m−2 sec−1. In contrast to this, lines expressing the directed dimer NB-containing PSR constructs show hypersensitivity to low fluences of R (<0.1 μmol m−2 sec−1), but do not show an increase in response when exposed to progressively higher levels of light (Figure 5a). Immunoblot analysis of seedlings grown under varying levels of continuous R shows that the fusion proteins are present at similar levels under the various R fluences (Figure S5a). Therefore, NB–GAL and NB–Cdc/NB–Bem dimers mediate stronger inhibition of hypocotyl elongation than full-length WT phyB under low R, but cannot further activate that response upon a 104-fold increase in R fluence. In a similar way, NBC357S–Cdc/NB–Bem photo-heterodimers mediate hypersensitivity of cotyledon unfolding under low-fluence R (Figure 5b), where WT phyB is not activated. Hence, very small amounts of NB Pfr/Pr dimers in the nucleus appear to be sufficient to initiate signaling for unfolding. Unlike the NB–GAL construct, the ND–GAL fusion does not mediate R hypersensitivity but shows restricted activity over the whole fluence range (Figure 5a).

Figure 5.

R light hypersensitivity of NB dimer-expressing transgenic lines, and the effect of constitutive nuclear localization on phyB activity. (a) Fluence response of hypocotyl length to continuous R. Seedlings of WT, the phyB mutant, phyB(NB–Cdc/NB–Bem), phyB(NB–GALH), phyB(NB–GALM), phyB(ND-GALH), phyB(PHYB–Myc6) line 210 and phyB(His6PHYB) line 336 were grown for four days at 20°C under the indicated fluences of R (values are means ± SE;= 15–30). (b) R light-fluence response of the angle between the two cotyledons in seedlings of WT, phyB mutant, phyB(NB–Cdc/NB–Bem) and phyB(NBC357S–Cdc/NB–Bem) lines grown for 4 days at 20°C (values are means ± SE;= 15–25). (c) Fluence response of hypocotyl length to continuous R in 4-day-old seedlings of the phyB(35S:FLphyB–YFP) line 267–5 and phyB(35S:FL phyB–NLS-YFP) line 341–4 (values are means ± SE;= 15–20). The inset shows immunoblot analysis of the levels of the transgene products using anti-phyB antibody (top) and anti-GFP antibody (bottom).

A potential explanation for the hypersensitivity of lines expressing NB PSR fusion constructs to low R is that the NB constructs carry the SV40 NLS so they do not require light-induced nuclear translocation in order to interact with nuclear signaling pathways. If phy translocation is limiting at low R in WT cells, the activity of WT phyB at these fluences should be increased by addition of the SV40 NLS. Figure 5(c) shows that a 35S:phyB–YFP transgene causes hypersensitivity to R fluences above but not below 0.1 μmol m−2 sec−1, whereas a 35S:phyB–NLS–YFP transgene induces strong hypersensitivity at low R levels. This suggests that, under light conditions whereby only a small proportion of phy subunits are photo-converted, partial light-induced cytoplasm-to-nucleus relocalization of phyB Pfr may limit signaling and responses.

Specific type II phy PSR heterodimers mediate different R sensitivity patterns

Arabidopsis type II phytochromes form multiple heterodimeric combinations, but it is difficult to assess which dimer forms are most active in different R responses (Clack et al., 2009). To address this, NX–Cdc transgenes (where X = C, D or E) containing the PSRs of phyC–phyE were constructed and introduced into the phyB mutant and phyB(NB–Bem), generating doubly transgenic phyB(NX–Cdc/NB–Bem) lines and corresponding phyB(NX–Cdc) control lines (Figure 1a). As shown in Figure 6(a), the ND–Cdc protein is expressed well, and, unlike NB–Cdc, is as stable in the absence of a binding partner NB–Bem protein as in its presence. The NE–Cdc protein is weakly expressed on its own, but is stabilized when co-expressed with NB–Bem to a level not much below that of its partner protein. The NC–Cdc protein is present in extracts in very low amounts when expressed either alone or with NB–Bem. The various doubly transgenic lines express the NB–Bem dimerization partner protein at similar levels (Figure 6a). SEC analysis shows that, when co-expressed, the NB–Bem and ND–Cdc proteins form a discrete heterodimer in vivo, and co-immunoprecipitation demonstrates that the ND–Cdc and NE–Cdc proteins bind to NB–Bem (Figure 2). An immunoblot comparing the levels of the His6-tagged NB–GALH, NB–Cdc, NBC357S–Cdc, ND–GALH and ND–Cdc proteins in independently isolated examples of key lines is shown in Figure S5(b), confirming that they are present in transgenic lines at comparable steady-state levels.

Figure 6.

Expression levels and R light responses of singly transgenic phyB(NX–Cdc) monomer and doubly transgenic phyB(NX–Cdc/NB–Bem) heterodimer lines for the phyC, phyD and phyE PSRs. (a) Immunoblot analysis and relative NX chimeric protein levels in transgenic phyB(ND–Cdc), phyB(ND–Cdc/NB–Bem), phyB(NC–Cdc), phyB(NC–Cdc/NB–Bem), phyB(NE–Cdc) and phyB(NE–Cdc/NB–Bem) lines. Total protein extracts from 5-day-old white light-grown seedlings were fractionated by SDS–PAGE, blotted, and probed with anti-His6 or anti-Myc6 antibodies. Two independent lines are shown, indicated by numbers above the lanes. Numbers below the lanes indicate the expression levels in the transgenic lines relative to the normal wild-type phyB level. (b) Seedling morphologies of the indicated lines grown for 5 days under continuous R (25 μmol m−2 sec−1). Scale bar = 3 mm. (c) Hypocotyl lengths and cotyledon areas of lines grown under continuous R (25 μmol m−2 sec−1) for 5 days (values are means ± SE;= 20–30). (d) Cotyledon angles of lines grown under continuous R (25 μmol m−2 sec−1) for 5 days (values are means ± SE;= 20–30).

Figure 6(b) shows seedling phenotypes of transgenic lines expressing NX–Cdc monomers, NX/NB heterodimers and NX–GAL homodimers (where X = C, D or E). Quantification of these phenotypes is shown in Figure 6(c,d). The parental monomer phyB(NB–Bem) line shows very little response to R (Figure 3), as do the NC–Cdc, ND–Cdc and NE–Cdc monomer control lines (Figure 6). In contrast, phyB(ND–Cdc/NB–Bem) lines are strongly complemented for hypocotyl elongation, cotyledon expansion and cotyledon opening. This demonstrates that heterodimeric NB/ND is highly active in R sensing/signaling. The phyB(NE–Cdc/NB–Bem) lines are not complemented for hypocotyl elongation under R, but show expanded cotyledons compared to the monomer control lines (Figure 6c). Surprisingly, phyB(NC–Cdc/NB–Bem) lines, which contain very low levels of the NC–Cdc subunit, show marked opening of their cotyledons compared with their monomer control lines. This is also the case in comparison with the phyB(NE–Cdc/NB–Bem) lines, which are more active in cotyledon expansion (Figure 6d). Seedlings expressing the NC–GAL, ND–GAL and NE-GAL homodimers are included in Figure 6 for comparison. We conclude that various combinations of phy PSRs, associated with each other by directed dimerization, have distinct but overlapping regulatory activities in seedlings.

Figure 6(c,d) shows that cotyledon expansion and opening in the phyB(ND–Cdc/NB–Bem) lines respond strongly to continuous high-fluence R. To determine whether NB/ND heterodimers, like NB/NB homodimers, mediate hypersensitivity to low levels of R, the fluence responses of phyB(ND–Cdc/NB–Bem) seedling phenotypes were characterized. Figure 7 shows that the NB/NB and NB/ND lines are both hypersensitive to R below 0.1 μmol m−2 sec−1 for cotyledon expansion, cotyledon angle and hypocotyl elongation. However, while cotyledon area in the two directed dimer lines increases in a R fluence-dependent manner and their cotyledons are open at even the lowest fluence (Figure 7a), hypocotyl length in the two directed dimer lines shows almost no change with increasing R (Figure 7b). Therefore, like WT phyB, directed NB/NB and ND/NB PSR dimers are capable of converting the increasing Pfr/Ptotal ratios to a progressively greater cotyledon expansion response, but, unlike WT phyB, they do not mediate a fluence-dependent change in hypocotyl elongation.

Figure 7.

Hypersensitivity to low R mediated by directed ND/NB heterodimers in seedlings. (a) Seedling morphologies of WT, phyB, phyB(NB–Cdc/NB–Bem) and phyB(ND–Cdc/NB–Bem) lines grown for 5 days under a 104-fold range of continuous R fluences. Scale bar = 3 mm. (b) Fluence response of hypocotyl length to continuous R. Five-day-old seedlings of the lines shown in (a) were grown in the dark or under varying continuous R fluences (values are means ± SE;= 20–30).

Directed phy PSR dimers remain active as plants mature. The rosette morphologies of control and dimer-producing lines following growth for 40 days under continuous R are shown in Figure S6(a). First true leaves emerge from the apical meristems in all lines, including the phyB parent; however, all of the PSR dimer-producing lines exhibit more robust rosettes with more and larger leaves. The phyB(NBC357S–Cdc/NB–Bem) line is complemented for many aspects of adult morphology compared to its control monomer lines, demonstrating a broad capacity for chromo-NB/achromo-NB photo-heterodimers to mediate R responsiveness. In addition, like over-expressors of full-length phyB (Bagnall et al., 1995), lines expressing NB/NB, ND/ND or ND/NB dimers flower earlier than the phyB parent and control monomer lines (Figure S6b).


Plants have evolved sophisticated photosensory systems to optimize their growth and development in response to their light environment. Previous work has shown that Arabidopsis cells contain mixed populations of homodimers and heterodimers of the phyA–phyE phytochromes (Sharrock and Clack, 2004; Clack et al., 2009). Monogenic loss-of-function mutations in the PHYA or PHYB genes cause strong photomorphogenic phenotypes, and, based on these phenotypes, phyA and phyB have been recognized as the major receptors for the very low fluence and FR high-irradiance responses and for the low-fluence and shade avoidance responses, respectively (Mathews, 2006; Rockwell et al., 2006). The functions of the less abundant phyC, phyD and phyE homo- and heterodimers are more difficult to define, partly because they have more subtle roles but also because mutational removal of one of these phy proteins results in loss of multiple dimer forms (Clack et al., 2009). We have developed an in vivo expression system, directed dimerization, and used it to pursue functional studies of the type II phy homo- and heterodimers. In addition, as no phy mutations that interfere selectively with dimerization without severely altering other properties of the molecule have been described, it has not previously been possible to directly test the requirement for dimerization in phy function. The directed dimer approach has allowed this question to be addressed.

Because phytochromes are modular in structure and function, with separable N- and C–terminal regions, they are amenable to application of quaternary structure engineering. Wagner et al. (1996) showed that, when expressed in transgenic plants, the N–terminal 651 amino acid PSR of phyB (NB) binds chromophore and is fully photoactive but is monomeric and lacks regulatory activity. However, Matsushita et al. (2003) showed that this NB region is highly functional in R sensing and signal transduction when it is fused in a reading frame with the GFP and GUS coding sequences and the constitutive SV40 NLS sequence. The light regulatory activities of fusion constructs of this 651 amino acid PSR region (PAS-GAF-PHY) and a smaller 450 amino acid phyB fragment (PAS-GAF) have been characterized with respect to a variety of seedling and mature plant light responses (Matsushita et al., 2003; Oka et al., 2004; Palagyi et al., 2010). We have extended these studies to the other Arabidopsis type II phytochromes and the heterodimer combinations they form with phyB.

The yeast GAL4 dimerization domain (Hidalgo et al., 2001) was used to assemble homodimers of the phyB–phyE PSRs (NX) in vivo, and the yeast Bem and Cdc PB1 protein–protein interaction domains were used to produce specific heterodimers. The Bem and Cdc PB1 domains are small and have a high affinity for each other but do not interact with themselves (Figure S1) (van Drogen-Petit et al., 2004). We have shown by SEC analysis that the NB–Bem–Myc–NLS and NB–Cdc–His6–NLS fusion proteins expressed individually in transgenic plants are monomeric, whereas co-expression results in formation of the predicted dimer. In addition, the NB–Bem and NX–Cdc fusion proteins are co-immunoprecipitated from seedling extracts, and several are stabilized by co-expression with their dimerization partners. Therefore, Bem/Cdc domain-mediated dimerization is efficient and stable in transgenic plants. All of the constructs used in these experiments were expressed from the 35S promoter, and the steady-state protein levels of NB–GAL, ND-GAL, NB–Bem, NB–Cdc, NBC357S–Cdc and ND–Cdc only vary by approximately twofold (Figure S5b). Therefore, the differences in regulatory activity observed among these constructs probably reflect differences in the extents to which they are able to activate downstream signaling mechanisms. The NC–Cdc, NC-GAL and NE–Cdc proteins were expressed at lower levels than the other constructs, and this must be considered when evaluating their activities.

Using directed dimerization, we have made observations regarding the in vivo activities of assembled phy PSRs. First, NB–Bem and NB–Cdc monomers are inactive or only very weakly active in R light sensing/signaling, whereas NB–Bem/NB–Cdc directed dimers or NB–GAL homodimers confer robust R sensitivity for many seedling and mature plant responses. This demonstrates that dimerization is an essential process in phy biogenesis and molecular signaling mechanisms. Second, NB–Bem/NBC357S–Cdc dimers, consisting of one chromophore-containing monomer and one chromophoreless monomer, are functional in R-induced cotyledon expansion and opening and in promotion of post-seedling true leaf development, but have no effect on hypocotyl elongation or time to flowering. This suggests that different responses are activated to different extents by phyB Pfr/Pr photo-heterodimers, and has implications for plant growth under low light or FR-enriched conditions, where the majority of phy Pfr subunits are present in Pfr/Pr photo-conformational dimers. Previously, the activity of full-length phy Pfr/Pr dimers, their photochemical stability and their capacity to be transported into the nucleus compared to Pfr/Pfr dimers have been assessed indirectly (Furuya and Schäfer, 1996). Chen et al. (2003) calculated expected in vivo phyB Pfr/Ptotal ratios at various R fluences and R:FR ratios, and concluded that phyB:GFP Pfr/Pr dimers may be imported into the nucleus but do not associate in visible nuclear bodies and only weakly regulate hypocotyl elongation, whereas Pfr/Pfr dimers form nuclear bodies and lead to strong R inhibition of hypocotyl elongation. In our experiments, NBC357S–Cdc (Pr)/NB–Bem (Pfr) heterodimers formed under R illumination are highly active in mediating cotyledon and true leaf responses but inactive in hypocotyl and flowering responses. This indicates that environmental light conditions that result in formation of a small amount of phyB Pfr/Pr dimer, which is transported to the nucleus but does not associate with nuclear bodies (Chen et al., 2003), differentially trigger a subset of phy responses. One hypothesis to explain this is that components of some signaling pathways downstream of Pfr may be activated by the single Pfr N–terminus in a Pfr/Pr dimer, whereas regulators of other responses require recognition or activity of a Pfr/Pfr NB dimer. Alternatively, the formation and function of specific types of nuclear bodies may be required for activation of pathways that require Pfr/Pfr dimers.

Third, heterodimerization of NB–Bem with the NC–Cdc, ND–Cdc or NE–Cdc PSRs generates receptor chimeras with different regulatory activities. This is consistent with full-length phyC, phyD and phyE heterodimerizing with phyB to generate functionally diverse components of the normal phy array (Sharrock and Clack, 2004; Clack et al., 2009). Indeed, phyC and phyE form only heterodimers in vivo, so the phenotypic effects of phyC and phyE null mutations likely result from loss of specific heterodimers (Clack et al., 2009). The data presented here indicate that the NB/ND directed dimer is similar in its signaling activity to the NB–GAL or NB–Bem/NB–Cdc dimers for the seedling and adult plant R responses tested. In contrast, the NB/NE dimer has a considerable effect on cotyledon area under R but only a small effect on cotyledon opening. The NB/NC dimer, although produced in very low amounts, shows the opposite signaling specificity from NB/NE. Null phyC, phyD and phyE mutations have phenotypic effects under R that are consistent with the observed activities of the NB/NX dimers, including reduced cotyledon area and opening in phyC and phyD and elongated hypocotyls in phyD (Aukerman et al., 1997; Devlin et al., 1998, 1999; Franklin et al., 2003; Monte et al., 2003).

Fourth, in the directed dimer system, homodimers of the ND PSR reduce the time to flowering to a similar extent as NB dimers, and mediate increased cotyledon expansion and opening under R, consistent with the reduced cotyledon size and opening seen in the phyD mutant (Aukerman et al., 1997). However, GAL4 domain-mediated homodimers of the NC or NE PSRs are unable to appreciably restore R sensitivity in the phyB mutant background. Overall, these findings support a model for phy evolution and function that includes differential activities for the large number of homodimer and heterodimer phytochrome forms. Such a diverse collection of R/FR sensors may facilitate fine-tuning and integration of the control of responses to R and FR light in different plant organs and at different times during development.

Compared to WT, the phyB(NB–GAL), phyB(NB–Bem/NB–Cdc) and phyB(NB–Bem/ND–Cdc) lines show pronounced hypersensitivity of hypocotyl growth inhibition under low R (<0.1 μmol m−2 sec−1) but hyposensitivity under high R (>1 μmol m−2 sec−1). This indicates that fewer R photons are required for chimeric PSR receptor forms to trigger this response than for full-length phyB, but that chimeric receptor signaling in the hypocotyl is saturated by very low R. As a result, the PSR fusion constructs used here do not confer any fluence responsiveness for this phenotype. Other seedling characteristics, such as cotyledon opening and expansion, while also hypersensitive to low R in the NB PSR fusion lines, do exhibit fluence responsiveness. Why hypocotyl elongation inhibition saturates at a very low level of NB Pfr while other R responses are activated proportionally to NB Pfr concentration is not known. Moreover, the question arises as to how NB PSR fusion proteins confer such abnormally strong sensitivity to low R. Matsushita et al. (2003) reported hypersensitivity of NB–GFP–GUS–NLS fusion-expressing lines at all R fluences. They attributed the general increase in signaling efficiency by the chimeric receptor to absence of the normal phyB C–terminus, and postulated a negative regulatory role for C–terminal phyB apoprotein sequences. In contrast to this, our comparison of the activities of phyB–EYFP and phyB–NLS–EYFP showed that hypersensitivity to low R may be conferred to full-length phyB by inclusion of the NLS. Therefore, eliminating the requirement for light to induce nuclear re-localization of phy differentially affects seedling response to low R fluences, indicating that nuclear translocation may be limiting for phyB signaling under those light conditions. This may be a separate, low R fluence-specific effect that is additive or synergistic with the effects of loss of the C–terminus (Matsushita et al., 2003).

Protein oligomerization is a ubiquitous phenomenon in living systems, and is fundamental to many biological activities and functions. Diversity of receptor structure and function in cells is frequently generated by combinatorial quaternary associations involving a set of subunit partners. However, few efforts have been made previously to direct the quaternary structure of proteins in vivo to produce oligomers of specific subunit composition and activity. The protein structures of the directed phy PSR dimers described here are highly engineered, and positioning of the PSRs relative to each other may be altered or constrained in the chimeric dimers compared to native full-length phy molecules. Nevertheless, it is clear that synthetic dimerization of phy PSRs both activates their signaling functions and results in differential activities depending upon which PSRs are paired. Further analysis of these lines may address questions relating to the effects of specific phy PSR directed dimers on light-regulated gene expression and light-induced formation of nuclear bodies, and additional type II phy-mediated physiological functions such as shade avoidance responses. Moreover, in addition to providing insights into the structure and mechanism of phytochromes, the use of small heterologous binding domains such as Bem and Cdc to engineer protein interactions in vivo represents a potential mechanism for designing and implementing synthetic cellular signaling pathways in the future.

Experimental Procedures

Plant material, growth conditions and seedling measurement

The Arabidopsis thaliana wild-type (WT), phyB–1 mutant and all transgenic lines were in the Nossen (No–0) background (Aukerman et al., 1997). Seeds were surface-sterilized and planted on Murashige & Skoog (MS) medium (Sigma-Aldrich, containing 0.8% agar with or without 3% sucrose. The plates were incubated at 4°C in the dark for 3–4 days, exposed to fluorescent light at room temperature for 3 h to induce germination, and then transferred to the light conditions described in the figure legends. R (670 nm) and FR (735 nm) light were supplied by light-emitting diodes in an E–30 LED growth chamber (Percival, For adult plant experiments, R light was provided by red plastic-wrapped fluorescent bulbs (Sharrock and Clack, 2002) in a controlled-temperature growth chamber (Conviron, at 20°C. For hypocotyl length and cotyledon area and angle measurements, seedlings were grown on MS plates without sucrose at 20°C for the durations specified in the figure legends, laid out on agar plates under the light conditions under which they were grown or under green safe-light conditions, photographed, and measured using ImageJ software ( Values for seedling cotyledon angles are the measured angle between the two cotyledons. For rosette leaf morphology and flowering time experiments, seeds were sown directly into pots, treated at 4°C for 4 days, and grown at 20°C under continuous low R (15 μmol m−2 sec−1) or high R (90 μmol m−2 sec−1).

Plasmid construction and plant transformation

All plant transformation plasmids were constructed in previously described pBI or pGNT vectors with kanamycin resistance and gentamicin resistance markers and Myc6 or His6 epitope tags, respectively (Clack et al., 2009). The N–terminal regions of the PHYB, PHYBC375S, PHYC, PHYD and PHYE cDNA sequences were translationally fused to Myc6-tagged Bem or GAL4 or His6-tagged Cdc or GAL4 regions, with the SV40 NLS at their C–termini (Figure 1a). Diagrams of these constructs are shown in Figure S7, and the PCR primers and restriction enzymes used in cloning are listed in Table S1. The chimeric coding sequences were transcribed from the CaMV 35S promoter. Plant transformations were performed by the floral dip method using Agrobacterium tumefaciens strain GV3101, and transformed plants were selected on MS agar plates containing 55 μg ml−1 kanamycin or 175 μg ml−1 gentamicin plus 200 μg ml−1 carbenicillin. For each transgene or transgene combination, 15–30 independent lines were initially identified, and three to six homozygous, protein-expressing T3 or F3 lines were identified and characterized in depth.

Yeast two-hybrid assay

Yeast two-hybrid assays were performed using materials and protocols from Matchmaker GAL4 System 3 (Clontech, Regions from the yeast genomic DNA sequences were PCR-amplified using forward primers containing EcoRI sites and reverse primers containing BamHI sites (Table S1), and cloned into the pGAD–T7 and pGBK–T7 vectors (Clontech) to generate clones containing Bem–317 (N235–I551), Bem–128 (D424–I551), Bem–82 (A470–I551), Cdc–182 (E673–Y854) or Cdc–95 (S760–Y854) in both vectors. Colorimetric yeast two-hybrid assays were performed under room lights, and colony growth assays were performed in a dark incubator at 28°C.

Protein extraction, SDS–PAGE, immunoblot analysis and immunoprecipitation

Seedling protein extracts were prepared in extraction buffer (50 mm Tris pH 8.0, 150 mm NaCl, 0.1% Nonidet P-40) as described previously (Sharrock and Clack, 2004). Proteins were fractionated on 8% SDS–PAGE gels and transferred to Hybond–ECL membranes (Fisher Scientific, Membranes were blocked overnight at 4°C using blocking buffer (5% non-fat dry milk, 0.2% Tween 20 in TBS–T buffer, pH 7.6). Membranes were probed in blocking buffer containing the following primary monoclonal antibodies: anti-phyB B3B6 (Hirschfeld et al., 1998), anti-Myc 9E10 (1:1000; gift from Seth Pincus, Louisiana State University, Baton Rouge, LA) and anti-RGS-His (Qiagen, After three–five min washes with TBS–T buffer at room temperature (24°C), chemiluminescent detection of primary antibodies was performed using horseradish peroxidase-conjugated secondary antibody and Supersignal West Pico reagents (Thermo Fisher Scientific, Total protein was analyzed by the Bio–Rad protein assay (Bio Rad, Immunoprecipitation was performed as described previously (Sharrock and Clack, 2004). For immunoprecipitation experiments, a set of lanes on an SDS gel was loaded with seedling protein extracts on the basis of protein concentration to illustrate the input to the immunoprecipitation reactions. A set of adjacent gel lanes was loaded with the corresponding immunoprecipitated samples.

Size-exclusion chromatography

Protein extracts of light-grown phyB(NB–Bem), phyB(NB–Cdc), phyB(NB–Bem/NB–Cdc), phyB(NB–GALH) and phyB(NB–GALH) seedlings were prepared as described above, and 1.0 ml samples containing approximately 1 mg of total soluble protein were applied to a Superose 6 gel filtration column (25 ml bed volume) (GE Healthcare Life Sciences, The column was eluted with extraction buffer at 4°C at a rate of 0.5 ml min−1, and 1.0 ml fractions were collected. Immunoblot analysis of column fractions was performed as described above. The column was calibrated with high molecular weight gel filtration protein standards and immunoblots were scanned and analyzed using ImageJ software.

Accession Numbers

The Arabidopsis Genome Initiative or GenBank/EMBL accession numbers for the sequences used in this paper are NC_001134 (locus tag ‘YBR200W’, yeast Bem1), BK006935 (locus tag ‘YAL041W’, yeast Cdc24), BK006949 (locus tag ‘YPL248C’, yeast GAL4), At2g18790 (PHYB), At5g35840 (PHYC), At4g16250 (PHYD) and At4g18130 (PHYE).


We thank Ted Clack for technical assistance and Brian Eilers (Department of Chemistry and Biochemistry, Montana State University) for advice and help in performing SEC experiments. This work was supported by a grant to R.A.S. from the US National Science Foundation (IOS-0920766).