Fluorescence of Phytochrome Adducts with Synthetic Locked Chromophores

Abstract

We performed steady state fluorescence measurements with phytochromes Agp1 and Agp2 of Agrobacterium tumefaciens and three mutants in which photoconversion is inhibited. These proteins were assembled with the natural chromophore biliverdin (BV), with phycoerythrobilin (PEB), which lacks a double bond in the ring C-D-connecting methine bridge, and with synthetic bilin derivatives in which the ring C-D-connecting methine bridge is locked. All PEB and locked chromophore adducts are photoinactive. According to fluorescence quantum yields, the adducts may be divided into four different groups: wild type BV adducts exhibiting a weak fluorescence, mutant BV adducts with about 10-fold enhanced fluorescence, adducts with locked chromophores in which the fluorescence quantum yields are around 0.02, and PEB adducts with a high quantum yield of around 0.5. Thus, the strong fluorescence of the PEB adducts is not reached by the locked chromophore adducts, although the photoconversion energy dissipation pathway is blocked. We therefore suggest that ring D of the bilin chromophore, which contributes to the extended π-electron system of the locked chromophores, provides an energy dissipation pathway that is independent on photoconversion.

Introduction

Phytochromes are photoreceptors of plants, bacteria, and fungi, which are most sensitive to red and far-red/near-infrared light. A typical phytochrome consists of an N-terminal chromophore module that bears the PAS (Per-Arnt-Sim (1)), GAF (cGMP phosphodiesterase, adenylyl cyclase and transcriptional factor FhaA (2)), and PHY (phytochrome-specific) domains and a C-terminal output module with a histidine kinase or a histidine kinase-like domain (Fig. 1). Each phytochrome protein subunit carries one bilin chromophore that is covalently bound to a conserved cysteine residue. The PAS/GAF/PHY chromophore module is sufficient for chromophore incorporation and full spectral activity (3). Light absorption triggers the photoconversion between the so-called Pr³ and Pfr forms, which have respective absorption maxima in the red and far-red spectral regions. The first step in the photoconversion is a Z to E isomerization around the double bond between pyrrole rings C and D of the bilin chromophore. In darkness, both forms are kinetically stable or have rather long interconversion times (4). The type of bilin chromophore differs between species. In cyanobacteria, two chromophore binding modes are realized: the natural chromophore of the Cph1/CphA-type phytochromes is PCB, whereas CphB-type phytochromes incorporate BV. This bilin also constitutes the chromophore of most bacterial and probably all fungal phytochromes. A third bilin, PΦB, serves as a chromophore in plant phytochromes. In BV-binding phytochromes the chromophore binding cysteine residue lies in the N-terminal part of the PAS domain, whereas PCB- and PΦB-binding phytochromes have their chromophore-binding cysteine in the GAF domain (5). Apart from this major difference, the chromophore pockets of BV- and PCB-binding phytochromes are very similar as judged from the x-ray structures of recently crystallized representatives (6–11). Sequence alignments also show that amino acid residues of the chromophore pocket are highly conserved.

FIGURE 1.

Domain arrangement of Agrobacterium phytochrome Agp1, the truncated mutant Agp1-M15, Agp2, and the domain-swap mutant Agp2_Phy1. Note that the PHY domains of Agp1 and Agp2 are presented in different shades of gray. The positions of the chromophore-binding Cys residues are indicated, as well as His-280 and Val-249 of the Agp1 mutant and Asp-197 of the Agp1-M15 mutant.

BV, which is synthesized from heme by a one-enzyme reaction, is regarded as the most ancient chromophore. PΦB and PCB are synthesized via additional enzymes that reduce one and two double bonds of BV, respectively. These reductions result in a less conjugated system and hence in a blue shift of the respective adducts. The absorption maxima of typical BV-, PΦB-, and PCB-binding phytochromes are at 700, 665, and 655 nm, respectively. This kind of spectral tuning has probably evolved as an adaptation to a chlorophyll a-rich environment. Many variations of the prototypical phytochrome have been described. Several cyanobacterial phytochrome-like proteins utilize, for example, the GAF domains for chromophore binding, which they have probably inherited from a typical phytochrome (12). Domain rearrangements have also been described for fungal and bacterial BV-binding phytochromes (13, 14). Most phytochromes have a Pr ground state, but bacterial phytochromes with a Pfr ground state, so called bathy phytochromes, have been found in Bradyrhizobium sp., Rhodopseudomonas palustris (15), Agrobacterium tumefaciens (16), and Pseudomonas aeruginosa (17) as well as other rhizobial species (18).

Early studies on plant phytochromes have shown that the fluorescence quantum yield of Pr is low at ambient temperature but is increased significantly under cryoconditions where photoconversion is blocked (19). Adducts that fluoresce strongly at ambient temperature have been obtained by chromophore replacement. Phycoerythrobilin (PEB), a bilin chromophore of the phycoerythrin antenna pigment of cyanobacteria and red algae, lacks the C15=C16 double bond between pyrrole rings C and D (Fig. 2). This bilin is incorporated into apophytochromes, but the adducts are photoinactive and highly fluorescent (20). A quantum yield of 0.82 has been reported for the PEB adduct of cyanobacterial phytochrome Cph1 (20). Feeding PEB to plant mutants with an impaired chromophore biosynthesis enabled the detection of phytochrome by fluorescence microscopy within the cell (20, 21) and studies on dimer subunit interactions of Cph1 by means of fluorescence resonance energy transfer (22). An adduct of P. aeruginosa phytochrome PaBphP with 15,16-dihydrobiliverdin, which also lacks the ring C-D-connecting double bond, is also fluorescent (17). The inhibition of photoconversion by low temperature or chromophore replacement correlates with an increase of Pr fluorescence, indicating that photoisomerization provides a major route of energy dissipation. Fluorescence is also increased in phytochrome mutants with impaired photoconversion; the Y166H mutant of cyanobacterial phytochrome Cph1, in which photoconversion is inhibited, has a fluorescence quantum yield of >0.1 (23). The homologous mutation resulted in the same fluorescence increase in plant phytochromes but not in BV-binding bacterial phytochromes (24). Increased fluorescence has been reported for mutants of Deinococcus radiodurans and P. aeruginosa phytochromes DrBphP and PaBphP, respectively, in which a highly conserved Asp residue (Asp-207 in DrBphP) was replaced by Ala (9, 25). The D197A mutant of the DrBphP PAS/GAF fragment has been used for directed evolution employing DNA shuffling. A 13-residue mutant resulting from that approach had an increased fluorescence quantum yield of 0.07. This variant has been successfully employed as a marker protein with advantageous properties in mammalian organs (26).

FIGURE 2.

Chromophores and chromophore abbreviations used in the present study.

Adducts have been prepared with locked chromophores in which rings C and D are held in a fixed geometry by an additional carbon bridge. They cannot undergo photoconversion and are thus comparable with the above mentioned PEB and 15,16-dihydrobiliverdin adducts. However, such adducts with locked chromophores have longer wavelength absorption maxima because of their extended π-system, a property that is in line with the demand for novel fluorescent dyes. In this article, we have characterized the fluorescence properties of several different adducts with bacterial phytochromes and locked chromophores. These studies show that the lock increases fluorescence significantly but not to the same extent as the loss of a C15=C16 double bond. This difference suggests that the D ring provides a radiationless energy dissipation pathway.

MATERIALS AND METHODS

Phytochrome Expression Vectors, Protein Expression, and Purification

All proteins (Cph1, Agp1, Agp1-V249C, Agp1-M15-D197A, Agp1-H280A, Agp2, and Agp2_Phy1) were expressed in Escherichia coli Xl1blue or BL21(DE3) cells. The expression vectors encode for proteins with a C-terminal poly-His tag. Expression vectors pF10.his, pAg1, and pAg2, which encode for Cph1, Agp1, and Agp2, respectively, are described elsewhere (27–29). The expression vectors for the Agp1-V249C mutant, which binds phycocyanobilin in a covalent manner, and for the Agp1-M15-D197A mutant are described elsewhere (30, 31). An expression vector for the Agp1-H280A mutant was made with the QuikChange site-directed mutagenesis kit (Stratagene). The expression vector pAg2-Phy1 encodes for a domain-swap mutant termed Agp2_Phy1. In this protein, the PHY domain of Agp2 (amino acids 315–501) is replaced by the PHY domain of Agp1 (amino acid 315 to 495). The vector pAgp2-Phy1 was cloned as follows. The sequence coding for the PHY domain of Agp1 was amplified by PCR by using Phusion polymerase (Finnzymes, Espo, Finland) and the following phosphorylated primers: BZ22.1 5′-ACCATCAACCACGCCCATGC-3′ and BZ22.2 5′-CTCGCTGTGGTGGAACGCCAC-3′. Then, another PCR reaction was performed with the pET21b(+) (Novagen, Madison, WI)-based Agp2 expression plasmid, pAgp2 (3), using the primers BZ21.1 5′-TTGATGGCAGGCGAAAGAGAGC-3′ and BZ21.2 5′-GTCGAGGCGGCGCTTCTGCTTC-3′. In this reaction, the entire plasmid DNA except for the sequence coding the PHY domain of Agp2 was amplified. Both blunt end PCR products were ligated with the T4 ligase (New England Biolabs, Beverly, MA) overnight at 15 °C (32) and transformed directly into the E. coli expression strain BL21(DE3). The correctness of the new expression vector was confirmed by DNA sequencing.

For protein expression, E. coli cells containing the desired plasmid were grown in LB medium with ampicillin at 37 °C until the A_{600 nm} reached about 0.6. Then specific protein expression was induced by the addition of isopropyl 1-thio-β-d-galactopyranoside (20 μg/ml). The cultures were further grown at 16 °C until the A_{600 nm} reached about 3. Thereafter, cells were collected by centrifugation, washed with basic buffer (50 mm Tris-HCl, 300 mm NaCl, 5 mm EDTA, pH 7.8), and extracted with basic buffer. Proteins of the soluble fraction were precipitated with ammonium sulfate and resuspended in 10 mm imidazole, 50 mm Tris-HCl, 300 mm NaCl, pH 7.8. The His-tagged phytochrome proteins were then purified by Ni²⁺ affinity chromatography. Following elution from the column with a buffer containing 250 mm imidazole, the phytochrome-containing fractions were again subjected to ammonium sulfate precipitation, and the protein was finally suspended in basic buffer. Details on the purification steps are given in earlier publications (28, 33–35).

Chromophores and Chromophore Assembly

BV was obtained from Frontier Scientific. PEB was extracted from Porphyridium cruentum and purified by high pressure liquid chromatography as described elsewhere (35, 36). Synthesis of the locked biliverdin chromophores 15Za and 15Ea (37, 38), of the doubly locked chromophores 5Za15Ea and 5Ea15Ea (39), and of 5Zs15Za (40) is described elsewhere. All chromophores were dissolved in dimethyl sulfoxide at concentrations of about 4 mm. The exact stock concentrations were estimated by measuring the absorption of the samples diluted in methanol/HCl. The following extinction coefficients (at λ_max) were used: ϵ_BV = 30,800 m⁻¹cm⁻¹ (41), ϵ_PEB = 25,200 m⁻¹cm⁻¹ (42), ϵ_15Za = 40,300 m⁻¹cm⁻¹, ϵ_15Ea = 51,300 m⁻¹cm⁻¹, ϵ_5Za15Ea = 61,488 m⁻¹cm⁻¹, ϵ_5Ea15Ea = 41,461 m⁻¹cm⁻¹, and ϵ_5Zs15Za = 83,400 m⁻¹cm⁻¹. The values for locked chromophores were obtained from exactly weighed, pure, solid material after dissolution in methanol/HCl.

Chromophore assembly was initiated by adding 2–3 μl of chromophore from the stock solution to 1 ml of protein, which was adjusted to about 10 μm to achieve an equimolar concentration of protein and chromophore. The assembly process was followed by measuring absorption spectra at different time points until no further absorption changes were evident. Thereafter, free chromophore was separated from the protein by using desalting columns (NAP-10 columns, GE Healthcare). Most if not all chromophore remained bound to the protein after the separation. The columns were equilibrated with basic buffer, and 1 ml of holoprotein was loaded on the column. The flow-through was discarded. Thereafter, 1.5 ml of basic buffer was applied on the column, and the flow-through, containing the protein, was collected. Free chromophore was not eluted because of its low molecular weight and its tight interaction with the matrix. Spectra were recorded before and after the separation.

UV-visible Spectroscopy, Fluorescence Spectroscopy, and Quantum Yield

All UV-visible measurements were performed at 20 °C using a JASCO V-550 photometer. Fluorescence spectra were recorded by using a Yobin-Yvon FluoroMax2 spectrofluorimeter at 20 °C. To avoid cross-talk and scattering artifacts, the emission monochromator was shielded with an appropriate cut-off filter (RG570, RG665, or RG695, thickness of 3 mm; Schott, Mainz, Germany). Slit widths of excitation and emission were set between 2 and 7 nm according to the fluorescence intensities. Corrected excitation and emission spectra were obtained considering the spectral characteristics of the lamp, the filter, and the detector. The fluorescence quantum yield Φ of each sample was calculated by integrating its emission spectrum and referring it to fluorescein as a standard (Φ = 0.95 (43)). Agp2-BV was irradiated with far-red light (780 nm from a light-emitting diode, irradiance ∼200 μmol photon m⁻² s⁻¹) and fully converted to Pr before measuring the fluorescence.

RESULTS

Fluorescence Properties of Free Chromophores

Because of the free mobility of the pyrrole rings, the fluorescence of free bilin molecules is very weak (44, 45). Incorporation into the chromophore pocket of a biliprotein such as phycocyanin or phytochrome restricts this mobility and results in an increase of the fluorescence quantum yield. We initially measured the fluorescence properties of free chromophores under the buffer conditions that were later been used for measurements of phytochrome adducts (Table 1). The fluorescence of BV and the locked chromophore 15Za was not measurable under these conditions, we assume that the fluorescence quantum yields are below 10⁻⁶. The other chromophores had quantum yields in the range of 10⁻⁵ to 10⁻⁴, and the quantum yield of free PEB was around 10⁻³. Because in the chromophore pocket of phytochromes the bilins are in the protonated form, the fluorescence of free chromophores was also measured after acidification by HCl. This treatment changed the absorption and fluorescence properties of all tested chromophores. As stated by Falk (45), acidification of bilins is generally accompanied by a red shift of the absorption maximum. All chromophores in the present study, with the exception of BV, follow this rule, although the red shifts of 5Za15Ea and 5Zs15Za are only 3 and 13 nm, respectively. The blue shift of BV upon acidification is in contrast to other work (46) but might be related to the aqueous solvent used in the present control measurements.

TABLE 1.

Absorption and fluorescence characteristics of free chromophores in aqueous buffer at pH 7.8 and after the addition of 0.5 m HCl from a 10 m stock solution at pH 1

The absorption, excitation, and emission maxima of the Q-band are given. Fluorescein (43) was used as a reference for calculating the quantum yields (Ф). ND, not detectable.

Chromophore	Absorption λmax	Excitation λmax	Emission λmax	Stokes shift	Ф
	nm	nm	nm	nm
Fluorescein	490	488	574	86	0.95
PEB, pH 7.8	531	535	579	44	0.0011
PEB, pH 1	587	580	613	33	0.0016
BV, pH 7.8	673	ND	ND
BV, pH 1	629	ND	ND
15Za, pH 7.8	618	ND	ND
15Za, pH 1	678	ND	ND
15Ea, pH 7.8	631	632	708	76	0.000003
15Ea, pH 1	679	640	683	43	0.000013
5Za15Ea, pH 7.8	653	654	661	7	0.000026
5Za15Ea, pH 1	656	655	672	17	0.000006
5Ea15Ea, pH 7.8	609	647	669	22	0.00064
5Ea15Ea, pH 1	644	640	668	28	0.00089
5Zs15Za, pH 7.8	693	660	674	14	0.000064
5Zs15Za, pH 1	706	ND	ND

The fluorescence quantum yields of the locked chromophores are generally higher than that of BV but lower than that of PEB. The fluorescence increase over BV is most likely due to the lock and the enhanced rigidity. Note that the introduction of a lock does not reduce mobility completely; it is assumed that seven-member rings still provide some flexibility (47), and the same is probably true for eight-member rings. Acidification led to a (slight) fluorescence increase of three chromophores, whereas two showed a decrease. Hence, there was no general pattern for the effect of protonation on fluorescence yields.

Phycoerythrobilin and Biliverdin Adducts

As a reference for our new phytochrome chromophore adducts, we measured absorption and fluorescence spectra of the PEB adducts of Cph1 and Agp1 (Fig. 3). In our hands the quantum yield of PEB-Cph1 was 0.54 (Table 2) and thus slightly lower than the 0.82 value reported by others (20, 48). The bacterial phytochrome Agp1 contains BV as a natural chromophore, bound via the vinyl side chain of ring A to the protein; the protein can also incorporate PEB into its chromophore pocket. However, PEB, PCB, and PΦB, which carry an ethylidene side chain on ring A, are incorporated in a noncovalent manner, and the PEB adduct has an atypical, broad absorption spectrum (28). The mutant V249C of Agp1 incorporates PCB (30) as well as PEB⁴ in a covalent manner. The absorption and fluorescence spectra of the PEB-Agp1-V249C adduct (Fig. 3)⁴ are almost indistinguishable from those of PEB-Cph1. The excitation and emission maxima for the Cph1 and Agp1-V249C adducts lie at 579/580 and 590/588 nm (Table 2), respectively. Hence, the broad spectrum of the PEB adduct with wild type Agp1 (28) results from the noncovalent binding mode. The fluorescence quantum yield of PEB-Agp1-V249C is 0.67 (Table 2), which is even higher than that of our PEB-Cph1 control.

FIGURE 3.

PEB adducts of Cph1 (A) and Agp1-V249C (B). Absorption spectra (dotted lines) and fluorescence emission (solid lines) and excitation (dashed lines) spectra are shown.

TABLE 2.

Absorption and fluorescence characteristics of the phytochrome adducts used in the present study

The absorption, excitation, and emission maxima of the Q-band are given. Fluorescein (43) was used as a reference for calculating the quantum yields (Ф). ND, not detectable.

Chromophore Adduct	Absorption λmax	Excitation λmax	Emission λmax	Stokes shift	Ф
	nm	nm	nm	nm
Fluorescein	490	488	574	86	0.95
PEB-Cph1	580	579	590	11	0.5369
PEB-Agp1-V249C	580	580	588	8	0.6706
BV-Agp1	703	700	701	1	0.0002
BV-Agp2	703	703	713	10	0.0013
BV-Agp1-M15-D197A	700	684	714	30	0.0022
BV-Agp1-H280A	701	705	713	8	0.0012
BV-Agp2_Phy1	693	686	711	25	0.0016
15Za-Agp1	715	714	718	4	0.0131
15Za-Agp2	713	712	718	6	0.0140
15Za-Agp1-M15-D197A	713	715	716	1	0.0158
15Za-Agp1-H280A	714	705	713	8	0.0123
15Za-Agp2_Phy1	742	715	715	0	0.0024
15Ea-Agp1	736	712	721	9	0.0027
5Za15Ea-Agp1	685	682	696	14	0.0196
5Za15Ea-Agp2	704	702	707	5	0.0006
5Ea15Ea-Agp1	689	670	697	27	0.0015
5Ea15Ea-Agp2	701	683	697	14	0.0022
5Zs15Za-Agp1	711	709	712	3	0.0099
5Zs15Za-Agp2	707	705	712	7	0.0118
15Ea-Agp2	749	ND	ND

As expected, the fluorescence of wild type phytochromes Agp1 and Agp2 assembled with the natural chromophore BV is very weak (Fig. 4). The fluorescence of both adducts was measured in the Pr form, because thus far no Pfr fluorescence has been detected (19). The bathy phytochrome Agp2 converts into Pfr in darkness; hence this sample was photoconverted to Pr by 780 nm light prior to the fluorescence measurements. We obtained quantum yields for BV-Agp1 and BV-Agp2 of 0.0002 and 0.0013, respectively (Table 2). The BV adduct of Agp1 showed not only the weakest fluorescence of all, but excitation and emission maxima were also indistinguishable from each other.

FIGURE 4.

Spectra of BV adducts with Agp1 (A), Agp2 (B), Agp1-M15-D197A (C), Agp1-H280A (D), and Agp2_Phy1 (E). Absorption spectra (dotted lines) and fluorescence emission (solid lines) and excitation (dashed lines) spectra are shown.

Several phytochrome mutants have been described that exhibit inefficient photoconversion combined with an elevated fluorescence quantum yield (23, 24, 26). Here, we tested the fluorescence properties of three different mutants, Agp1-M15-D197A (31) and two new mutants, Agp1-H280A and the domain-swap mutant Agp2_Phy1 (Fig. 4). In the Agp1-M15-D197A mutant, the highly conserved Asp residue at position 197, which is located close to rings B and C of the tetrapyrrole chromophore, is mutated to Ala. The mutation results in a slower chromophore incorporation, a weaker chromophore protonation, and a less efficient photoconversion as compared with the wild type (31). For fluorescence studies, we chose the D197A mutant of the PAS/GAF/PHY fragment of Agp1 (which is termed Agp1-M15) lacking the histidine kinase module. As the histidine kinase has almost no effect on the spectral properties of Agp1, and because the fluorescence properties of a comparable Cph1 fragment are indistinguishable from the full-length variant (49), we assumed that the observed altered properties of Agp1-M15-D197A were due to the single amino acid exchange at position 197. We obtained a fluorescence quantum yield of 0.0022 for the BV adduct of Agp1-M15-D197A (Table 2), which is about 10 times higher than that of the wild type BV adduct. Hence, our result is qualitatively comparable with the increased fluorescence of the corresponding mutants of DrBphP (25) and PaBphP (9), although information about these quantum yields is lacking.

The excitation maximum of the Agp1-M15-D197A mutant adduct at 684 nm is shifted by 16 nm to the blue as compared with the absorption maximum at 700 nm. We therefore assumed that a subfraction of the adduct fluoresces more strongly than the dominating species absorbing at 700 nm. The 684 nm subfraction could be related to species with a deprotonated chromophore that does not undergo photoconversion.

In the Agp1-H280A mutant, the highly conserved His-280, which is in close contact to ring D of the chromophore and is involved in a chromophore/protein hydrogen-bonding network (6, 9), was mutated to Ala. As a result of this mutation, BV assembly was slowed down (Fig. 5A). Red light induces only a very weak photoconversion (compare the difference spectrum with the absorption spectrum (Fig. 5A)). The fluorescence excitation spectrum matches well with the absorption spectrum, and the fluorescence quantum yield is 0.0012, which is 6 times higher than that of wild type BV-Agp1.

FIGURE 5.

A, Agp1-H280A absorption spectra during assembly with BV (solid lines, left ordinate) and photoconversion difference spectra (dashed line, right ordinate) of the BV adduct. Absorption spectra are shown for the apoprotein and at 1, 10, 90, and 100 min after the addition of BV. The arrows indicate the direction of spectral changes. Note that there is no difference between the 90 min and 100 min spectra, indicating completion of assembly at <90 min. B, Agp2_Phy1 absorption spectra during assembly with BV (solid lines, left ordinate) and photoconversion difference spectra (dashed line, right ordinate). Absorption spectra are given for the apoprotein and at 1 min, 6 h, and 13 h after the addition of BV. The arrows indicate the direction of the assembly. The mutant exhibits an unusual transient far-red absorption peak.

In the domain-swap mutant Agp2_Phy1, the PHY domain of Agp2 is replaced by the PHY domain of Agp1 (Fig. 1). The PHY domain is known to be required for the spectral integrity of the phytochrome and for the complete photoconversion into Pfr (3). The replacement of the PHY domain switched the bathy phytochrome Agp2 (with a Pfr ground state) to a “normal” phytochrome with a Pr-like ground state (Fig. 5). The absorption profile is flat and broad, suggesting that the chromophore is misplaced. The BV assembly of this mutant is characterized by the transient appearance of a long wavelength absorption peak. The maximum of this peak is at 744 nm, whereas the maximum of the final Pr adduct is at 693 nm. Because the position of the assembly intermediate absorption maximum coincides with that of Pfr, one may speculate that the protein might transiently go through a Pfr state during chromophore incorporation. However, the formation of Pfr during the dark assembly of BV-Agp2 or during dark conversion from Pr takes hours to reach completion. Hence, a rapid formation of a Pfr chromophore during BV assembly of Agp2_Phy1 is very unlikely. The long wavelength intermediate might thus represent, e.g. a noncovalent Pfr-like adduct.

Red light irradiation of BV-Agp2_Phy1 resulted in only a weak photoconversion of the BV adduct of this mutant. The shape of the difference spectrum (Pr minus Pfr) has the same characteristics as that of Agp2, i.e. a small Pr peak and a large Pfr valley (50, 51). The fluorescence excitation maximum of Agp2_Phy1 was also blue-shifted as compared with the absorption maximum, in this case by 7 nm. Again, the fluorescence might be related to a subfraction with a deprotonated chromophore. The quantum yield of the adduct at 0.0016 was within the range of BV-Agp1-M15-D197A and BV-Agp1-H280A (Table 2).

Fluorescence of Adducts with Singly Locked Chromophores

In the next step, we tested the fluorescence properties of adducts with singly locked chromophores. In the 15Za and 15Ea chromophores, the C15=C16 methine bridge of the chromophore is fixed in either Z anti or E anti stereochemistry, which correspond to the Pr and Pfr form of phytochrome, respectively (Fig. 2). It has been shown that both chromophores assemble with Agp1 and Agp2 and produce adducts with Pr- and Pfr-like properties. As in the PEB adducts, photoisomerization of the 15Za and 15Ea adducts is blocked (29).

The 15Za chromophore was assembled with Agp1, Agp2, Agp1-M15-D197A, Agp1-H280A, and Agp2_Phy1. In all cases the fluorescence quantum yield was higher than that of the corresponding BV adducts. The increase was only minimal for Agp2_Phy1, whereas 15Za-Agp1-M15-D197A and 15Za-Agp1-H280A reached quantum yields of 0.016 and 0.012, respectively. The quantum yields of the 15Za adducts with wild type Agp1 and Agp2 were within the same range, namely 0.013 and 0.014, respectively (Table 2). In all of the 15Za adducts the excitation and emission spectra overlap with indistinguishable peak positions (Fig. 6). The fluorescence spectra of the Agp1, Agp2, Agp1-M15-D196A, and Agp1-H280A adducts match well with the absorption spectra, whereas in the Agp2_Phy1 adduct, absorption and excitation spectra are again clearly different; the excitation maxima lie at shorter wavelengths than the absorption maxima, respectively (Fig. 6). In addition, the fluorescence spectra of 15Za-Agp1-M15-D197A and 15Za-Agp1 resemble each other closely. This spectral coincidence speaks for a fluorescence of the protonated subfraction of 15Za-Agp1-M15-D197A.

FIGURE 6.

Spectra for Agp1 and Agp2 wild type and mutant adducts with singly locked chromophores. A, 15Za-Agp1; B, 15Za-Agp2; C, 15Za-Agp1-M15-D197A; D, 15Za-Agp1-H280A; E, 15Za-Agp2_Phy1; F, 15Ea-Agp1. Dotted lines, absorption spectra; dashed and solid lines, excitation and fluorescence emission spectra, respectively.

We obtained a weak fluorescence for the 15Ea-Agp1 adduct, but the excitation maximum (712 nm) and the emission maximum (721 nm) were significantly blue-shifted compared with the absorption maximum (736 nm). We could not measure a fluorescence signal in 15Ea-Agp2. Other 15Ea adducts were not analyzed.

Fluorescence of Adducts with Doubly Locked Chromophores

We also tested phytochrome adducts with doubly locked chromophores in which both the C5 and C15 methine bridges are locked by additional carbon chains (52). The 5Za15Ea and 5Ea15Ea chromophores do not form a covalent link with the protein, because of the missing vinyl side chain of the ring A (Fig. 2). Both chromophores form adducts with Agp1 and Agp2, which absorb in the red region of the spectrum, whereas adducts with two other doubly locked chromophores, 5Es15Ea and 5Zs15Ea, absorb in the blue spectral range only (52). Adducts with the latter two chromophores were not considered further here.

Both 5Ea15Ea adducts show weak fluorescence with excitation maxima, which are blue-shifted as compared with their absorption maxima (Fig. 7). The 5Za15Ea-Agp2 adduct exhibits a very weak fluorescence, which is even lower than that of the BV adduct, whereas 5Za15Ea-Agp1 has a rather high quantum yield of 0.02 (Table 2). This is surprising, as the other 5Za15Ea/5Ea15Ea adducts showed only very weak fluorescence as well. We had proposed earlier that the stereochemistry of the Agp1 Pfr chromophore could be ZZEasa (i.e. the double bonds around C5, C10, and C15 are in the Z, Z, E configuration, and the single bonds are in the anti, syn, anti conformation) (29, 52). The 5Za15Ea chromophore thus might fit well into the chromophore pocket of Agp1, where it appears to be kept in a more rigid conformation. Another possibility is that the chromophore is oriented in a reversed manner within the chromophore binding pocket of Agp1, i.e. as 5Ea15Za. This would be in line with the rather high fluorescence of the other 15Za adducts, although the fluorescence of the Agp2 adduct with the doubly locked chromophore is not as strong.

FIGURE 7.

Spectra for Agp1 and Agp2 adducts with doubly locked chromophores. A, 5Za15Ea-Agp1; B, 5Za15Ea-Agp2; C, 5Ea15Ea-Agp1; D, 5Ea15Ea-Agp2; E, 5Zs15Za-Agp1; F, 5Zs15Za-Agp2. Dotted lines, absorption spectra; dashed and solid lines, excitation and fluorescence emission spectra, respectively.

The new doubly locked chromophore 5Zs15Za has a ring A vinyl side chain (Fig. 2) and could thus bind covalently to Agp1 and Agp2. Both adducts exhibit a characteristic Pr-like absorption spectrum with maxima at 711 nm for 5Zs15Za-Agp1 and 707 nm for 5Zs15Za-Agp2. Covalent binding of the chromophore to Agp1 and Agp2 was demonstrated using desalting columns and SDS (0.1%)-treated protein adducts (53) (data not shown).

The fluorescence excitation maximum of both 5Zs15Za adducts matches closely with the absorption maximum, and the excitation-emission red shift was 3 and 7 nm for the Agp1 and Agp2 adducts, respectively. The quantum yields were in the range of 1% (Table 2). The Agp2 adduct again revealed a higher quantum yield than that of the Agp1 adduct.

DISCUSSION

Phytochromes are biliproteins that exhibit only very weak fluorescence under ambient conditions. This property distinguishes them from the strongly fluorescent phycobiliproteins of cyanobacteria and red algae (54). However, phytochrome fluorescence intensity can be increased by lowering the temperature, by chromophore replacement, or by site-directed mutagenesis, which leads to an inhibited photoconversion. One future aim is to use phytochrome derivatives as fluorescence markers for cellular studies in a similar manner as the green fluorescent protein. One of the advantages of phytochromes lies in their long wavelength excitation and emission maxima. The use of biliverdin-binding bacterial phytochromes might be a superior choice over PCB- or PΦB-binding cyanobacterial and plant phytochromes (26), because the heme degradation product, biliverdin, is present in many cells including mammalian cells, whereas the other chromophores are found only in plant and cyanobacterial cells.

In this study we investigated the fluorescence properties of the two phytochromes Agp1 and Agp2 from A. tumefaciens and mutants thereof, which were assembled with different natural and synthetic chromophores. According to the fluorescence quantum yield the resulting phytochrome adducts may be divided into four groups: (i) adducts of the wild type proteins with the natural chromophore, which exhibit only very weak fluorescence; (ii) BV adducts of mutants with an impaired Pr to Pfr photoconversion, which have a fluorescence quantum yield in the range of 0.1 to 0.2%; (iii) adducts with locked chromophores, in which the quantum yield is in the range of 1%; and (iv) PEB adducts with a high fluorescence quantum yield of around 50% (Table 2).

We were surprised to see that the fluorescence of all locked chromophore adducts is weaker than that of the PEB adducts. It is generally agreed that the fluorescence of PEB adducts is high, because photoisomerization as the proposed major route of energy dissipation is blocked. However, the same argument also holds for all locked chromophore adducts investigated here. The forced conformation of the locked chromophores could impose structural restraints, and the qualitative difference between PEB adducts and locked chromophore adducts could result from different arrangements of the bilins within the chromophore pocket. In many of the adducts with doubly locked chromophores, the chromophore indeed appears distorted and thus more flexible, as indicated by the low extinction coefficients and broad absorption spectra. (We assumed that distortions should generally result in lowered extinction coefficients, because the high extinction coefficients established during the evolution of phytochromes reflect an optimized situation.) Notably, however, the 15Za adducts of Agp1 and Agp2, and of all PEB adducts used here, are characterized by sharp absorption peaks and high extinction coefficients. Therefore, these chromophores are most likely firmly integrated into the chromophore pocket. The only plausible explanation for the lower quantum yield of 15Za-Agp1 as compared with PEB-Agp1 is to suppose that the energy in the 15Za adducts is dissipated via ring D of the chromophore, which contributes to the extended π-system in 15Za but not in PEB. It is interesting to note that an energy dissipation pathway via ring D has been proposed for phycoerythrocyanin. The phycoviolobilin chromophore of phycoerythrocyanin undergoes light-triggered Z to E isomerization around the C15 methine bridge. The fluorescence yield with the E- configured chromophore is very low (55).

For Agp1, we propose two possibilities for energy dissipation via ring D. (i) Thermal motions of ring D: As noted above, the seven- and eight-member rings of the locked chromophores still provide some flexibility (25, 47). It is unclear whether inside the protein this flexibility is high enough to provide an energy dissipation pathway for >60% of absorbed photons. (ii) Excited state proton transfer: In a recent publication on bacteriophytochrome BphP3 from R. palustris, it has been put forward that the excited state of phytochrome might be deactivated by a rapid (350 ps) proton transfer that finally ends up in a back-conversion to Pr (56). Such a reaction would compete with fluorescence that has a lifetime in the ns range (22). This excited state proton transfer could be a major route of energy dissipation in phytochromes that has not been considered previously, although such a possibility has been discussed for bilins in general (45). The difference between locked chromophore and PEB adducts could result from differences in their excited state proton transfer; in PEB adducts this effect could be lost but still be present in the locked chromophore adducts. In this view, the excited state proton transfer would originate from ring D of the chromophore, which is decoupled from the major conjugated π-system in PEB.

In the H280A mutant, the hydrogen-bonding network around ring D should be disturbed, but a water molecule close to the ring D (7) might serve as proton acceptor for the excited stated proton transfer in this mutant. Newly designed chromophores can help to decide between these possibilities of energy dissipation.

The effect of chromophore distortion can be studied on the Agp2_Phy1 mutant. This protein has been assembled with BV and 15Za, and in both adducts the absorption profiles are broad and flat as compared with the equivalent Agp1 or Agp2 adducts. In the BV adduct of Agp2_Phy1, these distortions are accompanied by an inhibition of photoisomerization. This inhibition is most likely the reason for the elevated fluorescence quantum yield of BV-Agp2_Phy1 over BV-Agp1 or BV-Agp2. The quantum yield of 15Za-Agp2_Phy1 is slightly higher than that of BV-Agp2_Phy1 but lower than that of 15Za-Agp1 or 15Za-Agp2. Thus, distortions of the 15Za chromophore result in a lower fluorescence quantum yield in this case.

In the Agp1-M15-D197A mutant, the situation is different; the 15Za adduct fluoresces stronger than 15Za-Agp1 or 15Za-Agp2, although the comparison of absorption and fluorescence spectra shows that only a subfraction of the adduct is fluorescent. The chromophore of the Agp1-M15-D197A mutant is only partially protonated at the slightly basic pH used here, whereas in wild type Agp1 the chromophore is completely protonated under these conditions (57). Because the excitation and emission spectra of 15Za-Agp1-M15-D197A are remarkably similar to those of 15Za-Agp1, we propose that fluorescence is related to the protonated chromophore. In BV-Agp1-M15-D197A, the excitation spectrum is, however, blue-shifted relative to the absorption spectrum. This speaks for a fluorescence of the deprotonated chromophore. Thus, the Agp1-M15-D197A mutation increases the fluorescence of BV and 15Za in two different ways. The BV fluorescence increases because photoisomerization of the deprotonated chromophore is inhibited. The larger 15Za fluorescence in Agp1-M15-D197A versus Agp1 cannot result from an inhibition of photoisomerization. We assume that the fluorescence increase is related to changes in the hydrogen-bonding network. In wild type Agp1, Asp-197 is part of a hydrogen-bonding network that involves rings B and C of the chromophore (7, 11, 31). Disturbing this network by an Asp-197 replacement could also impact the hydrogen-bonding network around ring D and, in this way, inhibit the energy dissipation pathway proposed above (56).

Altogether, our data may be interpreted in the following manner: in adducts with the natural chromophore, BV, fluorescence can be increased by inhibiting photoisomerization. Such inhibition can, for example, result from a disturbance of the chromophore hydrogen-bonding network, deprotonation of the chromophore, or chromophore displacement. These same effects, which finally result in increased quantum yields, can, however, lower fluorescence by other means. Such effects become evident from the locked chromophore adducts, in which photoisomerization is always inhibited. We thus propose that the combination of mutagenesis and synthesis of new locked chromophores can result in high fluorescence quantum yields of bacterial phytochrome derivatives.