Báo cáo khoa học: Probing protein–chromophore interactions in Cph1 phytochrome by mutagenesis potx

We generated 12 site-directed mutants of Cph1D2, focusing on conserved residues which might be involved in chromophore– protein autoassembly and photoconversion.. Abbreviations BV, biliv

Probing protein–chromophore interactions in Cph1

phytochrome by mutagenesis

Janina Hahn1, Holger M Strauss1, Frank T Landgraf2, Hortensia Faus Gimene`z2, Gu¨nter Lochnit3, Peter Schmieder1and Jon Hughes2

1 Forschungsinstitut fu¨r Molekulare Pharmakologie, Berlin, Germany

2 Pflanzenphysiologie, Fachbereich Biologie & Chemie, Justus-Liebig-Universita¨t, Giessen, Germany

3 Biochemisches Institut, Fachbereich Medizin, Justus-Liebig-Universita¨t, Giessen, Germany

Phytochrome photoreceptors play a central role in the

regulation of plant development Phytochromes are

red⁄ far-red photochromic proteins with a covalently

bound linear tetrapyrrole (bilin) prosthetic group In

the Pr ground state the chromophore preferentially

absorbs red light, this leading to a Zfi E

isomeriza-tion around the C15–C16double bond between rings C and D Further conformational changes culminate in the formation of Pfr, the signalling state This prefer-ably absorbs far-red light which converts the pigment back to Pr [1] The active photoreceptor is formed by the apoprotein taking up and covalently attaching an

Keywords

biliprotein; photoreceptor; phytochrome; site

directed mutagenesis; structure–function

studies

Correspondence

J Hahn, Forschungsinstitut fu¨r Molekulare

Pharmakologie, Robert Ro¨ssle Str 10,

D-13125 Berlin, Germany

Fax: +49 30 94793169

Tel: +49 30 94793316

E-mail: hahn@fmp-berlin.de

(Received 28 October 2005, revised 27

January 2006, accepted 3 February 2006)

doi:10.1111/j.1742-4658.2006.05164.x

We have investigated mutants of phytochrome Cph1 from the cyanobacter-ium Synechocystis PCC6803 in order to study chromophore–protein inter-actions Cph1D2, the 514-residue N-terminal sensor module produced as a recombinant His6-tagged apoprotein in Escherichia coli, autoassembles

in vitro to form a holoprotein photochemically indistinguishable from the full-length product We generated 12 site-directed mutants of Cph1D2, focusing on conserved residues which might be involved in chromophore– protein autoassembly and photoconversion Folding, phycocyanobilin-bind-ing and Prfi Pfr photoconversion were analysed using CD and UV–visible spectroscopy MALDI-TOF-MS confirmed C259 as the chromophore attachment site C259L is unable to attach the chromophore covalently but still autoassembles to form a red-shifted photochromic holoprotein H260Q shows UV–visible properties similar to the wild-type at pH 7.0 but both Pr and Pfr (reversibly) bleach at pH 9.0, indicating that the imidazole side chain buffers chromophore protonation Mutations at E189 disturbed fold-ing but the residue is not essential for chromophore–protein autoassembly

In D207A, whereas red irradiation of the ground state leads to bleaching

of the red Pr band as in the wild-type, a Pfr-like peak does not arise, impli-cating D207 as a proton donor for a deprotonated intermediate prior to Pfr UV-Vis spectra of both H260Q under alkaline conditions and D207A point to a particular significance of protonation in the Pfr state, possibly implying proton migration (release and re-uptake) during Prfi Pfr photo-conversion The findings are discussed in relation to the recently published 3D structure of a bacteriophytochrome fragment [Wagner JR, Brunzelle

JS, Forest KT & Vierstra RD (2005) Nature 438, 325–331]

Abbreviations

BV, biliverdin IXa; Cph1D2, the N-terminal 1–514 residue sensory module of Cph1 from Synechocystis PCC6803; e, extinction coefficient; FTRR, Fourier transform resonance Raman spectroscopy; FWHM, full width half maximum; IPTG, isopropyl thio-b- D -galactoside; LED, light-emitting diode; MeOH, methanol; Pr ⁄ Pfr, red ⁄ far-red absorbing form of phytochrome; PCB, phycocyanobilin; PFB, phytochromobilin; SAR, specific absorbance ratio; SEC, size-exclusion chromatography; k max , wavelength of the absorbance maximum.

appropriate bilin from the cytoplasm: this process is

called autoassembly [2] Phytochromes are exceedingly

effective photoreceptors on account of their high

extinction coefficients in the red⁄ far-red region, low

fluorescence losses, high resistance to photobleaching

and use of a thermodynamically stable signalling state

to activate their response pathway The molecular

pro-cesses underlying autoassembly, hyper- and

photochro-micity and signal transduction are thus of considerable

interest

The unexpected discovery of a prokaryotic

phyto-chrome, Cph1 [3,4], fundamentally changed our view

of evolution and function of this class of

photorecep-tors, relating them to histidine sensor kinases, a

pro-tein family involved in a wide variety of perception

systems in prokaryotes, fungi and plants [5] Cph1 has

numerous features in common with plant

phyto-chromes Furthermore, large amounts of pure, highly

concentrated holoCph1 can easily be produced by

apo-protein overexpression in Escherichia coli and in vitro

autoassembly with an appropriate bilin [6] HoloCph1

can also be produced in E coli by coexpressing haem

oxygenase and appropriate bilin reductase genes

together with Cph1 [7,8] Cph1 is thereby well suited

to studies of autoassembly as well as of the

photocon-version mechanism Numerous related photoreceptors

have subsequently been identified in prokaryotes,

nota-bly bacteriophytochrome from Deinococcus

radiodu-rans, DrBphP, the 3D structure of whose N-terminal

domain was recently published [9]

Phytochrome sequences show highly conserved

regions probably representing functionally essential

subdomains [6,10] The UV-Vis absorbance and

vibra-tional spectroscopic characteristics of phytochromes

assembled with the same chromophore are remarkably

similar, whereas significant and characteristic changes

are associated with subtle changes in the bilin

pros-thetic group It was thus expected that the pocket in

which the chromophore is held is constructed from

various functionally conserved subdomains reflected at

the sequence level in all phytochromes The new X-ray

structure [9] indeed bears this out although it must be

born in mind that DrBphP differs functionally from

plant-like phytochromes in many respects and that the

fragment crystallized is photochemically impotent

In oat phytochrome A the phytochromobilin (PFB)

chromophore is attached by a thioether link to C322#380

[11–13], a residue conserved in plant-type phytochromes

including Cph1 but not in bacteriophytochromes (the

residue number is that of the named phytochrome,

# indicating its position in the alignment at www

uni-giessen.de/gf1251/Phytochrome/align2x.htm)

Phyco-cyanobilin (PCB) is probably the native Cph1

chromophore [14], but no direct evidence for its expec-ted attachment at C259#380 has been published [5] A substitution at this putative ligation site should abolish covalent attachment, but not necessarily other protein– bilin interactions, as studies with blocking reagents and

of autoassembly kinetics have implied [15–17] In free PCB at neutral pH 7 the two central ring nitrogens share a single proton, but a second is added under acid conditions Protonation occurs during phytochrome autoassembly too, but the donor is unknown Con-versely, homology studies implied that a basic residue homologous to R254#375close to the presumed chromo-phore attachment site interacts with the propionate side chain of chromophore ring B [18,19] Additionally, the strength and position of the dominant red and far-red absorbance bands of Pr and Pfr, respectively, are pH-dependent, an H residue near the chromophore being implicated [20] H260#381adjacent to the putative ligation site is perfectly conserved and hence a prime candidate for this function

Such conserved interactions probably central to phytochrome action can be probed by modifying the protein moiety via site-directed mutagenesis of the cog-nate gene [21–25], with the important proviso that, except in the case of null phenotypes, all conclusions based purely on site-directed mutagenesis are confoun-ded by unknown possible side-effects on folding Ide-ally, the mutations are guided by 3D structural data Such information for phytochrome [9] were not avail-able at the time of this study

The N-terminal 514 residue sensory module of recombinant Cph1) that is, Cph1D2 ) is photochemi-cally autonomous We generated 12 amino acid replacement mutants in Cph1D2 and analysed their expression, autoassembly, UV-Vis absorbance, photo-chromicity and thermal reversion properties We also used CD spectroscopy to detect gross changes in sec-ondary structure: only correctly folded products were considered to offer interpretable information We pro-vide the first direct epro-vidence that the PCB chromo-phore is ligated to C259#380, analogously to oat phyA, and also describe the effects of mutations at this resi-due The H260#381Q mutant showed massive, reversible

pH effects on the absorbance spectra, obliterating the characteristic Pfr peak, implying that the imidazole side chain buffers chromophore protonation, partic-ularly in the case of Pfr A perhaps related effect was seen for the conserved acidic residue D207#328: when this was replaced by A, although Pr (reversibly) photo-bleached, no Pfr-like peak was formed in its place Mutations of R254#375 had similar small effects on UV-Vis properties: that the R residue is nevertheless perfectly conserved implies a role in signal

transduc-tion We also show that E189#130 is not required for

covalent autoassembly, as had been proposed We

dis-cuss these findings in relation to phytochrome function

and the bacteriophytochrome X-ray structure [9] which

has subsequently become available

Results

Characterization of Cph1D1 and D2

At the start of this work two deletion clones with

C-terminal His6-tags were created, Cph1D1 and

Cph1D2, in which the N-terminal 492 and 514

resi-dues, respectively, of Cph1 were overproduced as

apo-proteins in E coli, autoassembled with and purified

by nickel affinity chromatography Although most

Cph1D1 was expressed as insoluble inclusion bodies

so that the final yield of soluble apoprotein was only

 500 lgÆL)1 culture, addition of PCB resulted in

covalent autoassembly (as apparent from Zn2+

-induced bilin fluorescence in SDS⁄ PAGE) and red ⁄

far-red photochromicity The Pfr-like band was weak

and significantly blue-shifted, however, in comparison

to full-length Cph1 (Fig 1); an effect was also seen

in a similar deletion mutant [26] In contrast,

Cph1D2 yielded up to 80 mg apoproteinÆL)1 culture

and showed a difference spectrum almost identical to

that of full-length Cph1 (Fig 1), confirming the

results of Yeh et al [4] Coomassie-stained SDS⁄

PAGE indicated a purity of  80% for Cph1D2 at

this stage Further purification via Superose 200

(Amersham Pharmacia⁄ GE) size-exclusion

chromato-graphy (SEC) yielded essentially pure holoprotein

Cph1D2, unlike full-length Cph1, shows no tendency

to aggregate in vitro (data not shown)

The extinction coefficient of Cph1D2 Pr was

82 mm)1Æcm)1 at 654 nm (kmax), a value of

85 mm)1Æcm)1 for full-length Cph1 confirming that earlier reported [16] Further UV-Vis data are summar-ized in Table 1 Extinction coefficients for free PCB were 16, 20, 29, 30 and 46 mm)1Æcm)1at each kmax in Tris⁄ HCl pH 7.8, MES pH 5.5, sodium acetate

pH 3.0, 0.5 m HCl pH 0.3 and CH3Cl⁄ HCl (1 : 19), respectively The relevant UV-Vis spectra are shown in Fig 2 The maximal 654 nm⁄ 280 nm specific absorb-ance ratio (SAR) of Cph1D2 Pr obtained was 1.3, sig-nificantly higher than that for full-length Cph1 at equivalent purity (1.0 [16]) e280 nm was calculated to

be 59 and 83 mm)1Æcm)1 for Cph1D2 and full-length Cph1 apoproteins, respectively (Vector NTI, Infor-max) The contribution of PCB attached to the holo-protein is about 5 mm-1Æcm-1 [16], yielding 64 and

88 mm)1Æcm)1 at 280 nm for the holoproteins Taking the Pr ekmax in the red region from Table 1, the max-imal SAR would be 1.28 and 0.98, respectively, in close agreement with that of our purest samples Quantum efficiencies of photoconversion were not measured directly, but kinetics under red and far-red irradiation were similar for Cph1D2 and full-length Cph1 holoproteins As the e-values are similar, we thus expect quantum efficiencies to be similar too, that is

 0.16 in each direction [6,20] A maximal 0.70 mole fraction of Pfr at photoequilibrium in red light is seen in Cph1D2 as in full-length Cph1: the calculated UV-Vis spectrum for 100% Pfr derived from this is identical to that of purified Cph1D2 Pfr (unlike full-length Cph1, Cph1D2 as Pr is monomeric except at very high (> 10 mm) concentrations; the Pfr form, however, homodimerises, readily allowing it to be purified by SEC [27]) Dark reversion is insignificant: none was detected after 2 weeks at 20
C (data not shown)

λ [nm]

-1.0

-0.5

0.0

0.5

1.0

Fig 1 UV-Vis difference spectra of full-length Cph1 (n) and deletion

mutants Cph1D1 (d) and Cph1D2 (m).

Table 1 Summary of UV-Vis absorbance data for full-length Cph1 and the deletion mutant Cph1D2 ND, not determined; ibp, isosbe-stic point.

Parameter

Cph1 (full-length) Cph1D2 (N514D)

k max, red 656 nm 704 nm 654 nm 702 nm

kmax, UV⁄ A 359 nm ND 358 nm ND

e at kmax, red 86 m M )1Æcm)1 ND 82 mM)1Æcm)1 ND

k DA,max 655 nm 707 nm 655 nm 707 nm

Molecular mass (Apoprotein + His tag)

Direct determination of the chromophore

attachment site

Tryptic fragments of PCB-Cph1D2 holoprotein were

separated by HPLC, the chromopeptide eluting as

a single peak with a spectrum closely fitting that

expec-ted for protonaexpec-ted PCB covalently linked via ring A to

the peptide (Fig 3A) The chromopeptide showed weak

214 nm absorbance, implying a poor release efficiency

MALDI-TOF MS of this fraction showed major peaks

([M + H]+) with a typical isotopic profile

correspond-ing to predicted tryptic fragments 98–112 (m⁄ z 1753.849

and in the methionine oxidized form at m⁄ z 1769.851),

64–80 (m⁄ z 1951.986) and 399–420 (m ⁄ z 2534.271) of

Cph1 (Fig 3B) The expected PCB-coupled

chromopep-tide SAYHC*HLTYLK (residues 255–279) is predicted

to have a molecular mass of 1920.923 Da A small but distinct double peak corresponding to the expected [M + H]+ at m⁄ z 1921.9384 and to [M–H]+ at m⁄ z 1919.929 was seen, the latter probably presenting an oxidized derivative (a similar effect was seen in a study

of Agp1 where the Biliverdin IXa (BV) chromopeptide ion detected was also 2 Da lighter than expected [28])

MS2 analysis of the double peak showed fragment ions

at m⁄ z 585.991 ⁄ 587.939 (reflecting the expected and

A

B

Fig 2 pH dependence of PCB UV-Vis absorption spectra (A)

Absorption spectra of free PCB in different buffers at different

pH-values Spectra are plotted for PCB in 100 m M Tris ⁄ HCl pH 7.7

(n), 10 m M MES pH 5.5 (d), 5 m M sodium acetate pH 3.0 (m),

0.5 M HCl pH 0.3 (.), HCl ⁄ MeOH (1 : 19) (e) and CH 3 Cl ⁄ HCl

(1 : 19) (n) For comparison the absorption spectrum of Cph1D2 in

the Pr state is shown (dotted line) (B) pH difference spectra for

free PCB Absorbance changes are plotted for pH values 5.5, 3.0,

0.3 (solid lines) and HCl ⁄ MeOH (dashed line) after subtraction of

the pH 7.7 spectrum.

A

B

C

Fig 3 Tryptic profiles and MALDI spectra of Cph1D2 (A) HPLC elution profiles at 214 nm (peptide absorbance, upper panel) and

370 nm (bilin UV ⁄ A absorbance, lower panel); inset: UV-Vis-spec-trum of chromopeptide peak (B) MALDI-TOF specUV-Vis-spec-trum of chromo-peptide fraction; inset: enlarged (C) MALDI-TOF ⁄ TOF spectrum; inset: enlarged.

oxidized forms of the cleaved chromophore) and

1336.915 (reflecting the peptide backbone) (unlike

full-length Cph1, Cph1D2 as Pr is monomeric except at very

high (> 10 mm) concentrations; the Pfr form, however,

homodimerizes readily allowing it to be purified by SEC

[28]) Edman microsequence data (not shown) from the

same fraction were consistent with the sequences of

the fragments identified in MALDI: the fifth residue of

the chromopeptide) C259 ) was absent, as would be

expected for a cysteinyl–PCB complex Taken together

these data show that the PCB chromophore is ligated to

C259 via a thioether bond

Characterization of Cph1D2 site-directed mutants

To determine the role of specific conserved residues in

the Synechocystis phytochrome Cph1, 12 site-directed

mutations were introduced into the N-terminal sensory

module Cph1D2 The mutants were heterologously

expressed as C-terminally His6-tagged apoproteins in

E coli, purified and tested for PCB-binding,

apopro-tein folding, Pr–Pfr photochromicity and thermal

reversion using SDS⁄ PAGE ⁄ zinc fluorescence and CD

and UV-Vis spectroscopy The appropriate data is

summarized in Table 2 and in Figs 4 and 5

Y257#378, h258#379, l261#382

These residues lie close to the chromophore binding

site but are not conserved in other phytochromes and

are thus, in contrast to conserved residues, probably not functionally important Indeed, the Y257H, H258F and L261A holoproteins showed no significant differences in chromophore autoassembly, UV-Vis or

CD properties relative to the wild-type (Table 2)

C259#380

As MALDI studies showed, this is the residue in Cph1

to which PCB becomes attached via a thioether bond Thus mutations at C259 should abolish covalent attach-ment and have dramatic effects on photochemistry Both C259M and C259L mutants autoassembled with PCB to give red⁄ far-red photochromic holoproteins although, as expected, covalent attachment did not occur (Figs 4 and 6) The autoassembly reaction was much slower than in the wild-type especially under nonreducing conditions, taking many hours for chromophore binding and photochromicity to become saturated even with a large PCB molar excess (as seen

in [29]) After brief incubation of apoprotein with a small molar excess of PCB under nonreducing condi-tions, holoC259L showed an almost symmetrical differ-ence spectrum following red irradiation, with lowest energy bands at 674 and 735 nm, representing a

 25 nm bathochromic shift relative to the wild-type Subsequent irradiation with FR did not repopulate the Pr-like species, however (Fig 6A,B) HoloC259L allowed to assemble to completion under reducing con-ditions showed similarly shifted lowest energy bands

Table 2 Characterization of Cph1D1, Cph1D2 and Cph1D2-mutants ND, not determined; ibp, isosbestic point.

Mutation

Soluble expression

relative to Cph1D2 a

CD like wild-type Cph1D2

Covalent PCB attachment

Difference spectrum Absorption spectrum

k max (Pr) [nm]

k DA,0 ibp

[nm]

k max (Pfr) [nm]

k max (Pr) [nm]

k max (Pfr) [nm]

(+ E196G)

a Expression yield of Cph1D2: 80 mgÆL)1culture: + + +, 100 – 50%; + +, 40–10%; +, < 10%; –, insoluble expression.

The Pr peak was much weaker than that of Pfr, but this

photochromicity was now stable (Fig 6C,D) Attempts

to remove unbound bilins by chromatography lead to

chromophore escape as all reversibility was lost

H260#381 This residue is perfectly conserved in all phytochromes, even those in which the canonical C#380 attachment site itself is missing The H260Q mutant of Cph1D2 bound PCB covalently (Fig 4A) to give an only slightly blue-shifted Pr absorbance maximum at

639 nm (Fig 4B) As an H residue imidazole side chain was expected to be involved in (de)protonation

of the chromophore [20], we measured absorption spectra of Cph1D2 wild-type and H260Q holoproteins

at different pH values following far-red and red irradi-ation, uncovering a remarkable phenotype (Fig 7) At

pH 7 the spectra of the mutant and wild-type Pfr forms were similar (kmax 700 nm and 703 nm for the lowest energy bands, respectively) while mutant Pr was

 14 nm downshifted (kmax 641 nm and 656 nm, respectively) At pH 9, however, the mutant behaved differently from the wild-type: Pr-typical red absorb-ance band weakened almost 10-fold while that of Pfr disappeared completely Weaker bands at 549 nm and

577 nm, respectively, appeared in their place The effect was fully reversed by returning the pigment to

pH 7 Not surprisingly, far-red irradiation of the bleached form at pH 9 induced no photochemistry, whereas 550 nm irradiation of the bleached ground state did lead to photoconversion as the product revealed itself as Pfr once the pH 7 was restored (data not shown)

D207#328 This acidic amino acid is conserved in all phyto-chromes and might be involved in chromophore proto-nation CD spectroscopy showed that the D207A replacement was similarly folded to the wild-type, binding PCB covalently to form an apparently normal

Pr state (kmaxat 653 nm, Figs 4 and 5) Upon red irra-diation the Pr band bleached as in the wild-type, but

no Pfr-like peak appeared The Pr-like form reap-peared in darkness, however, reversion being complete within an hour (Fig 8) The extinction coefficient of D207A was estimated to be 62.9 mm)1Æcm)1 A D207N⁄ E196G double mutant behaved similarly (Table 2)

R254#375

As this residue is perfectly conserved in plant as well

as prokaryotic phytochromes, it is likely to be func-tionally important Therefore R254 was mutated to K and to A Whereas the CD spectrum of the conserva-tive K mutant was almost identical to that of the

wild-Coomassie

wt D207A R254K H260Q C259L E189Q

Zn-fluorescence

A

B

Fig 4 Cph1D2 and site-directed mutants (A) Coomassie stain and

Zn2+fluorescence of Cph1D2 and of selected Cph1D2 site-directed

mutants after SDS ⁄ PAGE (B) UV-Vis difference spectra of Cph1D2

and of selected Cph1D2 site-directed mutants Absorbance

differ-ence maxima and isosbestic points are given The dotted vertical

lines are drawn through the absorption maxima of the Pr and Pfr

state of Cph1D2 to highlight shifts in the mutants.

type, the CD spectrum of R254A implied slight folding

differences (Fig 5) Nevertheless, both bound PCB

covalently to form a red⁄ far-red photochromic

holo-protein with the red kmax of Pr  10 nm downshifted

but Pfr spectra indistinguishable from that of the

wild-type

E189#310

This acidic residue was the focus of earlier mutagenesis

studies which inferred a central role in bilin ligation

[25] Our E189A mutant was expressed as insoluble

protein bodies, attempts at refolding by solubilization

in urea followed by slow dialysis proving unsuccessful

The E189Q mutation was better tolerated although the

expression yield of soluble protein was much lower

than for the Cph1D2 wild-type CD spectroscopy

implied, furthermore, that folding was significantly

dif-ferent from that of the wild-type However, when this

mutant apoprotein was presented with PCB, a low

level of covalent attachment accompanied by a weak

Pr-like band at 665 nm was seen (Figs 4A and 9),

implying normal protonation of a thioether-linked

bilin No photochromicity signal associated with red⁄

far-red irradiation was measurable, however

Discussion

In this study we focused on Cph1D2, the N-terminal

514 residue sensory module of Cph1 The smaller dele-tion product, Cph1D1 (N1–492) was funcdele-tionally compromised (Table 1 and Fig 1), showing very poor solubility and a weak Pfr-like absorbance typical

of phytochromes in which the PHY subdomain (see http://www.sanger.ac.uk//Software/Pfam) is incom-plete On the other hand the UV-Vis absorbance prop-erties of holoCph1D2) whose C-terminus corresponds exactly to that of the PHY subdomain) closely resem-ble those of full-length holoCph1 (see Taresem-ble 1 and Fig 1) Thus the sensory module is photochemically autonomous, as implied in an earlier study [25] Cph1D2 can therefore be used as a convenient model for investigating phytochrome functions, as) unlike full-length Cph1) it does not aggregate under normal

in vitroconditions Indeed, as pure Pfr can be obtained

by SEC [27], it might also be possible to obtain struc-tural data for that form too

UV-Vis absorbance properties of bilins and other tetrapyrroles are determined both by their protonation state and by the extent and linearity of the conjugated

p orbital system Coiled bilins (like free PCB) show a

Fig 5 Circular dichroism spectra of

selec-ted mutants For comparison Cph1D2-wt

spectra are shown (dotted lines) and molar

ellipticities were calculated.

strong UVA band but weak absorbance at longer

wavelengths, while in linear bilins the situation is

reversed, the dipole moment perpendicular to the long

molecular axis giving strong red absorbance at the expense of the UV⁄ A band (the UV ⁄ A and lowest energy red⁄ far-red bands are sometimes called Soret

Fig 6 UV-Vis absorbance properties of Cph1D2-C259L in the presence of excess PCB Absorbance and difference spectra under nonreducing (A,B) and reducing (C,D) conditions (A,C) (n) Pr[1], after autoassem-bly with PCB in the dark; (d) Pfr (7 ⁄ 3 Pfr ⁄ Pr mixture) after R irradiation; (h) Pr[2], after

FR irradiation (B, D) (n) Pr[1]–Pfr; (d) Pfr– Pr[2].

A

B

Fig 7 pH-dependence of UV-Vis absorbance properties of Cph1D2

and Cph1D2-H260Q after far-red (100% Pr) and after red (7 ⁄ 3

Pfr ⁄ Pr photoequilibrium) irradiation (A) Cph1D2 wild-type at pH 7

(after far-red n, after red d) and pH 9 (after far-red h, after red s).

(B) Cph1D2-H260Q at pH 7-start (after far-red n, after red, m), pH 9

(after far-red h, after red n) and at pH 7-end (after far-red r, after

red ).

A

B

Fig 8 UV-Vis absorbance properties of Cph1D2-D207A (A) Spectra recorded after autoassembly with PCB in the dark (n), after 90 s red irradiation (d) and after 90 s far red irradiation (m) (B) Thermal reversion of Cph1D2-D207A After PCB assembly in the dark (n) and saturating red irradiation (d), the sample was kept in the dark and absorption spectra were recorded after 10 (m), 20 (.), 30 (r) and 45 (b) min.

and Qy in analogy to closed-ring tetrapyrroles; this

possibly misleading terminology has been avoided

here) [30,31] While the ratio of the two ‘oscillator

strengths’ changes with uncoiling, the total absorptivity

remains constant FTRR (e.g [32] for oat phyA) gives

more specific information about the conformation of

the chromophore, as of course can more direct

meth-ods like NMR (e.g [33,34] and Rohmer T and

Matysik J., University of Leiden, the Netherlands,

unpublished data) and X-ray crystallography (e.g [9])

On the other hand, protonation has only subtle effects

in the UV⁄ A region, while the red peak strengthens

approximately threefold at low pH [34,35] In fact, in

PCB a new band at 688 nm appears and strengthens

with protonation, but its kmax does not shift, while the

broad, weaker shoulder centred at 615 nm (also seen

in phytochrome spectra) remains unchanged (Fig 2)

Recently, Go¨ller et al (2005) [36] have successfully

modelled the role of protonation in strengthening ered

by enhancing electronic coupling between PCB rings,

while NMR studies [33] have proved that all four rings

are fully protonated in both Pr and Pfr states of Cph1

The current model for the autoassembly reaction

[15,16,37], important to the present study, envisages

three-steps: (1) an initial chromophore recognition

pro-cess (< 1 ms) of the unprotonated, coiled bilin with

weak absorbance in the orange region; (2) entrance

into a pocket within the protein (100–200 ms) during

which uncoiling and protonation occur, leading to a

fourfold hyperchromicity of the lowest energy

absor-bance band in red⁄ far-red and the appearance of

photochromicity in that region; (3) a final covalent

ligation (1–10 S) to a C residue through the formation

of a thioether bond While pioneering studies showed

that PUB is attached to oat phyA at C322#380 [13] at

least some bacteriophytochromes attach a BV chromo-phore at a C residue close to the N-terminus [9,28,38], contradicting earlier data [39] A further important dif-ference is that bacteriophytochromes covalently ligate

to the ring A vinyl side chain of BV, forming a two-carbon linker, while in oat phyA the ring A ethylidene side chain of PFB yields a single-carbon linker Here

we present direct evidence that in Cph1 PCB is simi-larly ligated to C259#380 (Fig 3) As shown by our C259L mutant, if step 3 of autoassembly is prevented

by mutating this residue, many holophytochrome-like features appear, but kmax values are shifted  25 nm bathochromically (Fig 4) as would be expected if the ethylidene group double bond was left intact to contri-bute to the PCB delocalized p-electron system This is seen also in the wild-type if C residues are blocked nonspecifically by iodacetamide [16] No effect is seen with blocked Agp1 and BV, however, in accordance with the vinyl group double bond in that case not being connected to the p-system [40] Redox conditions seem to be important in autoassembly steps 1 and⁄ or

2 PCB binding was weak under nonreducing condi-tions, requiring at least 15 lm PCB, and even then only a single round of R⁄ FR photoconversion was possible – as though the chromophore was lost as a consequence of photoconversion (Fig 6A,B) Under reducing conditions (Fig 6C,D) the relative strengths

of the UV⁄ A and red bands were approximately equal, implying a more chiral chromophore conformation than in the wild-type, a conclusion consistent with experiments using methoxy-PCB [41] Thus conforma-tional changes leading to uncoiling are associated with both step 2 and step 3 of autoassembly The long wavelength band in the Pfr-like state of the C259L mutant was even weaker than that of the ground state

R#375 near the ligation site is perfectly conserved amongst all known phytochromes, bacteriophyto-chromes and even several other biliproteins It thus might be expected to be important in chromophore binding and conformation Indeed, X-ray structural data shows that it forms a salt bridge with the pro-pionate side chain of ring B, apparently pulling on the chromophore from deep within the protein [9,17–19] However, R#375I and T mutants of pea phyA showed only  5 nm hypsochromic shifts [22] Here we mutated R254#375 to K (likewise a basic residue) and

to A (a smaller, moderately hydrophobic residue) Both mutants fold similarly to the wild-type and bind PCB effectively (Figs 4A and 5) Whereas their Pfr absorbance characteristics match those of the wild-type almost exactly, the Pr peak shows a 10-nm hypsochro-mic shift in both cases (Table 2, Fig 4B) This might

Fig 9 UV-Vis absorbance properties of Cph1D2-E189Q Spectra

recorded after autoassembly with PCB in the dark (n) and

subse-quent red irradiation (d).

arise from a rotation around the C5–C6 bond specific

to the Pr form (H Scheer, LMU Munich, personal

communication), although such a shift is also predicted

for (de)protonation of the propionate [36] Either way,

the UV-Vis shift is much too subtle to explain the

degree of conservation seen, thus it is very likely that

R254#375 instead plays a central role in signal

trans-duction: indeed, the 1ZTU structure implies that even

a slight movement of the chromophore would break

the salt bridge, fitting with the UV-Vis phenotype of

our R254K mutant Overlooked to date, the X-ray

structure shows an intriguing 2.5-A˚ diameter channel

leading from the salt bridge to the other side of the

protein, wide enough for water molecules or

hydroxo-nium ions It would in any case be worthwhile

investi-gating R254#375mutants at the physiological level

Go¨ller et al [36] also calculated that ring B ⁄ C

pro-tonation of chiral PCB leads to dramatically increased

electron coupling associated with a bathochromic shift

of emax of 130 nm: we observe a shift of 160 nm on

acidification of free PCB compared with 150 nm for

holoCph1 prior to ligation (i.e in C259L), a

reason-able fit considering likely conformation differences

Protonation requires a donor with a pKa of < 4.6,

and is thus just possible for E or D carboxyl side

chains [35] None of the 12 highly conserved E or D

residues are near C259#380 in the primary sequence,

furthermore, the recent 1ZTU structure shows that

only two of these E189#310 and D207#328 are

posi-tioned anywhere near the bilin Coincidentally, these

are the two residues we had focused on in the present

study As we show, E189Q mutations are better

toler-ated than others [25,42], UV-Vis properties being

con-sistent with a ground state resembling protonated Pr

(Figs 4 and 9, Table 2) Thus the proposed central role

for E189#310 in ligation [25] is unlikely, neither is it

likely to be the proton donor in autoassembly step 2

Unfortunately, D207#328 too is most unlikely to fulfil

this role The mutant apoprotein bound PCB

covalent-ly (Fig 4A) to yield a ground state with a similar emax

and kred compared to wild-type, implying a partially

coiled, protonated chromophore Furthermore, 1ZTU

shows that the carboxyl group of D207#328 is directed

away from the chromophore, forming a hydrophilic

acid patch exposed to the solvent, at least in this BphP

deletion mutant The main chain carbonyl oxygen of

D207#328interacts with the nitrogens of rings A, B and

C, so that they might share their protons – but this is

a proton acceptor, not a donor, and of course any

mutation at this site could fulfil this role Our mutants

imply that D207#328 is important in Pfr formation

Red irradiation leads to bleaching (as in the wild-type),

but no Pfr-like band appeared, rather a broad peak

centred at 590 nm remained Not surprisingly, FR irra-diation had no effect, but the bleached form reverts thermally to the Pr-like state D207#328N behaved simi-larly (Table 2) As proton exchange is probably associ-ated with photoconversion (see below), D207#328might

be involved in reprotonation prior to Pfr formation Such a role would not be apparent from 1ZTU because this cannot form bona fide Pfr It seems clear, however, that neither E189#310 nor D207#328 can be the proton donor in step 2 of autoassembly Thus the donor for bilin protonation, even in the light of the 1ZTU structure, remains unknown

Although it is now certain that all four chromophore nitrogens are protonated in neutral buffers in both Pr and Pfr states [33], transient deprotonation of the chro-mophore seems to be a feature of Pr fi Pfr photocon-version [20,43,44] Significant deprotonation of both Pr and Pfr can be induced by shifting the pH, however, one pKa component being close to neutral, thus possibly representing an imidazole (H side chain) and⁄ or a direct effect on PCB [20] In the present study we show that H260#381 plays a crucial role in this process Mutagen-esis of H260#381to L, R, F, G and Q has already been reported for recombinant phytochrome A from pea and oat In the first four cases covalent ligation and photo-chromicity were obliterated probably because of misfolding, whereas H260#381Q retained covalent attachment and photochromicity [21,22,24] While Q resembles H regarding its steric demands and its hydro-gen bonding ability [45], its buffering capacity is very weak While our H260Q mutant is wild-type like in its folding, photochromicity and covalent autoassembly under normal conditions (Table 2; Figs 4 and 5), it shows dramatically increased sensitivity to buffer pH (Fig 7), the long wavelength absorbance peak of Pr weakening drastically and that of Pfr disappearing com-pletely at pH 9.0, much weaker, broader bands centred

at 549 nm and 577 nm, respectively, appearing in their place The UV⁄ A bands show smaller changes for both states These effects are fully pH-reversible We con-clude that H260#381plays a crucial role in buffering both

Pr and Pfr protonation As it is easy to titrate chromo-phore protonation in the mutant, this offers a poten-tially useful degree of freedom for more sophisticated analytical methods

H260#381 is likely to be important according to the 1ZTU X-ray structure [9] The chromophore nitrogens

of rings A, B and C are hydrogen bonded to the D207#328 backbone nitrogen, perhaps sharing their protons On the other side of the pocket, the d1 nitro-gen of H260#381and chromophore ring C nitrogen are separated by 3.3 A˚, just outside van der Waals’ con-tact, but a hydrogen bonding bridge is provided by