Targeted knockout in Physcomitrella reveals direct actions of phytochrome in the cytoplasm

Abstract

The plant photoreceptor phytochrome plays an important role in the nucleus as a regulator of transcription. Numerous studies imply, however, that phytochromes in both higher and lower plants mediate physiological reactions within the cytoplasm. In particular, the tip cells of moss protonemal filaments use phytochrome to sense light direction, requiring a signaling system that transmits the directional information directly to the microfilaments that direct tip growth. In this work we describe four canonical phytochrome genes in the model moss species Physcomitrella patens, each of which was successfully targeted via homologous recombination and the distinct physiological functions of each gene product thereby identified. One homolog in particular mediates positive phototropism, polarotropism, and chloroplast movement in polarized light. This photoreceptor thus interacts with a cytoplasmic signal/response system. This is our first step in elucidating the cytoplasmic signaling function of phytochrome at the molecular level.

Plants possess sophisticated photoperception systems to regulate their physiology. The phytochrome system is particularly important, controlling, for example, germination, hypocotyl and stem extension, pigment accumulation, and even flowering in higher plants (1). Although phytochrome is known to regulate transcription, this cannot explain very fast phytochromemediated responses in higher and lower plants or phytochrome-mediated directional responses (2, 3).

In marked contrast to higher plants in which cells are organized as complex tissues, lower plants are well suited to studies of intracellular sensing of and response to light direction. For example, in moss protonemata, the filaments produced after spore germination, numerous directional responses within the cell are mediated by phytochrome. The tip cells use phytochrome to sense light direction and adjust their direction of growth accordingly (4). Chloroplast position is also regulated by phytochrome, as is the developmental axis of regenerating protoplasts (5, 6). Importantly, such responses are often very sensitive to light polarization. For these effects, phytochrome molecules must be fixed in coherent patterns in the cell, and their transduction chains must faithfully transmit the directional information defined by the light beam to the site of the response. On the other hand, abundant evidence indicates that phytochrome is freely mobile in the cytosol, entering the nucleus to regulate transcription (1). How can these two very different concepts of phytochrome action be reconciled?

To answer this question, the molecules involved in light detection and signaling must be identified. To this end, we have carried out an extensive molecular genetic and physiological study of phytochromes in the model moss species Physcomitrella patens: this is the only well characterized plant system in which high levels of homologous recombination allow specific chromosomal genes to be targeted and their functions thereby studied (7). Moreover, protonemal filaments (Fig. 1) are gametophytic and thus haploid, allowing the results of genetic lesions to be observed immediately. Physcomitrella has for these reasons recently emerged as an important organism for plant research (8). Here we use gene-targeting methods to mutate each of four canonical phytochromes in the moss Physcomitrella to identify the photoreceptor molecule involved in several directional responses, thereby providing direct evidence for a currently unknown phytochrome signaling system within the cytoplasm.

Fig. 1.

Protonemata of P. patens.(A and B) Wild-type caulonemal filaments growing negatively gravitropically in darkness on vertical agar surface. (C) Light-grown chloronemal filament; note the numerous large green chloroplasts and the transverse cell wall. (D) Positive phototropic bending of tip cell after 3 h R from right. [Bars, 50 μm(A) and 15 μm(B–D).]

Materials and Methods

Plant Material and Cloning of PP1–4. P. patens (Hedw.) wild type, originally collected in Gransden Wood, Huntingdonshire, U.K. (9), was used. The four Physcomitrella phytochrome genes and cDNAs were cloned by using PCR-based methods (see Fig. 7 and Supporting Text, which are published as supporting information on the PNAS web site).

Phylogenetic Analysis of the Phytochrome Gene Products. Predicted amino acid sequences were aligned with clustalx (Institut de Génétique et de Biologie Moléculaire et Cellulaire, Strasbourg, France) and phylogenetic trees derived from sensory, PAS (Period, Amt, Singleminded homology), and kinase domains by using distance-matrix methods with the software package phylip (10).

Gene Targeting and Selection of Targeted Knockouts. The four phytochrome genes, Phypa;PHY;1–4 (PP1–4) were targeted by using P-35S-driven eukaryotic hygromycin (11) or G418 (12) resistance cassettes. PP1 was targeted with a circular insertional knockout construct (7) based on a 1.2-kb HindIII fragment cloned into the unique HindIII site of pGL2 (11); in this case, an additional NcoI site was inserted into the recombinant gene to provide additional proof of targeted knockout in Southern analysis. PP2–4 were targeted with replacement knockout constructs (13) in which the resistance cassette is flanked on both sides by ≈1 kb of gene-specific DNA. For PP2 knockout, a 2.1-kb genomic PCR product obtained with primers 5′-GGGAAGATGTCGACTCCCAAGAAG-3′ (binding despite mismatches) and 5′-GCTGAGCTCATCCATGTC-3′ was cloned into the EcoRV site of pBluescript II (KS^–) (Stratagene). After digestion with EcoRV and deletion of a 196-bp internal EcoRV fragment, a hygromycin resistance cassette obtained by PvuII digestion of pGL2 (11) was inserted in antisense orientation. Transformation was done with plasmid DNA after digestion with XhoI that cuts near the 5′ end of the PP2 sequence in the pBluescript backbone and StuI that cuts in the PP2 sequence near its 3′ end. For PP3 knockout, an ≈5-kb genomic PCR product obtained with primers 5′-GGGCATCTGAAGACACTCG-3′ and 5′-CTCTCCGTCCACACTTCAG-3′ was cloned into pT-Adv (Clontech). A 2.7-kb EcoRI-AflIII restriction fragment was set free, blunted with T4 DNA polymerase, and cloned in SacI-SmaI digested and blunted pNEB193 (NEB, Beverly, MA). A G418 resistance cassette was obtained by EcoRI digestion of pHP23 (12) and blunting with T4 DNA polymerase. This marker was inserted in sense orientation after digestion with SacI and blunting with T4 DNA polymerase. Transformation was done with BsmI-digested plasmid DNA. For PP4 knockout, a 2.2-kb genomic PCR product obtained with primers 5′-GGGATGTCGACCACCAAGTTGGC-3′ and 5′-GTCGCTGTCTCTATCAATCTA-3′ was cloned into the EcoRV site of pBluescript II (KS^–). The resulting plasmid was linearized with EcoRV and a G418 resistance cassette (see above) inserted in sense orientation. The SalI-StuI digested plasmid was used for transformation. The gene-targeting approaches, PCR primers, and product positions are shown in Figs. 8–11, which are published as supporting information on the PNAS web site, and the primer sequences are shown in Supporting Text. After transformation of protoplasts and regeneration on selective medium (14, 15), stable transformants were screened for the expected homologous recombination events via PCR. In each case, three PCRs were carried out, one spanning the targeted region (I) and two between the marker cassette and the chromosomal regions 5′ and 3′ of the transforming constructs (II and III, respectively).

Expression Analyses of Knockouts. For RT-PCR of wild-type and each selected knockout line, 1 μg of total RNA was copied with RNaseH^– Moloney murine leukemia virus reverse transcriptase (BRL) with a lock-docking oligo-d(T) header-primer (5′-GGGCTCGGCCTGACCGGCCTTTTTTTTTTTTTTTTTTV-3′). TaqDNA polymerase and 1/40 of the resulting first-strand cDNA were used for each RT-PCR with oligonucleotides binding in the 5′ and 3′ untranslated regions of PP1–4, respectively: PP1: 5′-TCCTGGAGTTTCGAGATTCTG-3′, 5′-AACCGTGAAAAGGAGTTGAAC-3′; PP2: 5′-TGGAGGTTGGACAAGCACAG-3′, 5′-CAAGGAAAAGATCAGAGTCTA-3′; PP3: 5′-GGGCATCTGAAGACACTCG-3′, 5′-CTCTCCGTCCACACTTCAG-3′; and PP4: 5′-GATTGAGATTAGTTGAGGAGAA-3′, 5′-AAGGGTTACTAGCATAGCTCTC-3′.

Difference Spectroscopy and Western Blotting. Phytochrome protein in wild-type and each knockout line was extracted (16) and analyzed by difference spectroscopy (Kontron, Zurich, Uvikon 930) with actinic light at 660 and 730 nm [red (R) and far-red (FR), respectively] from light-emitting diodes and by Western blots by using affinity-purified polyclonal APC1 antibody (a kind gift of Tilman Lamparter, Freie Universität Berlin) raised against the highly conserved phytochrome N terminus (17).

Northern Blotting. For Northern blotting, ≈500 bp of each phytochrome cDNA, mainly the 3′ UTR, was cloned and amplified by PCR, purified, and α[³²P]ATP-labeled with the Megaprime kit (Amersham Pharmacia). The probes were purified with Nucleotrap push columns (Stratagene). Total RNA was extracted from light-grown filaments or after 4 days of growth in darkness by using the TRIzol method (GIBCO/BRL). Five micrograms of each extract was denatured, separated on 1.2% agarose formaldehyde gels, and blotted onto a nylon membrane. Hybridization was carried out in 10% dextran sulfate (D8906, Sigma)/6× SSC/50% formamide/0.5% SDS/50 mg/ml salmon sperm DNA at 42°C for 12 h. Blots were washed twice for 5 min with 2× SSC/0.1% SDS at 42°C and once for 15 min with 1× SSC, 0.1% SDS at 55°C, then exposed for 8 h to Kodak Biomax film at –80°C.

Phototropism and Polarotropism of Dark-Adapted Filament Tip Cells in Physcomitrella. Physcomitrella protonemata were grown in axenic culture on nutrient agar according to standard procedures (18). Caulonemal filaments were induced by incubation for 16 h in 50 μmol·m^–2·s^–1, photosynthetically active radiation, white light from above with the Petri dishes standing vertically. These were then dark-adapted for 10 days before light treatment. For analysis of phototropism, unilateral collimated monochromatic red light (R) obtained from halogen projectors (Zeutschel, Tübingen-Hirschau, Germany) fitted with interference filters [type DAL, Schott, Mainz, Germany; 661 nm, full width at half height (FWHH) 20 nm] over a wide range of fluence rates was applied for 24 h. In the fluence effect curves, each data point represents at least 150 filaments from five or more independent experiments. The procedure contrasts with that of Jenkins and Cove (19), who used primary chloronemata emerging from spores that had been induced to germinate by white light; we analyzed responses of dark-adapted caulonema tip cells (Fig. 1 A, B, and D; see also ref. 20), allowing phototropic effects to be analyzed without being influenced by a preceding illumination. The gravitropic response also provided a vector against which the phototropic response could be measured. R/FR light reversibility of phototropism was demonstrated with filaments illuminated with either R for 5 min per hour or with R followed by FR (730 nm DAL, FWHH 25 nm) for 5 min each (fluence rates of 1.5 and 4 μmol·m^–2·s^–1, respectively).

For polarotropic analysis, R-emitting diodes (660 nm, full width at half height, 20 nm) at 0.03, 0.5, and 0.8 μmol·m^–2·s^–1) equipped with a polarizing filter (Edmund Industrial Optics, Barrington, NJ) were used. The vibration plane of the electrical vector of the polarized light was oriented at –45° to the axis of the negatively gravitropically grown filaments.

Immediately after the treatments, curvature of individual filament tips was measured by using microvideo imaging systems and imagep (H+H Messsysteme, Berlin) or imagetool (University of Texas Health Science Center, San Antonio) software. The data were analyzed by using excel and, in the case of phototropism, plotted as circular histograms by using a program written by Tilman Lamparter.

Chloroplast Photorelocation. For chloroplast photorelocation (5) in nonpolarized light, R-grown filaments on vertical plates were irradiated laterally in R for 5 days followed by 24hRfrom below (5 μmol·m^–2·s^–1). For analysis of chloroplast movement in polarized light, filaments grown in R (2.5 μmol·m^–2·s^–1) on horizontal agar surfaces in glass cuvettes were irradiated from above for 24 h with polarized R (900 nmol·m^–2·s^–1) with the vibration plane of the electrical vector parallel to the cell axis.

Results

Phototropism of the Physcomitrella Wild-Type. Following Ceratodon (20, 21), we began by quantifying the phototropic response of dark-adapted filament tip cells of wild-type Physcomitrella. Caulonemata growing negatively gravitropically on vertical agar (Fig. 1 A and B) were irradiated with unilateral monochromatic R parallel to the agar surface. In this approach, the light vector is at right angles to that of gravity, providing coordinates for the phototropic response (Figs. 1D and 5A).

Fig. 5.

Quantitative assay of phototropism. (A) Experimental setup. Caulonemata grown vertically on nutrient agar in darkness for 10 days (negative gravitropism). R from right induces a phototropic response in the tip cell. The bending angle, θ, is measured and the data classes shown in angular histograms (inner circle, 0%; outer circle, 50%). (B) Phototropic responses of the wild-type and pp1–4 knockouts in R (661 ± 20 nm) from right. Fluence rate, response curves (mean bending angle) for the wild-type (C) and pp1 and pp2 (D), pp3 (E), and pp4 (F) knockouts (wild-type curves included for comparison). Error bars (±SE) are included for each mean.

Whereas hourly R pulses induced a distinct phototropic response, no reaction was seen if these were followed by monochromatic FR (Fig. 12, which is published as supporting information on the PNAS web site), implying the involvement of phytochrome. Blue light did not result in any phototropic response at the fluence rates tested (data not shown).

We quantified the phototropic response after 24-h R treatment, revealing a complex fluence rate dependence (Fig. 5 B and C). Up to ≈5 nmol·m^–2·s^–1 filaments showed no phototropic response, whereas between 15 and 150 nmol·m^–2·s^–1, most grew negatively phototropically (away from the light source). At ≈500 nmol·m^–2·s^–1, individual tip cells displayed a clear tendency to grow either away from or toward the light, whereas at 1.5 μmol·m^–2·s^–1, almost all were strongly positively phototropic. Above 5 μmol·m^–2·s^–1, however, the filaments bent away from the light.

Thus, caulonemal tip cells of wild-type Physcomitrella display three clearly distinct tropic responses to R: a weak negative phototropism at low fluence rates, a positive phototropism from 0.5 to 5 μmol·m^–2·s^–1, and an avoidance reaction in bright light. Using essentially the same experimental procedure as that followed here, Ceratodon caulonema showed a similar behavior at low fluence rates. Interestingly, however, an avoidance reaction is not seen in Ceratodon (20, 21).

Whether these responses derive from separate photoreceptors and/or from different signaling modes was at this point unclear.

The Phytochrome Gene Family in Physcomitrella. We cloned and sequenced four Physcomitrella phytochrome genes (PP1–4) (for details, see Fig. 7 and Supporting Text). All four as well as the Ceratodon genes Cerpu;PHY;2 and Cerpu;PHY;3 (here CP2 and CP3) encode canonical plant phytochromes lacking the N-terminal extension typical of the PHYB family in higher plants. The moss phytochromes show ≈80% identity and represent a distinct slowly evolving clade (Fig. 2). PP1 and PP3 together with CP2 represent one subgroup, whereas PP2 and PP4 as well as CP3 correspond to another subgroup.

Fig. 2.

Phytochrome phylogeny (phylip Fitch–Margoliash tree based on the N-terminal sensory domain; the numbers at the branches are bootstrap values in percent). Trees from the PAS and kinase modules show the same structure as do analyses using other algorithms.

We and others have failed to detect further aberrant phytochrome-like sequences similar to Cerpu;PHY;1 or Adica;PHY;3 (22, 23) in Physcomitrella or Ceratodon (or indeed higher plants). Thus, although mosses might contain such genes, we considered the possibility that the R/FR reversible phototropic responses in Physcomitrella derive from one or more of the canonical phytochromes we cloned. Although Physcomitrella EST databases contain several phytochrome gene-like sequences additional to those reported here (gc63d12.y1, gb82f07.y1, and BI206149), extensive PCR-based searches failed to correlate any of them with a gene sequence. Hence, we expect PP1–4 to represent the complete phytochrome family of Physcomitrella.

Expression Analysis. Northern blotting showed that all four phytochrome genes are expressed in wild-type Physcomitrella (Fig. 3A). Transcript levels for PP1 and PP3 are lower in light than in darkness, a pattern resembling the down-regulation of higher plant phytochrome A by light and the decrease of phytochrome levels after R irradiation in Ceratodon (17, 24). Unusually, PP2 and PP4 show the opposite pattern, their transcript levels being higher in light-grown filaments. The gene products are too similar to allow differential detection in Western blots using polyclonal antibodies (Fig. 3B).

Fig. 3.

Expression of phytochrome genes. (A) Transcript analysis. Gel (Upper, loading control) and Northern blot hybridized with 3′ UTR probes specific for PP1–4 (Lower). Extracts from dark- (D) and light-grown (L) filaments. (B) Western blot of extracts from dark-grown wild-type and pp1–4 knockouts probed with polyclonal antibody raised against the conserved sensory domain. (C) R/FR difference spectrum of partially purified phytochrome from dark-grown wild-type filaments. ΔΔA × 10^–4. Arrows, wavelengths (nm) of maxima and isosbestic points. Difference spectra from each knockout were indistinguishable from that of wild-type.

Knockout of PP1–4. The function of a specific gene can often be deduced if that locus is mutated; ideally, it is targeted directly by homologous recombination. Up to now, however, Physcomitrella is the only plant in which this elegant method is readily available (7, 13). Thus we attempted to disrupt each PHY gene via homologous recombination. Despite the high sequence similarities of PP1–4, PCR tests showed that we were successful in accurately targeting each gene in ≈30–95% of 20–100 transgenic lines screened. In each case, one knockout line was selected for detailed analysis; additional knockouts each were used to confirm selected aspects of these results. Detailed PCR and Southern analyses (Figs. 4A and 8–11) showed that in each case, the targeted locus had been disrupted. RT-PCR showed that all four PHY transcripts are present in the wild-type, whereas only that of the inactivated gene was missing in the knockout lines (Fig. 4B). Precise gene targeting was thereby demonstrated unequivocally.

Fig. 4.

Targeted knockout of four phytochrome genes. (A) PCRs of genomic DNA from wild-type and pp1–4 knockouts. I, product spanning the disruption site; II, product from untargeted 5′ region to insertion cassette; III, product from insertion cassette to untargeted 3′ region; M, marker. In all cases except pp3, the gene is disrupted by concatemers of the mutating DNA. (B) RT-PCRs of RNA from knockouts pp1–4 and wild-type, products spanning entire message from 5′ to 3′ untranslated regions of PP1–4 transcription products. In each case, only the transcript of the targeted gene is missing.

According to both Western analysis and difference spectroscopy (Fig. 3 B and C), total phytochrome levels in each knockout were similar to that in wild-type. These data suggest that, despite the differences in transcript levels in light and darkness (Fig. 3A), none of the four phytochromes is quantitatively predominant in dark-grown filaments. The isosbestic point and difference maximas are close to those of higher plant phytochromes.

Phototropism in the Knockouts. The phototropic responses of knockout lines were then measured quantitatively to correlate each gene with a potential function in direction sensing (Fig. 5 B and D–F). Whereas all four knockout lines behaved indistinguishably from the wild-type below 15 nmol·m^–2·s^–1, each showed a distinct phenotype at higher fluence rates. pp1 and pp2 showed similar changes in the fluence-response curves (Fig. 5D), the positive phototropic response seeming to be slightly depressed in both cases. In the case of pp1, this was confirmed by a complete fluence-rate response analysis for an additional knockout line (data not shown). pp3 showed a very different phenotype in which the avoidance response at high fluence rates was weakened dramatically (Fig. 5E); indeed, even at the highest fluence rate tested, the phototropism was still predominantly positive. The strongest phenotype, however, was shown by the pp4 knockouts (Fig. 5F and Table 1). Again, the negative bending in dim light was similar to the wild-type, but at higher fluence rates the positive phototropic response was missing. Thus each phytochrome has a specific role in Physcomitrella phototropism.

Table 1. Phototropic response of wild-type and different phy4 lines at 1.5 μmol·m^-2·s^–1 (24-h irradiation).

Line	Mean bending ± SE of tip cell, °	n
Wild-type DC (standard wt)	45.0 ± 1.3	517
phy4 (standard line; F161DC)	-0.9 ± 2.2	217
phy4 (F161.19)	1.0 ± 1.4	76
phy4 (F161.21)	2.3 ± 1.5	75

Polarotropic Response. According to the dichroism model for phytochrome regulation of phototropism (see Discussion), the genetic lesions that led to the phenotypes seen in Fig. 5 should also manifest themselves as defects in phytochrome-mediated polarotropism. Wild-type Physcomitrella shows pronounced R/FR-reversible polarotropism at 500 nmol·m^–2·s^–1 (19). Even at 30 nmol·m^–2·s^–1 of polarized R (Table 2), we observed a strong polarotropic response in wild-type tip cells. When the vibration plane of the electrical vector was oriented at –45° to the cell axis, the tip cells showed an average bending of +12.7° after 24-h irradiation at this fluence rate. A separate wild-type line (wt TG, kindly provided by Thomas Girke, University of Hamburg, Hamburg) responded similarly. Whereas under these conditions the pp1–3 knockouts also resembled the wild-type with mean bending angles between +9.1 and +11.9°, polarotropism in the pp4 knockout lines was lost, the average bending angle for different pp4 lines not differing significantly from the controls in darkness. The pp4 knockout does show polarotropism at higher fluence rates, however (Table 2).

Table 2. Polarotropic bending of Physcomitrella wild-type and phytochrome knockout lines.

Line	Mean bending ± SE of tip cell, °	n
At 30 nmol·m-2·s-1
Wild-type DC (standard wt)	12.7 ± 1.6	118
Wild-type TG	10.6 ± 2.8	15
phy1	9.1 ± 1.2	82
phy2	11.7 ± 2.7	40
phy3	11.9 ± 2.2	50
phy4 (standard line; F161DC)	2.9 ± 1.4	13
phy4 (F161.7)	2.2 ± 0.9	48
phy4 (F161.18)	0.0 ± 1.3	38
phy4 (F161.19)	0.5 ± 0.8	75
phy4 (F161.26)	3.1 ± 1.5	38
Wild-type DC dark control	2.2 ± 1.0	39
At 800 nmol·m-2·s-1
Wild-type DC	24.3 ± 1.5	15
phy4 (F161DC)	19.1 ± 1.7	40

The vibration plane of the electrical vector was -45°, based on the cell axis of the filaments after negatively gravitropical growth. Polarotropic bending of the tip cells after 24-h irradiation.

Chloroplast Photorelocation. An additional phenomenon induced by polarized R in mosses is chloroplast movement. Chloroplasts tend to accumulate at the cross walls or along the side walls in the center of the filament cells after 24-h irradiation in polarized R with the electrical vector perpendicular or parallel to the protonemal axis, respectively (5). We repeated this work using polarized R at 900 nmol·m^–2·s^–1. In our hands, the response was slow, with no apparent effect on chloroplast position even after 3-h irradiation (data not shown). After 24-h irradiation, however (Fig. 6 A–D), in not only the wild-type but also the pp1–3 knockouts, numerous intercalary cells showed the expected relocation response. No relocation was seen in different pp4 knockout lines, however. We also confirmed that relocation can be induced by directional nonpolarized light (5); however, here we found no differences between any of the four knockouts and the wild-type (data not shown), implying that no single phytochrome is responsible for this response.

Fig. 6.

Responses to polarized light. Chloroplast photorelocation of R-grown wild-type and pp1–4 filaments after 24-h irradiation with polarized R. ↔, E vector. (Bars, 10 μm.)

Discussion

In this work, we have mapped a number of photophysiological responses of P. patens in which the direction or polarization of the light stimulus is reflected in vectorial responses. In the wild-type, the phototropic response shows a complex fluence rate-dependent behavior in monochromatic R (Fig. 5 A–C). Polarotropism is also seen (Table 2), as are chloroplast relocation responses in polarized (Fig. 6) and nonpolarized R. On account of their R/FR reversibility, these responses are likely to be mediated by phytochrome. We expect to have cloned the complete Physcomitrella PHY gene family, PP1–4. Although each apparently encodes a canonical plant phytochrome, the family is nevertheless cladistically distinct from higher plant phytochromes, a fact reflected in intron structure (Fig. 7) and differential transcript regulation in light and dark (Fig. 3A). Crude extracts of dark-adapted Physcomitrella show R/FR difference spectra (Fig. 3C) similar to those from higher plants, however, implying phytochromobilin (PΦB) chromophores in similar molecular environments. We used the high rates of homologous recombination available in Physcomitrella to target individual PHY genes, generating numerous knockout lines in each case (Figs. 4 and 8–11). Each knockout showed a specific defect in the phototropic response, pp4 in particular showing no significant positive phototropism (Fig. 5 B and D–F and Table 1). Whereas the other knockouts did not differ significantly from the wild-type in polarotropism and chloroplast relocation, the pp4 lines showed physiological defects in both cases (Table 2 and Fig. 6). As discussed below, it is not at all clear how phytochrome can mediate these vectorial responses; thus, we hope in this work to have created tools for uncovering the underlying molecular physiology.

In vitro phytochromes are soluble dimeric biliproteins with a characteristic photochromicity. The ground state (Pr) absorbs R (λ_max ≈660 nm) and is thereby photoconverted to the thermodynamically stable Pfr form. This physiologically active state absorbs in the FR region (λ_max ≈730 nm) and is thereby photoconverted back to Pr. Because photoconversion is rather efficient (ε_max ≈100 mM^–1·cm^–1, quantum efficiency ≈0.16), an irradiance-independent steady-state level of Pfr is reached rapidly even at low irradiances. Many photoresponses in plants are mediated by phytochrome regulation of transcription. GFP fusions imply that higher-plant phytochromes are freely cytosolic as Pr but become sequestered upon photoconversion to Pfr, which then moves to the nucleus (25) where it interacts with transcriptional factors such as PIF3 (26). Here, the location of phytochrome in the cytoplasm is of no particular importance; similarly, responses based on transcriptional regulation carry no directional information.

Whereas it is clear that phytochrome regulates transcription, it has long been known that many responses both in lower and higher plants (reviewed in ref. 2) are too fast to be explained in this way. It has, for example recently been shown that cytoplasmic motility in the higher plant Vallisneria is stimulated by Pfr formation within 2.5 sec of irradiation (27). R/FR reversible changes in the surface potential of Avena or Hordeum coleoptiles (28, 29) can be detected in <1 min, whereas the transmembrane potential in Salmanea pulvini cells responds to phytochrome within 2 min (30). The phototropic response of moss filaments is clearly initiated within 10 min of irradiation (3).

Because R/FR photochromicity is characteristic for phytochrome, directional responses like those studied in the current work have been explained via dichroism models (31–33). Here, Pr molecules are fixed anisotropically at the plasma membrane and change the orientation of their chromophores upon photoconversion to Pfr, a situation that would allow Pfr to appear locally. This could then interact more or less directly with the cytoskeleton, leading to chloroplast movement or changes in growth direction closely correlated with the direction or polarization of the incident light. For want of an alternative explanation, a 2-fold paradox thereby emerges: (i) freely mobile cytosolic phytochrome molecules are bound anisotropically to a structure close to the plasma membrane, and (ii) Pfr moves into the nucleus to act as a transcriptional regulator while orchestrating directional photoresponses within the cytoplasm.

We note that it is possible that all plant phytochromes have physiological roles in the cytoplasm compatible with the dichroism model. The complex tissues of higher plants scatter light strongly, precluding most photobiological experiments which might demonstrate this role, even if an appropriate response were to exist. In fact, the broader literature indicates rather clearly that phytochromes interact with cytoplasmic components in higher plants. In addition to nuclear-localized transcription factors, yeast two-hybrid and other studies have shown that higher plant phytochromes bind to cytosolically localized PKS1 (34) and to NDPK2, which is present in both nucleus and cytosol (35). The significance of sequestered Pfr “speckles” in the cytoplasm and, subsequently, the nucleus is still unknown; we note, however, that a significant proportion of phyB Pfr does not show this behavior, remaining in the cytoplasm even in continuous R. Earlier, Neuhaus et al. (36) and Bowler et al. (37) used microinjection techniques in tomato aurea mutants, concluding that phytochrome A utilizes both G protein (but see ref. 38) and Ca/CAM signaling systems, both of which are restricted to the cytoplasm. In the context of lower plants, an association between phytochrome and actin microfilaments is likely (3). Thus, despite the fashionable view of phytochrome as a regulator of transcription, the broad data imply rather than exclude a physiological role for phytochrome in the cytoplasm of both higher and lower plants.

From the current work, it is clear that pp4, a canonical phytochrome, plays a predominant role in cytoplasmic perception/responses in Physcomitrella. This is intriguing with respect to recent findings in the fern Adiantum (39). Here an unusual phytochrome-related protein, Adica;PHY;3, a phytochrome–phototropin chimera, mediates R-induced phototropism and chloroplast photorelocation. Phototropins are soluble but are thought to interact with membrane-associated components of the signaling system (40), consistent with the dichroism model for detecting light direction. However, we and others have failed to find an Adica;PHY;3 homolog in mosses. It is in any case clear from our data that canonical phytochromes are necessary and sufficient as photoreceptors to mediate phototropic and polarotropic responses in Physcomitrella.

Here we have shown that Physcomitrella displays physiological responses deriving from a currently unknown cytological phytochrome signaling system particularly but not exclusively associated with pp4. Using methods for revealing protein–protein interactions together with gene targeting, it should now be possible to piece together the signal transduction system and thereby gain further insight into this neglected but perhaps universal aspect of phytochrome biology.