Identification of CDF1 as an Interacting Partner of pPORA During Plastid Import.

We have previously shown that pPORA interacts with at least 10 different proteins during its Pchlide-dependent import (15). Here, we sequenced additional proteins that were present in terms of import intermediate-associated proteins (IAPs) formed with the bacterially expressed, hexa-histidine (His₆)-tagged pPORA in isolated barley and Arabidopsis chloroplasts at 0.1 mM Mg-ATP and 0.1 mM Mg-GTP (15). At these nucleoside triphosphate concentrations, pPORA inserts across the outer plastid envelope membrane and gets in touch also with components of the inner plastid envelope membrane (15). When the protein mixture that was obtained after detergent solubilization of the IAPs and subsequent Ni-NTA agarose chromatography was subjected to 2D-SDS/PAGE, basically the same pattern was obtained as reported previously (Fig. S1A) (15). Protein sequencing of a ∼30-kDa spot revealed that this spot represents a previously uncharacterized component (Fig. S2).

The ∼30-kDa protein interacting with the pPORA is most likely identical to a previously identified protein termed CDF1 (19, 20). This protein is encoded by At5g23040 in Arabidopsis thaliana. Two more possible CDF1 homologs have been identified in the Arabidopsis genome: At3g51140 and At2g20920 (19, 20). At5g23040 encodes a polypeptide of 258 amino acids and has an estimated molecular mass of 28.8 kDa. The CDF1 protein displays sequence homology to the proteins encoded by At3g51140 (32.9%) and At2g20920 (22.4%). CDF1-related genes are present in plants and mosses, and a counterpart was found in Synechocystis, named slr1918 (23.5%). CDF1 proteins have no reported homology with any known animal protein, except for the presence of a J-like domain characteristic of DnaJ/Hsp40-type cochaperones (21, 22). Hydrophobicity analyses revealed that CDF1 and the protein encoded by At3g51140 contain three trans-membrane domains, whereas other CDF1-related proteins, including Synechocystis slr1918 and At2g20920, possess four trans-membrane domains (19, 20, 23).

An antibody was raised against the ∼30-kDa IAP coisolated with pPORA. This antibody recognized a single protein spot on 2D-SDS/PAGE (Fig. S1B). Consistent with structure predictions made using different algorithms, CDF1 was localized to chloroplasts (Fig. S1C). Fractionation experiments led to somewhat surprising results because CDF1 was detected in inner envelopes, thylakoids, and the plastid stroma. Western blot analyses with antibodies against respective marker proteins revealed that these CDF1 signals were not due to cross-contaminations (Fig. S1C). CDF1 was partially released from isolated inner envelope membranes by extraction with either 1 M NaCl or 0.1 M Na₂CO₃, pH 11 (Fig. S1 D, a). These results demonstrated that CDF1 behaves, in part, as a peripheral and, in part, as an integral membrane protein of the inner plastid envelope membrane. By contrast CDF1 represents a classical intrinsic membrane protein in the thylakoids of the chloroplast (Fig. S1 D, b). CDF1 is a constitutively expressed protein that did not undergo detectable changes in abundance during greening of etiolated seedlings (Fig. S1E).

POR Expression in CDF1-Depleted Plants.

Attempts to isolate homozygous knockout plants from public mutant collections (GK-852F12-025757; GABI-Kat) failed because of the lethality of the mutation (20, 23). We therefore examined in vivo effects of CDF1 deficiency in Arabidopsis by using the dexamethasone (DEX)-inducible RNA interference (RNAi) approach developed by Lee et al. (23). DEX-inducible AtCdf1 RNAi Arabidopsis plants were cultivated on agar plates supplemented with DEX added right from the beginning of seed germination and kept either in continuous white light (photomorphogenesis) or darkness (skotomorphogenesis) (Fig. 1A). In a parallel experiment, we used 4.5-d-old etiolated seedlings that had been sprayed with DEX 12 h before seedling harvest—that is, at day 4—and then kept for another 12 h in darkness before being exposed to white light (Fig. 1B). Protein extracts were prepared, separated by SDS/PAGE, and blotted onto nitrocellulose membrane. Western blot analyses then were carried out with antibodies against POR. Fig. 1 depicts the results of these different experiments. When DEX-induced AtCdf1 RNAi seedlings were cultivated in continuous white light, provoking photomorphogenesis, PORB was the only detectable POR protein species (Fig. 1A). This result confirmed previous expression data demonstrating a lack of PORA in illuminated plants (summarized in ref. 6). On the other hand, the seemingly unaffected accumulation of PORB under photomorphogenetic growth conditions at first glance seemed to disprove a role of AtCDF1 for PORB protein accumulation. When PORB protein levels were analyzed for DEX-induced AtCdf1 RNAi plants that had been kept under skotomorphogenetic conditions, however, such an effect became apparent (Fig. 1A). DEX-induced AtCdf1 RNAi seedlings grown in the dark contained drastically reduced amounts of PORB as well as PORA, both of which are normally present in etiolated seedlings (Fig. 1A).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

POR protein levels in CDF1-depleted seedlings during skotomorphogenesis and photomorphogenesis (A) and during greening (B). DEX-inducible RNAi lines were germinated either in the presence of DEX in the dark or in white light for variable periods (in days) (A) or first grown in darkness for 4 d in the absence of DEX, briefly sprayed with a DEX solution, and transferred back to darkness for another 12 h, before being exposed to white light (in hours) (B). Western blot analyses then were carried out with POR antiserum. The positions of the PORA and PORB are indicated.

When similar experiments were carried out with 4.5-d-old etiolated AtCdf1 RNAi seedlings that had been sprayed with DEX 12 h before seedling harvest—that is, at day 4—and then kept for another 12 h in darkness before being exposed to white light (Fig. 1B), a more refined picture appeared. In such seedlings, the light-induced decline in PORA protein levels was enhanced by DEX, leading to CDF1 depletion, compared with the mock-incubated (ethanol-treated) control (Fig. 1B). By contrast, PORB protein levels seemed to be unaffected by DEX-induced AtCDF1 depletion (Fig. 1B). Together, these results suggested (i) an effect of CDF1 on overall POR accumulation in seedlings germinating in darkness (skotomorphogenesis) and (ii) a more specific effect on PORA in seedlings germinating in white light (photomorphogenesis) (Fig. 1A). Furthermore, silencing of AtCDF1 by DEX spraying 12 h before seedling harvest accelerated the decrease in PORA abundance in etiolated seedlings during greening, compared with the non–DEX-treated control (Fig. 1B). These changes were not due to respective decreases in PORA and PORB transcript abundances, as demonstrated by Northern blot analyses (Fig. S3).

In Vitro Import of pPORA and pPORB Into Etioplasts of AtCDF1-Deficient Plants.

Kim and Apel (24) have elegantly shown that the Pchlide-dependent import of the pPORA is operative in Arabidopsis. It is, however, confined to etiolated seedlings. Import of a pPORA-GFP used as a reporter into chloroplasts of light-grown plants was substrate-independent (24). We provided evidence that this Pchlide-independent pathway involved cytosolic 14:3:3 proteins that bound to the mature region of the pPORA and guided the precursor from the Pchlide-dependent import site to the general import site and that this default pathway was not operational in etiolated seedlings (25).

Taking into account these previous results, in vitro import experiments were carried out using isolated Arabidopsis etioplasts and [³⁵S]methionine-labeled pPORA and pPORB. Etioplasts were isolated from 4.5-d-old dark-grown plants that had been pretreated with DEX 12 h before seedling harvest and thus did not express any CDF1 protein (23) (Fig. S3). Henceforth, these plants are designated Atcdf1. As a control, etioplasts were isolated from 4.5-d-old wild-type (WT) plants. Fig. 2 shows that already at time point 0 there was some artificial processing of the ³⁵S-pPORA and ³⁵S-pPORB in the wheat germ extract used to produce the pPORA and pPORB by coupled transcription/translation of respective cDNA clones. When the levels of the precursor and mature PORs were compared, a reduction of the former and simultaneous increase of the latter was observed for etioplasts from WT plants. By contrast, the precursor levels stayed constant for pPORA in the case of etioplasts that had been isolated from AtCDF1-deficient plants. In fact, no mature PORA was detectable in AtCDF1-deficient plants, indicating that pPORA was not imported (Fig. 2A). For the pPORB, a different result was obtained. Whereas the level of the precursor declined, which is suggestive of import, no mature PORB was found in etioplasts of Atcdf1 plants (Fig. 2B). At first glance, this result suggested that CDF1 is also needed for import of pPORB. It could not be excluded, however, that the imported pPORB underwent proteolytic degradation in Atcdf1 etioplasts because of the lack of its interacting partner PORA.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

In vitro import of [³⁵S]methionine-labeled pPORA (A) and pPORB (B) into etioplasts from Atcdf1 and WT seedlings. Etioplasts were isolated from 4.5-d-old etiolated WT seedlings and 4.5-d-old etiolated Atcdf1 that had been sprayed with DEX 12 h before seedling harvest. In vitro import reactions were conducted for 15 min in darkness, and the amounts of the ³⁵S-pPORA and ³⁵S-pPORB and their processed, mature forms were determined by SDS/PAGE and autoradiography. Positions of precursors and mature proteins are indicated.

Low-Temperature Pigment Fluorescence Measurements.

Two different spectral forms of Pchlide are detectable in the membranes of etioplasts by fluorescence spectroscopy at 77 K (26). The predominating species are termed photoactive Pchlide (Pchlide-F655) and photoinactive Pchlide (Pchlide-F635), respectively (26). Photoactive Pchlide can be converted into Chlide by a 1-ms flash of white light, whereas photoinactive Pchlide cannot. Accumulation of photoactive Pchlide is a reflection of the presence of the PORA and PORB and their assembly into larger light-harvesting structures, whereas the peak named photoinactive Pchlide represents a mixture of free pigments and unassembled POR–Pchlide–NADPH ternary complexes (6, 26).

To trace Pchlide-F655 and Pchlide-F635, we carried out low-temperature pigment fluorescence analyses at 77 K on etiolated Atcdf1 plants. When the low-temperature spectra were compared for WT and Atcdf1 seedlings, they were fundamentally different. Atcdf1 plants were devoid of Pchlide-F655 but contained large amounts of Pchlide-F635 (Fig. 3A). WT plants did contain both spectral Pchlide forms of which Pchlide-F655 was photoconvertable (Fig. 3A and Table S1). Pigment extractions with acetone and subsequent pigment quantification revealed a large, ∼fivefold excess of total Pchlide in Atcdf1 versus WT plants (Table S1).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

(A) Low-temperature fluorescence analysis of pigments in dark-grown Atcdf1 and WT seedlings. CDF1-depleted seedlings were generated as described in Fig. 2B and used for low-temperature pigment fluorescence emission measurements at an excitation wavelength of 440 nm. For comparison, the low-temperature spectrum is shown for WT seedlings. (B) Cell death in etiolated WT, flu, and Atcdf1 Arabidopsis seedlings during greening. CDF1-depleted seedlings were generated as described in Fig. 2B and exposed to white light for different periods (in hours). Cell viability was assessed by tetrazolium staining. Percentages refer to the total number of viable seedlings at the beginning of illumination, set as 100. For comparison, cell death was scored for etiolated WT and flu seedlings.

Light-Triggered Cell Death in Atcdf1 Seedlings During Greening.

The large amount of free, non–POR-bound Pchlide in Atcdf1 seedlings suggested that the porphyrin pigment could act as a photosensitizer and trigger the production of singlet oxygen once the seedlings were illuminated. Kim and Apel (27) have shown that singlet oxygen operates both as a powerful signaling compound and cytotoxin and that both effects contribute to cell death execution.

Tetrazolium staining (28) was used to score cell death in Atcdf1 versus WT seedlings. Once again, the same type of etiolated, Atcdf1 seedlings as before was used. Such seedlings were pretreated with DEX 12 h before seedling harvest and then exposed to continuous white light of 125 µmol photons m⁻²·s⁻¹ for different periods. Cell viability was scored by counting the number of seedlings that were no longer able to retain the tetrazolium dye. For comparison, the fluorescent mutant of Arabidopsis, flu (5), was analyzed. Fig. 3B shows the respective cell death curve for Atcdf1, flu, and WT seedlings. Obviously, cell death in Atcdf1 plants (t_1/2 = 2 h) was much faster than that in flu (t_1/2 = 6 h) seedlings. For the WT seedlings, no signs of cell death were found under the tested conditions (Fig. 3B).

Singlet Oxygen Production in Atcdf1 Plants.

The cell death phenotype in Atcdf1 seedlings after irradiation suggested that free Pchlide molecules not bound to POR operated as a photosensitizer and caused the production of singlet oxygen. To test the involvement of singlet oxygen in light-triggered cell death in Atcdf1 seedlings, singlet oxygen measurements were carried out with the DanePy reagent. DanePy is a dansyl-based singlet oxygen sensor undergoing quenching of its fluorescence upon reacting with singlet oxygen (29, 30). DanePy has a broad emission peak at around 530 nm. Upon reacting with singlet oxygen, this peak is reduced, with the amount of fluorescence drop reflecting the amount of singlet oxygen produced (29, 30).

To test singlet oxygen evolution, 4.5-d-old dark-grown plants that had been pretreated with DEX 12 h before seedling harvest and thus did not express any CDF1 protein were used. In three replicate experiments, batches of 120 seedlings each were exposed to white light of 125 µE m⁻²·s⁻¹ for 30 min. As a reference, 4.5-d-old etiolated WT and flu seedlings were analyzed in parallel. Fig. 4 displays DanePy fluorescence emission spectra of Atcdf1 and WT seedlings. These spectra revealed the accumulation of singlet oxygen in Atcdf1 plants but not in WT seedlings, as evidenced by the quenching of DanePy fluorescence that was collected between 425 and 600 nm. Kinetic measurements (Fig. 4, Inset) additionally showed that Atcdf1 plants produced approximately twofold higher levels of singlet oxygen than irradiated flu seedlings. In WT seedlings, there was very little, if any, DanePy fluorescence quenching indicative of the generation of singlet oxygen (Fig. 4).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Singlet oxygen production in WT, flu, and Atcdf1 seedlings during greening. The same type of CDF1-deficient Arabidopsis plants as those used in Fig. 2B was infiltrated with DanePy and irradiated for 30 min with white light. At different time intervals, DanePy fluorescence emission was collected between 425 and 600 nm. Inset shows kinetics of the drop in DanePy fluorescence indicative of singlet oxygen production during greening.

Interaction of CDF1 with pPORA During Plastid Import.

The fact that CDF1 does not operate in either Pchlide or NADPH binding to PORA (Figs. S4 and S5; Table S2) suggests that it may play a rather passive role as a “holdase.” Holdases are a particular kind of molecular chaperone that assists the noncovalent folding of proteins in an ATP-independent manner (31). Examples of holdases are DnaJ and Hsp33. Holdases bind to protein-folding intermediates to prevent their aggregation but without directly refolding them (31). Consistent with the presence of a J-like domain in CDF1’s predicted primary amino acid sequence (23), such holdase activity could permit the binding of Pchlide to pPORA’s transit sequence, thereby triggering the import step (32, 33). If so, stable pools of CDF1-bound (p)PORA–Pchlide–NADPH ternary complexes should accumulate in the stroma.

To explore this possibility, time course import experiments were conducted using isolated WT Arabidopsis chloroplasts. To exclude any mistargeting of the pPORA to the general import pathway, a fusion protein was used in these studies that consisted of the transit sequence of pPORA (transA) and a cytosolic dihydrofolate reductase (DHFR) reporter protein of mouse (32, 33). As shown previously, this chimeric precursor is imported into the plastids in a Pchlide-dependent manner and engages the PTC complex (16, 18).

The [³⁵S]methionine-labeled transA-DHFR was bound to the plastids at 0.1 mM Mg-ATP and 0.1 mM Mg-GTP during preincubation; then, the plastids were reisolated and supplemented with 2.5 mM Mg-ATP and 5-ALA to (i) drive intraplastidic Pchlide synthesis and (ii) power molecular chaperones operating as import motors. After different time intervals, stromal transA-DHFR complexes were precipitated with an antiserum against the DHFR or coimmunoprecipitated with the antiserum against CDF1 and the pulled-down complexes were analyzed by SDS/PAGE and autoradiography.

As shown in Fig. 5 A–C, significant amounts of ³⁵S-transA-DHFR were coprecipitated by the CDF1 antiserum. This stromal pool of the ³⁵S-precursor also contained Pchlide, as shown in Fig. S6. Interestingly, an antiserum against chloroplast chaperonin 60 (CPN60) also pulled down ³⁵S-transA-DHFR. However, the time courses of interaction between ³⁵S-transA-DHFR and CDF1 versus CPN60 were clearly distinct. Only the unprocessed ³⁵S-transA-DHFR interacted with CDF1 (Fig. 5 C, b). By contrast, the processed DHFR reporter interacted only with CPN60 (Fig. 5 C, c), indicating that both chaperone systems (CDF1 and CPN60) act sequentially in import of the used model precursor.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

CDF1 binding to unprocessed ³⁵S-transA-DHFR in the chloroplast stroma. (A–C) Import of ³⁵S-transA-DHFR into isolated Arabidopsis chloroplasts. To allow for the detection of unprocessed ³⁵S-transA-DHFR, the incubations were carried out at 15 °C. At each time point indicated, aliquots were withdrawn from total assays (A) and stromal extracts that had been prepared from lysed chloroplasts (B), separated by SDS/PAGE and detected by autoradiography. (C) Stromal extracts were subjected to coimmunoprecpitations with antisera against the DHFR (a), CDF1 (b), or CPN60 (c). After SDS/PAGE, the gels were blotted and the resulting membranes probed with anti-CPN60 antiserum using an ECL system. (D) Mtx-dependent release of ³⁵S-transA-DHFR from Atcdf1 chloroplasts tested at different Mtx concentrations. Inset depicts time courses of ³⁵S-transA-DHFR released from Atcdf1 chloroplasts analyzed by Western blotting. (Upper) Mock incubations were conducted for the indicated periods (in min) in the absence of Mtx. (Middle and Lower) ³⁵S-transA-DHFR was detected in the supernatant (Sup) and chloroplast (CP) fractions, respectively, obtained after subjecting the assays to centrifugation. (E) Scatchard plot of the data shown in D.

As a final experiment, chloroplasts from CDF1-depleted (Atcdf1) plants were allowed to bind transA-DHFR during a preincubation at 0.1 mM Mg-ATP and 0.1 mM Mg-GTP. Then, the plastids were reisolated on Percoll, resuspended in import buffer, and fed 2.5 mM Mg-ATP and 5-ALA to trigger Pchlide synthesis and precursor translocation. At the same time, methotrexate (Mtx) was added to provoke the release of ³⁵S-transA-DHFR from the plastids to the incubation mixture. Mtx is a folate analog that tightly binds to the DHFR (34). Immediately after adding Mtx, the plastids were sedimented by centrifugation and the amount of plastid-bound and free ³⁵S-transA-DHFR was determined by SDS/PAGE and autoradiography as well as liquid scintillation counting.

Fig. 5 D and E shows that no import of ³⁵S-transA-DHFR was detectable for plastids of Atcdf1 plants. The ³⁵S-transA-DHFR protein levels remained constant in plastids, and no mature DHFR appeared (Fig. 5 D, Upper). When Mtx was added, however, ³⁵S-transA-DHFR was quantitatively released into the supernatant obtained after centrifugation of the assays (Fig. 5 D, Middle and Lower). At saturating Mtx concentrations, ∼400 ³⁵S-transA-DHFR molecules were released per chloroplast (Fig. 5E). This number was identical to the number of binding sites determined previously for the pPORA and Pchlide-free chloroplasts (17). Scatchard analysis (17) allowed calculating the K_d for ³⁵S-transA-DHFR’s binding to the plastid envelope. This K_d was 8.5 mM and thus laid in a range identical to the K_d reported previously for the binding of the pPORA to Pchlide-free chloroplasts (17). We concluded that CDF1 does not participate in the binding of transA-DHFR and pPORA to the chloroplast but is involved in the subsequent translocation of either precursor into the stroma. Hereby, CDF1 operated most likely as a holdase, permitting Pchlide binding to pPORA’s transit sequence and blocking movement of the translocating polypeptide chain back to the cytosol.