Versatile roles of Arabidopsis plastid ribosomal proteins in plant growth and development


  • Isidora Romani,

    1. Dipartimento di Bioscienze, Università degli studi di Milano, I-20133 Milano, Italy
    Search for more papers by this author
    • These authors contributed equally to this work.

    • Present Address: Lehrstuhl für Molekularbiologie der Pflanzen (Botanik), Department Biologie I, Ludwig-Maximilians-Universität München, D-82152 Planegg-Martinsried, Germany.

  • Luca Tadini,

    1. Lehrstuhl für Molekularbiologie der Pflanzen (Botanik), Department Biologie I, Ludwig-Maximilians-Universität München, D-82152 Planegg-Martinsried, Germany
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Fabio Rossi,

    1. Dipartimento di Bioscienze, Università degli studi di Milano, I-20133 Milano, Italy
    Search for more papers by this author
  • Simona Masiero,

    1. Dipartimento di Bioscienze, Università degli studi di Milano, I-20133 Milano, Italy
    Search for more papers by this author
  • Mathias Pribil,

    1. Lehrstuhl für Molekularbiologie der Pflanzen (Botanik), Department Biologie I, Ludwig-Maximilians-Universität München, D-82152 Planegg-Martinsried, Germany
    Search for more papers by this author
  • Peter Jahns,

    1. Plant Biochemistry, Heinrich-Heine-University Düsseldorf, Universitätsstrasse 1, D-40225 Düsseldorf, Germany
    Search for more papers by this author
  • Martin Kater,

    1. Dipartimento di Bioscienze, Università degli studi di Milano, I-20133 Milano, Italy
    Search for more papers by this author
  • Dario Leister,

    Corresponding author
    1. Lehrstuhl für Molekularbiologie der Pflanzen (Botanik), Department Biologie I, Ludwig-Maximilians-Universität München, D-82152 Planegg-Martinsried, Germany
    Search for more papers by this author
  • Paolo Pesaresi

    1. Dipartimento di Bioscienze, Università degli studi di Milano, I-20133 Milano, Italy
    Search for more papers by this author



A lack of individual plastid ribosomal proteins (PRPs) can have diverse phenotypic effects in Arabidopsis thaliana, ranging from embryo lethality to compromised vitality, with the latter being associated with photosynthetic lesions and decreases in the expression of plastid proteins. In this study, reverse genetics was employed to study the function of eight PRPs, five of which (PRPS1, -S20, -L27, -L28 and -L35) have not been functionally characterised before. In the case of PRPS17, only leaky alleles or RNA interference lines had been analysed previously. PRPL1 and PRPL4 have been described as essential for embryo development, but their mutant phenotypes are analysed in detail here. We found that PRPS20, -L1, -L4, -L27 and -L35 are required for basal ribosome activity, which becomes crucial at the globular stage and during the transition from the globular to the heart stage of embryogenesis. Thus, lack of any of these PRPs leads to alterations in cell division patterns, and embryo development ceases prior to the heart stage. PRPL28 is essential at the latest stages of embryo–seedling development, during the greening process. PRPS1, -S17 and -L24 appear not to be required for basal ribosome activity and the organism can complete its entire life cycle in their absence. Interestingly, despite the prokaryotic origin of plastids, the significance of individual PRPs for plant development cannot be predicted from the relative phenotypic severity of the corresponding mutants in prokaryotic systems.


Plant growth and development are controlled by the concerted actions of many signalling pathways, which are triggered by developmental and metabolic cues. Plastids play an important role in plant development. On the one hand, they display a variety of interconvertible differentiated forms which are closely associated with different cell types (Waters and Pyke, 2005; Hsu et al., 2010). In addition, many essential metabolic processes take place in plastids (Neuhaus and Emes, 2000; Yamaguchi and Kamiya, 2000; Seo and Koshiba, 2002; DellaPenna and Pogson, 2006) that also serve as sources of signals to the nucleus (plastid or retrograde signalling) which regulate plastid biogenesis and coordinate cell differentiation and tissue architecture (Lopez-Juez and Pyke, 2005; Lopez-Juez, 2007; Pesaresi et al., 2007; Tejos et al., 2010).

Embryogenesis also depends on plastid function and differentiation, and is usually divided into two phases. During embryo morphogenesis, the basic body plan is established, whereas embryo maturation includes cell growth and expansion together with the accumulation of macromolecules that allow the embryo to withstand the desiccation that accompanies seed formation and enable seedling growth after germination (Goldberg et al., 1994). Embryo morphogenesis begins with a single-celled zygote which, in Arabidopsis thaliana, undergoes a stereotypical series of cell divisions giving rise in turn to pre-globular, globular, heart and torpedo, linear- and bent-cotyledon stages and mature green embryos. Plastids in embryonic cells of A. thaliana remain undifferentiated and non-photosynthetic until the late globular stage, when grana become visible (Hsu et al., 2010; Tejos et al., 2010). Subsequently, the number of chloroplasts increases at the torpedo stage, before entry into the maturation phase, during which chloroplasts contribute to seed metabolism by producing NADPH and ATP for fatty acid biosynthesis, supplying oxygen to otherwise hypoxic seeds and refixing respiratory CO2 via the Calvin cycle (Rolletschek et al., 2002; Ruuska et al., 2004).

The precise roles of proplastids and chloroplasts during embryogenesis are not fully understood. Nevertheless, genetic screens in Arabidopsis have demonstrated that impairment of plastid functions can perturb embryogenesis and even result in embryo lethality (Hsu et al., 2010; Bryant et al., 2011; Muralla et al., 2011; Lloyd and Meinke, 2012). Indeed, among the 400 Arabidopsis genes so far identified as essential for embryo development, about 30% encode chloroplast-localised proteins (see the SeedGenes Project database, Such plastid proteins are frequently involved in metabolite biosynthesis or, like aminoacyl-tRNA synthetases and plastid ribosomal proteins (PRPs), play a role in plastid protein synthesis. Thus, of the nuclear-encoded elements of the plastid ribosome, three components of the small subunit (PRPS5, -S9 and -S13) and eight of the large subunit (PRPL1, -L4, -L6, -L10, -L13, -L18, -L21 and -L31) have been suggested to be essential for embryogenesis (Hsu et al., 2010; Bryant et al., 2011; Muralla et al., 2011; Lloyd and Meinke, 2012; Yin et al., 2012). However, only in five cases (PRPS5, -S13, -L1, -L6 and -L21) has the gene–phenotype relationship been unambiguously confirmed by allelism tests or genetic complementation assays (Bryant et al., 2011; Lloyd and Meinke, 2012; Yin et al., 2012). In the absence of PRPL21, embryo development is arrested at the globular stage (Yin et al., 2012), whereas the loss-of-function phenotypes of PRPS5, -S13, -L1 and -L6 were analysed in the course of large reverse genetics screens without further detailed characterisation. Nevertheless, these four PRPs are annotated in the SeedGenes Project database as essential at the pre-globular or globular stage of embryo development, similar to mutants defective in other aspects of plastid protein synthesis (Berg et al., 2005; Muralla et al., 2011).

Interestingly, not all ribosomal subunits are equally important for embryo development. For instance, Arabidopsis lines without PRPL11 show normal germination rates and are characterised by reduced photosynthetic performance, pale green leaf colour and a drastic reduction in growth rate under greenhouse conditions, in association with diminished levels of protein synthesis in plastids (Pesaresi et al., 2001). Similarly, lack of PRPS21 in the ghs1 (glucose hypersensitive1) mutant increases sensitivity to glucose, together with a reduction in plastid protein synthesis, altered photosynthetic performance and impaired chloroplast development (Morita-Yamamuro et al., 2004). The Arabidopsis ore4-1 mutant was identified on the basis of its extended leaf longevity, and shown to exhibit reduced expression of the PRPS17 gene (Woo et al., 2002). In this mutant, as well as in lines lacking PRPL24 (Tiller et al., 2012), growth rate and leaf pigment content are decreased, as a consequence of altered plastid protein synthesis. Although analysis of the five plant-specific ribosomal proteins PSRP2–6 is hampered by the lack of knock-out lines for PSRP2, -4 and -5, preliminary analyses have indicated that none of the five is essential for embryo development (Tiller et al., 2012).

In the present study, a reverse genetic approach targeted to several PRPs associated with different ribosome domains was employed to further dissect the function of plastid ribosomes during embryogenesis and plant development. Interestingly, although plastids originated from free-living prokaryotes, we found little correlation in phenotypic severity between homologous plastid and prokaryotic mutants.


Isolation of mutants for PRPS and PRPL proteins in Arabidopsis

Arabidopsis lines carrying T-DNA insertions in single-copy nuclear genes coding for a total of nine protein components of the small (PRPS proteins) and large (PRPL proteins) subunits of the plastid ribosome were identified. Of these, five (PRPS1, -S20, -L27, -L28 and -L35) have not been functionally characterised previously in A. thaliana. PRPS17 had been analysed using leaky alleles or RNA interference (RNAi) lines only (Woo et al., 2002; Tiller et al., 2012), while PRPL1 and PRPL4 have been reported to be essential for embryo development (Bryant et al., 2011) but their mutant phenotypes have not been subjected to thorough investigation. Only prpl24-1 has been characterised in detail before (Tiller et al., 2012), and is used here as a control.

One insertion mutant allele only was found for each of three PRPS genes (PRPS1/At5g30510, PRPS17/At1g79850 and PRPS20/At3g15190) and four PRPL genes (PRPL24/At5g54600, PRPL27/At5g40950, PRPL28/At2g33450 and PRPL35/At2g24090) (Figure 1). Two mutant alleles each were obtained for PRPL1/At3g63490 and PRPL4/At1g07320.

Figure 1.

 T-DNA tagging of PRPS and PRPL genes. Exons are indicated as numbered white boxes, introns as black lines. Arrowheads indicate the positions of translation initiation and stop codons. Sites, designations and orientations of T-DNA insertions are indicated (RB, right border; LB, left border). For PRPL1, PRPL4 and PRPL24, the intron–exon structures of the most abundant transcript variants are indicated. The T-DNA insertions are not drawn to scale.

Interestingly, two splicing variants have been reported for PRPL1 and PRPL24, and up to four different transcripts have been predicted for PRPL4 (see Real-time PCR analyses indicate that both PRPL1 splicing variants are present in leaves and developing siliques, with PRPL1.1 being more abundant than PRPL1.2 (Figure S1 in Supporting Information). Conversely, only one transcript variant was detectable for PRPL4 (PRPL4.1) and PRPL24 (PRPL24.1) in both leaves and siliques.

Mutations in PRPS1, PRPS17 and PRPL24 affect growth and photosynthesis

Only in the case of PRPS1 and PRPS17, and the prpl24-1 allele used as a control for lines with a reduced plastid translation efficiency (Tiller et al., 2012), could homozygous mutants be identified by PCR-based genotyping. The T-DNA insertions in the PRPS17 and PRPL24 loci completely suppressed the accumulation of the corresponding transcripts, whereas residual amounts [about 8% of the level in Columbia-0 (Col-0)] of PRPS1 transcripts were detected by Northern analysis in prps1-1 homozygotes (Figure 2). An agarose gel stained with ethidium bromide and showing the different rRNA molecules was used as a loading control, allowing us to discern any effects of loss of PRPL24 or PRPS17 on rRNA accumulation. Indeed, specific reductions in levels of 23S rRNA and 5S-4.5S rRNA were found in prpl24-1 leaves, whereas the amount of 16S rRNA was decreased in prps17-1 leaves (Figure 2).

Figure 2.

 Northern analyses of wild type (Col-0 and Ler) and mutant (prpl24-1, prps1-1 and prps17-1) plants. Total leaf RNA was fractionated on a denaturing agarose gel, transferred to a nylon membrane, and hybridised with the corresponding PRP cDNA probes, reported on the left side of the panel. An agarose gel stained with ethidium bromide and showing the different rRNA molecules (from 25S to 4.5S; Zybailov et al., 2009) was used as a loading control.

All three prp homozygotes were characterised by pale green cotyledons and leaves, and reduction in overall size (Figure 3a), and the mutant phenotype segregated as a single-locus recessive trait. Quantification of growth rates by non-invasive image analyses under optimal growth-chamber conditions showed that the size of 4-week-old prpl24-1 and prps1-1 mutant plants was reduced by about 80% and 60%, respectively, relative to the corresponding wild type (WT) background (Col-0) (Figure 3b). For prps17-1, a reduction of approximately 55% in size in comparison to the WT [Landsberg erecta (Ler)] was measured.

Figure 3.

 Phenotypes of mutant (prpl24-1, prps1-1 and prps17-1) and wild-type (Col-0 and Ler) plants.
(a) The different genotypes were grown for 4 weeks in a growth chamber.
(b) The growth kinetics of the different genotypes was measured from 4 to 28 days after germination (d.a.g.). Each point is based on the determination of mean leaf area from at least 10 individuals ( 10). Bars indicate standard deviations.
(c,d) The photosynthetic parameter FV/FM (c) and ΦII (d) of the different genotypes were measured as described in the Experimental Procedures. Signal intensities for FV/FM and ΦII are indicated according to the colour scale at the bottom of the figure.

Photosynthetic performance was characterised by monitoring chlorophyll (Chl) a fluorescence. The data showed a clear decrease in maximum quantum yield of photosystem II (PSII) (FV/FM, ratio of variable to maximum fluorescence) in prpl24-1 and a somewhat less pronounced effect in prps1-1 and prps17-1 plants (Col-0, 0.83 ± 0.01; Ler, 0.83 ± 0.01; prpl24-1, 0.47 ± 0.03; prps1-1, 0.69 ± 0.01; prps17-1, 0.66 ± 0.01; see also Figure 3c). A similar picture emerged when the effective quantum yield of PSII (ΦII) was taken into account. In this case too, prpl24-1 plants were markedly impaired with respect to Col-0, whereas differences between prps1-1 and prps17-1 and the corresponding WT plants were less marked (Col-0, 0.77 ± 0.01; Ler, 0.77 ± 0.01; prpl24-1, 0.43 ± 0.02; prps1-1, 0.63 ± 0.01; prps17-1, 0.60 ± 0.02; see also Figure 3d). To quantify the alteration in leaf coloration in prpl24-1, prps1-1 and prps17-1 plants, leaf pigments were analysed by HPLC. As expected, mutant plants contained only 58% (prps1-1), 50% (prps17-1) and 68% (prpl24-1) of WT levels of total chlorophyll (Chl b) (Table 1). The Chl a/b ratio was also clearly decreased in prpl24-1, and to a lesser extent in prps1-1 (Table 1), indicating either a higher PSII/PSI ratio or, rather more likely in light of the nature of the mutations, an increase in the size of the Chl b-binding peripheral antenna (which is made of nuclear-encoded subunits) relative to the Chl a-binding reaction centres, synthesis of which requires plastid ribosomes. Interestingly, the Chl a/b ratio was not markedly altered in prps17-1 with respect to Ler plants, suggesting that in this case the reduction in amounts of plastid-encoded reaction centres might be accompanied by a similar decrease in antenna complexes.

Table 1. Levels of leaf pigments in light-adapted mutant (prpl24-1, prps1-1, prps17-1) and wild-type (Col-0 and Ler) plants at the six-leaf rosette stage. Leaf pigments were determined by HPLC and are reported in pmol mg−1 leaf fresh weight. Mean values ± SD are shown
 Leaf pigment content (pmol mg−1 leaf fresh weigh)
NxLutChl bChl aβ-CarVAZChl a + bChl a/b
  1. Nx, neoxanthin; Lut, lutein; Chl b, chlorophyll b; Chl a, chlorophyll a; β-Car, β-carotene; VAZ, violaxanthin + antheraxanthin + zeaxanthin.

Ler36 ± 3117 ± 13249 ± 221007 ± 95100 ± 1051 ± 41251 ± 984.05 ± 0.03
Col-039 ± 4124 ± 11246 ± 23952 ± 76985 ± 843 ± 41198 ± 983.87 ± 0.03
prpl24-1 39 ± 41075 ± 11191 ± 20625 ± 6132 ± 253 ± 5815 ± 793.27 ± 0.11
prps1-1 30 ± 287 ± 9146 ± 10549 ± 5337 ± 344 ± 4694 ± 633.76 ± 0.09
prps17-1 26 ± 277 ± 6123 ± 11497 ± 3934 ± 350 ± 1619 ± 514.04 ± 0.06

All mutant phenotypes could be rescued by Agrobacterium tumefaciens-mediated transformation of homozygous mutants with either the appropriate coding sequence (PRPS1 and PRPS17) fused to the 35S promoter of cauliflower mosaic virus (35S-CaMV), or the genomic sequence including a 1-kbp fragment of the promoter (PRPL24), corroborating a direct correspondence between genotype and phenotype (see Table S1). Thus, when PRPS1 and PRPL24 were introduced into the corresponding mutants, a complete rescue of the mutant phenotype was observed, whereas a marginal reduction of ΦII values and leaf pigment content with respect to WT remained after complementation of prps17-1 lines with the WT PRPS17 gene (Table S1). Taken together, the data indicate that PRPS1, PRPS17 and PRPL24 are required for optimal plastid performance in terms of photosynthesis and growth, but their loss is compatible with plant viability.

Plastid protein synthesis and thylakoid composition are perturbed in prps1-1, prps17-1 and prpl24-1 plants

To determine whether the defect in photosynthetic performance described above is associated with alterations in the protein composition of thylakoids, two-dimensional blue native (2D BN)/SDS-PAGE and one-dimensional (1D) SDS-PAGE were performed on thylakoids and total protein extracts, respectively (Figure 4). Quantification of banding patterns on Coomassie-stained 2D BN/SDS polyacrylamide (PA) gels revealed marked reductions in levels of the plastid-encoded PSII core subunits PsbA, PsbB, PsbC and PsbD in prpl24-1, prps1-1 and prps17-1 thylakoids, with amounts corresponding to 25–27, 51–59 and 40–44%, respectively, of those seen in the WT (Figure 4a, Table 2). Accumulation of PsbB was also quantified by 1D SDS-PAGE and immunoblot analysis, and values comparable to those from stained 2D SDS-PA gels were obtained (Figure 4b, Table 2). Other plastid-encoded subunits of thylakoid multiprotein complexes, including the reaction centre of PSI (PsaA/B), the β-subunit of ATPase (ATPase β), cytochrome b6 (PetB) and the large subunit of Rubisco (RbcL) also showed marked declines in level in mutant plants, particularly in prpl24-1 homozygotes. The reduced accumulation of the PSI reaction centre was accompanied by a decrease in amounts of nuclear-encoded PSI subunits, such as PsaD, PsaF and PsaO, indicating that the entire PSI complex is destabilised as a result of the mutations (Figure 4b, Table 2). Other nuclear-encoded thylakoid proteins, including the PSI antenna protein Lhca2, and Lhcb4 and Lhcb6 (the minor antenna of PSII), behaved like PSI and PSII, while the nuclear-encoded components of LHCII (Lhcb1-3) were least affected (Figure 4a, Table 2), most probably because they can accumulate independently of PSI and PSII (Caffarri et al., 2005).

Figure 4.

 Two-dimensional blue native (BN) SDS-PAGE separation and one-dimensional SDS-PAGE analysis of thylakoid proteins from wild-type (Col-0 and Ler) and mutant (prps1-1, prps17-1 and prpl24-1) leaves.
(a) Thylakoid protein complexes were fractionated by BN-PAGE in the first dimension and then by 15% SDS-PAGE, followed by staining with colloidal Coomassie Blue (G 250). The identity of relevant proteins is indicated by arrows. Note that Ler behaved like Col-0.
(b) Total leaf proteins were fractionated by SDS-PAGE, and blots were probed with antibodies raised against individual subunits of photosystem I (PsaD, PsaF, PsaO), photosystem II (PsbB), LHCI (Lhca2), the minor antenna of photosystem II (Lhcb4, Lhcb6), the chloroplast ATP synthase (β-subunit), the Cyt b6/f complex (PetB) and the large subunit of Rubisco (RbcL). Decreasing levels of wild-type proteins were loaded in the lanes marked 0.5× Col-0 and 0.25× Col-0. Immunodecoration with an actin-specific antibody was employed to control for equal loading.

Table 2. Quantification of thylakoid proteins in light-adapted mutant plants (prps1-1, prps17-1 and prpl24-1). Wild-type levels are set to 100%. Average values were calculated from three independent 2D polyacrylamide gels and protein gel blots (see Figure 4)
Protein prpl24-1 prps1-1 prps17-1
2D PAGEImmunoblot2D PAGEImmunoblot2D PAGEImmunoblot
  1. nd, not determined.

  2. aProteins encoded by plastid genes.

PsaA/Ba0.31 ± 0.03nd0.59 ± 0.05nd0.46 ± 0.04nd
PsaDnd0.32 ± 0.03nd0.44 ± 0.04nd0.39 ± 0.04
PsaFnd0.34 ± 0.03nd0.52 ± 0.04nd0.45 ± 0.03
PsaOnd0.28 ± 0.02nd0.37 ± 0.03nd0.32 ± 0.04
PsbAa0.27 ± 0.02nd0.51 ± 0.03nd0.42 ± 0.02nd
PsbBa0.25 ± 0.020.34 ± 0.030.54 ± 0.040.45 ± 0.040.41 ± 0.040.48 ± 0.04
PsbCa0.25 ± 0.02nd0.55 ± 0.02nd0.40 ± 0.03nd
PsbDa0.27 ± 0.02nd0.59 ± 0.04nd0.44 ± 0.04nd
Lhca2nd0.31 ± 0.03nd0.27 ± 0.02nd0.36 ± 0.02
Lhcb1/b2/b30.67 ± 0.04nd0.74 ± 0.05nd0.53 ± 0.04nd
Lhcb4nd0.32 ± 0.03nd0.45 ± 0.04nd0.38 ± 0.03
Lhcb6nd0.39 ± 0.03nb0.48 ± 0.04nd0.45 ± 0.04
ATPase βa0.29 ± 0.030.26 ± 0.030.51 ± 0.050.46 ± 0.040.42 ± 0.030.45 ± 0.04
PetBand0.18 ± 0.02nd0.79 ± 0.05nd0.39 ± 0.03
RbcLand0.19 ± 0.02nd0.27 ± 0.03nd0.21 ± 0.02

To study whether the putative defects in plastid protein synthesis which, in the case of prps17-1 and prpl24-1, are already suggested by the changes in rRNA accumulation (see Figure 2), might be mitigated by adaptive mechanisms at the transcriptional and/or post-transcriptional level, the effects of prps1-1, prps17-1 and prpl24-1 mutations on accumulation and processing of plastid transcripts were studied by Northern analysis (Figure 5a). The steady-state level of the psbA transcript was increased by a factor of almost two in mutant leaves relative to WT plants. A similar increase was also observed for rbcL mRNA and for transcripts of the psaA–psaB and atpBatpE operons, monitored by employing psaB- and atpB-specific probes. These results exclude the possibility that the reduced accumulation of plastid-encoded proteins (see Figure 4) can be ascribed to reduced transcription of plastid genes. To investigate this issue further, plastid protein synthesis was measured by monitoring the rate of incorporation of [35S]methionine into plastid proteins in young leaves of WT and mutant (prps1-1, prps17-1 and prpl24-1) plants in the presence of light and inhibitors of cytoplasmic protein synthesis for 5, 15 and 30 min. Subsequently, total leaf proteins were extracted and fractionated by SDS-PAGE (Figure 5b). In three independent experiments, the amount of PsbA and RbcL labelled in prpl24-1 plants was decreased on average to 20% of WT levels after 30 min of [35S]methionine incubation. More moderate reductions in the levels of labelled PsbA and RbcL proteins (to about 40 and 30% of WT) were observed in prps1-1 and prps17-1 leaves, respectively.

Figure 5.

 Translation efficiency of chloroplast-encoded proteins in wild-type (WT; Col-0 and Ler) and mutant (prps1-1, prps17-1 and prpl24-1) leaves.
(a) Analysis of transcripts from WT and mutant leaves. Total leaf RNA was fractionated by denaturing agarose gel electrophoresis, blotted onto nylon membrane, and hybridised with the probes reported on the left side of the panel. A replicate agarose gel, stained with ethidium bromide and showing the different rRNA molecules, was used as a loading control.
(b) Incorporation of [35S]methionine into total leaf proteins isolated from six-leaf-rosette plants at low light (20 μmol photons m−2 sec−1). After pulse labelling with [35S]methionine for 5, 15 and 30 min in the presence of cycloheximide, total leaf proteins were isolated, fractionated by SDS-PAGE and detected by autoradiography. As loading control, a portion of the Coomassie Brilliant Blue (C.B.B.)-stained SDS-PAGE, corresponding to the LHCII migration region, is shown. Levels of [35S]methionine incorporation into RbcL and PsbA proteins were quantified and reported in the bar-plot. Values were normalised to the maximal signal intensities obtained in WT leaves (Col-0 and Ler) after 30-min labelling.

Taken together, our results imply that the phenotypic behaviour of prpl24-1, prps1-1 and prps17-1 mutants with respect to the reduction in growth rate and photosynthesis is caused by a decrease in the accumulation of photosynthetic proteins. This can be attributed to defects in ribosome function, as indicated by reductions in rRNA levels and translation of plastid mRNAs.

PRPS20, -L1, -L4, -L27, -L28 and -L35 are required for normal embryonic development

Unlike the three ribosomal mutations described above, no homozygous mutant plants could be identified in the case of prps20-1, prpl4-1, prpl27-1, prpl35-1 and prpl1-1, the last of which was used as a control for an embryo-lethal phenotype (Bryant et al., 2011). Homozygous WT (PRP/PRP) and heterozygous mutant (PRP/prp) plants segregated in a 1:2 manner, indicating that the corresponding gene products are essential during the early stages of embryo and/or seed development.

To further characterise the developmental phenotype of these mutant lines, silique length, seed-setting numbers and the ratio of normal to abnormal seeds in siliques were quantified in homozygous WT (PRP/PRP) and heterozygous mutant plants (PRP/prp). Whereas seed set and silique morphology were very similar in all PRP/PRP and PRP/prp plants (Table S2), albino seeds were readily distinguishable at 6–7 days after fertilisation (DAF) in all siliques of PRP/prp plants bearing each of the five different mutant alleles (Figure 6a, left panel). In mature siliques, the albino seeds eventually turned into shrunken, dark brown structures that were unable to germinate on either MS medium or soil (Figure 6a, right panel). In all mutant plants analysed, the mean percentage of albino/aborted seeds ranged between 23.7 and 26.9%, whereas only 1.7 and 2.1% aborted seeds were observed in WT Ler and Col-0 sister plants, respectively (Table S2). These values are consistent with a 3:1 (normal:aborted seeds) segregation ratio, indicating that the abnormal seed phenotype is a recessive trait controlled by a single locus. To confirm that the abnormal seed phenotypes were indeed caused by mutation of the PRP genes, an additional mutant allele (in the case of PRPL1 and PRPL4) was phenotypically characterised or the respective WT PRP gene was introduced into heterozygous mutant plants to obtain in the next generation homozygous mutants containing the transgenic PRP gene. As expected, the prpl1-2 and prpl4-2 alleles behaved phenotypically like the prpl1-1 and prpl4-1 alleles, respectively (Table S2). Moreover, viable plants with normal seeds and WT-like photosynthetic performance and leaf pigment content were obtained by introducing the WT PRP genes into the homozygous mutant background, demonstrating that the seed phenotypes are indeed caused by mutation of the PRP genes (Tables S1 and S2).

Figure 6.

 Effects of loss of plastid ribosomal proteins (PRPs) on early plant development.
(a) Morphological characterisation of seed development in siliques of wild-type (WT; Col-0) and heterozygous PRPL27/prpl27-1 and PRPL28/prpl28-1 plants. In WT siliques at 10 days after fertilisation (DAF), all developing seeds are green, whereas in PRPL27/prpl27-1 and PRPL28/prpl28-1 siliques around 25% of seeds are albinotic. A very similar phenotype was observed in heterozygous PRPS20/prps20-1, PRPL1/prpl1-1, PRPL4/prpl4-1 and PRPL35/prpl35-1 plants. In mature WT siliques (at 20 DAF) all seeds are round and yellowish, whereas 25% of seeds are shrunken and aborted (in PRPL27/prpl27-1) or much paler (in PRPL28/prpl28-1). Note that PRPS20/prps20-1, PRPL1/prpl1-1, PRPL4/prpl4-1 and PRPL35/prpl35-1 siliques behaved like PRPL27/prpl27-1 siliques.
(b) Characterisation of different stages of embryo development. Top panel, cleared whole mount of WT (Col-0) seeds containing embryos at different developmental stages including pre-globular (1 DAF), globular (3 DAF), transition globular to heart (3–4 DAF), torpedo (4–5 DAF) and linear cotyledons (6 DAF). Bottom panel, 25% of embryos from PRPS20/prps20-1, PRPL1/prpl1-1, PRPL4/prpl4-1, PRPL27/prpl27-1 and PRPL35/prpl35-1 siliques stopped developing, although they retained the capacity for cell division, and remained arrested at a disordered globular stage. For each genotype, the left image was taken at 3–4 DAF and the right picture at 4–5 DAF. Bars = 20 μm.
(c) Analysis of cell division pattern of mutant embryos from WT (Col-0) and heterozygous PRPS20/prps20-1 plants. Siliques were subjected to a modified pseudo-Schiff propidium iodide (mPS-PI) staining technique aimed at visualising the cellular organisation of embryos. The transverse cell division plane in the mutant is indicated by arrowheads. Insets show the cell division planes typical of WT embryos. Note that the cell division pattern of prpl1-1, prpl4-1, prpl27-1 and prpl35-1 mutant embryos was very similar to the one of prps20-1 mutant embryos. Bars = 20 μm.
(d) Images of isolated fully mature embryos (bent cotyledon stage) from WT (Col-0) and prpl28-1 seeds. Bars = 20 μm.

To determine whether embryo development was affected in the prps20, prpl1, prpl4, prpl27 and prpl35 mutants, optical sections of cleared seed whole-mounts at different developmental stages were analysed by microscopy (Figure 6b). At 3–4 DAF, WT embryos had reached the heart stage (Figure 6b, top panel), whereas about 25% of the seeds from PRP/prp plants were retarded in their development and the embryos arrested at the globular stage (Figure 6b, bottom panel, left picture of each genotype). In the WT, heart-stage embryos undergo an ordered series of cell divisions that allow them to traverse through the torpedo and linear cotyledon stages to the fully mature embryo stage (Figure 6b, top panel). In contrast, although they retained the capacity for cell division, mutant embryos at 4–5 DAF exhibited a disordered globular-like structure (Figure 6b, bottom panel, right picture of each mutant), and began to disintegrate at about 15 DAF. To visualise the cellular organisation of mutant and WT embryos, representative siliques of heterozygous PRPS20/prps20-1 plants at 3–4 DAF were subjected to a modified pseudo-Schiff propidium iodide (mPS-PI) staining technique (Figure 6c; see Experimental procedures). The mutant embryos showed a disordered globular-like organisation marked by abnormal cell division patterns. In particular, mutant globular embryos showed additional transverse cell division planes (Figure 6c, arrowheads) while lacking the usual two longitudinal division planes (Figure 6c, inset). Therefore, it appears that while the epidermis, the first element of embryo radial pattern (Jenik et al., 2007) which becomes visible in the 16-cell embryo (early globular), can differentiate in mutant embryos, provascular cells fail to differentiate. Similar differences were also observed in the other mutant genotypes, indicating that, in the absence of PRPS20, -L1, -L4, -L27 or -L35, embryo development is perturbed due to a change in cell division patterns, which prevents embryos from progressing through the globular to the heart stage and beyond.

Mutation of the gene coding for PRPL28 results in a phenotype which differs from that described above. Although siliques of heterozygous PRPL28/prpl28-1 plants were also characterised by the presence of albino seeds at 6–7 DAF (Figure 6a, left panel), these seeds retained their very pale colour even in mature siliques (Figure 6a, right panel) and accounted for about one-quarter of all seeds, which is typical of a monogenic recessive trait (Table S2). The pale seeds contained fully mature albino embryos (Figure 6d) that were able to germinate on both soil and MS medium, but they did not survive past the cotyledon stage when grown under photoautotrophic conditions (Figure S2). This observation, together with the complete restoration of the WT phenotype by introduction of the WT PRPL28 gene into the prpl28-1 background (see Table S1), implies that PRPL28 is essential for the latest stages of embryo–seedling development, during the greening process.


Decreases in plastid translation rates can affect photosynthesis

Certain ribosomal proteins are not essential for ribosomal function, but their removal reduces the translational capacity and decreases the photosynthetic performance of plastids. Thus, Arabidopsis mutants that lack PRPS21, -L11 or -L24, or have reduced amounts of PRPS17, are viable but display a marked drop in rates of photosynthesis and growth (Pesaresi et al., 2001; Woo et al., 2002; Morita-Yamamuro et al., 2004; Tiller et al., 2012). In this study, characterisation of Arabidopsis lines lacking PRPS1 or -S17, together with the prpl24-1 mutant as a control, confirmed that these loss-of-function mutants are able to complete their entire life cycle (Figure 7), although all three mutant genotypes showed reductions in growth rate, leaf pigment content and photosynthetic performance. The altered photosynthetic performance is attributable to the marked decrease in chloroplast translational activity in these mutants, as shown by reduced incorporation of [35S]methionine into PsbA and RbcL. Are PRPS17 and PRPL24 dispensable for the basal activity of plastid ribosomes (for PRPS1 this cannot be unequivocally concluded because residual expression of PRPS1 remains in prps1-1 homozygotes) or are there alternative explanations available for the viability of these prp mutants? Theoretically, dual targeting of their nuclear-encoded mitochondrial counterparts (At1g49400 and At3g18880 in the case of PRPS17; At5g23535 in the case of PRPL24) to mitochondria and chloroplasts could account for the viability of prps17 and prpl24 mutants. However, the Ambiguous Targeting Predictor (; Mitschke et al., 2009), a machine-learning implementation that predicts dual-targeted organelle proteins, attributed very low scores to At1g49400 (0.53), At3g18880 (0.53) and At5g23535 (0.44), whereas PRORS1, an experimentally verified dual-located protein (Pesaresi et al., 2006), was unambiguously predicted into both plastids and mitochondria (0.95). Moreover, proteomic studies failed to detect At1g49400, At3g18880 and At5g23535 within chloroplasts [see also ‘The Subcellular Location of Proteins in Arabidopsis Database (SUBA; and the Plant Proteome Database (]. Therefore, it is unlikely that dual targeting of nuclear-encoded mitochondrial ribosomal proteins contributes to the phenotypes observed for the prps17 and prpl24 mutants.

Figure 7.

 Schematic representation of functions of plastid ribosomal proteins (PRPs). The proteins PRPS5, -S13, -S20, -L1, -L4, -L6, -L21, -L27 and -L35 allow the transition from the globular to the heart stage. PRPL28 is essential for the greening of embryos and seedlings, whereas PRPS1, -S17, -S21, -L11 and -L24 play a major role in adult plants, being pivotal for optimal ribosome activity. Note that the role of the following subunits has been reported in previous publications: PRPS5, -S13, -L1, -L6 (Bryant et al., 2011; Lloyd and Meinke, 2012); PRPL21 (Yin et al., 2012); PRPS21 (Morita-Yamamuro et al., 2004); PRPL11 (Pesaresi et al., 2001) and PRPL24 (Tiller et al., 2012).

Plastid translation is required for embryo development in Arabidopsis

Plastid differentiation during embryogenesis, which generates specific patterns of chloroplast-containing cells in specific cell layers at specific stages of embryogenesis, must be tightly regulated (Tejos et al., 2010), but how this is achieved at the molecular level remains unclear. Functional plastids are certainly essential for embryogenesis in Arabidopsis. Thus, based on the work of Bryant et al. (2011) and Yin et al. (2012), and the data in the present study, it can be concluded that at least nine nuclear-encoded PRPs (PRPS5, -S13, -S20, -L1, -L4, -L6, -L21, -L27 and -L35) are essential for embryo development in Arabidopsis. Our analysis of lines lacking PRPS20, -L1, -L4, -L27 or -L35 indicate that these PRPs become indispensable at the globular stage of embryo development (see Figures 6 and 7), corroborating previous analyses of lines lacking PRPL21 (Yin et al., 2012). Moreover, such mutant embryos are characterised by perturbations in cell division patterns, leading to the formation of highly disordered globular-like structures. The resulting failure to differentiate specific cell layers prevents the progression of embryo development beyond this stage. This defect in the patterning of cell division could be due to the lack of specific molecules, such as hormones produced by plastids (Peltier et al., 2006; Santner and Estelle, 2009; Santner et al., 2009) that either directly influence cell division or are involved in the coordination of nuclear and plastid gene expression.

Which plastid process is essential for embryogenesis in Arabidopsis?

The possibility that the embryo lethality observed in some prp mutants is caused by interference with the photosynthetic process can be excluded, because mutants devoid of components essential for thylakoid electron flow can still reach the seedling stage (see, for instance, Maiwald et al., 2003; Weigel et al., 2003; Ihnatowicz et al., 2004). Intraspecific variation with respect to the sensitivity of plant embryogenesis to the loss of plastid translation has shed further light on the role of plastids in embryogenesis. In most cases, mutations that interfere with chloroplast translation efficiency in barley, maize or Brassica napus generally do not disrupt embryogenesis, but allow the formation of albino seedlings instead (Hess et al., 1994; Zubko and Day, 1998). For instance, lack of the plastid-encoded PRPS12 results in embryo lethality in Arabidopsis, whereas the corresponding maize mutant is able to germinate and produce albino seedlings (Ostheimer et al., 2003; Asakura and Barkan, 2006). Several lines of evidence suggest that a relative lack of the plastid-encoded accD-subunit of the multimeric acetyl-CoA carboxylase required for fatty acid biosynthesis might be one of the causes responsible for the lethality of Arabidopsis embryos defective in plastid translation (Bryant et al., 2011). Indeed, grass species and B. napus contain a plastid-located monomeric acetyl-CoA carboxylase that, differently from Arabidopsis, is encoded in the nucleus and translated in the cytosol (Schulte et al., 1997; Chalupska et al., 2008). Therefore fatty acid biosynthesis (and embryogenesis) can continue even when plastid protein synthesis is affected in these species. The essential nature of fatty acid biosynthesis in embryo development is also known from the disruption of a nuclear-encoded subunit of the plastid multimeric acetyl-CoA carboxylase (At5g16390; Li et al., 2011), as well as from the arrest of fatty acid biosynthesis by preventing the accumulation of the S-malonyltransferase enzyme (At2g30200; Bryant et al., 2011). This, together with the observed block in embryogenesis at the globular stage in prp mutants (this study and Yin et al., 2012), suggests that at the globular stage, when pro-plastids start to differentiate into chloroplasts (Mansfield and Briarty, 1991), fatty acid biosynthesis might become essential for embryo development (Kobayashi et al., 2007). Certainly, disruption of other essential plastid functions can also lead to the arrest of embryogenesis. For instance, loss of function of key players in the TOC-TIC machinery required for plastid protein import, such as TOC75, TIC20, TIC110, results in embryo lethality (Jarvis, 2008; Inaba and Ito-Inaba, 2010; Kasmati et al., 2011). Likewise, mutation in the gene encoding the key enzyme of galactolipid biosynthesis, MGD1, compromises proper embryogenesis. Mutation in other essential proteins, such as chaperons, proteases and aminoacyl-tRNA synthases, in plastids also compromise embryogenesis (Inaba and Ito-Inaba, 2010). However, with the exception of the ClpP protease (Shikanai et al., 2001), these genes are encoded by the nuclear genome; therefore disruption of plastid translation cannot directly interfere with their expression.

Can one predict which plastid ribosomal proteins are essential for embryogenesis?

The nine plastid ribosomal subunits identified so far as being essential for embryogenesis in Arabidopsis (PRPS5, -S13, -S20, -L1, -L4, -L6, -L21, -L27 and -L35) contribute to different ribosomal domains in either the 30S or the 50S subunit (Stelzl et al., 2001). Moreover, the essential/non-essential nature of the role of PRPs in Arabidopsis embryogenesis cannot be predicted on the basis of studies on prokaryotes. This is outlined in the following three scenarios in which the phenotypic effects of the removal of individual subunits from Escherichia coli are compared with those of deletion of their counterparts in chloroplast ribosomes in Arabidopsis.

  • 1 Lethality in E. coli/embryo lethality in Arabidopsis: (P)RPL4 and-L27 (this study; Table S3; Hashimoto et al., 2005; Bryant et al., 2011). Only in these cases is the essential nature of the Arabidopsis PRP reflected in the E. coli mutant phenotype.
  • 2 Viability in E. coli/embryo lethality in Arabidopsis: (P)RPS20, -L1 and -L35 (this study; Table S3; Hashimoto et al., 2005; Bryant et al., 2011). Interestingly, loss of PRPL35 also results in aberrant embryo morphogenesis and non-viable seeds in maize (Magnard et al., 2004), indicating that a certain level of plastid protein synthesis is also needed for normal embryo development in grasses.
  • 3 Lethality in E. coli/unperturbed embryogenesis in Arabidopsis: (P)RPS1, -S17, -L24, -L28 (this study; Table S3; Hashimoto et al., 2005). The Arabidopsis prps1-1, prps17-1 and prpl24-1 mutants are viable, albeit with reduced photosynthetic performance. However, a fundamental role of PRPS1 in Arabidopsis cannot be entirely excluded, because some residual PRPS1 gene expression can still be observed in prps1-1 plants. Interestingly, the maize high chlorophyll fluorescence 60 mutant, which lacks PRPS17, displays a seedling-lethal phenotype (Schultes et al., 2000), in contrast to the corresponding Arabidopsis mutant which can complete its life cycle. Although Arabidopsis plants without PRPL28 are seedling lethal, they still show embryo and seed formation (this study). Given the altered pigmentation of prpl28-1 mutant seeds, it can be speculated that the formation of the photosynthetic machinery, which is associated with the greening process, is disturbed. For instance, a specific role of PRPL28 in the translation process on the surface of thylakoid membranes might be hypothesised (Minami and Watanabe, 1984; Hurewitz and Jagendorf, 1987; Zhang et al., 1999). Indeed, stromal ribosomes are recruited into thylakoid polysomes, which are active in synthesising thylakoid proteins that are essential for the biogenesis of the photosynthetic apparatus. In this context, PRPL28 might play a major role in the recruitment of ribosomes to thylakoids. Alternatively, the presence of PRPL28 might be specifically required for thylakoid-bound ribosomes.

Taken together, this cross-kingdom comparison of mutant phenotypes clearly suggests that the impact of specific ribosomal proteins on embryogenesis in Arabidopsis cannot be predicted on the basis of their mutant phenotypes in E. coli. In principle, this can be explained by changes in the function of these proteins during evolution of the chloroplast from a cyanobacterial endosymbiont, or more probably by the combination of changes in protein and rRNA sequences.

The sensitivity of embryogenesis to plastid gene expression in Arabidopsis allows one to identify PRPs which are essential for embryogenesis. Further studies have to clarify whether these PRPs are also essential for ribosomal function, or whether plastid ribosomes that lack such PRPs retain a basal activity that is insufficient to meet the need for plastid protein synthesis during embryogenesis.

Experimental procedures

Plant material, propagation and growth measurements

Mutant alleles were identified by searching the T-DNA Express database ( and mutant lines were obtained from the SALK collection (Alonso et al., 2003) (prps20-1/Salk_094710; prpl4-1/Salk_117563; prpl4-2/Salk_094226; prpl24-1/Salk_010822; prpl28-1/Salk_142282), the SAIL collection (Sessions et al., 2002) (prps1-1/Sail_560_B02; prpl1-1/Sail_295_A02; prpl35-1/Sail_367_E07), the John Innes Centre collection (Tissier et al., 1999) (prps17-1/GT_5_19055; prpl1-2/GT_5_101962) and the GABI-KAT collection (Rosso et al., 2003) (prpl27-1/GABI_123H12). With two exceptions, mutant alleles are in the Col-0 genetic background; prps17-1 and prpl1-2 are derived from Ler. T-DNA insertions were confirmed by sequencing PCR products obtained using gene- and T-DNA-specific primers (Table S4). Arabidopsis thaliana Heynh. WT (Col-0 and Ler) and mutant plants were grown under controlled growth chamber conditions as described (Pesaresi et al., 2009). Phenotypic analyses were also conducted on plants grown on Murashige and Skoog (MS) medium (Duchefa, with or without 1% (w/v) sucrose. Growth measurements are described elsewhere (Leister et al., 1999).

Plant transformation and isolation of transgenic lines

For complementation analyses, the PRPS1 and PRPS17 coding sequences were recombined into the Gateway plant transformation destination vector pB2GW7 (Flanders Interuniversity Institute for Biotechnology, Gent, Belgium), under the control of the 35S promoter from the Cauliflower Mosaic Virus (CaMV) (see Table S5 for primer sequences). For PRPS20, PRPL24, PRPL27, PRPL28 and PRPL35 the corresponding genomic sequences, together with 1 kbp of promoter regions, were recombined into pB2GW7, devoid of the 35S-CaMV promoter. Plants were transformed according to Clough and Bent (1998) and independent transgenic plants were selected on the basis of their resistance to Basta.

Nucleic acid analysis

Arabidopsis thaliana DNA was isolated as described (Ihnatowicz et al., 2004). For RNA analysis, total leaf RNA was extracted from fresh tissue using the TRIzol reagent (Invitrogen, Northern analysis was performed under stringent conditions, according to Sambrook and Russell (2001). Probes complementary to nuclear and chloroplast genes were used for the hybridisations. Primers used to amplify the probes are listed in Table S6. All probes used were cDNA fragments labelled with 32P. Signals were quantified with a phosphoimager (Typhoon; GE Healthcare, using the program imagequant.

For quantitative real-time PCR (qRT-PCR) profiling, 4-μg aliquots of total RNA, treated with DNase I (Roche Applied Science, for at least 30 min, were utilised for first-strand cDNA synthesis using iScript reverse transcriptase (Bio-Rad, according to the supplier’s instructions. The qRT-PCR profiling was carried out on an iCycler iQ5 real-time PCR system (Bio-Rad), using the oligonucleotide sequences reported in Table S6. Data from three biological and three technical replicates were analysed with Bio-Rad iQ5 software (version 2.0).

PAGE and immunoblot analyses

Leaves were harvested from plants at the six-leaf rosette stage, and thylakoids were prepared as described (Bassi et al., 1985). For BN-PAGE, thylakoid samples equivalent to 100 mg of fresh leaf material were solubilised and fractionated as described in Pesaresi et al. (2009). For 2D PAGE, BN-PAGE lanes were subsequently fractionated on denaturing Tricine-SDS gels (15% PA gel) and the protein content was stained with colloidal Coomassie Blue (G 250).

For immunoblot analyses total proteins were prepared from plants at the six-leaf rosette stage (Martinez-Garcia et al., 1999), then fractionated by SDS-PAGE (on 12% PA gels) (Schägger and von Jagow, 1987). Subsequently, proteins were transferred to poly(vinylidene difluoride) membranes (Ihnatowicz et al., 2004), and replicate filters were immunodecorated with appropriate antibodies. Signals were detected by enhanced chemiluminescence (GE Healthcare) and quantified using image quant for Macintosh (version 1.2; Molecular Dynamics,

In-vivo translation assay

The in-vivo translational assay was performed essentially as in Pesaresi (2011). Twelve leaf discs of 4 cm diameter were incubated in a buffer containing 20 μg ml−1 cycloheximide, 1 mm K2HPO4–KH2PO4 (pH 6.3), and 0.1% (w/v) Tween-20 to block cytosolic translation. The [35S]methionine was added to the buffer (0.1 mCi ml−1) and the material was vacuum-infiltrated. Leaves were exposed to light (20 μmol photons m−2 s−1) and four leaf discs were collected at each time point (5, 15 and 30 min). Total proteins were extracted as described above and loaded on Tricine SDS-PAGE (12% PA). Signals were detected and quantified using the phosphorimager and the imagequant program as described above.

Chlorophyll fluorescence and pigment analyses

In vivo Chl a fluorescence of leaves was measured using the Dual-PAM-100 (Walz, as described (Pesaresi et al., 2009). Five plants of each genotype were analysed and average values plus standard deviations were calculated. Plants were dark-adapted for 30 min and minimal fluorescence (F0) was measured. Then pulses (0.8 sec) of saturating white light (5000 μmol photons m−2 sec−1) were used to determine the maximum fluorescence (FM), and the ratio (FM − F0)/FM = FV/FM (maximum quantum yield of PSII) was calculated. A 10-min exposure to actinic light (80 μmol photons m−2 sec−1) served to drive electron transport between PSII and PSI. Then steady-state fluorescence (FS) was measured, and inline image was determined after exposure to further saturation pulses (0.8 sec, 5000 μmol photons m−2 sec−1). The effective quantum yield of PSII (ΦII) was calculated as the ratio inline image.

In vivo Chl a fluorescence of whole plants was recorded using an imaging chlorophyll fluorometer (Imaging PAM; Walz) by exposing dark-adapted plants to a pulsed, blue measuring beam (1 Hz, intensity 4; F0) and a saturating light flash (intensity 4) to obtain FV/FM. A 10-min exposure to actinic light (80 μmol photons m−2 sec−1) was then used to calculate ΦII.

Pigments were analysed by reverse-phase HPLC (Färber et al., 1997).

Whole-mount preparation and microscopy

To analyse defects in seed development, siliques of WT and heterozygous PRPS20/prps20-1, PRPL1/prpl1-1, PRPL4/prpl4-1, PRPL27/prpl27-1, PRPL28/prpl28-1 and PRPL35/prpl35-1 plants were manually dissected and observed using a Zeiss LUMAR.V12 stereomicroscope ( To follow defects during embryo development, siliques from the same genotypes were cleared as reported (Yadegari et al., 1994). Developing seeds were observed using a Zeiss Axiophot D1 microscope equipped with differential interface contrast (DIC) optics. Images were recorded with an Axiocam MRc5 camera (Zeiss) using the Axiovision program (version 4.1). Modified pseudo-Schiff propidium iodide (mPS-PI) embryo staining was performed as described by Truernit et al. (2008). Whole seeds were observed with a Leica TCS-SP5 confocal laser scanning microscope (Leica Microsystems, The excitation wavelength for mPS-PI-stained samples was 488 nm, and fluorescence emission was collected between 520 and 720 nm.


This work was supported by the Italian Ministry of Research, special fund for basic research (PRIN 2008XB7774B) to PP and by the Deutsche Forschungsgemeinschaft (SFB-TR1, TP B8 and FOR 804) to DL. We thank Paul Hardy for critical reading of the manuscript.