We performed a phylogenetic analysis of Arabidopsis CRWN proteins and their homologues in other species to begin our investigation of the potential diversification within this family. The predicted Arabidopsis proteome contains four closely related CRWN proteins (CRWN1 through 4) that share 30-40% amino acid identity; no other Arabidopsis proteins with extended regions of significant amino acid identity to CRWN proteins were found. Similar proteins were found in other plant species, but interestingly, no fungi or animal CRWN homologues were identified from searches of protein databases. In addition, CRWN homologues were absent in the predicted proteome of the green algae Chlamydomonas and Volvox.
We constructed a phylogram of CRWN proteins and related plant homologs using a maximum likelihood algorithm (Figure 1 and Additional file 1: Table S1). The tree features two major clades distinct from CRWN homologues in two basal plants, Selaginella moellendorffii and Physcomitrella patens. One clade includes three of the Arabidopsis paralogs, CRWN1, CRWN2 and CRWN3, while CRWN4 belongs to the other clade. Within each clade, the monocot proteins, represented by maize, sorghum and rice, group independently from the dicot proteins. Only two CRWN paralogs exist in these monocots – one CRWN1-like and one CRWN4-like. However, certain dicot species, such as Arabidopsis, poplar, grape, and castor bean, contain multiple copies of CRWN1-like proteins. The dicot CRWN4-like proteins are also distinct from their monocot counterparts in lacking a conserved amino acid motif at the extreme C-terminus (yellow inset in Figure 1 and Additional file 2).
Genetic redundancy in the CRWN family
The inference that CRWN4 and related proteins are divergent from members of the CRWN1-containing clade was supported by genetic analyses to dissect the functions of the CRWN paralogs. We used Agrobacterium T-DNA insertion alleles to study the effects of inactivating different combinations of CRWN genes [17]. Previously, we demonstrated that the crwn1-1 and crwn2-1 T-DNA alleles severely reduce or eliminate transcription downstream of the T-DNA insertion [5]. Here, we performed transcript analysis by RT-PCR for the crwn3-1 and crwn4-1 alleles used in this study (Additional file 3). For crwn3-1, some transcript was detected downstream of the insertion; however, no transcript could be detected using primers that flanked the insertion. The crwn4-1 insertion blocked transcription downstream of the T-DNA. The lack of full-length CRWN transcripts from homozygous mutant lines indicates that all four mutations used in this study are likely to be loss-of-function alleles. We note that the CRWN genes have similar developmental gene expression patterns: the steady-state abundance of transcripts for all four paralogs peak in proliferating tissues (http://bar.utoronto.ca/efp/cgi-bin/efpWeb.cgi) [18].
Mutant plants carrying single insertions were intercrossed and progeny carrying homozygous insertions in different combinations were recovered. Figure 2 shows the whole-plant phenotype of the viable mutants at the rosette stage just after the transition to flowering. Plants carrying a mutation in any single CRWN gene had phenotypes similar to wild-type Columbia plants, as did the double crwn2 crwn3 and crwn3 crwn4 mutants. The crwn2 crwn4 and crwn1 crwn4 double mutants exhibited slightly smaller rosettes, while the remaining double mutants, crwn1 crwn2 and crwn1 crwn3, displayed markedly smaller rosette sizes. We were able to recover only two of the four triple mutants - crwn1 crwn2 crwn4 and crwn1 crwn3 crwn4, both of which were extremely stunted and set few seed.
Our inability to isolate a mutant combining alleles in all four CRWN genes indicates that at least one functional CRWN protein is required for viability. Triple mutant plants carrying only CRWN2 or CRWN3 were extremely stunted, but still viable. This result suggests that CRWN2 or CRWN3 alone can cover the minimum requirements for the entire CRWN protein family. However, plants carrying only CRWN1 or only CRWN4 were not recovered, suggesting that CRWN1 and CRWN4 are specialized and that neither protein alone can express the full range of functions of the CRWN protein family.
CRWN proteins are required to maintain proper nuclear size and shape
We next observed crwn mutant nuclei from adult leaf tissue to determine the role of different CRWN proteins in specifying nuclear size and shape. Among the single mutants, a deficiency of CRWN1 or CRWN4 reduced nuclear size (Figures 3 and 4A; Additional file 4), while loss of CRWN2 or CRWN3 had no effect. Combining a crwn1 mutation with a crwn2 or crwn3 mutation had a synergistic effect on nuclear size, suggesting that CRWN1 function overlaps, at least partially, with those of CRWN2 and CRWN3. Double mutant combinations containing crwn4 and either crwn2 or crwn3 did not show additive phenotypes but rather resembled crwn4. In contrast, combination of a crwn1 with a crwn4 mutation had an additive effect on nuclear size. These findings indicate that CRWN1 and CRWN4 are the major determinants of nuclear size among the CRWN paralogs. Further, the additive effects of crwn1 and crwn4 mutations suggest CRWN4 acts independently from CRWN1, consistent with their distinct phylogenetic grouping (Figure 1) and the genetic analysis shown in Figure 2.
CRWN proteins are required for development or maintenance of the elongated spindle shapes which characterize larger nuclei in differentiated wild-type cells [19]. We previously reported that a deficiency of CRWN1 causes nuclei in all cells to adopt the spherical shape characteristic of proliferating tissue at root and shoot apices [5]. The present study confirmed the importance of CRWN1 for nuclear shape differentiation and also uncovered a similar role for CRWN4 (Figure 3), a conclusion also reached recently by Sakamoto and Takagi [7]. Nuclei from crwn4 leaf tissue often have irregular margins and are more spherically shaped, compared to wild-type nuclei (Additional file 5). However, crwn4 nuclei are less uniformly round in comparison to crwn1 nuclei, particularly larger crwn4 nuclei. Further, crwn4 nuclei occasionally contain thin projections that appear to be drawn from the nuclear surface (arrowheads in Figure 3).
Loss of CRWN proteins affects nuclear DNA packing density
The direct correlation between endopolyploidy and nuclear size in wild-type Arabidopsis cells [20] prompted us to examine this relationship within the crwn mutants. We measured the average endopolyploidy level of nuclei from the same adult leaves harvested for the nuclear size analysis shown in Figure 4A (see also Additional file 4). Some crwn mutants showed a decrease in endopolyploid levels, particularly the crwn triple mutants and the crwn1 crwn2 double mutant, but the remaining crwn genotypes had average endopolyploidy levels that approached wild-type levels (Figure 4B). The dashed line in Figure 4B depicts the expected nuclear size change in response to a reduction in endopolyploidy based on the established one-to-one relationship between nuclear volume (approximated by nuclear area in our measurements of isolated and flattened leaf cell nuclei, Additional file 6) and DNA content in wild type plants. With the exception of crwn2 and crwn3, the crwn mutations caused a more pronounced reduction in nuclear size than predicted from the observed endopolyploidy level. As a consequence, crwn mutants display a spectrum of nuclear DNA densities, ranging from wild-type values in crwn2 and crwn3 mutants to four-fold higher densities in crwn1 crwn2 double mutants and the two viable crwn triple mutants.
We then investigated the relationship between nuclear size and DNA content by examining the effects of different crwn genotypes on nuclear size in leaf guard cells, a diploid cell type where endopolyploidy is not a factor [21]. crwn1 mutant guard cell nuclei were smaller than nuclei in wild type cells with an area approximately one-half of the wild type value, corresponding to a volume difference of approximately threefold assuming a roughly spherical shape to nuclei in the cell (Figure 5). Double and triple mutants lacking CRWN1 displayed nuclear sizes similar to the crwn1 single mutant. Consistent with their effects on nuclear size shown in Figure 4A, neither the crwn2 nor crwn3 mutation affected nuclear size in guard cells. Interestingly, the size of nuclei in crwn4 guard cells was also unaffected, in contrast to the effect seen in a population of adult leaf cells (Figure 4A). However, crwn2 crwn3, crwn2 crwn4, and crwn3 crwn4 double mutants had nuclei approximately 20% smaller than those seen in wild-type guard cells, suggesting some functional redundancy among CRWN2, CRWN3 and CRWN4 proteins. Overall, our results indicate that CRWN1 plays the major role in affecting nuclear size in the absence of changes in endopolyploidy.
CRWN4 maintains interphase chromocenter integrity and organization
Considering the dramatic effects of crwn mutations on nuclear size and morphology, we turned our attention to the role of CRWN proteins on the internal organization of the nucleus. A conspicuous feature of Arabidopsis interphase nuclei are discrete foci of heterochromatin, or chromocenters, visualized as bright spots after staining with fluorescent DNA-intercalating dyes [22]. A typical interphase nucleus contains approximately ten chromocenters corresponding to the number of diploid chromosomes (2n = 10) [23]. Chromocenter number remains fairly constant over a wide range of nuclear sizes and endopolyploid levels (2n to 16n), most likely as a result of lateral association of sister chromatids after endoreduplication [24, 25]. We found that the average chromocenter number in crwn1, crwn2 and crwn3 leaf cell nuclei was similar to that seen in wild-type leaf cell nuclei (Figure 6A) and did not change dramatically as a function of nuclear size. In crwn4 nuclei, however, chromocenter number was strongly correlated with nuclear size (Figure 6A): smaller nuclei contained fewer chromocenters than the wild-type value of ~9, while larger, presumably endopolyploid, crwn4 nuclei exhibited a wide range of chromocenter numbers (2–27). A similar pattern was observed in double mutants containing the crwn4 mutation (Additional file 7). In contrast, double mutants containing the crwn1 allele paired with another crwn mutation displayed a reduced average chromocenter number with a weaker association with nuclear size (Figure 6A and Additional file 7).
To explore the chromocenter phenotype in more detail, we developed a statistic, referred to as an aggregation index (AI) (see Methods), to characterize the distribution of visible DAPI-bright spots within interphase nuclei. The value of this index ranges from 0 to 1, reflecting both the number of distinct chromocenter spots and the uniformity of their size distribution. The expected AI for wild-type nuclei containing 10 equally sized chromocenters is 0.1, while clustering of chromocenters into fewer but larger aggregates will lead to a higher AI value. A dispersal of chromocenters into smaller heterochromatic satellites will push the AI lower. For a given chromocenter number, a skewed CC size distribution is associated with a larger AI compared to when each CC is equally sized. As shown in Figure 6B, the AI index of wild-type nuclei averaged close to 0.1 and was not affected significantly by nuclear size. The absence of a significant correlation between AI and nuclear size indicates that chromocenter organization remains constant across different endopolyploidy levels in wild-type nuclei. A similar pattern was observed for the crwn1, crwn2, and crwn3 mutant samples. In contrast, combining crwn1 and crwn2 mutations led to an approximately two-fold higher AI over a range of nuclear sizes, consistent with the two-fold reduction in chromocenter number via aggregation in crwn1 crwn2 mutants. A different pattern was displayed in the crwn4 sample, which displayed a negative correlation between AI and nuclear size. This result suggests a tendency for chromocenters to aggregate in smaller crwn4 nuclei and to become dispersed in larger crwn4 nuclei. The reduction in chromocenter number in crwn1 crwn2 and crwn4 mutants with smaller nuclei could reflect the aggregation of individual chromocenters in the limited confines of these nuclei, but a similar clustering does not occur in small wild-type nuclei, arguing that small nuclear dimensions alone are insufficient to cause clustering. The variability in chromocenter size and number in crwn mutant nuclei suggests that CRWN proteins are required for proper organization of heterochromatin in interphase nuclei.
We tested this hypothesis by visualizing the spatial arrangement of chromocenter-associated genomic regions in crwn1 crwn2 and crwn4 mutants. Arabidopsis chromocenters are comprised of large segments of repetitive DNA such as the tandemly-arrayed centromeric and 5S RNA repeats located within pericentromeric regions [23]. Using fluorescent in situ hybridization (FISH), we examined the spatial organization of the major 180-bp centromeric tandem repeat and the 5S RNA gene arrays in both large and small nuclei from wild-type, crwn1 crwn2 and crwn4 plants (Figure 7A, B). The centromeric and 5S RNA repeats were co-localized with the DAPI-bright spots in both small and large wild-type nuclei, confirming previous reports that these sequences are normally compartmentalized within chromocenters at the nuclear periphery [26] (see Additional file 8: Movie S1). It was common to find a decondensed centromere signal at several chromocenters in wild-type nuclei; however, decondensed centromeric repeat clusters were infrequently observed in crwn1 crwn2 nuclei and the total number of clusters was reduced (Figure 7C) (also see Additional file 8: Movie S2). These findings indicate that there is a compaction of the centromere repeat arrays within coalesced chromocenters in crwn1 crwn2 nuclei. In contrast, the number of discrete centromere repeat clusters visible in crwn4 nuclei was more variable, and decondensed signals were often seen in nuclei with numerous clusters. This pattern is consistent with the hypothesis that chromocenters become dispersed in larger crwn4 nuclei. A similar but more pronounced trend was seen for the 5S RNA gene arrays (Figure 7B, D), which were dispersed outside chromocenter aggregates in roughly one-half of the crwn4 nuclei. We note that the dispersed 5S RNA gene signal remained localized to the nuclear periphery (see Additional file 8: Movie S3). The apparent dispersal of chromocenters in larger crwn4 nuclei and the mis-positioning of centromeric and 5S RNA repeats outside of the chromocenter indicates that higher-order organization of heterochromatin breaks down in interphase in the absence of CRWN4.