Arabidopsis ribosomal proteins RPL23aA and RPL23aB are functionally equivalent

Background: In plants, each ribosomal protein (RP) is encoded by a small gene family but it is largely unknown whether the family members are functionally diversified. There are two RPL23a paralogues genes ( RPL23aA and RPL23aB ) found in Arabidopsis thaliana . Knock-down of RPL23aA using RNAi impeded growth and led to morphological abnormalities, whereas knock-out of RPL23aB had no observable phenotype, thus these two RPL23a paralogous proteins have been used as examples of ribosomal protein paralogues with functional divergence in many published papers. Results: In this study, we characterized T-DNA insertion mutants of RPL23aA and RPL23aB . A rare non-allelic non-complementation phenomenon was found in the F1 progeny of the rpl23aa X rpl23ab cross, which revealed a dosage effect of these two genes. Both of RPL23aA and RPL23aB were found to be expressed almost in all examined tissues as revealed by GUS reporter analysis. Expression of RPL23aB driven by the RPL23aA promoter can rescue the phenotype of rpl23aa , indicating these two proteins are actually equivalent in function. Interestingly, based on the publicly available RNA-seq data, we found that these two RPL23a paralogues were expressed in a concerted manner and the expression level of RPL23aA was much higher than that of RPL23aB at different developmental stages and in different tissues. Conclusions: Our findings suggest that RPL23aA and RPL23aB proteins actually have equal function and presence of paralogous genes for the RPL23a protein in plants might be necessary to maintain its adequate dosage. as exemplified by the single mutant phenotypes. Our findings suggest that the two paralogous RPL23a proteins had equivalent function and the presence of multiple genes for individual RPs in plants might be necessary to maintain adequate ribosome dosage at least for some ribosomal protein families. paralogues with functional divergence in many published papers. In this study, our findings provided four convincing evidences demonstrating duplicated RPL23a genes actually have redundant function (without functional specialization), thus are necessary to provide a threshold dose: 1) The non-allelic non-complementation phenomenon between rpl23aa and rpl23ab suggests RPL23aA and RPL23aB are dosage dependent genes; 2) RPL23aA and RPL23aB genes are expressed in the same tissues; 3) RPL23aB could rescue the phenotype of rpl23aa, demonstrating RPL23aA and RPL23aB protein have equal function; 4) RPL23aA and RPL23aB genes are transcribed in a concerted manner without spatiotemporal divergence. Since the presence of multiple genes for individual RP widely exists in plants, our findings will have implications to study other paralogous RP genes.

RPL23aB had no observable phenotype, thus these two RPL23a paralogous proteins have been used as examples of ribosomal protein paralogues with functional divergence in many published papers.

Results:
In this study, we characterized T-DNA insertion mutants of RPL23aA and RPL23aB. A rare non-allelic non-complementation phenomenon was found in the F1 progeny of the rpl23aa X rpl23ab cross, which revealed a dosage effect of these two genes. Both of RPL23aA and RPL23aB were found to be expressed almost in all examined tissues as revealed by GUS reporter analysis. Expression of RPL23aB driven by the RPL23aA promoter can rescue the phenotype of rpl23aa, indicating these two proteins are actually equivalent in function. Interestingly, based on the publicly available RNA-seq data, we found that these two RPL23a paralogues were expressed in a concerted manner and the expression level of RPL23aA was much higher than that of RPL23aB at different developmental stages and in different tissues.

Conclusions:
Our findings suggest that RPL23aA and RPL23aB proteins actually have equal function and presence of paralogous genes for the RPL23a protein in plants might be necessary to maintain its adequate dosage.

Background
Ribosome, a ribonucleoprotein complex formed from a large and a small subunit, is responsible for polypeptide synthesis in all living cells. In plants, the large ribosomal subunit is composed of 28S, 5.8S and 5S rRNAs together with 48 RPL (Ribosomal Protein of Large subunit) proteins, whereas the small subunit is composed of 18S rRNA and 33 RPS (Ribosomal Protein of Small subunit) proteins [1]; [2]. In E. coli, genes encoding RPs are arranged in about 20 operons, with approximately half of the genes mapping to a single locus [3]; [4]. In mammals, although there are about 2,000 sequences which may encode RPs, most of them are predicted to be pseudogenes, and most functional RPs are 3 encoded by a single copy [5]. In yeast Saccharomyces cerevisiae, two-thirds of the RPs are encoded by gene families with more than one member. Although there is substantial redundancy between duplicate genes, there is a degree of non-redundant functions for some RPs [6]; [7]; [8].
Plants have even more gene members encoding a single RP than yeast. In Arabidopsis thaliana, each RP is encoded by a small gene family with several family members, which share between 65% and 100% amino acid sequence identity [9]. Assessment of cognate EST (expressed sequence tag) numbers of RP genes suggested that RP gene family members were differentially expressed in Arabidopsis [9]. Microarray data also revealed that transcripts of genes encoding RPs within the same family were accumulated at different levels in Arabidopsis [10]. Under various stimuli, while the transcript level for most ribosomal protein genes remain unchanged, some RP genes show significantly increased or decreased transcript levels [10]. Many studies have investigated the functional consequence of absent/reduced expression of a single RP paralogue in Arabidopsis.
Disruptions in any one of the RP protein genes, RPL3A, RPL8A, RPL19A, RPL23C, RPL40B, and RPS11A, is embryo lethal [11]. Less severe phenotypes were reported for mutations in several other RPs.
In Arabidopsis, the RPL23a family consists of two paralogues (RPL23aA and RPL23aB), which encode proteins with 95% amino acid identity. Both paralogues are transcribed and translated, and protein products of either paralogue can be incorporated into the ribosome [17]; [18]. Knock-down of the RPL23aA gene through RNAi results in severe developmental defects, whereas knock-down, or even knock-out, of RPL23aB has no phenotypic consequences [19], which could be the basis for the argument that RPL23aA and RPL23aB had specialized functions [10]; [11]; [19]; [20]; [21].
With the general question of why plant RPs are encoded by paralogous genes in mind, we sought to study the functional relationship between RPL23aA and RPL23aB. With T-DNA insertion mutants in RPL23aA and RPL23aB, we found a rare non-allelic non-complementation phenomenon, indicating that two (out of the four) gene copies in a diploid plant being functional is not sufficient to produce a wildtype phenotype. We showed that expression of RPL23aB driven by the RPL23aA promoter can rescue the phenotype of rpl23aa, demonstrating that RPL23aA and RPL23aB proteins are functionally equivalent. Furthermore, interrogation of RNA-seq data from several developmental stages and in different organs showed that although the level of RPL23aA transcripts is much higher than that of RPL23aB, the fluctuations in expression of the two genes were well matched, suggesting that these two genes were coordinately regulated. These results revealed that duplicated RPL23a genes contribute to ribosome dosage necessary for plant growth and development.

Plant material and growth conditions
Arabidopsis thaliana wild type Columbia-0 (Col-0) and the T-DNA insertion lines, SALK_005448 (named here rpl23aa) and SAIL_597_B08 (named here as rpl23ab), were obtained from the Arabidopsis Biological Resource Center (ABRC). Seeds were first treated for 2 minutes in 75% ethanol, then treated for 6 minutes in commercial bleach and rinsed at least 3 times with sterile distilled water.
Solid medium consisted of 2.2 g/L Murashige and Skoog basal salt mixture (Phyto Tech Labs), 10 g/L sucrose, and 8 g/L agar. pH was adjusted to 5.6 with KOH before autoclaving. When required, BASTA (GOLDBIO) was added at a final concentration of 125 µg/L. Seeds were sown in a water suspension, using a 1.5 mL pipette tip, in 150 mm Petri dishes filled with 120 ml of solid culture medium, at a density of 150 regularly spaced seeds per plate. Once inoculated, the Petri dishes were sealed with Micropore Scotch 3M surgical tape, which prevented contamination but allowed gaseous exchange, and placed in 4℃ for 24 h. Growth was allowed to proceed at 22℃ in Percival tissue culture chambers under long day conditions (16 hours light and 8 hours dark). 10-day seedlings were then transplanted to pots containing a 1:2:2 mixture of perlite, vermiculite and soil at 22℃ under long day conditions from a combination of incandescent and fluorescent lamps (10000 lux). Plants were watered twice a week with nutrient solution.
RNA isolation and RT-PCR 5 50 mg seedlings from 14-day-old Col-0, rpl23aa, and rpl23ab were harvested and immediately frozen in liquid nitrogen. RNA was extracted using RNAiso Plus(TAKARA BIO INC). In the elution step, RNA was resuspended in DEPC-treated water. cDNA was obtained by reverse transcription of 1 µg of RNA with the PrimeScriptTMRT reagent Kit with gDNA Eraser(TAKARA BIO INC).

Plasmid construction and generation of transgenic plants
In order to construct the pRPL23aA::RPL23aA and pRPL23aB::RPL23aB plasmids, we amplified the fulllength coding sequences (including the promoter region) of RPL23aA (AT2G39460) and RPL23aB (AT3G55280) genes from Col-0 genomic DNA using Phusion polymerase (Thermo Scientific). The primers used are shown in Table S1 (Additional file 10). The amplified DNA sequences were cloned in pEG301 [22] to result in pRPL23aA::RPL23aA and pRPL23aB::RPL23aB. The plasmids were used to transform rpl23aa. For pRPL23aA::RPL23aB construction, the promoter region (about 1.5 kb) of RPL23aA plus the coding region of RPL23aB were synthesized by a commercial company(GENEWIZ SuZhou), then the synthesized DNA fragment was sequenced and was cloned in pEG301. The promoter regions of RPL23aA (AT2G39460) and RPL23aB (AT3G55280) were cloned into pMDC162 [22] to generate the plasmids pRPL23aA::GUS and pRPL23aB::GUS, which were then used to

Characterization of rpl23aa and rpl23ab mutants
The Arabidopsis genome contains two RPL23a paralogous genes RPL23aA (At2g39460) and RPL23aB (At3g55280), which encode proteins with 95% amino acids identity (see Additional file 1). We acquired T-DNA insertion lines of RPL23aA and RPL23aB, namely SALK_005448 and SAIL-597-B08, respectively (hereafter referred to as rpl23aa and rpl23ab). PCR-genotyping confirmed that both rpl23aa and rpl23ab are homozygous T-DNA insertion alleles (see Additional file 2). Sequencing results revealed that rpl23aa contains a T-DNA insertion in the 3' UTR region, 10 bp downstream of the stop codon of the RPL23aA gene (Fig. 1A), while rpl23ab contains a T-DNA insertion in the second exon of RPL23aB (Fig. 1B). A semi-quantitative RT-PCR assay was used to detect transcripts from RPL23aA and RPL23aB in these T-DNA lines. As shown in Fig. 1C, the 3' region around the stop codon of the RPL23aA mRNA was disrupted in the mutant. Because the majority of the RPL23aA mRNA from the T-DNA line was intact, we suspect that SALK_005448 is a hypomorphic allele. rpl23ab is likely null mutant, because no RPL23aB mRNA was detected (Fig. 1D). Absence of dosage compensation by RPL23aA in Arabidopsis was reported following loss of RPL23aB [24]. As shown in Fig. 1D, there is also no dosage compensation by RPL23aB in the rpl23aa mutant.
The rpl23aa mutant exhibits pleiotropic defects, including pointed leaves, retarded root growth, and reduced plant size (Fig. 2B). These phenotypes are similar to those of a previously reported RNAi line [19]. An incompletely penetrant tricotyledon phenotype (less than 5% of the total population) was observed in rpl23aa mutant plants (see Additional file 3). However, we didn't observe appreciable defects in terms of growth rate, morphology, flowering or fertility in the rpl23ab mutant (Fig. 2D), which is consistent with published work [24]. We amplified genomic DNA encompassing the promoter plus the coding region of RPL23aA from wild-type plants and fused it to the sequence encoding the HA epitope tag. When this transgene was introduced into rpl23aa, the developmental defects were fully rescued (Fig. 2C), suggesting that dysfunction of RPL23aA was responsible for the developmental defects in rpl23aa.

RPL23aA and RPL23aB are dosage-dependent genes
In order to study the genetic interaction between RPL23aA and RPL23aB, we crossed rpl23aa with rpl23ab. To our surprise, the doubly heterozygous plants (RPL23aA/rpl23aa; RPL23aB/rpl23ab) in the F1 progeny all have pointed first true leaves (Fig. 3B). Siliques of the doubly heterozygous plants are much shorter than siliques of rpl23aa or rpl23ab (Fig. 3I). We dissected siliques from RPL23aA/rpl23aa; RPL23aB/rpl23ab plants and found many aborted ovules (Fig. 3G and 3H). An F2 population was generated by selfing the above F1 plants. We genotyped 144 F2 plants but did not find double homozygous (rpl23aa /rpl23aa; rpl23ab /rpl23ab) plants. In fact, we did not even detect any genotypes with a single functional allele from either gene (RPL23aA/rpl23aa; rpl23ab /rpl23ab or rpl23aa /rpl23aa; RPL23aB/rpl23ab) ( Table 1), although these genotypes are collectively expected to appear in 31.25% (5 out of 16) of the F2 plants. We suspected this non-allelic non-complementation phenomenon between rpl23aa and rpl23ab is probably due to gene dosage effects. Table 1 Genetic interactions between rpl23aa and rpl23ab.

Genotype
First leaf phenotype  lines uncovered a ubiquitous expression pattern for both genes with particularly intense GUS staining in young and actively proliferating tissues, such as developing leaves, floral buds and root apices (Fig. 4). Similar expression patterns of RPL23aA and RPL23aB support our hypothesis that the nonallelic non-complementation phenomenon between these two genes is the consequence of overlap in expression (and function) of RPL23aA and RPL23aB in the same cells.
RPL23aA and RPL23aB proteins are functionally equivalent It has been reported that some paralogous ribosomal proteins have evolved specialized functions in yeast [7]. As mentioned above, dysfunction of RPL23aA results in severe developmental defects, whereas knock-out of RPL23aB has no phenotypic consequences in Arabidopsis. It's natural to assume that these two paralogous ribosomal proteins have undergone functional specialization.
We RPL23aA and RPL23aB genes are transcribed in a concerted manner with higher expression level of RPL23aA than RPL23aB Since the above results indicated that RPL23aA and RPL23aB proteins have equivalent function, we suspected that the difference in phenotype between rpl23aa and rpl23ab is due to the difference in the expression levels of these two genes. The expression of RPL23aA may be much higher than RPL23aB, so the impacts on ribosomes by the rpl23aa mutation are higher than the rpl23ab mutation thus leading to much severe morphological defects. We compared the transcript levels of RPL23aA and RPL23aB at different developmental stages and in different organs by analyzing published RNAseq data [25]. As shown in Fig. 6 and Figure S5(Additional file 5), transcript levels of RPL23aA are much higher than those of RPL23aB at all developmental stages and in all the examined tissues.
Strikingly, the spatial and temporal patterns of expression of these two paralogous genes are well matched, suggesting that they are similarly regulated at differently developmental stages in all examined tissues. ACT2, which is a house keeping gene, was included as a control. Transcript levels of RPL23aB are higher than ACT2 in some organs, and total amount of PRL23a transcripts is much higher than ACT2 in most examined organs (Fig. 6C, 6E), indicating that RPs are in great demand for plant development. In this study, we characterized the RPL23a gene family containing two highly homologous family members. The hypomorphic T-DNA insertion allele of RPL23aA exhibits pleiotropic defects. However, knock-out of RPL23aB has no appreciable phenotypic impacts. We crossed mutants of RPL23aA and RPL23aB and found a non-allelic non-complementation phenomenon in their F1 progeny. This phenomenon is also found in other RP coding gene families such as RPL5 [26], RPL36a [27], and RPS6 [28]. However, mutations in the paralogues within RPL5, RPL36a, and RPS6 families caused almost the same phenotype, indicating that the paralogues are functionally equivalent. In the case of the RPL23a family, phenotypes of the single mutants suggest unequal functions of the two paralogues.
The non-allelic non-complementation phenomenon may be due to a dosage problem -reduced dosage at one of the paralogues still supports the wild phenotype but simultaneous reduction of dosage at both paralogues could not sustain the wild phenotype. For the dosage effect hypothesis to be true, there must be at least some overlap in the expression of the gene family members. Indeed, promoter-GUS experiments demonstrated that both RPL23aA and RPL23aB were ubiquitously expressed.
Phenotypical differences between members of an RP within a family might result from diversification of protein function or variation in levels and patterns of expression. We demonstrated that RPL23aA and RPL23aB proteins had equal function, as expression of RPL23aB driven by the RPL23aA promoter could rescue the phenotype of the rpl23aa mutant. We found that the expression level of RPL23aA was much higher than that of RPL23aB according to the publicly available RNA-seq data. Thus, the difference in expression levels might be the reason why disruption of RPL23aA and RPL23aB had different consequences. It is interesting that despite the difference in expression levels, the temporal and spatial patterns of expression of the two paralogous genes were nearly identical. These results suggested that RPL23aA and RPL23aB genes are transcribed in a coordinated manner.
Posttranscriptional and translational regulation may also play a role in RPL23aA and RPL23aB expression [29]. Subcellular localization specialization could be another factor that causes differences in functional effects between paralogous RPs [30]. Previous studies revealed that both of RPL23aA

Consent for publication
Not applicable

Availability of data and materials
The original RNA-seq data of A. thaliana different organs and developmental stages were downloaded from NCBI Sequence Read Archive (project ID PRJNA314076 for samples except meristem and project ID PRJNA268115 for the meristem samples).
analysis and interpretation of the data and in writing the manuscript.

Author contributions
XC and BM designed experiments; WX, XZC, and CZ carried out experiments; WX, JZ, and TL analyzed the RNA-seq data; XW and LL analyzed experimental results; XW, XZC, XC and BM wrote the manuscript. All authors agree to be accountable for the content of the work.  The non-allelic non-complementation phenomenon between rpl23aa and rpl23ab. 9-day-old plants of (A) rpl23aa × Col-0 (F1 generation), (B) rpl23aa × rpl23ab (F1 generation).