Synopsis of the SOFL Plant-Specific Gene Family

SUPPRESSOR OF PHYB-4#5DOMINANT (sob5-D) was previously identified as a suppressor of the phyB-4 long-hypocotyl phenotype in Arabidopsis thaliana. Overexpression of SOB5 conferred dwarf phenotypes similar to those observed in plants containing elevated levels of cytokinin (CK) nucleotides and nucleosides. Two SOB-FIVE- LIKE (SOFL) proteins, AtSOFL1 and AtSOFL2, which are more similar at the protein level to each other than they are to SOB5, conferred similar phenotypes to the sob5-D mutant when overexpressed. We used protein sequences of founding SOFL gene family members to perform database searches and identified a total of 289 SOFL homologs in genomes of 89 angiosperm species. Phylogenetic analysis results implied that the SOFL gene family emerged during the expansion of angiosperms and later evolved into four distinct clades. Among the newly identified gene family members are four previously unreported Arabidopsis SOFLs. Multiple sequence alignment of the 289 SOFL protein sequences revealed two highly conserved domains; SOFL-A and SOFL-B. We used overexpression and site-directed mutagenesis studies to demonstrate that SOFL domains are necessary for SOB5 and AtSOFL1’s overexpression phenotypes. Examination of the subcellular localization patterns of founding Arabidopsis thaliana SOFLs suggested they may be localized in the cytoplasm and/or the nucleus. Overall, we report that SOFLs are a plant-specific gene family characterized by two conserved domains that are important for function.

displayed reduced apical dominance, conferred smaller rosette leaves, retarded root growth and delayed senescence phenotypes (Zhang et al. 2006(Zhang et al. , 2009. In addition, these transgenic lines also contained elevated levels of cytokinin (CK) nucleotides and nucleosides; transzeatin riboside (tZR), trans-zeatin riboside monophosphate (tZRMP), and N 6 -(Δ 2 -isopentenyl)adenosine monophosphate (iPRMP) (Zhang et al. 2006(Zhang et al. , 2009. These results suggested that the founding SOFL gene family members may have similar or overlapping CK-related functions. However, the specific details of what roles, if any, intermediate CK nucleotide/nucleoside species play during plant growth and development are not fully understood and will have to be further investigated in the future. Zhang et al., (2006), reported that the three founding SOFLs constituted a small, novel and intron-less Arabidopsis thaliana gene family. In addition, SOFL homologs and expressed sequence tags were also identified in a few monocotyledonous and dicotyledonous species; Malus x domestica, Brassica napus, Oryza sativa and Populus trichocarpa (Zhang et al. 2006). Due to few sequenced genomes at the time it was difficult to obtain a comprehensive phylogenetic overview of the SOFL gene family. Nonetheless, based on these preliminary results we speculated that SOFL homologs also existed in genomes of other eukaryotic and/or prokaryotic species (Zhang et al. 2006).
Since Arabidopsis SOFLs were first identified, more genome sequence information has been released (Fang et al. 2013;Wheeler et al. 2008;Kersey et al. 2016). We have taken a phylogenetic approach to update our understanding of the evolution and diversification of the SOFL gene family. Database searches using the three founding SOFL members' proteins sequences as queries revealed at least 289 SOFL homologs in 89 angiosperm species. During this process, we also discovered that the Arabidopsis thaliana genome contains four additional SOFLs that were previously unreported. In addition, phylogenetic analysis showed that the 289 SOFL homologs evolved into four main clades. Finally, to gain further insights into this gene family, we performed subcellular localization, tissue expression and domain mutational studies in three founding SOFL members. This updated synopsis of the SOFL gene family provides basic background information needed to design future studies to eventually fully characterize this novel gene family.

Plant materials and growth conditions
All Arabidopsis thaliana lines used in this manuscript are in the Columbia (Col-0) background. For phenotypic analysis, all plants were grown on soil in pots in growth chambers set at 21°under white light (200 mmol m-2 sec-1 ) and 60-70% humidity, and under 16 hr light and 8 hr darkness.
Sequence alignment and phylogenetic analysis SOB5, AtSOFL1 and AtSOFL2 protein sequences were downloaded from NCBI (Wheeler et al. 2008;Sayers et al. 2009). The sequences were used as queries to search NCBI and Phytozome (Goodstein et al. 2011) databases for SOFL homologs using BLASTP option using a , 1 X 10-2 cut-off value. The amino acid sequences were aligned using Probalign V1.3 (Roshan 2014) on CIPRES Science Gateway Server (Miller et al. 2015). Bayesian inference analysis was performed with the Mr. Bayes 3.2.1 on XSEDE tool on CIPRES Science Gateway for 30 million generations with 8 chains to run. Generations were sampled every 10,000 generations with the first 25% generations used as a burn-in.

Sequence logo analysis
Sequence logo analysis of conserved SOFL-A and SOFL-B domains was performed using the online WebLogo tool (Crooks et al. 2004). The 289 SOFL sequences retrieved from NCBI (Sayers et al. 2009) and Phytozome (Goodstein et al. 2011) were used to generate the sequence logo using default settings.

Overexpression and point-mutation analysis of SOFL domains in SOB5 and AtSOFL1
Site-directed point mutations were generated using the QuikChangeÒ Lightning mutagenesis kit according to manufacturer's instructions (Agilent Technologies). Mutagenesis primers were designed using Agilent's web-based QuikChangeÒ Primer Design Program (http://www.genomics.agilent.com/ primerDesignProgram.jsp) (Table S1). GatewayÒ compatible entry vectors, pENTR223 (ABRC), carrying SOB5 and AtSOFL1 coding sequences, were used as templates in mutagenesis PCR reactions. Each point mutation was confirmed via sequencing. Once mutagenesis was confirmed, SOB5 or AtSOFL1 coding sequences carrying point mutations were cloned via LRÒ reactions into pCHF3 (Zhang et al. 2006) and pEarlyGate100 (Earley et al. 2006) binary vectors, respectively. Destination binary vectors carrying mutated SOB5 and AtSOFL1 genes under control of the constitutive 35S promoter were transformed into Arabidopsis thaliana Col-0 wildtype plants using the floral dip method and the Agrobacterium tumafaciens strain GV3101 (Clough and Bent 1998). Transformants were screened on 1/2· Linsmaier and Skoog media plates containing 25 mg L -1 Basta (pEarleyGate100) and 30 mg L -1 kanamycin (pCHF3). Homozygous lines were identified in the T 3 generation. Three independent transgenic lines representing each construct were selected for analysis. All plants were grown and analyzed in a growth chamber.

RNA extraction and real-time PCR
Total RNA was extracted from tissues pooled from three independent plants (n = 3 for all tissues) of six-day old seedlings, 10-day old juvenile plants, rosette leaves from a 20-day old plant, floral tissue, siliques, a 25-day old adult plant, and roots from a 25-day old plant using the Qiagen RNeasyÒ Mini Kit (Qiagen). On-column DNase digestion was performed using the RNase-Free DNase Set (Qiagen, Valencia, CA). Complementary DNA (cDNA) was further generated using the iScriptÒ Reverse Transcription Supermix for RT-qPCR (Bio Rad, Hercules, CA). Identical amounts of cDNA input were used in PCR reactions using either ubiquitin10 (UBQ10) or gene-specific primers (Table S1). Amplification using UBQ10 primers was done in 30 cycles while 45 cycles were used for gene-specific primers. Negative control, no reverse transcriptase (No RT), samples were prepared using the same reaction conditions and reagents (minus reverse transcriptase enzyme) used to make cDNA.

Data availability
Genetic material used in this manuscript is available upon request. Primers used in the study are listed in Table S1. Total numbers of SOFL homologs identified in various species are listed in Table S2. Table S3 shows results from protein localization prediction analysis. Alternatively-spliced AtSOFL1 product sequences (AtSOFL1.1 and AtSOFL1.2) are provided in File S2. Figure S1 shows a cartoon and gel image of AtSOFL1's alternative splicing products. Sequences extracted from database searches using the founding SOFL protein sequences as queries are provided in File S1.

RESULTS
SOFLs are a plant-specific gene family To identify SOFL homologs, we performed BLASTP searches against the NCBI database (Sayers et al. 2009) and Phytozome database (Goodstein et al. 2011) using SOB5, AtSOFL1 and AtSOFL2 sequences as queries. We extracted 289 SOFL homologs from 89 flowering plant species, of which 15 were monocotyledons and 74 were dicotyledons (Table S2). To gain insight into the evolutionary history of the SOFL gene family, we next performed phylogenetic analysis using the 289 SOFL protein sequences extracted via database searches. Our analysis suggested that SOFLs likely evolved into four major clades: Clade I, II, III and IV (Figure 1a and 1b). We hypothesize that SOFLs within each clade have greatly expanded through evolution, with Clades I and -IV showing the largest expansion ( Figure 1a and 1b). However, since the tree was constructed only with the existing SOFL genes in the plant species examined and no outgroup used, we cannot rule out gene loss events. Annotation data from NCBI (Sayers et al. 2009), Phytozome (Goodstein et al. 2011) and individual species genome databases revealed that 60 SOFL homologs contained introns and another 226 did not (Table S2). There was no annotation data for Arabis alpina and Erythranthe guttata (Mimulas guttatus) SOFL homologs.

Arabidopsis thaliana genome contains seven SOFL members
It was previously reported that the Arabidopsis genome contained three SOFL members (Zhang et al. 2006(Zhang et al. , 2009. However, latest database search results revealed that the Arabidopsis thaliana genome contained a total of seven SOFL members (File S1). In addition, we also identified seven SOFLs in Arabidopsis lyrata (File S1). We have designated the newly identified Arabidopsis thaliana SOFL genes as AtSOFL3 (AT3G30580), AtSOFL4 (AT5G38790), AtSOFL5 (AT4G33800) and AtSOFL6 (AT1G58460). SOB5 is in Clade IV of the phylogenetic tree ( Figure 1a and 1b), compared to AtSOFL1 and AtSOFL2 which were in Clade I. Interestingly, the newly identified SOFL members, AtSOFL3, AtSOFL4, AtSOFL5 and AtSOFL6 are in Clade IV (Figure 1b), suggesting that genetic redundancy may exist among these genes and SOB5. All Arabidopsis SOFLs are intron-less except AtSOFL1 (recently classified by TAIR as intron-containing), AtSOFL5 and AtSOFL6.
We next assessed the expression patterns of Arabidopsis thaliana SOFL homologs. RNA was extracted from seedlings, juvenile plants, adult rosette leaf, floral structures, siliques, an entire flowering plant and adult plant roots. SOB5, AtSOFL1 and AtSOFL2 transcripts were detected in all samples tested except roots (Figure 2). AtSOFL3, AtSOFL4, AtSOFL5 and AtSOFL6 were expressed at varying levels in seedlings, juvenile plants, flowers, siliques and whole plant ( Figure 2). AtSOFL3, AtSOFL4, AtSOFL5 were the only ones detected in roots.
AtSOFL6 showed little to no expression in rosette leaves and roots, and similarly, AtSOFL5 showed little to no expression in seedlings and rosette leaves (Figure 2). AtSOFL4 was expressed in all samples tested but had the lowest transcript levels overall. The varying transcript accumulation levels and differential tissue expression patterns may suggest unique functional roles among the seven Arabidopsis thaliana SOFLs during plant development ( Figure 2). Surprisingly, during efforts to amplify a full-length AtSOFL1 (At1g26210) cDNA we identified a previously unreported splice variant. We have designated the original 447 bp transcript AtSOFL1 (At1g26210.1), and the newly identified 771bp second transcript as (At1g26210.2) ( Figure S1a, b).

SOFL homologs are characterized by two conserved domains
Multiple sequence alignment analysis is a useful tool used to infer relationships among gene family members. Alignment data can provide functional information through the identification of key conserved domains and other important features. We aligned SOFL homologs protein sequences and identified two conserved domains in the N-terminal region (Figure 3a), which were previously reported by Zhang et al. (2006), albeit based on fewer protein sequences. We have designated the two conserved domains, SOFL-A and SOFL-B (Figure 3a). When only Arabidopsis SOFL sequences were aligned ( Figure 3b) and compared to an alignment of all 289 SOFL homologs (Figure 3c, 3d) we observed slight differences in the SOFL-B domain. In an alignment of Arabidopsis SOFL sequences only, SOFL-B domain contains eight 100% conserved residues SM·SDASS·P ( Figure 3b). However, the same domain only contains four 100% conserved residues (S··SDA) when all 289 SOFLs homologous sequences were aligned ( Figure 3d). In general, based on all SOFL ortholog sequences identified so far, SOFL-A domain contains an SGWT·Y motif and SOFL-B domain contains an S··SDA motif (· = amino acid residue that is not 100% conserved) (Figure 3c, d). There were no areas of high sequence similarity in the C-terminal region, except for an increased presence of basic amino acid residues in most SOFL homologs (Table S3).
Conserved amino acid residues in SOFL domains are required for the manifestation of SOB5 and AtSOFL1 overexpression phenotypes Previously, Zhang et al., (2009), investigated the requirement of some of the 100% conserved residues in SOFL-A and SOFL-B domains for the manifestation of AtSOFL2's overexpression phenotype via sitedirected mutagenesis. Constructs carrying AtSOFL2 gene harboring individual point mutations in each of the two conserved domains were used to generate plants overexpressing an aberrant protein. Resultant transgenic mutant plants overexpressing AtSOFL2 gene harboring T21I and D80N point mutations lost the phenotype that is typically associated with the overexpression of the wild type AtSOFL2 gene (Zhang et al. 2009). These results implied that the conserved amino acid residues were important for the manifestation of AtSOFL2 overexpression phenotype.
To further investigate the biological importance of some of the 100% conserved residues in the other two founding SOFLs, SOB5 and AtSOFL1, we generated a series of site-directed point mutations in their respective SOFL-A and SOFL-B domains. Wild-type Arabidopsis thaliana plants were separately transformed with constructs carrying mutations in SOB5 (T21I and P61R) and AtSOFL1 (T23I and P84R) coding sequences. Transgenic plants overexpressing SOB5 (T21I, P61R) and AtSOFL1 (T23I, P84R) mutated genes lost the dwarf/ semi-dwarf phenotypes typically observed when wild-type versions were overexpressed (Figure 4a, b). We also performed site-directed mutagenesis on non-conserved amino acid residues; D33H and D53H for SOB5 and AtSOFL1, respectively, which are not located in the two conserved domains (Figure 3b). Transgenic plants overexpressing mutated SOB5 (D33H) and AtSOFL1 (D53H) exhibited phenotypes similar to those observed when wild-type genes are constitutively expressed (Figure 4a, b) (Zhang et al. 2006(Zhang et al. , 2009. These data suggest that less conserved amino acid residues are not required for the overexpression phenotypes.

SOFL subcellular localization
We used two publicly available online prediction programs, Wolf PSORT (Horton et al. 2007) and SeqNLS (Lin and Hu 2013), to gain insights into subcellular localization patterns of the 289 SOFL homologs. Wolf PSORT converts protein amino acid sequences into numerical localization features based on sorting signals, amino acid composition and functional domains such as DNA-binding motifs (Horton et al. 2007). SeqNLS uses a sequential pattern mining algorithm to effectively identify potential nuclear localization signals (NLS) in protein sequences (Lin and Hu 2013). Wolf PSORT predicted 274 SOFL homologs to localize to the nucleus, with the remainder predicted to localize to chloroplast, cytoplasm,  (Table S3). In contrast, the SeqNLS program detected nuclear localization signals in only 156 SOFLs with a statistically significant prediction score of at least 0.89 (Table S2) (Lin and Hu 2013). Interestingly, SeqNLS did not detect nuclear localization signals in 13 SOFLs which were predicted to localize to the nucleus by Wolf PSORT (Table S3). 119 SOFLs scored below the default cutoff threshold to be classified as containing a NLS. Overall, at least 156 SOFL homologs were predicted to localize to the nucleus by both Wolf PSORT and SeqNLS algorithms (Table S3).
To further assess some of the subcellular prediction data, we examined the intracellular distribution of the founding Arabidopsis thaliana SOFL members. We fused SOB5, AtSOFL1 and AtSOFL2 to the carboxyl end of a cyan fluorescent protein (CFP) and transiently expressed the fusion proteins in onion epidermal cells. CFP-SOB5, CFP-AtSOFL1 and CFP-AtSOFL2 fluorescent signals were observed in both the cytoplasm and nucleus of the onion epidermal cells ( Figure 5). We also examined whether the fluorescent signals of all three fusion proteins were localized to the cytosol and not the cell wall. This was achieved by inducing plasmolysis through exposure of onion epidermal peels to 0.8 M mannitol. Ensuing the mannitol treatment, CFP signal remained restricted to the edges of the plasma membrane and/or cytoplasm, suggesting that CFP-SOB5, CFP-AtSOFL1 and CFP-AtSOFL2 fusion proteins were not localized to the cell wall.

SOFL homologs emerged in angiosperms
The founding members of the SOFL gene family were first identified via an activation tagging screen in Arabidopsis thaliana (Zhang et al. 2006). Based on the limited number of sequenced genomes available at the time, it was initially concluded that SOB5, AtSOFL1 and AtSOFL2 were a small three-member intron-less Arabidopsis thaliana gene family. To expand on earlier studies and acquire a basic level understanding of this poorly characterized gene family, we took advantage of an increasing number of sequenced genomes to perform database searches using the founding Arabidopsis thaliana founding SOFL members as queries. We have retrieved a total of 289 SOFL sequences from 89 angiosperm species (Table S2). However, we cannot rule out the possibility that  Partial N-terminal amino acid sequence alignment of Arabidopsis thaliana SOFL proteins. The alignment was obtained using JalView Probcons with Defaults program (Malek 2001;Zhang 2003). Semi-conserved amino acids are indicated with a light purple shade. Conserved amino acids are indicated by a darker purple shade. Red squares indicate amino acid residues that were selected for site-directed mutagenesis. WebLogo© analysis of conserved, (c) SOFL-A domain, and (d) SOFL-B domain, using all 289 SOFL protein sequences (Lin and Hu 2013). Red asterisk denotes 100% conserved amino acid residues among all 289 SOFL sequences.
SOFL homologs are present in unsequenced organisms or were lost in other genomes. In addition, we expect additional SOFL homologs to be identified in other species as more genomes are sequenced. Furthermore, as genome sequence annotation improves, it is possible that some of the SOFL homologs currently classified as intron-less or introncontaining may be re-categorized in the future.
No SOFL homologs were identified in Volvox cartari (green algae), Chlamydomonas reinhardtii (green algae), Ostreococcus lucimarinas (single-celled water algae), Micromonas pusilla (water algae), Physcomitrella patens (non-vascular bryophyte, moss) and Selaginella moellendorffii (member of an ancient vascular plant lineage) (Lamesch et al. 2012), strongly suggesting that SOFLs may have emerged after the separation of lycophytes, pterophytes and seed plants. Out of the four main plant groups: bryophytes, seedless vascular (pteridophytes/lycophytes), gymnosperms and angiosperms, SOFLs have so far only been identified in angiosperms. We hypothesize that the SOFL gene family emerged and expanded during the evolution of flowering plant species.

SOFL gene family is comprised of intron-containing and intron-less genes
Approximately 20% of SOFL homologs identified so far contain introns and 80% are intron-less (Table S2). These results are inconsistent with overall data from rice and Arabidopsis genomes which revealed the presence of only 19.9% and 21.7% intron-lacking genes, respectively (Jain et al. 2008). Our latest database search results also showed that Arabidopsis thaliana contains seven SOFL genes, in contrast to previous reports of three intron-less family members (Zhang et al. 2006(Zhang et al. , 2009. Out of the four newly discovered Arabidopsis SOFLs, only AtSOFL5 and AtSOFL6 contain introns, a departure from the previous designation of the Arabidopsis SOFL family as being comprised of members lacking introns (Zhang et al. 2006(Zhang et al. , 2009. Genes lacking introns are a characteristic of prokaryotes and are a useful resource for studying the evolution of gene architecture in eukaryotes, but information on their biological significance remains limited (Yan et al. 2014). Considering that SOFLs seem to have emerged during the evolution of angiosperms, it is surprising that the majority of the genes in this family are intronless, a trait expected in early species (Zou et al. 2011). On the other hand, this result can be explained by the fact that majority of the intronless SOFLs could have potentially arisen because of gene duplication events (Yan et al. 2016;Lecharny et al. 2003). Gene prediction and gene functional studies data suggests that intron-less genes may play unique roles in growth and development, including translation and energy metabolism in maize, rice and Arabidopsis as well as cell envelope and amino acid biosynthesis in rice and Arabidopsis (Yan et al. 2014;Jain et al. 2008). In addition, gain-of-function studies from Zhang et al. (2006Zhang et al. ( , 2009 suggested that the three-founding intron-lacking SOFLs may be involved in CK-related functions. According to Zhao et al. (2014) the presence of introns enhances the transcription of associated genes. Therefore, to further explore the biological significance of intron-less genes, studies that include gain-of-function, loss of function, gene expression pattern, gene expression level analysis and subcellular localization will need to be performed in the future.
SOFL-A and SOFL-B domains are important for SOB5 and AtSOFL1's overexpression phenotypes Previously, Zhang et al. (2009) demonstrated via site-directed mutagenesis experiments that certain amino residues in the SOFL domains were necessary for the manifestation of AtSOFL2's overexpression phenotype. In our study, we have similarly shown that specific conserved amino acid residues in SOFL-A and SOFL-B domains are also necessary for both SOB59s and AtSOFL1's overexpression phenotypes ( Figure 4). These results, together with sequence logo analysis showing that certain amino acid residues in SOFL-A and SOFL-B domains are 100% conserved in all 289 SOFLs, further strengthen the hypothesis that they are important for function. These results should, however, be interpreted with caution because point mutations can potentially cause proteins to fold incorrectly. Nonetheless, our results demonstrated, at least, that not all point mutations lead to a loss of protein function. Future experiments in which all conserved residues are replaced with alanines or entire domains are deleted may provide answers to whether all conserved residues in SOFL-A and SOFL-B domains are important for function. It is not yet clear what biological role these highly conserved domains play. However, conserved domains typically play crucial roles in protein-protein interactions, DNA binding, and other important cellular processes.
Founding Arabidopsis thaliana SOFL members localize to the nucleus and cytoplasm To begin to examine the intracellular distribution of SOFLs homologs we first used nuclear localization signal (NLS) detection and sub-cellular localization programs, SeqNLS (Lin and Hu 2013) and Wolf PSORT (Horton et al. 2007). The two programs predicted that majority of SOFLs contained NLSs and may localize to the nucleus among other cellular locations and organelles (Table S3). This hypothesis was further supported by CFP-SOFL protein fusion subcellular localization experimental data, which showed that SOB5, AtSOFL1 and AtSOFL2 localize to the nucleus and the cytosol ( Figure 5). However, SOB5 subcellular localization results were at odds with findings by Zhang et al., (2006) who reported that SOB5 only localized to the cytoplasm and/or plasma membrane, but not the nucleus. These contrasting conclusions could be due to a different experimental design and fluorescence detection methods used in both studies. The SOB5-GFP C-terminal fusion protein in Zhang et al. (2006) was missing five C-terminal amino acids from SOB5, whereas in our case we used a full-length SOB5 protein fused to the carboxyl end of the CFP tag. Second, we used confocal microscopy to detect CFP fluorescence, which is more reliable compared to fluorescence microscopy used by Zhang et al. (2006). In addition, we generated and analyzed N-terminal fusions in contrast to the C-terminal fusion described in Zhang et al., (2006). C-terminal and N-terminal tagged proteins have been shown to display opposite  localization patterns (Palmer and Freeman 2004), a possibility that may also explain SOB5-GFP and CFP-SOB5 localization pattern. In the future, both C-and N-terminal fusions of each protein should be examined to avoid such potential problems.
Even though we used a plasmolysis assay to show that the three founding SOFLs were not localized to the cell wall, we could not distinguish whether they were localized in the plasma membrane, cytosol or both. To try and answer this question, we used a bioinformatics approach by running the 289 SOFL homolog sequences against web-based transmembrane protein topology prediction algorithms. TMHMM (Kahsay et al. 2005) and TMMOD (Kahsay et al. 2005). Both hidden Markov prediction models, did not predict any transmembrane helices in all 289 SOFL ortholog sequences. This outcome could be verified experimentally by discriminating between cytosolic and cell membrane proteins using osmotic disruption of the protoplast vacuole in hypotonic solution (Serna 2005). This method results in the diffusion of the GFP signal from the cell periphery to the central part of the cell volume, an outcome that will not occur when the protein under study is attached to the cell membrane (Serna 2005). Overall, these data suggest that at least, the three founding Arabidopsis thaliana SOFLs and possibly several other SOFL homologs may localize to the nucleus and/ or cytoplasm.

Conclusion
The identification of at least 289 SOFL homologs creates an opportunity for the study and characterization of this novel gene family in at least 89 flowering plant species. A combination of gain-of-function, lossof-function, protein-protein interaction and CK quantitation studies will go a long way toward answering various questions regarding the function of Arabidopsis SOFLs raised by Zhang et al. (2006Zhang et al. ( , 2009. One critical question is whether CK nucleosides and nucleotides have biological activity, as suggested by Zhang et al., (2006Zhang et al., ( , 2009. Gainof-function and CK-quantitation studies in selected species may be used to test the hypothesis that overexpression of SOB5, AtSOFL1 and AtSOFL2 homologs can cause similar CK-related phenotypes reported by Zhang et al., (2006Zhang et al., ( , 2009. Results from such studies can then be compared to data from higher order null mutants in which putative redundant SOFLs from the same phylogenetic clades are knocked out. In addition, our site-directed mutagenesis studies suggested that conserved SOFL-A and SOFL-B domains were important for function. Similar studies involving mutagenesis of additional conserved amino acid residues in both Arabidopsis and other species will likely provide new functional details regarding these domains. Finally, even though our latest database searches suggest that SOFLs are a plant-specific gene family, the ever-increasing number of sequenced genomes will continue to test this hypothesis.