Babo1, formerly Vop1 and Cop1/2, is no eyespot photoreceptor but a basal body protein illuminating cell division in Volvox carteri

In photosynthetic organisms many processes are light dependent and sensing of light requires light-sensitive proteins. The supposed eyespot photoreceptor protein Babo1 (formerly Vop1) has previously been classified as an opsin due to the capacity for binding retinal. Here, we analyze Babo1 and provide evidence that it is no opsin. Due to the localization at the basal bodies, the former Vop1 and Cop1/2 proteins were renamed V.c. Babo1 and C.r. Babo1. We reveal a large family of more than 60 Babo1-related proteins from a wide range of species. The detailed subcellular localization of fluorescence-tagged Babo1 shows that it accumulates at the basal apparatus. More precisely, it is located predominantly at the basal bodies and to a lesser extent at the four strands of rootlet microtubules. We trace Babo1 during basal body separation and cell division. Dynamic structural rearrangements of Babo1 particularly occur right before the first cell division. In four-celled embryos Babo1 was exclusively found at the oldest basal bodies of the embryo and on the corresponding d-roots. The unequal distribution of Babo1 in four-celled embryos could be an integral part of a geometrical system in early embryogenesis, which establishes the anterior–posterior polarity and influences the spatial arrangement of all embryonic structures and characteristics. Due to its retinal-binding capacity, Babo1 could also be responsible for the unequal distribution of retinoids, knowing that such concentration gradients of retinoids can be essential for the correct patterning during embryogenesis of more complex organisms. Thus, our findings push the Babo1 research in another direction.

Volvox shows a complete germ-soma division of labor between approximately 16 asexual reproductive cells (gonidia) and approximately 2000 somatic cells. The small somatic cells are arranged as a monolayer at the surface of a transparent sphere of extracellular matrix (ECM), whereas the large reproductive cells are embedded in the ECM just beneath the somatic cells. The mortal somatic cells are equipped with two flagella and an eyespot apparatus for light perception. These cells are thus responsible for light-regulated movement of the spheroid, whereas the potentially immortal reproductive cells represent the germline.
In Volvox many cellular processes are light dependent, including photosynthesis, phototaxis, sexual reproduction, circadian clock, and developmental processes such as initiation of cell division, cellular differentiation, and cell cycle control (Starr, 1980;Kirk and Kirk, 1985;Kirk, 1998;Kianianmomeni and Hallmann, 2014). To explore the molecular basis of light perception, 13 putative photoreceptor genes have been identified so far in the Volvox genome (Kianianmomeni, 2015). Most of these genes show a celltype specific expression in somatic cells that could imply a function in phototaxis and light-dependent orientation. Only one of the putative photoreceptor genes shows a celltype specific expression in reproductive cells and it is even highly overexpressed in this cell-type (Ebnet et al., 1999;Kianianmomeni, 2015;Klein et al., 2017). This gene has previously been called vop1 (Ebnet et al., 1999;Kianianmomeni, 2015;Klein et al., 2017) in Volvox and its homolog in the related algae Chlamydomonas was called Chlamyrhodopsin, cop or cop1/2 (Deininger et al., 1995;Fuhrmann et al., 1999;Fuhrmann et al., 2001;Greiner et al., 2017). However, we now felt obliged to rename vop1 and cop1/2 as babo1 because, as described below, the corresponding protein is no opsin-based photoreceptor at the eyespot but a basal body protein. From this point forward, therefore, we use only the new name 'babo1'.
The babo1 gene of V. carteri and the corresponding Babo1 amino acid sequence (formerly Vop1) (Ebnet et al., 1999) have been identified through the amino acid sequence similarity between V.c. Babo1 and the earlier known Chlamydomonas reinhardtii Babo1 (formerly Cop1/ 2, Cre01.g002500; not to be confused with the E3 ubiquitinprotein ligase Cop1, Cre02.g085050) (Tilbrook et al., 2016). Even if three splice variants have previously been predicted for C.r. babo1 (formerly cop1/2), only one of these variants shows reasonable expression (Fuhrmann et al., 2003) and this is also the only splice variant that is indicated in the current version of the C. reinhardtii genome (v5.5; gene ID Gene ID: Cre01.g002500). Therefore, we here refer to the latter variant, which corresponds to GenBank entry AF295371. The Babo1 amino acid sequences of C. reinhardtii and V. carteri share 71% sequence identity and 83% similarity. Babo1 (Cop1/2) protein of C. reinhardtii was initially purified from eyespot membrane preparations based on the binding of [ 3 H]retinal (Deininger et al., 1995). Because only a single band appeared after gel electrophoresis of [ 3 H]retinal-labeled cell extracts and subsequent fluorography, Babo1 (Cop1/2) has been considered to be the first and only retinal-binding protein of C. reinhardtii (Kr€ oger and Hegemann, 1994;Deininger et al., 1995). Based on the fact that the polyene chromophore retinal is an integral component of rhodopsins and due to the much earlier result that the photoreceptor for phototaxis must be a rhodopsin (Foster et al., 1984), C.r. Babo1 was suggested to be the rhodopsin that triggers the organism's phototactic behavior (Kr€ oger and Hegemann, 1994;Deininger et al., 1995). However, later RNAi experiments showed that C. reinhardtii Babo1 (Cop1/2) is definitely not the photoreceptor that is required for phototaxis (Fuhrmann et al., 2001) and soon afterwards two retinal-binding channelrhodopsins were identified in C. reinhardtii (Nagel et al., 2002;Sineshchekov et al., 2002;Nagel et al., 2003;Suzuki et al., 2003), which actually mediate photomovement responses (Sineshchekov et al., 2002).
Once it was clear that Babo1 cannot be the photoreceptor for phototaxis, the intriguing question arose what was the real function of this protein. It also requires clarification that Babo1 proteins of C. reinhardtii and V. carteri were postulated to have, at most, four hydrophobic membranespanning segments (Deininger et al., 1995;Ebnet et al., 1999), even though all known rhodopsins have at least seven transmembrane helices (Gao et al., 2015). Our attention was also attracted by the fact that we were not able to detect any convincing sequence similarity between Babo1 proteins and experimentally confirmed photoreceptor domains. Another peculiarity of Babo1 is its cell-type specific expression: babo1 mRNA was shown to be approximately 10-fold overexpressed in reproductive cells when compared with somatic cells (Kianianmomeni and Hallmann, 2015) and a similar cell-type specific distribution has been shown for the Babo1 protein (Ebnet et al., 1999). This distribution is remarkable because both a whole transcriptome RNA-seq analysis of separated cell types (Klein et al., 2017) and a cell-type specific expression analysis of selected genes (Kianianmomeni and Hallmann, 2015) showed that the known photoreceptor genes of Volvox are predominantly expressed in somatic cells (or in rare cases show no cell-type-specific expression), whereas babo1 is the only (putative) photoreceptor gene that is overexpressed in reproductive cells.
In addition to these conspicuous features, the previous information about the localization of Babo1 within the cell is ambiguous. Initially, C.r. Babo1 was purified from eyespot membrane preparations (Deininger et al., 1995) and assigned with an eyespot localization also due to immunolocalization experiments (Deininger et al., 1995), GFP-tagging (Fuhrmann et al., 1999) and due to its identification within the eyespot proteome (Schmidt et al., 2006;Wagner et al., 2008). However, V.c. babo1 shows only weak expression in the eyespot-containing somatic cells of Volvox, whereas the reproductive cells, which actually have no eyespot at all, show strong expression of babo1 (Kianianmomeni and Hallmann, 2015). The situation becomes even more confusing by the fact that C.r. Babo1 was also identified in thylakoid-enriched fractions (Allmer et al., 2006) and was suggested to be part of the Ycf4-photosystem I assembly complex (Ozawa et al., 2009). It also was proposed that Babo1 is a sensory light receptor, which influences biosynthesis of chloroplast-related proteins and photosynthetic activity in a light-dependent manner (Kianianmomeni and Hallmann, 2014).
So far, no rigorous in vivo localization study of Babo1 has been performed and the previously suggested localizations of Babo1 mainly resulted from indirect observations that even contradicted each other.
In this study, we analyze the sequence of Babo1 (Vop1, Cop1/2) and provide evidence that Babo1 is no transmembrane protein and, thus, cannot be an opsin. We also reveal a large family of more than 60 Babo1-related proteins from a wide range of green algae species (Chlorophyta) and we perform a molecular phylogenetic analysis. Sequence alignments not only allow for a reassessment of the conservation of critical amino acid residues, but also for clarification of the previously supposed similarity to regular rhodopsins and retinal-binding pockets. Moreover, the detailed subcellular localization of fluorescence-tagged Babo1 protein is analyzed in vivo using confocal laserscanning microscopy (CLSM). We show that Babo1 is neither localized in the chloroplast nor in the eyespot of V. carteri. Instead, Babo1 clearly accumulates at the basal apparatus of both somatic and reproductive cells. The former names Vop1 (Volvoxopsin, Volvoxrhodopsin1) and Cop1/2 (Chlamyopsin1/2, Chlamyrhodopsin1/2) therefore turned out to be unsuitable and the proteins, therefore, were renamed V.c. Babo1 and C.r. Babo1 due to their localization at the basal bodies. The localization at the basal apparatus also verifies that Babo1 is no membrane protein and that it is not even associated with membrane structures. Eventually, we were able to trace fluorescencetagged Babo1 protein together with the basal apparatus during basal body separation and cell division. Overall, we can clarify previous contradictions regarding the structure and localization Babo1 and we provide implications for its function.

Gene and mRNA sequences of babo1
The mRNA and genomic sequences of babo1 of V. carteri (formerly vop1) were investigated for the first time in 1999 by the group of Peter Hegemann (Ebnet et al., 1999) (Gen-Bank entries Z69301 and Y11204). However, our babo1 sequencing results slightly deviate from these previous results, but they coincide with the corresponding sequences of the V. carteri genome project (v.2.1, Vocar.0024s0227) (Prochnik et al., 2010) in Phytozome 12 (Goodstein et al., 2012). The differences are highlighted in Figure S1. The babo1 gene is located on scaffold 24, the start codon is at nucleotide position 1733480 on the reverse strand, and its genomic size is approximately 3.1 kb including UTRs and promoter region. The predicted total length of the babo1 mRNA is 1689 bp, which is slightly larger than stated earlier (1566 bp, accession number Z69301, (Ebnet et al., 1999)). An alignment of RNA sequencing data (Klein et al., 2017) and EST data (Prochnik et al., 2010) to the babo1 gene revealed that the 5 0 UTR of babo1 has a length of 48 bp, which was predicted correctly in the V. carteri genome data of Phytozome 12, while the 5 0 UTR is shorter in Ebnet et al., 1999. The 3 0 UTR, however, is approximately 100 bp shorter than predicted by the V. carteri genome annotation, but corresponds instead to the 3 0 UTR presented by Ebnet et al. (1999). The comparison of mRNA and genomic sequences shows that the babo1 gene contains seven introns with sizes from 69 to 375 bp. The number of seven introns corresponds to the average number of introns per gene in the V. carteri genome (v.2.1), which is precisely 7.05 (Prochnik et al., 2010). However, the sizes of the babo1 introns are all below the average intron length of all introns in all nuclear genes, which is 399.5 bp (Hanschen et al., 2016). The below-average intron sizes might be relevant, because genes with basic cellular activity (housekeeping genes) frequently have shorter introns (Eisenberg and Levanon, 2003;Carmel and Koonin, 2009;Eisenberg and Levanon, 2013).

Sequence analysis of Babo1
The 735-bp coding sequence of the babo1 mRNA encodes a polypeptide of 244 amino acids with an expected molecular mass of 26.4 kDa. Babo1 thus belongs to the smaller proteins of V. carteri. The comparison of the amino acid composition of Babo1 with the amino acid composition of the V. carteri (v.2.1) proteome (Prochnik et al., 2010), which was deduced from all V. carteri genes in Phytozome 12 (Goodstein et al., 2012), revealed that the basic amino acid lysine is found much more frequently in Babo1 than in the average of all V. carteri proteins ( Figure S2). The lysine content of Babo1 is 17.2%, which corresponds to every sixth amino acid of Babo1 and 42 lysine residues in total, whereas the average lysine content of all proteins is just 3.5% (Figures 1 and S2). In addition to lysine, there are two other amino acids with side chains that can be positively charged in aqueous solution (at neutral pH): arginine and histidine. However, the percentage share of the two latter amino acids appears to be below average in Babo1, rather than increased ( Figure 1).
Previously, Babo1 (formerly Vop1) was identified as a 30 kDa protein in membrane fractions of V. carteri and, mainly based on sequence similarities, its presence in membrane fractions and due to results with antisense transformants, it was believed to be a membrane receptor of an algal opsin family (Ebnet et al., 1999). Due to inconsistencies that arose from our preliminary CLSM experiments that localized Babo1 away from any membrane structures and the fact that opsins or other membrane receptors are necessarily embedded in a membrane, we reinvestigated the Babo1 amino acid sequence with regard to transmembrane spanning segments. We used bioinformatics tools that searched for similarities between Babo1 and verified transmembrane proteins, utilized algorithms that were trained on transmembrane protein datasets, and also analyzed Babo1 only based on its sequence of amino acid residues. The battery of applied programs included TMSEG (Bernhofer et al., 2016), PolyPhobius (K€ all et al., 2005; K€ all et al., 2007), Phobius (K€ all et al., 2004; K€ all et al., 2007), MEMSAT3 (Jones et al., 1994), MEMSAT-SVM (Nugent and Jones, 2009), PHDhtm (Rost et al., 1995;Combet et al., 2000), TMHMM (Krogh et al., 2001), TMpred (Hofmann andStoffel, 1993), DAS-TMfilter (Cserzo et al., 2004), MINNOU (Cao et al., 2006), TBBpred (Natt et al., 2004), PRED-TMR2 (Pasquier and Hamodrakas, 1999), and the Kyte and Doolittle hydrophobicity plot (Kyte and Doolittle, 1982). However, no transmembrane spanning segments have been identified and all of these programs predicted that Babo1 is not a transmembrane protein. Moreover, when yellow fluorescent protein (YFP)-tagged Babo1 of C. reinhardtii (formerly Cop1/2) was expressed in Xenopus oocytes, YFP fluorescence was exclusively found in the soluble fraction of oocyte extracts but not in the membrane fraction (Tian et al., 2018). As a consequence of these results, Babo1 cannot be an opsin, which would necessarily require a transmembrane structure with generally seven transmembrane helices. Thus, the former name 'Vop1', which stood for Volvoxopsin 1, was no longer justified and we renamed the protein Babo1.

Identification of a large family of Babo1-related proteins
Babo1 of V. carteri (formerly Vop1) has previously been identified due to is similarity with Babo1 of C. reinhardtii (formerly Cop1/2) (Deininger et al., 1995;Ebnet et al., 1999). These homologous proteins show 71% identity and 83% similarity. The coding sequences of the corresponding genes show 77% identity and the number and even position of the seven introns is conserved between the two babo1 sequences. Due to this significant evolutionary conservation between Volvox and Chlamydomonas, we searched for further Babo1-related proteins in other species using transcriptome data of the 1000 plants project (1KP) (Matasci et al., 2014) and the database resources of the National Center for Biotechnology Information (NCBI Resource Coordinators, 2018). Identified sequences that were too short, had gaps, had low similarity values, or that showed any other types of ambiguity were excluded from further analysis. Despite this strict quality control, we identified previously unknown Babo1-related proteins in more than 60 species. All Babo1-related proteins were identified in green algae or, more precisely, within the core chlorophytes (Turmel et al., 2009). The identified Babo1-related proteins originated from morphologically and ecologically divers chlorophytes, which included (i) unicellular and multicellular genera (e.g. Chloromonas in Chlamydomonadales and Codium in Ulvophyceae); (ii) microalgae and macroalgae (e.g. Hafniomonas in Chlamydomonadales and Acrosiphonia in Ulvophyceae); (iii) freshwater, marine and saline algae (e.g. Eudorina in Volvocaceae, Halochlorococcum in Ulvophyceae and Dunaliella in Chlamydomonadales); and (iv) algae having cells with no, two, or even four flagella (e.g. Chlorella in Chlorellales, Gonium in Volvocaceae, and Tetraselmis in Chlorodendrophyceae). A multiple alignment of all Babo1-related proteins used in this study is shown in Figure S3. The alignment shows that the greatest degree of conservation among the Babo1-related proteins is found between amino acid positions 108 to 129 of the alignment in Figure S3. The amino acids Q 80 , P 83 , P 91 , and P 93 of V.c. Babo1 showed the highest degree of conservation among all Babo1-related proteins. However, these amino acids are not conserved in vertebrate opsins and the corresponding part of the sequence previously was not thought to be particularly significant (Deininger et al., 1995;Ebnet et al., 1999); instead it was considered that K 228 in the motif AKA 227-229 close to the C-terminus of V. carteri Babo1 is a conserved retinal binding lysine (Ebnet et al., 1999). Figure S3 clearly shows that K 228 is not conserved and the sequence area immediately around K 228 , which previously has been called retinal binding site or retinal-binding region (Ebnet et al., 1999), is the area with the lowest degree of similarity among all 64 Babo1-related proteins. Furthermore, for Babo1 of C. reinhardtii (formerly Cop1/2), C 21 and C 115 were suggested to V. carteri Babo1 Babo1-related proteins of the Volvocaceae Babo1-related proteins of the Volvox branch Babo1-related proteins of the Chlamydomonadales all Babo1-related proteins V. carteri proteome C. reinhardtii proteome Figure 1. Relative abundance of basic amino acids in Babo1, Babo1-related protein groups, and the proteomes of Volvox carteri and Chlamydomonas reinhardtii. The relative abundance of basic amino acids was calculated for V. carteri Babo1, all Babo1-related proteins shown in Figure 2, for subgroups within Figure 2, and for proteomes of V. carteri and C. reinhardtii. The subgroups are as follows: the Babo1-related proteins of the Volvocaceae family, the Babo1-related proteins of the Volvox branch and the Babo1-related proteins of the Chlamydomonadales order. The Babo1related proteins are listed in Table S4. Error bars represent the standard deviation. The relative abundance data for all 20 canonical amino acids are shown in Figure  form an intramolecular disulfide bridge (Deininger et al., 1995). Not only does the alignment show that there are no conserved cysteines at these positions or elsewhere, but even half of the Babo1-related proteins have less than two cysteines in their entire amino acid sequence. Therefore it is impossible that these proteins form intramolecular disulfide bridges.
To reveal evolutionary relationships within the family of Babo1-related proteins, we performed a molecular phylogenetic analysis. The generated unrooted bootstrap consensus tree of Babo1 and 63 Babo1-related proteins is shown in Figure 2. In this tree, V. carteri Babo1 (Volcar857) branches within the Volvocaceae family, as expected. Babo1 of C. reinhardtii branches close to the Volvocaceae family within the Chlamydomonadales order. Other Babo1related proteins build subgroups that are consistent with the evolutionary relationship of the corresponding organisms. The bootstrap values within quite a few subgroups show a good support (>70%). However, several of the deep branches are uncertain due to low bootstrap values. As an example, previous molecular phylogenetic analyses indicate that the Chlamydomonadales are somewhat more closely related to the Ulvophyceae than to the Chlorellales within Trebouxiophyceae (Leliaert et al., 2012) but our molecular phylogenetic analysis can neither support nor oppose this assumption.
In consideration of our BLAST search results and the molecular phylogenetic analysis, there is no evidence of lateral gene transfer or convergent evolution of Babo1-related genes. There is also no indication of gene loss in any of the subgroups of the core chlorophytes even if we excluded some Babo1-related sequences (and thus species) due to our strict quality rules. We identified however babo1 gene duplicates in the genomes of some analyzed species (marked in Figure 2). The distribution of species with babo1 gene duplicates is scattered among the analyzed core chlorophytes. Moreover, the gene copies within an affected genome are identical or almost identical to each other. Both observations suggest recent and independent babo1-duplication events.

Amino acid composition of Babo1-related proteins
The noticeably high lysine content of V. carteri Babo1 (17.2%) prompted us to investigate the amino acid composition of all Babo1-related proteins (Figures 1 and S2). The lysine content of both the Babo1-related proteins of the Volvocaceae family and of the whole Volvox branch within the Chlamydomonadales is approximately as high as in V. carteri Babo1 (Figures 1 and S2). Similarly, when looking at the Babo1-related proteins of the entire Chlamydomonadales order or even at all investigated Babo1related proteins, the average lysine content is not less than 13%. By contrast, the average lysine content of all proteins both in the V. carteri proteome and the C. reinhardtii proteome is less than 4% (Figures 1 and  S2). Although the high lysine content is well conserved among Babo1-related proteins, the exact amino acid position of most of the lysines seems to be less important because the lysine residues do not stand out from the multiple alignment ( Figure S3).
For a better assessment of the high lysine content of Babo1-related proteins, we sorted all proteins of both the V. carteri or C. reinhardtii proteomes by their lysine content. V. carteri Babo1 ranks 36th and C. reinhardtii Babo1 ranks 65th among more than 14 000 predicted proteins each. As expected, there are predominantly histones and ribosomal proteins among the most lysine-rich proteins (Tables S1 and S2) because positively charged lysines bind nucleic acids by interacting with the negatively charged phosphate moiety in their backbone. Apart from the charge-mediated binding potential, the e-amino groups of lysine residues allow for post-translational modifications. In fact, lysine is essentially the most highly post-translationally modified amino acid out of the 20 naturally encoded amino acids (Zee and Garcia, 2012).

Production of Volvox transformants expressing fluorescence-tagged Babo1 and b2-tubulin
With regard to a rigorous in vivo localization of Babo1, Volvox is much more suitable than Chlamydomonas because the reproductive cells of Volvox exceed the volume of Chlamydomonas cells by more than 100 times, which significantly facilitates the accurate localization. To visualize the expression of Babo1 in living cells of Volvox, a chimeric gene was constructed that allows for expression of a fusion protein in which the C-terminus of V.c. Babo1 is fused via a pentaglycine interpeptide bridge (Gly5) to a YFP ( Figure 3a). The chimeric gene is driven by the endogenous V.c. babo1 promoter region and terminated by the endogenous V.c. babo1 terminator region (Figure 3a). This construction allows for a babo1/yfp expression level that is comparable with the babo1 expression level under natural conditions. In a similar DNA construct, babo1/yfp is driven by the constitutive and strong LHCBM1 promoter region (Figure 3b). For in vivo co-localization of Babo1 and microtubules, another chimeric gene was constructed that allows for expression of a fusion protein in which the N-terminus of b2-tubulin is fused to a cyan fluorescent protein (CFP) (Figure 3c). The chimeric gene is driven by the endogenous tubB2 promoter region. In a similar DNA construct, cfp/tubB2 was brought under the control of the constitutive and strong LHCBM1 promoter region ( Figure 3d).
For stable nuclear transformation of V. carteri strain TNit-1013, three vectors were used simultaneously: babo1 fused to the yfp reporter gene (Figure 3a), the cfp reporter gene fused to tubB2 ( Figure 3c) and pVcNR15 as a selectable marker. The obtained transformants were investigated for stable genomic integration of the DNA constructs and expression of the desired proteins at sufficient levels, which was examined by fluorescence microscope-based screening. Twenty-nine percent of the transformants expressed only the nitrate reductase but showed no fluorescence. Fifty-nine percent only showed YFP fluorescence, which originates from the Babo1-YFP fusion protein. The remaining 14% of the transformants expressed all three plasmids to sufficient extent. The transformant strains were synchronized by a light-dark cycle and the integrity of cell division and embryogenesis was microscopically verified. . e p S . l y C Figure 2. Phylogenetic tree of Babo1-related proteins. Sequence relationship between Volvox carteri Babo1 (Volcar857) and 63 Babo1-related proteins from green algae (Chlorophyta), as listed in Table S4. The unrooted bootstrap consensus tree is based on 10 000 replicates calculated using the neighbor-joining method (Saitou and Nei, 1987). The bootstrap values of the branch points are indicated. Babo1 of V. carteri (Volcar857) can be found in the Volvocaceae subgroup, which is part of the Volvox branch (dark green circular arc) of the Chlamydomonadales. In most species, babo1 is a single-copy gene, however, in some species there are one (•) or two (••) additional, almost identical gene copies in the genome. Cha., Chaetopeltidales; Chlorod., Chlorodendrophyceae; Cyl. Babo1 is not located at the eyespot Previously, Babo1 was classified as an opsin and it was thought to be localized at the eyespot (Deininger et al., 1995;Ebnet et al., 1999;Fuhrmann et al., 1999). However, when transformed algae expressing Babo1-YFP (and TubB2-CFP) were excited with 514 nm light, the corresponding YFP signal appeared at the flagellar bases of somatic cells (Figure 4a,b,d). Such a signal did not occur in wild-type control cells (Figure 4g,h,j). In the eyespots of both transformants and wild-type control cells, only an autofluorescence signal was detectable (Figure 4b,h), which was much weaker than the YFP signal at the flagellar bases and, therefore, it was only visible under overexposure conditions. Excitation at 405 nm stimulated the TubB2-CFP fluorescence, which shows the cytoplasmic microtubules at the flagellar base and the axonemal microtubules of the flagella of somatic cells (Figure 4c,d). The chlorophyll fluorescence shows the position of the single, large chloroplast with its meshwork-like structure (Figure 4e,k).
In order to unequivocally distinguish YFP fluorescence from autofluorescence, lambda scans were performed that allow the separation of spatially overlapping emission signals. More precisely, mean fluorescence spectra for selected subcellular regions of somatic cells were measured and compared both with each other and with the YFP spectrum. This procedure allowed for an unambiguous assignment of emission signals to the corresponding fluorescent molecules. After excitation at 514 nm, the fluorescence emitted from the basal bodies peaked at approximately 530 nm and reached about 1700 rlu ( Figure 5a). Both the peak position and the shape of the spectrum is typical for the utilized YFP variant (Kremers et al., 2006). By contrast, the fluorescence emitted from the eyespot had its maximum at about 557 nm and it reached only about 235 rlu. Moreover, the eyespot fluorescence spectrum of Babo1-YFP transformants was identical to the eyespot fluorescence spectrum of untransformed wild-type cells (Figure 5b). A pictorial representation of these data was generated by spectral imaging, also known as lambda view. Under this method, a color palette, mimicking the emission wavelength of the channel, is automatically assigned to the individual lambda images which are then displayed in a merge-type display. In lambda view, the flagellar bases appeared in blue-green color (Figure 5d), whereas the eyespot fluorescence was clearly characterized by longer wavelengths and thus appeared as a yellowgreen color (Figure 5e). Thus, our results clearly demonstrated that only the basal bodies of transformants emitted the expected YFP spectrum of Babo1-YFP, whereas eyespots exclusively exhibit weak autofluorescence. Because  basal bodies of wild-type cells showed no autofluorescence at all, the total amount of detected basal body fluorescence of transformants can be assigned to Babo1-YFP.

Overexpression of Babo1 negatively affects the viability of transformants
One of our further objectives was to investigate the effects of an increased expression of Babo1. For this purpose, transformation of V. carteri was performed using a plasmid that drives the expression of Babo1-YFP with the constitutive and strong LHCBM1 promoter region ( Figure 3b) instead of the endogenous babo1 promoter region used before ( Figure 3a). However, with this modified approach we were not able to generate any transformants and this indicated that strong overexpression of Babo1 negatively affects the viability of transformants. Likewise, we obtained no transformants when we highly overexpressed the fluorescence-tagged TubB2 under the control of the LHCBM1 promoter region (Figure 3d). These results are in accordance with previous reports where overexpression of GFPa2-tubulin in S. pombe was lethal, whereas moderate expression had no negative effects (Ding et al., 1998). In the green alga C. reinhardtii, the attempt to express GFP fusions of 10 different basal body and flagella proteins was almost without success: only one construct resulted in viable transformants that showed successful expression of the desired fusion protein (Schoppmeier et al., 2005). Therefore, overexpression of basal body and cytoskeletal components may have a negative impact on cell division and, thus, also on the viability and survival rate of transformants.

High-resolution localization of Babo1 at the basal apparatus
The two distinct cell types of V. carteri significantly differ in structure and function (Kirk, 1998) (Figure 6d). In transformants expressing Babo1-YFP under the control of the babo1 promoter region (Figure 3a), Babo1 can be found in both cell types, the small biflagellated somatic cells and the large flagella-less reproductive cells (gonidia). Because we utilized the original, endogenous promoter region, it reflects the natural expression pattern of Babo1. Figure 6(a,b) shows young daughter spheroids of transformed algae expressing Babo1-YFP. When viewed from outside onto the surface of the spheroid, each somatic cell exhibits two distinct fluorescent dots at the center of the cells just below the plasma membrane that correspond to Babo1-YFP at the two basal bodies (Figure 6a). The basal bodies of neighboring cells show the same orientation, which is a fundamental precondition for ensuring that the flagella beat into the same direction and that the whole spheroid swims in one specific direction. Even though the reproductive cells are flagella-less, they still have basal bodies. Consequently, each reproductive cell exhibits two fluorescent dots, which correspond to the Babo1-YFP at the basal bodies ( Figure 6b). The basal bodies appear brighter in the large reproductive cells than in the small somatic cells and, therefore, seem to contain more Babo1-YFP protein.
The highest fluorescence intensity is detected in the center of each basal body, which might indicate, that Babo1 is located in the lumen of the basal bodies.
Even if most of the Babo1-YFP fluorescence co-localizes with the basal bodies, a smaller proportion of the Transformant Wild-type overexposure Babo1

Babo1
Babo1 + TubB2 chlorophyll trans-PMT TubB2  Figure 6(e) indicates the viewing direction for easier orientation. A schematic representation of the basal apparatus of V. carteri is shown in Figure 6(f). Each basal body is attached to two microtubular rootlets (MTRs): the two-membered droot and the four-membered s-root (Moestrup, 1978). Striated microtubule-associated fibers (SMAFs) overlay the proximal part of the four microtubular rootlets (Geimer and Melkonian, 2004). The distribution of SMAFs at the microtubular rootlets correlates with that of Babo1-YFP. This becomes even clearer at the stage immediately before the two basal bodies separate from each other (to be described later  Small amounts of Babo1 are detectable in ectosomes above the basal apparatus Before and during the first cell division, Babo1-YFP was also observed in a pair of tiny dots just above the strong signals at the basal apparatus (Figure 7). The localization of these tiny dots is outside of the plasma membrane within the extracellular matrix (ECM), which raised the question of how Babo1-YFP could get there. It is known that undifferentiated reproductive cells develop short transient flagellar stubs, which are surrounded by the plasma membrane and protrude out of the cell body into the ECM of the gonidial 'vesicle' (Figure 7d) (Kirk, 1998). In addition, it has been shown that the flagella of C. reinhardtii release small, protein-filled membrane vesicles, called ectosomes, into the surrounding space and this budding of vesicles frequently happens at the flagellar tips (Wood et al., 2013;Wood and Rosenbaum, 2015). The ectosomes contain both membrane and flagellar proteins and their release appears to be linked to the flagellar resorption (Long et al., 2016). Taken together, our results indicate that ectosomes with some Babo1-YFP protein are released from the flagellar stubs when these stubs are retracted from the maturing cell. The released ectosomes remain at their initial position within the ECM throughout the first cell division (Figure 7c,f). However, the distance between ectosomes and the basal bodies increases during cell  (Kirk, 1998;Geimer and Melkonian, 2004;Geimer and Melkonian, 2005). Each basal body (BB) consists of nine triplet microtubules that constitute the walls of a hollow cylinder. Basal bodies are connected both with a two-membered MTR (d-root, d for dexter) and a four-membered MTR (s-root, s for sinister) via a set of different fibers (light blue). Note that the proximal end of one s-root microtubule is located below the other three. SMAFs (yellow green) are overlaying only the proximal part of the MTRs, which are actually much longer than shown here. division because once the cleavage furrow appears and then deepens, the basal bodies move together with the midcell constriction of the plasma membrane. The diameter of the ectosomes is roughly about 50-200 nm and this number is in accordance with earlier reports in C. reinhardtii (Wood et al., 2013).

Babo1-YFP uncovers variations of basal apparatus morphology in maturing reproductive cells
Because Babo1-YFP stains significant parts of the basal apparatus, it allows the study of the morphology of the basal apparatus in more detail. As Volvox cultures can be maintained in synchronous growth and development under an 8 h dark/16 h light regime, synchronized individuals can be easily compared with each other. The structure of the basal apparatus was found to be very variable in reproductive cells at the stage shortly before onset of the first cell division (Figure 8). In some reproductive cells, a prominent central axis was visible (see arrows in Figure 8a). The number of fluorescent strands in one optical section of the CLSM varied greatly from three to six. In the course of the transition from interphase to mitosis, the microtubular cytoskeleton undergoes major structural rearrangements and the probasal bodies elongate into mature basal bodies (Gould, 1975). The observed variations in basal apparatus morphology are presumably linked to these processes.  Figure 7. Babo1-YFP localization in ectosomes above the basal apparatus of gonidia. (a) Shortly before the first cell division, a small amount of Babo1-YFP (green) was localized in a pair of ectosomes above the basal bodies, in addition to its localization in the lumen of the basal bodies and at the MTRs. The basal apparatus is also visible but appears blurry because it is localized below the focal plane. The viewing direction is from outside obliquely onto the surface of the gonidium. (b) Same mature gonidium and same viewing direction as in (a) but in this image the basal apparatus is in the focal plane. (c) Top view onto the area above the basal apparatus of a metaphase gonidium expressing Babo1-YFP. The ectosomes are in the focal plane, whereas the basal bodies and the spindle are located deeper inside the cell. The basal bodies and spindle of this gonidium are shown in Figure 10 When the first cell division approached, the Babo1-YFP signal was more concentrated at the two basal bodies, the fluorescent strands were hardly visible and a very distinct central axis was present in all cells (Figure 8b). Immediately before the two basal bodies separated from each other, the central axis disappeared and the four strands of rootlet microtubules became clearly visible (Figure 8c). At this developmental stage, the basal apparatus clearly showed point symmetry, as known from other species (Melkonian, 1978). The d-roots formed a straight line, whereas the s-roots were slightly displaced against each other. The Babo1-YFP signal appeared clearly brighter on the d-roots than on the s-roots (Figure 6f). This distribution is similar to the position of the SMAFs in C. reinhardtii, which are thicker on the d-roots and less pronounced on the s-roots (Geimer and Melkonian, 2004).

Babo1-YFP allows the monitoring of basal body separation during prophase
During prophase, the basal bodies separate from each other and the MTRs remain attached to their respective basal body. Babo1-YFP is again localized predominantly at the basal bodies and the two-membered d-roots (Figure 9). On the four-membered strands of microtubules (s-roots) Babo1-YFP exhibits an irregular, patchy pattern. However, the degree of patchiness on the s-roots varies between different dividing cells (compare Figure 9a-i,k). During the separation of basal bodies, the s-roots slide along each other and the basal bodies with their associated microtubular roots exhibit a clockwise rotation (Figure 9a-i). The basal bodies also move slightly with the midcell constriction of the plasma membrane into the cleavage furrow, as is indicated by the changing chlorophyll signal of the chloroplast. At first, the s-roots slide along each other with a parallel movement while keeping a distance of approximately 0.5 µm between each other ( Figure 9c). Later, they gradually come closer together (Figure 9d-f) until the s-roots contact each other ( Figure 9g). Initially, the angle by which the s-root and the d-root are connected to the corresponding basal body measures approximately 100° (  Figure 9a). During the movement it widens to approximately 150°coinciding with a significant curvature of the MTRs (Figure 9a-h). The idealized overlay in Figure 9(j) summarizes the movements of both basal bodies with their associated MTRs. During the separation of the basal bodies, their distance increases continuously from approximately 1 µm (Figure 9a) to approximately 8.7 µm (Figure 9i). In Figure 9(k), which shows the separation shortly after that in Figure 9(i), the distance reached approximately 9.7 µm. Later, during formation of the spindle, the distance grew to approximately 13.4 µm (see Figure 10a).

Localization of Babo1 during spindle and phycoplast formation
Later in cell division when the mitotic spindle forms, the spindle poles are localized somewhat below the MTOCs (Figure 10). Babo1-YFP is still localized at the basal bodies, on the d-roots and, to a lesser extent, on the s-roots. The dark area between the MTOC and the spindle pole (Figure 10a7) most probably corresponds to the nucleus-basal body connector (NBBC). During cytokinesis, the basal bodies with their associated MTRs are localized above the newly formed nuclei and close to the leading edge of the emerged division furrow (Figure 10b). At this stage, the d-roots are almost parallel to the division furrow, whereas the s-roots are roughly perpendicular to the division furrow (Figure 10b2). Figure 10(c,d) schematically shows top and side views of the cell division apparatus during metaphase to better illustrate the three-dimensional arrangements and spatial relationships of the basal bodies, MTOC, MTRs, and spindle. Remarkably, the new mature basal bodies, which must have reached their full length before basal body separation, did not show any Babo1-YFP fluorescence. The same applies to the d-and s-roots of the new basal bodies. Later in the two-celled embryos, when the first division was completed, the Babo1-YFP fluorescence was still limited to the oldest basal bodies and their roots. Neither the probasal bodies, nor the newly formed mature basal bodies, nor any of their corresponding MTRs contained detectable amounts of Babo1-YFP. Apparently, during and after the first cell division, no new Babo1-YFP was synthesized.

Distribution of Babo1 in four-celled embryos
Even after the second cell division, Babo1-YFP was exclusively found on the oldest basal bodies of the four-celled embryo and on the corresponding d-roots ( Figure 11). The position of the basal bodies and the orientation of the d-roots are typical for this developmental stage. Also during the second cell division no new Babo1-YFP was expressed and, thus, only two of the four cells contained Babo1-YFP. Again, neither the probasal bodies nor the newly formed mature basal bodies nor any of their corresponding MTRs showed Babo1-YFP fluorescence. All focal planes of four-celled embryos were repeatedly investigated to confirm this result. Thus, we were able to prove that there is an unequal protein distribution among the cells of a four-celled Volvox embryo.
Because the overall signal intensity of Babo1-YFP decreased during the progression of embryogenesis, we were not able to clearly monitor the localization of Babo1-YFP after the third cell division. Thus, Babo1-YFP either could be gradually degraded in these later embryonic stages or it could gradually detach from the basal bodies

Previously supposed opsin characteristics of Babo1
Earlier reports on Babo1 stated that this protein is an opsin that contains not only multiple transmembrane helices but also a conserved retinal-binding domain (Deininger et al., 1995;Ebnet et al., 1999;Fuhrmann et al., 1999;Deininger et al., 2000;Fuhrmann et al., 2001;Fuhrmann et al., 2003;Ozawa et al., 2009;Greiner et al., 2017). To re-examine this statement, we initially performed sequence similarity searches and we were able to identify more than 60 proteins from a wide range of algae species that are clearly related to Babo1. A multiple sequence alignment of the family of Babo1-related proteins demonstrates that the earlier proposed opsin-like retinal binding site of Babo1 (Ebnet et al., 1999) is not conserved. Even after a thorough investigation of the multiple alignment of Babo1-related proteins we could not identify those amino acids positions that are conserved or functionally important for microbial opsins. In addition, no seven transmembrane (7TM) helix core architecture has been identified in Babo1, which would be a decisive part of a rhodopsin. Actually, we were not even able to identify a single transmembrane spanning segment in Babo1 and all of the numerous applied programs predicted that Babo1 is not a transmembrane protein. It is fitting, therefore, that we localized Babo1 away from any membrane structures at the basal bodies. Furthermore, when C.r. Babo1 was expressed in Xenopus oocytes, it was not found in the membrane fraction (Tian et al., 2018). Given that all rhodopsins known to date have at least seven transmembrane helices and also contain a conserved retinal binding site (Gao et al., 2015), it is obvious that Babo1 is not an opsin. Because binding of [ 3 H]retinal led to the initial identification and purification of Babo1 (Deininger et al., 1995), it is still possible that Babo1 binds retinal by a mechanism and binding site different from known retinal binding proteins. It can be assumed that retinal molecules form Schiff bases with the e-amino groups of the abundant lysyl residues of Babo1. In this context, it should be noted that retinal binding of Babo1 has only been shown in vitro and, thus, its retinal-binding capacity should also be investigated under in vivo conditions to exclude possible artifacts.

Previously supposed eyespot localization of Babo1
By analyzing fluorescence spectra, we were able to prove that YFP-tagged Babo1 protein is clearly not localized at the eyespot area, but at the basal bodies. The eyespots of both babo1-expressing transformants and wild-type control cells only showed an autofluorescence signal but no YFP signal. Any quenching of Babo1-YFP fluorescence by carotenoid pigments of the eyespot apparatus is virtually impossible because all validated eyespot photoreceptors are located in the plasma membrane above the carotenoid layer (Melkonian and Robenek, 1980;Kreimer, 2009) and our viewing direction was from outside of the Volvox spheroid onto the eyespot of the corresponding somatic cell. Furthermore, the performed spectral imaging in lambda mode makes it possible to separate superimposed signals if they exist and, thus, any existing YFP at the eyespot would have been identified and clearly discriminated from the autofluorescence of the eyespot. In the past, technical difficulties in differentiating between reporter fluorescence and autofluorescence seem to have produced misleading in vivo localization results regarding Babo1 (Deininger et al., 1995;Fuhrmann et al., 1999). Possible causes might be that in this previous work an earlier, less advanced generation of CLSM devices without spectral imaging detectors was utilized and that Babo1 has previously been tagged with GFP. Unfortunately, eyespots of both Volvox and Chlamydomonas cells showed a quite high level of background autofluorescence when GFP is excited at its optimal wavelength. By contrast, this autofluorescence does not apply to YFP. In earlier experiments, Babo1 also was detected by FITC-conjugated antibodies using glutaraldehyde-fixed cells (Deininger et al., 1995), whereby glutaraldehyde is known to cause especially high autofluorescence (Lee et al., 2013). Moreover, FITC fluorescence intensity peaks at 541 nm, which strongly overlaps with eyespot autofluorescence. Thus, eyespot autofluorescence also appears to have been misinterpreted as FITC fluorescence.
Even though both previous Babo1 localization studies had made efforts to deal with autofluorescence issues (Deininger et al., 1995;Fuhrmann et al., 1999), it was technically not possible to separate autofluorescence from FITC or GFP fluorescence. The previous incorrect fluorescencebased localization of Babo1 at the eyespot has then been viewed as a confirmation of the circumstances in which Babo1 was originally extracted from eyespot membrane preparations of V. carteri and C. reinhardtii (Deininger et al., 1995;Ebnet et al., 1999). However, in the knowledge of our results, the presence of Babo1 in any eyespot fractions can also be explained by the fact that the eyespot and the basal bodies are connected with each other by the D4 MTR (Mittelmeier et al., 2015) and, thus, basal bodies could attach to the eyespot or components of it during eyespot isolation. It should also be noted that Babo1 has been identified in several other cellular fractions (Allmer et al., 2006;Goold et al., 2016;Long et al., 2016) and this might indicate that Babo1 tends to bind unspecifically to other cellular components during cell fractionation.

Functional implications of sequence analysis and subcellular localization of Babo1
Our results demonstrate that Babo1 clearly accumulates at the basal apparatus of both somatic and reproductive cells. More precisely, it is located predominantly at the basal bodies and, to a lesser extent, at the four strands of rootlet microtubules. Generally, basal bodies serve as microtubule organizing centers. The building block of a basal body or any microtubule is the tubulin subunit, a heterodimer of atubulin and b-tubulin. These tubulins both possess a glutamic acid-rich 'E-hook' at their C-terminal ends and these negatively charged segments protrude out of the microtubule filaments. The E-hooks can be bound by proteins (e.g. kinesins) with positively charged, poly-lysine containing K-loops (Okada and Hirokawa, 2000). In general terms, a high content of positively charged residues is a common feature of microtubule binding motives (Zhou et al., 2015). Markedly, the lysine content of Babo1 is at the considerable value of 17.2%, which corresponds to every sixth amino acid. Furthermore, sequence analysis using the Microtubule-Associated Protein Analyzer (MAPanalyzer) (Zhou et al., 2015) identified six different motifs that are known to be enriched in microtubule binding sites and that are spread over the entire amino acid sequence of Babo1. These findings suggest a direct interaction between Babo1 and the tubulins of microtubules or basal bodies.
In C. reinhardtii cells, and in the somatic cells of Volvox, basal bodies provide the basis for flagellar assembly because they serve to nucleate the growth of axonemes (Kirk, 1998;Dutcher and O'Toole, 2016). However, knockdown experiments in C. reinhardtii suggested that Babo1 is not essential for flagellar function (Fuhrmann et al., 2001). Furthermore, Babo1 of V. carteri is clearly overexpressed in the flagella-less reproductive cell-type when compared with the somatic cell-type (Ebnet et al., 1999;Kianianmomeni and Hallmann, 2015;Klein et al., 2017). Those aspects make it unlikely that Babo1 has a relevant role in flagellar assembly, function, or stabilization. The observed overexpression in reproductive cells suggests, however, a role for Babo1 in cell division or cell cycle-associated functions. In fact, basal bodies are related to metazoan centrioles, which aid in mitosis and nucleate and organize the centrosomes. Basal bodies and related centrioles are also known to be an important signaling center of the cell Loncarek and Bettencourt-Dias, 2018). Basal bodies of C. reinhardtii are also known to be relevant for the spatial and temporal coordination of karyokinesis and cytokinesis. Mutants of C. reinhardtii that lack basal bodies are not able to determine the right time and place for formation of the spindle and the division furrow (Ehler et al., 1995). Thus, Babo1 could play a critical role in influencing such processes. Our subcellular tracking of Babo1 during the first cell divisions showed that Babo1 is only present on the Four-celled embryos therefore show an unequal allocation of Babo1: Only two of the four cells contained Babo1-YFP-stained basal bodies and d-roots. This finding demonstrates a controlled temporal and unequal spatial distribution of Babo1 in dividing embryos. The structure, regulation, and inheritance pattern of basal bodies and related centrioles is conserved in all branches of the eukaryotic tree of life, from unicellular algae all the way to mammals (Holmes and Dutcher, 1989;Wang et al., 2009). In multicellular organisms, the strict allocation of these structures during division therefore offers a suitable option for unequal distribution of attached cellular components among daughter cells, which leads to asymmetric cell division and, finally, to different cell types (Nigg and Stearns, 2011). For example, when mammalian centrioles separate in embryonic divisions, a centrosomebased mechanism is responsible for asymmetric distribution of certain proteins between daughter cells (Fuentealba et al., 2008). Due to the controlled temporal and unequal, spatial distribution of Babo1 in four-celled embryos, Babo1 could serve as a marker for asymmetry at this early stage of embryonic development. The relevance of Babo1 for proper embryogenesis of V. carteri is also supported by the characteristics of a transgenic strain with knockeddown expression of babo1. The corresponding algae grew and divided only slowly and their division pattern has been described as totally uncoordinated (Ebnet et al., 1999). In view of the unequal distribution of Babo1, it might also be crucial that Babo1 has the capacity to bind retinal (Kr€ oger and Hegemann, 1994;Deininger et al., 1995). In vertebrates, retinal derivatives such as retinol and retinoic acid are well known signaling molecules that play a major role in embryogenesis. During embryonic development, a concentration gradient of retinoic acid is established along the body axis (Shimozono et al., 2013) that is essential for correct patterning of embryos (Niederreither and Dolle, 2008). In a close relative of Volvox, C. reinhardtii, the presence of endogenous retinal has been demonstrated (Beckmann and Hegemann, 1991). Furthermore, this alga can convert retinal into retinol, which was assumed to be a storage form of retinal (Beckmann and Hegemann, 1991). The reaction is catalyzed by retinol dehydrogenases and our genome searches confirmed that also V. carteri (Prochnik et al., 2010) has potential retinol dehydrogenase genes. At least one of these genes is well transcribed based on RNA-seq data (Klein et al., 2017). For conversion of retinal into retinoic acid, a retinal oxidase is required. Two potential retinal oxidase genes have been identified in the V. carteri genome and at least one of these genes is well transcribed (Klein et al., 2017). We also identified a well transcribed but yet unannotated gene, Vocar.0007s0453, which codes for a protein that shows similarity to the human retinoic acid induced protein 1. Thus, retinoids seem to be relevant for Volvox. If the concentration of retinoids in dividing Volvox embryos is decisive for the cell fate, Babo1 could be responsible for the unequal distribution of these retinoids.

Morphology of the basal apparatus and in vivo observation of basal-body separation
Both the morphology of the basal apparatus and the separation process of basal bodies or centrioles are significant fields of research in cell biology, in particular as numerous disorders are associated with centriolar, centrosomal, or ciliary dysfunctions (Chavali et al., 2014;Nigg et al., 2014;Kempeneers and Chilvers, 2018;Schatten and Sun, 2018;Wang and Dynlacht, 2018). Because Babo1 co-localizes with basal bodies and rootlet microtubules, Babo1-YFP can be utilized as a fluorescent stain for in vivo investigation of these cellular structures. Our method and equipment allow not only for high-resolution in vivo imaging of the fine structures of the basal apparatus but also for in vivo tracing of their developmental dynamics. By contrast, many previously used methods require chemical fixation, dehydration, cutting, and staining. Dyes may infiltrate insufficiently or can potentially interfere with the specimen and cause experimental artifacts. Immunostaining with antibodies conjugated with fluorescent ligands is also an error-prone method. Moreover, these earlier methods provided only a snapshot of the dynamic developmental processes and they did not allow continuous following of the fate of one and the same cell and its components. Basal-body separation of V. carteri embryos has never been observed in vivo before and the observations in the related unicellular alga C. reinhardtii did not reach our temporal and optical resolution (Kirk et al., 1991;Kirk, 1998;Lechtreck et al., 2002). Our results also show that Babo1-YFP forms a prominent central axis between the two basal bodies just before first basal body separation. Although there are several investigations of basal bodies in C. reinhardtii (Dutcher and O'Toole, 2016), this structure has not been reported previously. One reason why it remained undiscovered could be that the structure is detectable only for a very short period of time. We were also able to show that the angle, in which the two microtubular rootlets are connected to the corresponding basal body, widens from approximately 100°to approximately 150°during separation of basal bodies. This is probably attributable to the force that is applied to the rootlets at the regions where the two s-roots come close together and are likely to touch each other. We assume that this force pushes the basal bodies away from each other and it might also be sufficient to induce rotational movement.

CONCLUSION
The protein Babo1 (formerly Vop1 or Cop1/2) has previously been classified as an eyespot-photoreceptor, but is neither localized at the eyespot nor is it an opsin. It is not even a transmembrane protein. We revealed a large family of more than 60 Babo1-related proteins from a wide range of algae species and showed their relationship in a molecular phylogenetic analysis. Our high-resolution in vivo imaging demonstrates that Babo1 is localized at the basal bodies and, to a lesser extent, at the four strands of rootlet microtubules. Dynamic structural rearrangements of Babo1 particularly occur right before the first cell division. In four-celled embryos, Babo1 was exclusively found on the two oldest basal bodies of the embryo and on the corresponding d-roots. The coordinated asymmetric distribution of Babo1 in four-celled embryos is the first molecular evidence for differences in protein composition among cells of such a very early stage of embryonic development in V. carteri. The spatial and temporal distribution of Babo1, therefore, suggests a role in cell division. The unequal distribution of Babo1 in four-celled embryos could be an integral part of a geometrical system in early embryogenesis, which establishes the anterior-posterior polarity and influences the spatial arrangement of embryonic structures and characteristics. Due to its retinal-binding capacity, Babo1 could also be responsible for the unequal distribution of retinoids, knowing that concentration gradients of retinoids can be essential for the correct patterning of embryos. Our findings, therefore, push Babo1 research in another direction and provide a promising basis for functional analyses.

Strains and culture conditions
The wild-type Volvox carteri f. nagariensis strain Eve10 (female) (Starr, 1969;Starr, 1970;Adams et al., 1990;Kianianmomeni et al., 2008), which originates from Japan, was used to produce a nitrate-reductase deficient (nitA À ) descendant for transformation experiments. The nitA À descendant of Eve10 was generated by random mutagenesis and chlorate selection as previously described (Huskey et al., 1979;Harper et al., 1987;Adams et al., 1990). The obtained non-revertible mutant strain, TNit-1013 (Tian et al., 2018), contains a deletion of 1013 bp in the nitA gene and, therefore, is not able to grow in medium containing nitrate as the sole source of nitrogen. V. carteri strain TNit-1013 was therefore grown in Volvox medium (Provasoli and Pintner, 1959;Starr, 1969) supplemented with 1 mM ammonium chloride (NH 4 Cl) as a nitrogen source. Strain Eve10 was grown in standard Volvox medium (Provasoli and Pintner, 1959;Starr, 1969). Transformation and CLSM analyses required synchronization of development of Volvox cultures that was achieved by growth under a light-dark cycle. Synchronous cultures were grown at 28°C in a cycle of 8 h dark/16 h cool fluorescent white light (Starr and Jaenicke, 1974) at an average of approximately 100 µmol photons m À2 sec À1 photosynthetically active radiation. Cultivation was performed in glass tubes with caps that allow for gas exchange or in Fernbach flasks, which were aerated with approximately 50 cm 3 sterile air/min. For synchronous growth, culture density was no more than 10 spheroids/ml.

Construction of vectors for expression of fusion proteins in V. carteri
For construction of expression vectors carrying the V.c. babo1 gene (Vocar.0024s0227) fused to the yfp reporter gene (plasmid pBlue_Babo1_YFP; Figure 3a) and the cfp reporter gene fused to tubB2 (Vocar.0007s0229) (plasmid pBlue_CFP_TubB2; Figure 3c), recombinant PCRs were performed using the oligonucleotides shown in Table S3 as primers and both genomic DNA of V. carteri and plasmids carrying yfp or cfp genes as a template. The amplified PCR products were assembled within pBluescript II SK(À) (Stratagene) vector backbones. Both yfp (mVenus) (Kremers et al., 2006) and cfp (mCerulean3) (Markwardt et al., 2011) were previously engineered to match the codon usage of C. reinhardtii (Lauersen et al., 2015). On plasmid pBlue_Babo1_YFP, expression of babo1 is driven by the endogenous babo1 promoter. The corresponding DNA fragment is 215 bp in size, ends 15 bp upstream of the babo1 start codon, contains the remaining 33 bp of the 5 0 UTR and is flanked by artificial KpnI and ApaI sites (Figure 3a). The babo1 start codon is preceded by a 6-bp Kozak sequence. A 1959bp babo1 genomic fragment with all introns included (artificial ApaI to artificial EcoRV; Figure 3a) was amplified in two parts (ApaI to XhoI and XhoI to EcoRV; Figure 3a) making use of an endogenous XhoI site. Artificial EcoRV sites were added on both sides of the intronless yfp and, simultaneously, a short linker sequence, which codes for a flexible pentaglycine interpeptide bridge, was inserted in front of the yfp gene (0.7 kb, EcoRV to EcoRV; Figure 3a). The utilized babo1 terminator region is within a 1-kb fragment (artificial EcoRV to artificial BamHI; Figure 3a) containing 0.9 kb of 3 0 UTR and 100 bp of downstream sequence. In plasmid pBlue_CFP_TubB2, expression of tubB2 is driven by the endogenous tubB2 promoter. The corresponding DNA fragment is 462 bp in size, ends 19 bp upstream of the start codon, contains the remaining 99 bp of 5 0 UTR, and is flanked by two artificial XhoI sites (Figure 3c). The artificial XhoI site in front of the start codon is immediately (after 1 bp) followed by an artificial ClaI site. To facilitate cloning, artificial ClaI sites were added on both sides of cfp and, simultaneously, a short linker sequence, which codes for a flexible pentaglycine interpeptide bridge, was inserted behind the cfp gene (0.7 kb, ClaI to ClaI; Figure 3c). The cfp start codon is preceded by a 6-bp Kozak sequence. A 1879-bp tubB2 genomic fragment with all introns included was amplified (artificial ClaI to artificial BamHI, Figure 3c) and used for vector construction. The terminator region comes from the V. carteri LHCBM1 gene (Vocar.0001s0479) and is localized within a 0.3-kb fragment (artificial XbaI to artificial NotI; Figure 3c) containing 173 bp of the 3 0 UTR and 116 bp of downstream sequence.
Both fusion proteins were additionally brought under control of the LHCBM1 (Vocar.0001s0479) promoter (Figure 3b,d). The LHCBM1 gene is a chlorophyll a/b binding protein in the light-harvesting complex II. In previous RNA-seq studies, the LHCBM1 promoter demonstrated strong expression and the expression levels were similar in both cell types (Klein et al., 2017;Tian et al., 2018). A 0.9 kb DNA fragment containing the LHCBM1 promoter region was introduced in front of the fused genes using KpnI/ApaI for babo1/yfp and XhoI/XhoI for cfp/tubB2, resulting in the plasmids pBlue_LHCBM1_Babo1_YFP ( Figure 3b) and pBlue_LHCBM1_CFP_ TubB2 ( Figure 3d).

Stable nuclear transformation of V. carteri
The nitA À V. carteri strain TNit-1013 (Tian et al., 2018) was grown on a larger scale in Volvox medium supplemented with 1 mM NH 4 Cl. In preparation of particle bombardment, 3 mg of gold microprojectiles (1.0 µm in diameter, Bio-Rad, Hercules, CA, USA) were coated as previously described (Lerche and Hallmann, 2009;Lerche and Hallmann, 2013) using 5 µg of plasmid pVcNR15 (Gruber et al., 1996), which allows for expression of the nitA gene for selection, and 5 µg of each plasmid that expresses fused gene constructs for subcellular localization (Figure 3). The DNA-coated microprojectiles were suspended in 60 µl ethanol and kept at 4°C for use within 1 h. About 24 000 spheroids of strain TNit-1013 were harvested on a 40-µm stainless steel mesh and washed thoroughly with 3 L of standard Volvox medium, which lacks NH 4 Cl and, therefore, allows the selection of nitA + transformants. Transformation was performed using a Biolistic PDS-1000/He particle gun (Bio-Rad). The transformation procedure was as previously described (Schiedlmeier et al., 1994;Hallmann and Wodniok, 2006;Lerche and Hallmann, 2009;Lerche and Hallmann, 2013;Lerche and Hallmann, 2014), with the subsequent modifications. One-sixth of the suspension with DNA-coated microprojectiles (10 µl) was evenly spread on a macrocarrier (Bio-Rad) that was placed in a macrocarrier holder (Bio-Rad). The ethanol was allowed to evaporate from the surface of the macrocarrier. The burst pressure of the rupture disks was 900 psi, the rupture diskmacrocarrier distance was adjusted to 7 mm, the macrocarrierstopping screen distance was 8 mm, the stopping screen-target cell distance was 11 cm, and the bombardment chamber was evacuated to 28 inches of mercury. After each bombardment, the algae were briefly immersed in standard Volvox medium. The bombardment procedure was performed six times in total. The algae were then incubated in ammonium-free standard Volvox medium under standard conditions. From the fifth day on after particle bombardment, algae cultures were examined for green and living transformants (nitA + ) on a background of numerous bleaching, unaltered organisms (nitA À ). Each identified transformant was transferred to fresh selective medium for further culture. Aside from the expression of nitA, expression of the co-transformed fused gene constructs was verified by fluorescence microscopy.

Confocal laser scanning microscopy
For life cell imaging, algae were synchronously grown under standard conditions and examined using an inverted LSM780 confocal laser scanning microscope (Carl Zeiss GmbH, Germany) equipped with a 639 LCI Plan-Neofluar objective (Carl Zeiss GmbH). The confocal pinhole diameter was set to 1 Airy unit, which corresponds to an optical section of 0.8 µm. The YFP fluorescence was excited using an argon ion (Ar + ) laser at 514 nm and the emitted fluorescence was detected at 517-553 nm. The CFP fluorescence was excited by a diode laser (Diode 405-30) at 405 nm and the emitted fluorescence was detected at 460-500 nm. Chlorophyll fluorescence also was excited by the diode laser at 405 nm and detection was at 651-700 nm. Fluorescence intensity was recorded in bidirectional scan mode for YFP, CFP, and chlorophyll in three channels simultaneously. Transmission images were obtained in a fourth channel using a transmission-photomultiplier tube (trans-PMT) detector. An incubation device was used to keep the algae suspension on the microscopic slides at 28°C and to prevent evaporation. All images were captured with a bit depth of 12. Images were analyzed using the ZEN black digital imaging software (ZEN 2011, Carl Zeiss GmbH). Image processing and analysis was carried out using Fiji (ImageJ 1.51w) (Schindelin et al., 2012). Figures 6(b), 8, and 9 were gamma adjusted with a value of 0.5-0.7 to improve the overall visibility. The lambda scan function of ZEN and a gallium arsenide phosphide (GaAsP) QUASAR photomultiplier detector (Carl Zeiss GmbH) were used for simultaneous 20-channel readouts. Emission spectra between 517 and 695 nm were recorded for each pixel with a spectral resolution of 8.9 nm using a main beam splitter MBS 458/514 and 514-nm laser light for excitation. After data acquisition, spectral analysis for the regions of interest (ROIs) was performed.

Sequence database search
The TBLASTN and PSI-BLAST algorithms (Altschul et al., 1990;Altschul et al., 1997) were used to search for Babo1-related sequences in the databases of the NCBI, the China National Gene-Bank (CNGB) and Phytozome 12 (Goodstein et al., 2012). Most hits showing sequence similarity to Babo1 referred to translated transcriptome data produced within the framework of the 1000 plants project (Matasci et al., 2014). Redundant sequences, extremely short sequences and sequences with obvious flaws such as internal stop codons, were excluded from further analyses. However, a few incomplete transcript sequences were completed using genomic data and intron prediction, which was supported by sequence alignments with confirmed Babo1-related sequences. Whenever we found protein isoforms of differing lengths within a certain species, only the longest isoform was chosen for the final protein alignment. A unique, abbreviated protein identifier was assigned to all sequences based on the first three characters of both the genus and the species name (e.g. Volcar for Volvox carteri) and the last three digits of either the 1KP scaffold number, the NCBI accession number or the JGI identifier. The final list of Babo1 and Babo1-related proteins and corresponding identifiers is shown in Table S4.

Phylogenetic analysis
The protein sequences were aligned using the MUltiple Sequence Comparison by Log-Expectation program (MUSCLE) (Edgar, 2004). Minor manual optimization of the alignments, trimming, and management of multialigned data was done using BioEdit v7.0.5.3 (Hall, 1999). The alignments were illustrated using Gene-Doc 2.7 (Nicholas et al., 1997). The unrooted phylogenetic tree was calculated using the PHYLogeny Inference Package (PHYLIP) (Felsenstein, 1989). In these calculations, 10 000 bootstrap resamplings of multiply aligned sequences were generated using Seqboot. Distance matrices using Dayhoff's point accepted mutation (PAM) were computed with Protdist, trees were constructed using the neighbor-joining method (Saitou and Nei, 1987) as implemented in Neighbor, and finally, a consensus tree was built using Consense. Phylogenetic trees were drawn with iTOL 4.2 (Letunic and Bork, 2016).

Calculation of amino acid composition
Amino acid compositions were calculated for all available protein sequences in the proteomes of V. carteri and C. reinhardtii, as well as for the Babo1-related proteins shown in Figure 2 and Table S4. Proteome data refer to V. carteri v2.1 (Prochnik et al., 2010) and C. reinhardtii v5.5 (Merchant et al., 2007) in Phytozome 12 (Goodstein et al., 2012). The compositions of amino acids were calculated using the composition based protein identification (COPId) program (Kumar et al., 2008). Figure S3. Sequence alignment of Babo1-related proteins. Table S1. List of the most lysine-rich proteins of V. carteri. Table S2. List of the most lysine-rich proteins of C. reinhardtii.