A Musashi Splice Variant and Its Interaction Partners Influence Temperature Acclimation in Chlamydomonas

Three RNA metabolism proteins are part of an interaction network that integrates temperature information and confers acclimation to changes in ambient temperature in the green alga Chlamydomonas.

Abstract

Microalgae contribute significantly to carbon fixation on Earth. Global warming influences their physiology and growth rates. To understand algal short-term acclimation and adaptation to changes in ambient temperature, it is essential to identify and characterize the molecular components that sense small temperature changes as well as the downstream signaling networks and physiological responses. Here, we used the green biflagellate alga Chlamydomonas reinhardtii as a model system in which to study responses to temperature. We report that an RNA recognition motif (RRM)-containing RNA-binding protein, Musashi, occurs in 25 putative splice variants. These variants bear one, two, and three RRM domains or even lack RRM domains. The most abundant Musashi variant, 12, with a molecular mass of 60 kD, interacts with two clock-relevant members of RNA metabolism, the subunit C3 of the RNA-binding protein CHLAMY1 and the 5ʹ-3ʹ exoribonuclease XRN1. These proteins are able to integrate temperature information by up- or down-regulation of their protein levels in cells grown at low (18°C) or high (28°C) temperature. We further show that the 60-kD Musashi variants with three RRM domains can bind to (UG)₇ repeat-containing RNAs and are up-regulated in cells grown at a higher temperature during early night. Intriguingly, the 60-kD Musashi variant 12, as well as C3 and XRN1, confer thermal acclimation to C. reinhardtii, as shown with mutant lines. Our data suggest that these three proteins of the RNA metabolism machinery are key members of the thermal signaling network in C. reinhardtii.

Eukaryotic photosynthetic unicellular organisms (microalgae) are key contributors to carbon fixation on Earth (Field et al., 1998). Over the past decades, global warming has expanded and affected microalgal growth and photosynthetic capacities (Singh and Singh, 2015; Schaum et al., 2017). The consequences of this warming process already have dramatic effects on aquatic ecology. For example, thermal changes influence the bleaching of corals by disrupting the symbiosis between corals and their algal symbionts (Baker et al., 2008; Spalding and Brown, 2015; Hughes et al., 2017). In addition, the warming of the ocean has led to an increase of certain toxic algal blooms (Gobler et al., 2017).

The green freshwater alga Chlamydomonas reinhardtii (Merchant et al., 2007; Blaby et al., 2014; Jinkerson and Jonikas, 2015; Li et al., 2016; Cheng et al., 2017) and the marine diatom Phaeodactylum tricornutum (Bowler et al., 2008; Mann et al., 2017) emerged as model algae with sequenced genomes, many available molecular tools, and the potential to be used for synthetic biology (Scaife and Smith, 2016). Here, we focus on the biflagellate microalga C. reinhardtii, which has been studied for decades with regard to photosynthesis, light perception, flagella, and diseases related to flagellar defects (Merchant et al., 2007; Schmidts et al., 2015; Petroutsos et al., 2016). Also, temperature stresses below 7°C (cold stress) and above 40°C (heat stress) have been studied intensively in this microalga, and the involved molecular components, including HSP70A, HSP70B, and HSP90A, were characterized (Valledor et al., 2013; Schroda et al., 2015; Maikova et al., 2016). However, climate change does not necessarily lead to stress temperatures. Over the years, changes of only a few degrees Celsius will occur that still may be in the physiological range of the organism. In fact, biological circadian clocks are entrained not only by light/dark cycles but also by temperature cycles within the physiological range of the organism (Rensing and Ruoff, 2002). Differences as low as 2°C are sufficient to entrain the daily clock (Rensing and Ruoff, 2002). Therefore, it is of interest to identify proteins that sense different temperatures within the physiological range of the organism.

In C. reinhardtii, two subunits of the clock-relevant RNA-binding protein CHLAMY1, named C1 and C3, not only influence the phase and period of the circadian clock (Iliev et al., 2006) but also can integrate temperature changes in the physiological range (Voytsekh et al., 2008). Thus, the C1 subunit, which contains three K homology (KH) domains for RNA binding, is hyperphosphorylated when the algal cells are grown at a lower temperature (18°C) and hypophosphorylated when grown at a higher temperature (28°C). The C3 subunit, which bears three RNA recognition motifs (RRMs), is up-regulated at the protein level in cells grown at 18°C. Moreover, a clock-relevant interaction partner of C3, the 5ʹ-3ʹ exoribonuclease XRN1 (also known as Rhythm of Chloroplast86; Matsuo et al., 2008) that is subjected to degradation by the proteasome pathway, is up-regulated at the protein level in cells grown at 18°C in the presence of an inhibitor of the proteasome pathway (Dathe et al., 2012). Therefore, some of the algal clock components that are known currently (Noordally and Millar, 2015; Ryo et al., 2016) also are potential candidates for temperature control.

XRN proteins belonging to the 5ʹ-3ʹ exoribonucleases have been found in different cellular compartments and usually either are stored in ribonucleoprotein particles, the so-called processing bodies, or are involved in mRNA decay as part of the eisosome in yeast (Saccharomyces cerevisiae; Kulkarni et al., 2010; Vaškovičová et al., 2017). In the cytoplasm, these proteins play a vital role in the cotranslational degradation of mRNAs like XRN1 in S. cerevisiae, XRN1/PACMAN in Drosophila melanogaster, and XRN4 in Arabidopsis (Arabidopsis thaliana; Hu et al., 2009; Nagarajan et al., 2013). Some also are associated with chromatin in the cell nuclear exosome complex, including XRN2/RAT1 in human (Homo sapiens) and XRN3 in Arabidopsis (Nagarajan et al., 2013; Krzyszton et al., 2018).

Here, we have identified the RRM domain-containing RNA-binding protein Musashi as an interaction partner of XRN1 and the C3 subunit of CHLAMY1 in C. reinhardtii. Musashi proteins have been well investigated in D. melanogaster and mammals, where they fulfill a variety of biological functions, including translational control, polyadenylation, and signaling events, and also are involved in disease (Charlesworth et al., 2013; Bertolin et al., 2016; Kudinov et al., 2017; Weill et al., 2017). In contrast to this broad knowledge of Musashi functions in animals, little is known about their function in fungi and plants. The Arabidopsis AtRBP1 (At1G58470) seems to be a Musashi-like protein, since the RRMs on this protein are most similar to those of metazoan Musashi proteins (Lorković and Barta, 2002). In C. reinhardtii, 25 putative splicing variants of Musashi were found. The most abundant form is variant 12. It encodes a 60-kD Musashi protein that contains three RRM domains and interacts with (UG)₇ repeat-bearing RNAs, as CHLAMY1 does (Waltenberger et al., 2001; Zhao et al., 2004). The 60-kD Musashi is up-regulated at the protein level at early night in cells grown at a higher temperature (28°C). Using mutant lines, we show that XRN1, C3, and the 60-kD Musashi variant 12 are involved in the temperature acclimation of algal cells.

RESULTS

Search for Novel Interaction Partners of XRN1 in C. reinhardtii

Recently, we identified the interaction of two C. reinhardtii proteins that integrate temperature information: the clock-relevant RNA-binding protein C3 and XRN1 (Dathe et al., 2012). In C. reinhardtii, knowledge about XRN1 and the members of the downstream signaling cascade is limited. To study its biological role and its connection to temperature integration, we searched for interaction partners besides the already known C3 subunit of CHLAMY1. We performed a coimmunoprecipitation assay using purified anti-XRN1 antibodies. Proteins of a crude extract from wild-type cells and the xrn1 knockout mutant (Dathe et al., 2012; negative control) were precipitated, separated by SDS-PAGE, sliced into gel pieces covering the range of 17 to ∼260 kD, in-gel digested with trypsin, and subjected to liquid chromatography-electrospray ionization-tandem mass spectrometry (LC-ESI-MS/MS) analyses. The experiment was replicated twice. The list of potential interaction partners of XRN1 (Table 1; Supplemental Table S1) includes only candidates that were not detected in the negative control. Listed candidates include at least two different peptides in both biological replicates. The known interaction partner C3 that was identified with one and two different peptides, respectively, in the two replicates also is listed. We also found six potentially novel members of RNA-associated proteins (Table 1). One of them bears KH domains (Cre16.g672750.t1.2), while two of them (Cre16.g662702.t1.1 and Cre12.g560300.t1.2) contain RRM domains, suggesting that these proteins are RNA-binding proteins. The other three candidates of this category are predicted as splicing factors and do not contain KH or RRM domains (Table 1). Moreover, several candidates of three other categories were found (Supplemental Table S1): (1) six metabolism-related enzymes/proteins of different functions; (2) four flagella-associated proteins (mainly of the intraflagellar transport mechanism); and (3) two chaperone-like proteins, including the heat shock protein HSP70 (gene model Cre10.g439900.t1.2; Supplemental Table S1). A detailed list of all identified unique peptides of the candidates is provided in Supplemental Table S2.

Table 1. LC-ESI-MS/MS identifies RNA-associated and ribonucleoproteins as possible interaction partners of XRN1.

Database analysis is described in “Materials and Methods.”

Phytozome v12 (C. reinhardtii v5.5)	No. of Different Peptides in Co-IPa 1/2	Function/Homologies
Cre16.g672750.t1.2	3/4	Poly-C binding protein (PCBP), KH domain-containing RNA-binding protein
Cre12.g514450.t1.2	3/2	Splicing factor 3b, subunit 2
Cre12.g494100.t1.1	2/3	Splicing factor 3b, subunit 1
Cre16.g662702.t1.1	2/3	RNA-binding protein Musashi with RRM domains, splicing factor 3b, subunit 4
Cre14.g621000.t1.2	2/2	Pre-mRNA splicing factor
Cre12.g560300.t1.2	2/2	RNA-binding (RRM/RBD/RNP motifs) family protein, splicing factor 3b, subunit 4
Cre03.g177200.t1.2	2/1b	RNA-binding protein C3 with RRM domains, CELF family, subunit of CHLAMY1

^aTwo independent biological replicates of coimmunoprecipitation (Co-IP) experiments.

^bThe interaction of C3 with XRN1 was formerly verified by immunoblotting with anti-C3 antibody and yeast two-hybrid analysis (Dathe et al., 2012).

Evidence for Musashi Splice Variants in C. reinhardtii

So far, Musashi proteins have not been investigated in green microalgae. Cre16.g662702.t1.1 (Table 1) represents a Musashi protein (76.7 kD, referred to as 77 kD) with two RRM domains and two transmembrane domains (TMDs). Based on the gene model, we generated a codon-adapted, His-tagged 77-kD Musashi gene (Supplemental Fig. S1A), overexpressed the protein in Escherichia coli, and affinity purified it to generate antibodies (Supplemental Fig. S1B, left). In addition to the His-tagged 77-kD Musashi, we detected other bands of lower molecular mass (Supplemental Fig. S1, B and C). The proteins in these bands were tryptically digested and the peptides were analyzed by LC-ESI-MS/MS. All proteins from the investigated fractions (Supplemental Fig. S1C) belong to Musashi, indicating that they are degradation products (Supplemental Table S3). Immunoblots of these protein bands with the generated anti-Musashi antibodies confirmed this result (Supplemental Fig. S1B, center). In immunoblots with proteins from a crude extract of C. reinhardtii and anti-Musashi antibodies, several major bands were detected ranging from 57 to 78 kD (Supplemental Fig. S1B, right). These protein variants also may be degradation products or they may represent Musashi splicing variants, as found in mammals (Wuebben et al., 2012; MacNicol et al., 2017).

To determine whether Musashi splicing variants exist, the transcript sequence of the 77-kD Musashi (Cre16.g662702.t1.1) from Phytozome v12 (C. reinhardtii v5.5) was compared with RNA sequencing (RNAseq) data where RNAs from cells grown at 23°C and harvested throughout a 12-h-light/12-h-dark cycle (LD12:12) were analyzed (Zones et al., 2015). In total, 25 different Musashi sense transcripts were found covering the Musashi coding sequence (Fig. 1). Two transcript variants encode proteins without an RRM domain (variants 1 and 25), while in variant 25, two TMDs are encoded. Among the other 23 Musashi splice variants, 11 code for three RRM domains (variants 12–22), 10 code for two RRM domains (variants 4–11, 23, and 24), and two code for only one RRM domain (variants 2 and 3). Two splice variants (23 and 24) encode a protein with two RRMs as well as two TMDs. Variant 24 is the closest to the Joint Genome Institute (JGI) model of the 77-kD protein and is nearly identical in molecular mass (Supplemental Fig. S2). Seven amino acids at its N terminus are different, and one amino acid is missing in variant 24 compared with the JGI model. The calculation of the relative abundance of all 25 transcripts (Fig. 1) over the diurnal cycle (Supplemental Table S4) revealed that only one transcript (variant 12), encoding a protein of 60 kD, is present throughout all time points of the diurnal cycle and occurs overall at the highest rate. All other transcripts are present only at specific time points. Notably, transcript variant 24, the closest to the 77-kD Musashi (Cre16.g662702.t1.1), was found at only one time point (LD14.5), with an abundance of greater than 1.5% but still as low as 2% (Supplemental Table S4).

Figure 1.

Genomic and protein domain structures of 25 putative Musashi splicing variants based on RNAseq data (Zones et al., 2015). Splicing variants are ordered by domain structure and maximal abundance (Max abun.) within the analyzed LD12:12 (Supplemental Table S4). In the genomic DNA (gDNA), the introns are depicted as dashed lines and the exons as solid blocks. Gray color represents untranslated regions (UTRs) and purple color represents the open reading frames (ORFs). Domains were predicted by the online tool SMART (Simple Modular Architecture Research Tool). RRMs are marked in blue color, and TMDs are marked in yellow. Solid color bars symbolize the presence of the domain, while diagonal lines symbolize the absence of the domain in the corresponding splicing variant. Only variants with an abundance greater than 1.5% are shown. Blue arrows indicate the identified unique peptides from LC-ESI-MS/MS analysis of proteins from the eluate of the coimmunoprecipitation (Supplemental Table S5). MW, Molecular mass.

Consequently, we performed the MS evaluation from the coimmunoprecipitation with all 25 Musashi transcript variants (Supplemental Table S5). The originally identified peptides of the 77-kD Musashi (Cre16.g662702.t1.1) are present in most of the resulting protein variants (2–24 in Fig. 1). The highest number of unique peptides identified by MS was found in the variants bearing a third RRM domain. Thus, there is evidence that at least one of the three RRM-containing variants interacts with XRN1. It remains unclear whether the two-RRM-bearing 77-kD protein (Cre16.g662702.t1.1) and variant 24 interact with XRN1, since no C-terminal peptides were found that are specific for these variants.

The abundant 60-kD Musashi variant 12, encoding three RRM domains (Fig. 1; Supplemental Fig. S3A), also could be confirmed by two EST clones (AV639968 and AV641618 from the Kazusa DNA Research Institute; Asamizu et al., 1999; Supplemental Fig. S3B). In addition, a 60-kD band also was observed in the immunoblot with proteins from a C. reinhardtii crude extract (Supplemental Fig. S1B, right). The Kazusa EST clone AV639968 was used for a yeast two-hybrid analysis.

Three- and Two-RRM-Containing Musashi Variants Interact with the 5ʹ-3ʹ Exoribonuclease XRN1 and the RNA-Binding Protein C3

To verify the interaction between the 60-kD Musashi variant 12 and XRN1 or C3, we performed a yeast two-hybrid analysis. The mated yeasts carrying 60-kD Musashi variant 12, XRN1, and C3 in the vectors expressing GAL4-AD or GAL4-BD were used as negative controls (Fig. 2A). Fusions of XRN1 and 60-kD Musashi variant 12, as well as of 60-kD Musashi variant 12 and C3, resulted in yeast growth on medium lacking His (in both cases) and on medium lacking His and Ade (only in the case of XRN1 and Musashi; Fig. 2A). These data confirm that the 60-kD Musashi variant 12 interacts with XRN1 and with C3.

Figure 2.

The 60-kD Musashi variant 12 and the full-length 77-kD Musashi (Cre16.g662702.t1.1) interact with XRN1 and the C3 subunit of CHLAMY1. A, Protein-protein interactions verified by yeast two-hybrid assays. DNA encoding XRN1 and the 60-kD Musashi variant 12 were fused to the GAL4-AD (activation domain), and cDNAs encoding the 60-kD Musashi variant 12 and C3 were fused to the GAL4-BD (DNA binding domain), respectively. Empty fusion vectors served as controls. Diploid cells were grown to an OD₆₀₀ of 1 and spotted onto selective medium (SD-L,-W) to verify the presence of both GAL4-AD and GAL4-BD plasmids. SD-L,-W,-H medium plus 1 mm 3-amino-1,2,4-triazole was used to test for protein-protein interactions, and more restrictive medium also lacking adenine (SD-L,-W,-H,-A) was used to test for strong interactions. B, Expression of heterologous His-tagged 60-kD Musashi variant 12 in E. coli. Fifteen micrograms of proteins from a crude extract of E. coli cells (CE_{60-kDa Mus-ox}) and 5 µg of purified His-tagged 60-kD Musashi (60-kD Mus-_OX) were separated by 10% SDS-PAGE and stained with Coomassie Brilliant Blue (left); 5 µg of 60-kD Mus-_OX purified from E. coli cells was separated by 10% SDS-PAGE and immunoblotted with anti-Musashi (center) and anti-6×His (right) antibodies. The 62-kD His-tagged Musashi is indicated by the arrows. C and D, Protein-protein interactions verified by pull-down assays. His-tagged 60-kD Musashi variant 12 (C), 77-kD Musashi (D), and aCRY (negative control; D, right) were purified and coupled to Dynabeads. Proteins from a crude extract of C. reinhardtii wild-type cells grown at 18°C and harvested at LD14 with 10 μm proteasome inhibitor (MG132) added at LD13 were applied to the beads. Proteins from the last washing step (W4), the eluate (E), and the crude extract (CE) were analyzed by immunoblots with anti-XRN1 and anti-C3 antibodies using 50 µg of proteins of the CE as well as proteins present in 80 µL of W4 and 80 µL of E, separated by 9% (XRN1) or 10% (C3) SDS-PAGE.

To further verify these protein-protein interactions, we performed pull-down assays. For this purpose, a codon-adapted, His-tagged 60-kD Musashi DNA construct of variant 12 was synthesized (Supplemental Fig. S4), based on the sequence of the cDNA clone (AV639968) containing the full-length 60-kD Musashi, overexpressed in E. coli, and purified (Fig. 2B, left). The His-tagged protein was confirmed by immunoblots (Fig. 2B, center and right) and used for the pull-down assay (Fig. 2C). Proteins in the last washing step as well as in the eluates of the pull-down fraction were used for immunoblots with anti-XRN1 and anti-C3 antibodies, respectively. In both cases, a signal was observed in the eluate. Thus, the interaction of the 60-kD Musashi variant 12 with XRN1 and C3 was positively confirmed.

Another pull-down assay was conducted with the His-tagged 77-kD Musashi (Cre16.g662702.t1.1) protein overexpressed and purified from E. coli (Supplemental Fig. S1B, left). The detected protein interactions between the 77-kD Musashi with XRN1 and C3 (Fig. 2D) indicate that the N-terminal part of the first two RRM domains shared by Musashi variant 12 and the 77-kD Musashi (Cre16.g662702.t1.1; Fig. 1) may be sufficient for protein-protein interactions. As a negative control for the pull-down assays (Fig. 2D, right), we used His-tagged animal-like cryptochrome (aCRY), a photoreceptor of C. reinhardtii (Beel et al., 2012).

Three RRM-Containing Musashi Variants Bind to UG Repeat-Bearing RNAs

Previous research indicates that the CHLAMY1 complex is able to bind (UG)₇ repeat-bearing mRNAs (Waltenberger et al., 2001; Zhao et al., 2004). Homologs of C3 in mammals are known to bind to UG and CUG repeat sequences (Timchenko et al., 1996; Takahashi et al., 2000; Blech-Hermoni et al., 2016). To further study the function of Musashi and its interaction partners, we investigated if they are able to bind to (UG)₇ repeat-bearing mRNAs by using the 3′ UTR of glutamine synthetase2 (gs2), which contains seven UG repeats, as bait. As a control, we used a mutated gs2 having only two UGs (Zhao et al., 2004; Fig. 3A; see “Materials and Methods”). We used biotinylated CTP for in vitro transcripts along with proteins of a crude extract of C. reinhardtii cells harvested at early night (LD14), when the binding activity of CHLAMY1 was shown to be high (Mittag, 1996). A nonspecific RNA competitor, polyguanylic acid, was added to the binding assay. Proteins in the last washing steps and eluates were used for immunoblots with anti-C3, anti-Musashi, and anti-XRN1 antibodies. Our data confirmed that C3 (alone or in a complex) binds to the (UG)₇-bearing RNA but not to the (UG)₂-containing RNA (Fig. 3B). XRN1 is not able to bind to either of the two RNA samples (Fig. 3B).

Figure 3.

A 60-kD Musashi variant binds to (UG)₇ repeat-bearing RNA. A, Workflow of the in vitro RNA pull down using a CTP-biotinylated in vitro transcript. B, C3 and 60-kD Musashi but not XRN1 bind to the (UG)₇ repeat-bearing RNA. Fifty micrograms of crude protein extract (CE) as well as 80 µL of the last washing step (W11) and of the eluate (E) from the RNA pull-down experiment were separated by 10% (C3 and Musashi) or 9% (XRN1) SDS-PAGE and immunoblotted with anti-C3, anti-Musashi, or anti-XRN1 antibodies. C, Domain architecture of the 60-kD Musashi variant 12 and location of MS-identified peptides from the 60-kD band in B. aa, Amino acids. D, Phylogenetic analysis of Musashi homologs from plants, animals, and fungi. Bootstraps of 1,000 replicates were performed and shown as a percentage. The scale bar indicates amino acid substitutions per site. The C3 subunit of CHLAMY1 and its homolog proteins were taken as the outgroup. C3 and the 60-kD Musashi variant 12 from C. reinhardtii are indicated in boldface.

One or more 60-kD forms of Musashi were found to bind specifically to the (UG)₇-containing RNA (Fig. 3B), which also was observed in two other biological replicates. Nine Musashi variants have molecular masses close to 60 kD. They all have three RRM domains (variants 12–15, 17–20, and 22). Considering that the cells used for the assay were harvested at LD14, only two of the nine variants seem to be expressed at that time point (with transcript abundances of greater than 1.5%): variant 12 with high abundance (74.1%) and variant 20 with low abundance (2.4%; Supplemental Table S4). The specific binding of the 60-kD forms of Musashi was investigated further by cutting out the corresponding band of 60 kD, digesting it in gel with trypsin, and analyzing the peptides by LC-ESI-MS/MS (Supplemental Table S6). The 60-kD forms of Musashi were confirmed by seven different peptides that match all nine mentioned variants. For the reasons stated above, the three-RRM domain-bearing variants 12 and 20 are likely to bind to the (UG)₇-containing RNA, whereby variant 12 represents the major form (Fig. 3C).

We also cut out bands from all other molecular mass ranges (from ∼14 to 267 kD) and examined the resulting peptides by MS analysis against a database with the 25 putative Musashi variants. We did not detect any Musashi variants in any band other than the 60-kD band. This result suggests that the two-RRM domain-containing Musashi variants with a lower (variants 4–11) or higher (variants 23 and 24) molecular mass are not able to interact with the (UG)₇-containing RNA.

Finally, we performed an alignment and phylogenetic analysis of the most abundant 60-kD Musashi variant 12, which binds to (UG)₇-containing RNAs, with Musashi homologs from vascular plants, mosses, algae, fungi, and several model animal systems. In the investigated animal systems, the Musashi proteins bear two RRM domains, while those of fungi and photosynthetic organisms have either two or three RRM domains (Fig. 3D). The presence of several Met and Gly residues after the second RRM domain, as present in the ELAV-like family (CELF) protein C3 (Zhao et al., 2004), taken as the outgroup, also is obvious for the C. reinhardtii Musashi.

Temperature-Dependent Expression of XRN1, C3, and the 60-kD Splice Variant of Musashi

We investigated the expression of the 60-kD Musashi variants at two time points of the day/night cycle, LD4 (close to midday) and LD14 (early night), when the binding of CHLAMY1 to UG repeats was shown to be relatively low and high, respectively, at 24°C (Mittag, 1996). Moreover, we analyzed protein levels in cells grown at either 18°C or 28°C. Previous research indicates that the two subunits of CHLAMY1 can integrate temperature information in cells grown at different temperatures in the physiological range of C. reinhardtii (Voytsekh et al., 2008). We confirm that C3 is up-regulated at the lower temperature (18°C; Fig. 4A). The effect is stronger during daytime (LD4) compared with early night (LD14). We also investigated XRN1, which has been analyzed previously under these conditions in the presence of a proteasome inhibitor (Dathe et al., 2012). Here, we omitted the inhibitor and concentrated primarily on the full-size protein of ∼198 kD (predicted size) in addition to one prominent degradation product of ∼140 kD. In the absence of the proteasome inhibitor MG132, we see a significant difference in the abundance of the full-size XRN1 at different temperatures only during early night. Here, XRN1 is up-regulated in cells grown at the higher temperature (28°C; Fig. 4B). Intriguingly, the 60-kD forms of Musashi also are significantly up-regulated at early night in cells grown at the higher temperature (Fig. 4C). This up-regulation seems to be mediated at the translational level rather than at the transcriptional level, because Musashi transcript levels are not significantly different at night (LD14) in cells grown at different temperatures (Fig. 4D). However, the primers used cover most variants (Fig. 1). In general, there is little difference of Musashi transcript levels at LD4 and LD14. Their highest expression occurs around LD10 (Supplemental Fig. S5).

Figure 4.

C3, XRN1, and the 60-kD Musashi variants are expressed in a temperature-dependent manner during day (LD4) and/or early night (LD14). A to C, Cells from the wild-type strain SAG73.72 were grown under LD12:12 at either 18°C or 28°C and harvested at LD4 or LD14. Thirty micrograms of crude protein extract was separated by 10% (C3 and Musashi) or 9% (XRN1) SDS-PAGE and immunoblotted with anti-C3 (A), anti-XRN1 (B), and anti-Musashi (C) antibodies. As a loading control, unspecified bands from the polyvinylidene difluoride membrane stained with Coomassie Brilliant Blue (B) or from nitrocellulose membranes stained with Ponceau S (A and C) were used. The expression level of proteins from cells grown at 18°C and harvested at LD4 was set to 100% for each biological replicate (n = 3). D, Quantitative reverse transcription PCR (RT-qPCR) analysis of Musashi transcript levels. Transcript accumulation was quantified with RNA isolated from wild-type cells. Data shown represent transcript abundances for three biological replicates. Error bars represent sd. A two-tailed unpaired Student’s t test was performed. Asterisks indicate statistically significant differences: *, P < 0.05; **, P < 0.01; and ***, P < 0.001. n.s., Not significant.

XRN1 Mediates Temperature Acclimation in C. reinhardtii

The strong temperature dependence observed for the expression of the three proteins related to RNA metabolism, XRN1, C3, and the 60-kD Musashi forms, raised the question of whether the presence of these proteins is necessary to mediate thermal acclimation. Therefore, we performed a temperature-survival test of algal cells grown in LD12:12. Before the temperature shift, cells from the different cultures were diluted to the same cell density and then exposed for 12 h in the dark period to different temperatures. After this step, cultures were diluted to the same ratios and plated. Plates were incubated at 23°C under LD12:12, and colonies were counted after 6 d (Fig. 5A).

Figure 5.

XRN1 confers thermal acclimation to C. reinhardtii. A, Workflow of the temperature assays. The diagram shows an overview of the experimental approach to analyze cell survival rates after a 12-h exposure of cells to different temperatures (18°C, 28°C, 37°C, and 42°C) during the dark period. B, The xrn1 mutant (Dathe et al., 2012) was backcrossed to the wild-type (WT) strain SAG73.72 and verified in immunoblots for the absence of XRN1 along with the wild type (positive control). Thirty micrograms of crude protein extracts of cells harvested at LD4 with or without 10 μm MG132 (−PI or +PI) added at LD3 was separated by 9% SDS-PAGE and immunoblotted with anti-XRN1 antibodies. The loading control was done as described (Fig. 4B). C, The wild type as well as the xrn1 mutant were exposed to different temperatures (see A), and survival rates were determined afterward by counting colonies. Colony-forming units (cfu) of the C. reinhardtii cells exposed to different 12-h temperature shifts were compared with those of the wild type. The colony-forming units of the wild type treated for 12 h at 18°C was set to 100% for each replicate (n = 9). D, Immunoblot analysis of XRN1 in the wild type and ami_xrn1RNA lines labeled as #9 and #20. Immunoblotting was performed as described for B in the presence of MG132. E, For quantification of the XRN1 amount, the XRN1 level of the wild-type strain was set to 100% for each biological replicate (n = 3). Error bars represent sd. F and G, The ami_xrn1RNA lines #9 and #20 were exposed to different temperatures (see A) and evaluated as described for C. Error bars represent se. Asterisks indicate statistically significant differences calculated by a two-tailed unpaired Student’s t test: *, P < 0.05; **, P < 0.01; and ***, P < 0.001. n.s., Not significant.

First, we investigated the influence of XRN1 on temperature acclimation. For this purpose, we used an xrn1 knockout mutant (Dathe et al., 2012) backcrossed to the wild-type strain SAG73.72, used in most of our experiments, in order to make meaningful comparisons with the c3 and musashi knockdown or overexpressing lines in the same background. Immunoblots with anti-XRN1 antibodies confirmed the presence of XRN1 in the backcrossed wild type and its absence in the mutant, both with and without proteasome inhibitor treatment during cell harvest (Fig. 5B).

Exposure of wild-type cells for 12 h in the dark to 42°C (heat shock conditions) or 37°C (close to heat shock conditions) resulted in a strong decrease in the viability of C. reinhardtii cells compared with the viability of those kept at 28°C (Fig. 5C). Moreover, exposure to 28°C resulted in a higher number of viable cells than exposure to 18°C, albeit the number of viable cells was still higher at 18°C compared with 37°C. In the xrn1 knockout mutant, a small number of viable cells was observed under heat shock conditions (42°C), and no significant difference was observed compared with the wild type. However, significantly fewer cells of the xrn1 mutant survived at all other temperatures (18°C, 28°C, and 37°C) compared with the wild type (Fig. 5C), indicating that XRN1 has a strong influence on temperature acclimation in C. reinhardtii cells. To further verify our results, an xrn1-complemented mutant would be necessary, but the large size of the XRN1 protein (∼198 kD) renders the creation of a complementation vector bearing the full xrn1 gene, plus promoter sequences, UTRs, and a selection marker, very challenging. Therefore, we created artificial microRNA (amiRNA) lines and included them in the temperature assay. Immunoblots and quantifications showed that the XRN1 level was reduced to 18% and 5%, respectively, in two of the ami_xrn1RNA lines compared with the wild type (100%; Fig. 5, D and E). Next, we verified that the wild type shows similar survival rates to a vector control line used for the creation of the ami_xrn1RNA lines as well as for the previously generated RNA interference (RNAi) lines of c3 (Iliev et al., 2006; Supplemental Fig. S6A). Survival rate experiments with the ami_xrn1RNA lines (especially the strongly reduced xrn1 line, with only 5% XRN1 left) conducted at different temperatures confirmed the data obtained with the xrn1 knockout mutant (Fig. 5, F and G). In conclusion, we show that the reduction of XRN1 levels leads to a decrease in cell viability after the 12-h shifts at different temperatures, showing that XRN1 mediates temperature acclimation in C. reinhardtii.

C3 and the 60-kD Variant 12 of Musashi Also Are Involved in Temperature Acclimation

We utilized previously generated c3 RNAi lines with reduced C3 levels (Iliev et al., 2006) to investigate the influence of C3 on temperature acclimation. Because RNAi lines can be unstable over time, we measured the expression levels of C3 in the RNAi lines by immunoblots (Fig. 6, A and B). Low levels of C3 expression (10% and 30%, respectively) in these two RNAi lines compared with the wild type (100%) were confirmed. Similar to xrn1, silencing of c3 caused strongly reduced viabilities of C. reinhardtii cells exposed to different temperatures, which was most severe at 28°C and 37°C but also observed at 18°C. Under heat stress conditions (42°C), the number of surviving cells was very low in both the wild-type and silenced lines (Fig. 6C). These data show that the C3 subunit of CHLAMY1 contributes significantly to temperature acclimation in C. reinhardtii, in addition to XRN1.

Figure 6.

C3 and the 60-kD Musashi variant 12 confer thermal acclimation at different temperatures. A, The C3 protein levels in two c3 silencing strains (c3-sil₆ and c3-sil₇; Iliev et al., 2006) were confirmed. For this purpose, 50 µg of proteins from crude extracts (LD4, −PI; see Fig. 5B for details) was separated by 10% SDS-PAGE and immunoblotted with anti-C3 antibodies. B, Quantification of C3 protein levels. The amount of C3 in the wild type (WT) was set to 100% (n = 3) and evaluated as described for Figure 5E. C, The wild type as well as the c3-sil lines were exposed to different temperatures (Fig. 5A), and survival rates were determined by counting colony-forming units (cfu) as described for Figure 5C. D, Musashi protein levels in two musashi overexpressing (ox) lines (mus_ox #29 and #33) for the 60-kD form were analyzed. For this purpose, 30 µg of proteins from crude extracts (LD4, −PI; see Fig. 5B for details) was separated by 10% SDS-PAGE and immunoblotted with anti-Musashi antibodies. E, To quantify the amount of the 60-kD Musashi, its expression in the wild type was set to 100% (n = 3) and evaluated as described for Figure 5E. F, The wild type as well as the mus_ox #29 and #33 lines were exposed to different temperatures (Fig. 5A), and survival rates were determined by counting colony-forming units as described for Figure 5C. Asterisks indicate statistically significant differences calculated by a two-tailed unpaired Student’s t test: *, P < 0.05; **, P < 0.01; and ***, P < 0.001. n.s., Not significant.

We next analyzed the 60-kD Musashi, whose domain architecture is similar to that of C3 (Fig. 3D). To specifically target the 60-kD form with up-regulated expression at a higher temperature (Fig. 4C), we constructed an overexpression vector (Supplemental Fig. S7) and transformed it into C. reinhardtii. The survival rate of cells transformed with a control vector (VC2) was identical to that of the wild type, except for one small but significant difference at 37°C (Supplemental Fig. S6B). The three overexpressing lines had expression levels of 231%, 307% (Fig. 6, D and E), and 254% compared with wild-type 60-kD Musashi (Supplemental Fig. S8, A and B). In all cases, we observed major differences in the survival rates compared with those of the wild type after exposure to 28°C for 12 h. At this temperature, the mus_ox lines had a strongly reduced survival rate (Fig. 6F; Supplemental Fig. S8C). At 18°C, only a slight reduction in cell viability was visible compared with that of the wild type; at 37°C and 42°C, no significant difference between the three overexpressed lines and the wild type was observed (Fig. 6F; Supplemental Fig. S8C). These data indicate that the 60-kD Musashi variant 12 also influences temperature acclimation but suggest that this regulation only takes place at a restrictive temperature regime around 28°C, different from XRN1 (Fig. 5, C, F, and G) and C3 (Fig. 6C).

DISCUSSION

We have used a molecular approach to understand how microalgae react to changes in ambient temperature by identifying molecular components involved in the integration of temperature information in the model alga C. reinhardtii. These components influence the temperature acclimation (12-h temperature shifts during the dark period) of the microalgae, thus being linked directly to the survival rates under changing temperatures ranging from physiological (18°C, 28°C, and 37°C) to heat shock conditions (42°C) that start around 39°C for C. reinhardtii (Tanaka et al., 2000; Rütgers et al., 2017). These components include the three-RRM domain-containing RNA-binding proteins C3 and Musashi variant 12, as well as XRN1, highlighting the importance of RNA metabolism for thermal acclimation in C. reinhardtii. Our data suggest that these proteins are part of a temperature-controlled network that reacts to temperature changes within the cell. Thereby, a balance between the different members of the network being up-regulated either at 28°C or 18°C may be needed for intact thermal acclimation. Thus, their misregulation causes disturbances and reduces cell viability at ambient temperatures.

It appears unlikely that these components are the direct thermoreceptors of C. reinhardtii, albeit we cannot exclude this possibility. In general, eukaryotic thermoreceptors involve transmembrane ion channels, including members of the transient receptor potential (TRP) family or the anoctamin family of chloride channels (Sengupta and Garrity, 2013). C. reinhardtii encodes numerous members of the TRP family (Verret et al., 2010), many of which have unknown biological functions. In land plants, photoreceptors were connected recently to temperature regulation. Temperature dependence in the absence of accessory proteins has been observed for phytochrome B dark reversion and the photocycle of the light/oxygen/voltage domain of phototropin, suggesting that phytochrome B and phototropin themselves are temperature sensors (Casal and Qüesta, 2018). Notably, the C3 subunit of CHLAMY1 also seems to be involved in light-controlled signaling pathways, since its transcript levels are significantly up-regulated by blue and red light (Beel et al., 2012). This observation raises the possibility that light- and temperature-controlled signaling pathways are connected in C. reinhardtii. To date, one red light-sensing and several blue light-sensing receptors have been described in C. reinhardtii, indicating that there may be more than one light signaling pathway (Kottke et al., 2017). These receptors are involved in different cellular processes such as the life cycle or circadian clock, which can be entrained not only by light/dark cycles but also by temperature cycles.

Our results for the temperature integration of C3 confirmed earlier data (Voytsekh et al., 2008), where C3 was found to be up-regulated in cells grown at 18°C compared with 28°C during early day (LD2) and early night (LD14). This up-regulation at 18°C is more obvious at LD4 (2 h later during the day) compared with LD14 (Fig. 4A). In the case of XRN1, our data differ from former studies where cells (strain 4A+) were used in the presence of the proteasome inhibitor MG132 (Dathe et al., 2012). We find that XRN1 is significantly up-regulated in cells grown at higher temperature (28°C), occurring only at early night (LD14). To compare directly, we conducted these experiments in the same wild-type strain SAG73.72 background as all other generated mutant lines. Moreover, we omitted the previously used proteasome inhibitor (Dathe et al., 2012). Consistent with previous results, XRN1 is degraded rapidly by the proteasome pathway during the day (LD4). This degradation is visualized by detecting one of its major degradation products around 140 kD (Fig. 4B). Unexpectedly, we found that XRN1 degradation is reduced significantly during early night in cells grown at 28°C, showing that XRN1 degradation is influenced by both the time of day and temperature. While temperature-dependent up-regulation of C3 is regulated at the transcriptional level (Voytsekh et al., 2008; Seitz et al., 2010), XRN1 regulation seems to occur at the posttranslational level (Fig. 4B; Dathe et al., 2012).

A novel finding in this study is the identification of 25 putative Musashi splicing variants throughout the day/night cycle, some of which have only small differences in sequence (for details, see the alignments of amino acids sequences in Supplemental Fig. S9). Several of them have a molecular mass close to 60 kD. Intriguingly, a 60-kD Musashi binds to the (UG)₇-bearing RNA (Fig. 3B), and a proteome analysis identified several unique peptides that belong to most of the variants with a molecular mass of 60 kD. All these variants bear three RRM domains, suggesting that only the three-RRM-containing Musashi variants can bind to the (UG)₇ repeat sequence at LD14. Since variant 12 is the major form at LD14 (at least at 23°C; Supplemental Table S4) and the amount of Musashi transcript covering most variants does not change significantly from 18°C to 28°C (Fig. 4D), we hypothesize that the up-regulation of the 60-kD Musashi protein at LD14 at 28°C (Fig. 4C) is due to the thermosensitive translational regulation of variant 12.

Temperature-dependent translational regulation was reported recently. Mutations in the eukaryotic initiation factor eIF5B1 in Arabidopsis confer translational control in heat acclimation (Zhang et al., 2017). The underlying thermotolerance-defective hot3-1 mutant inhibits translational initiation during the recovery from heat stress. A hot3-2 mutant covering a second eIFB1 allele also has effects on plant growth and development. In D. melanogaster, initiation factor 4E-binding protein regulates the effects of ambient temperature on metabolism and lifespan (Carvalho et al., 2017). Overexpression of C. reinhardtii Musashi variant 12 and its effect on reduced cell viability also might involve the translational machinery, given that mammalian Musashi proteins can switch from a repressor to an activator of target cell-specific mRNA translation and regulate the cell cycle (MacNicol et al., 2011).

We do not know the role of the multiple variants of Musashi, nor whether they are all present at the protein level. The identified peptides from proteome analysis all map to variant 12. However, the peptides from the N-terminal portion of variant 12 also map to most of the other variants, which are overall quite similar (Fig. 1; Supplemental Fig. S9). Thus, the detected protein bands in the immunoblots with anti-Musashi antibodies could represent different Musashi variants in C. reinhardtii (Supplemental Fig. S1B, right), but they also may be posttranslationally modified versions of the variants. Alternatively, they may include other proteins, nonspecifically recognized by the polyclonal antibody. Notably, human Musashi2 is subject to site-specific phosphorylation, thus converting Musashi2 from a repressor to an activator of target mRNA translation. An alternatively spliced form of human Musashi2 (variant 2) was found to lack the regulatory phosphorylation sites and, thus, is not able to promote the translation of target mRNAs (MacNicol et al., 2017). Alternative pre-mRNA splicing events emerge as an important layer of regulation upon exposure to biotic and abiotic stresses as well as in the circadian system in plants (Meyer et al., 2015). Small changes in ambient temperature can influence alternative splicing in Arabidopsis (Streitner et al., 2013). A putative splice regulator, Porcupine, was identified recently as a temperature-specific regulator of development in Arabidopsis (Capovilla et al., 2018). Also, the central clock component Late Elongated Hypocotyl is subject to temperature-dependent splicing of its transcripts within the 5′ UTR (James et al., 2018). In silico analyses also have found that alternative splicing occurs in C. reinhardtii, whereby the most prominent splicing event was intron retention (Labadorf et al., 2010; Raj-Kumar et al., 2017).

It should be reemphasized that, initially, the 60-kD form of Musashi was not identified by LC-ESI-MS/MS experiments because the databases lacked information on its gene model. We later detected that this form indeed coimmunoprecipitates with XRN1. We did not find unique peptides in this assay belonging exclusively to the C terminus of the 77-kD form (Cre16.g662702.t1.1), and this predicted variant also was not found in the Musashi RNAseq data; only the closely related variant 24 was found (Fig. 1; Supplemental Fig. S2). With an artificial gene representing Cre16.g662702.t1.1, we showed that the 77-kD Musashi (Cre16.g662702.t1.1) also interacts with XRN1 and C3 similarly to the 60-kD Musashi variant 12. We conclude that the N-terminal region shared by these two variants, containing the first two RRM domains, is responsible for this interaction. The involvement of RNA-binding proteins in thermal regulation also has been shown in land plants. For example, the RNA-binding protein FCA from Arabidopsis is involved in the thermal adaptation of stem growth via auxin, in combination with PHYTOCHROME-INTERACTING FACTOR4 as well as YUCCA, a gene encoding an enzyme involved in auxin biosynthesis (Lee et al., 2014). The chloroplast RNA-binding proteins CP31A and CP29A are important for the resistance of chloroplast development to cold stress and are needed for the stability of several mRNAs at low temperature (Kupsch et al., 2012). Here, we focused on members of RNA metabolism in microalgae and their influence on thermal acclimation. This RNA metabolism protein network mediating thermal acclimation in C. reinhardtii is interesting in several respects. c3 and xrn1 knockdown and knockout mutant lines show that the lack of these proteins strongly affects the viability of C. reinhardtii when exposed to 18°C, 28°C, and 37°C in the dark (Figs. 5, C, F, and G, and 6C). Future work must determine if temperature shifts during the light lead to the same result.

In contrast to C3 and XRN1, overexpression of the 60-kD Musashi variant 12 affects the viability of C. reinhardtii to some extent at 18°C and especially at 28°C, but not at 37°C (Fig. 6F; Supplemental Fig. S8C). For the studies on Musashi, we focused on the 60-kD major splicing variant 12 and used overexpression lines to specifically target the 60-kD form that may have a different phenotype than the knockdown or knockout lines used for c3 and xrn1. It also is possible that Musashi affects thermal acclimation in a different temperature range than C3 and XRN1. In the future, additional knockdown or knockout strains for Musashi may be of interest.

The similarity between C3 and the 60-kD Musashi variant 12 is another interesting feature, both bearing three RRM domains and containing numerous Met and Gly residues in the linker region between the second and third RRM domains (Fig. 3D). They are both able to bind to (UG)₇ repeat RNAs and may share similar RNA-binding sites and affect similar or even the same RNAs in their metabolism. However, their mode of regulation is exactly the opposite. At the protein level, C3 is up-regulated in cells grown at 18°C, especially at LD4, whereas the 60-kD Musashi variant 12 is up-regulated at the protein level in cells grown at 28°C at LD14. This observation raises the possibility that C3 and the 60-kD Musashi variant 12 may act as counterbalancing members within a temperature-controlled network including XRN1. Their interaction with XRN1 might affect translational efficiency by changing the storage of UG-bearing mRNAs (e.g. in processing bodies) in a temperature-dependent manner, thus preventing mRNA degradation or their accessibility to the translational machinery. Similar to mammalian Musashi, C. reinhardtii Musashis also may act as activators or repressors of translation. Such scenarios will be of great interest in the future.

MATERIALS AND METHODS

Strains and Culture Conditions

Chlamydomonas reinhardtii strain SAG73.72 mt⁺ was routinely used as the wild type. Other strains were wild-type 4A+ and the underlying xrn1_mut strain (Dathe et al., 2012) as well as two c3 silencing strains (c3-sil₆ and c3-sil₇) in the background of SAG73.72 (Iliev et al., 2006). Cultures were grown under LD12:12 with a light intensity of 75 μmol m⁻² s⁻¹ in Tris-acetate-phosphate (TAP) medium (Harris, 1989) at the indicated temperatures. Cell harvest was either at LD4, 4 h after the beginning of the light period, or at LD14, 2 h after the beginning of the dark period.

Coimmunoprecipitation Assays for Detecting XRN1 Interaction Partners

Protein crude extracts of two biological replicates were prepared from wild-type strain 4A+ or the xrn1_mut grown at 18°C and harvested at LD14 with 10 μm proteasome inhibitor (MG132; Sigma) added at LD13. Immunoprecipitations were performed with the Pierce coimmunoprecipitation kit (Thermo Scientific) according to the manufacturer’s instructions with some modifications using 400 μg of purified anti-XRN1 antibody (Dathe et al., 2012). The elution steps were performed with 0.1 m Gly (pH 10.5) at room temperature followed by 2× SDS incubation for 5 min at 90°C.

In-Gel Tryptic Digestion, Nano-LC-ESI-MS/MS, and Data Analysis

Proteins present in the eluates from the coimmunoprecipitation experiments were separated by 10% SDS-PAGE containing 0.3% piperazine diacrylamide as a cross-linker (Wagner et al., 2004; Seitz et al., 2010). The gel was stained using the Colloidal Blue Staining Kit (Invitrogen). Whole lanes were cut into slices (1 cm × 0.5 cm) and destained; the proteins within those slices were in-gel tryptic digested as described previously (Schmidt et al., 2006). The tryptic peptides were resuspended in 5 μL of 5% (v/v) acetonitrile/0.1% (v/v) formic acid and subjected to nano-LC-ESI-MS using an UltiMate 3000 nano HPLC device (Dionex, now part of Thermo Fisher Scientific) with a flow rate of 300 nL min⁻¹ coupled online with a linear ion trap ESI mass spectrometer (Finnigian LTQ; Thermo Electron) as described (Schmidt et al., 2006; Eitzinger et al., 2015). The mass spectrometer cycled between one full MS and MS/MS scans of the four most abundant ions. After each cycle, these peptide masses were excluded from analysis for 3 min (Schmidt et al., 2006). Data were analyzed according to Boesger et al. (2009) using the Proteome Discoverer software (version 1.0 or version 1.4 for the Musashi variants) from Thermo Electron, including the SEQUEST algorithm (Link et al., 1999). Searches were completed for tryptic peptides allowing two missed cleavages. Detection of variable Met oxidation (+15.995) was enabled. The parameters for all database searches were set to achieve a false discovery rate ≤ 1% for each analysis. Data were originally searched against the JGI v3.1 database at Phytozome v9.1, and the accession numbers were updated using the C. reinhardtii v5.5 database hosted by the Phytozome v12 Web site. The protein sequences of the gene models were compared with the National Center for Biotechnology Information protein database by use of BLASTp to determine annotation and homologies of the proteins as well as the prediction of functional domains.

RNAseq Analysis

The RNAseq data (Zones et al., 2015) were uploaded and processed using the Galaxy (Afgan et al., 2016) Web platform (usegalaxy.org). The data were first filtered and clipped for quality control using Trimmomatic (Bolger et al., 2014) and the following parameters: paired-end mode; illumincaclip; adapter sequences, TruSeq3 (additional seqs); maximum mismatch count, 2; accuracy between the two adapter ligated, 30; accuracy between any adapter, 10; trim the leading and trailing nucleotides with less than a score of 20; sliding window of four nucleotides with at least a score of 17; minimum average read quality of 17; minimum length of 50 nucleotides.

The data for each time point and its duplicates were treated independently. The filtered data were then aligned to the gDNA of Musashi (Cre16.g662702) from Phytozome 12 (Goodstein et al., 2012) of the C. reinhardtii genome v5.5 using HISAT2 (Kim et al., 2015) with the default parameters of the software. The resulting alignments in BAM format were filtered by discarding the unmapped reads. The filtered BAM files then were processed with Cufflinks adjusting the following parameters: min isoform fraction, 0.01; pre-mRNA fraction 0; perform bias correction using Musashi gDNA (Cre16.g662702); inner mean distance, 250 nucleotides; inner distance sd, 150 nucleotides; intronic overhang tolerance, six nucleotides; minimal allowed intron size, 20 nucleotides.

For each processed file, in this case each LD time point, Cufflinks generated a GTF file. The GTF files of all time points were collapsed into a single file, including the information of all the GTF files and adding one extra first column of data containing the name of the file.

With an in-house script (Supplemental Text File S1), the collapsed GTF file was processed and converted to a GenBank file using the EMBOSS SeqRet engine (Rice et al., 2000). With the GenBank file, the mRNA sequence for each of the predicted splicing variants was extracted using the GenBank Feature Extractor (Stothard, 2000). The ORF, the coding sequence, and the protein sequence of each mRNA were predicted using OrfPredictor (Min et al., 2005).

Using another in-house script (Supplemental Text File S2), the mRNAs giving equivalent proteins were grouped together, and the abundance in transcripts per million was calculated from the fragments per kilobase of exon per million reads mapped given by Cufflinks using the formula as derived by Pachter (2011). Each of the unique proteins was assigned a number, and the expression for each time point was represented in Supplemental Table S4 for each splicing variant. Antisense transcripts (11 transcripts) were not considered for this analysis.

From the results obtained, all splicing variants with an abundance lower than 1.5% were omitted, as well as transcripts coding for peptides exclusively outside the coding sequence of the predicted 60-kD Musashi variant 12 (confirmed by cDNA clone AV639968; Supplemental Fig. S3B) or the 77-kD Musashi (Cre.g662702.t1.1) from Phytozome.

Phylogenetic Tree of Musashi Proteins

The 60-kD Musashi sequence of variant 12 (Supplemental Fig. S3A) was used as a query to search against the nonredundant database from the National Center for Biotechnology Information, and representative sequences of Musashi-like proteins from mammals, insects, green and red algae, fungi, higher plants, and others (Supplemental Fig. S10) were collected. C3 from C. reinhardtii and homolog proteins of other plants were used as the outgroup. Sequences of homologs from Physcomitrella patens were taken from Phytozome v12 (Physcomitrella patens v3.3) The MEGA 6 software package (Tamura et al., 2013) was used for the alignment of the protein sequences with its ClustalW algorithm and phylogenetic tree construction with default parameters: a gap opening penalty of 10 and gap extension penalties of 0.1 and 0.2 were chosen for pairwise and multiple alignments, respectively; the Gonnet weight matrix was set with residue-specific penalties ON, hydrophilic penalties ON, a gap separation distance of 4, end gap separation OFF, negative matrix OFF, and a delay divergent cutoff of 30%. Parameters for phylogenetic tree construction were Poisson correction model, complete deletion of gaps/missing data, and uniform rates among sites. The neighbor-joining method (Saitou and Nei, 1987) was employed for the construction of a bootstrap consensus tree. Bootstrap values were calculated with 1,000 replications and expressed as a percentage.

Yeast Two-Hybrid Assay

Plasmids used for yeast two-hybrid analysis are based on the host vectors pGADT7 and pGBKT7 from the Matchmaker GAL4 two-hybrid system 3 (Clontech). The preparation of plasmid constructs for expressing XRN1 fused to GAL4-AD was described previously (Dathe et al., 2012). To express C3 fused to GAL4-BD, C3 cDNA (Zhao et al., 2004) was amplified using oligonucleotides 5′-GTTATGGGAATGCAGAAGCC-3′ and 5′-AGGCCACAGCGCTATCATC-3′, ending 41 bp after the stop codon of C3, using pDI7 (Iliev et al., 2006) as a template. The PCR product was cloned into pBluescript KS(+) (Stratagene), which had been opened with SmaI and provided with a thymine overhang in the 3′ direction (Marchuk et al., 1991). The resulting plasmid was named pHD1.8. It was cut with BamHI and EcoRI, and the 1,379-bp C3 cDNA-containing fragment was introduced into pGBKT7, which had been opened with the same enzymes. The final plasmid was named pHD1.10 (GAL4-BD-C3). To express the 60-kD Musashi variant 12 in fusion to either GAL4-AD or GAL4-BD, a cDNA clone (AV639968) containing full-length 60-kD Musashi was obtained from the Kazusa DNA Research Institute (Supplemental Fig. S3B). The 60-kD Musashi variant 12 cDNA was amplified using the primers 5′-TAATATGAATTCCCCAGAATGGACAAAGAC-3′ carrying an EcoRI restriction site (underlined) and 5′-TAATATCTCGAGCCTAAACACAGCCAGAAC-3′ carrying an XhoI restriction site (underlined). The PCR product was digested with EcoRI and XhoI and ligated into pGADT7 that had been opened with the same restriction enzymes and into pGBKT7 that was opened with EcoRI and SalI. The obtained plasmids were named pHD1.25 (GAL4-AD-Musashi variant 12) and pHD1.26 (GAL4-BD-Musashi variant 12), respectively. Yeast was transformed according to the manual of the Matchmaker GAL4 two-hybrid system 3 containing yeast strains Y2HGold and Y187. Two-hybrid analysis was performed according to Dathe et al. (2012).

Heterologous Expression of Two Musashis (77-kD and 60-kD Variant 12) in Escherichia coli, Purification, and Antibody Production

The DNA sequence of codon-optimized His-tagged 77-kD Musashi (Supplemental Fig. S1A) was synthesized by Life Technologies and cloned into the NcoI and HindIII restriction sites of vector pET28a(+) (Novagen), resulting in expression vector pWS15. The vector was expressed in E. coli strain BL21(DE3) pLysE (Thermo Fisher Scientific) in Luria-Bertani medium with 50 mg L⁻¹ kanamycin. Musashi expression was induced with 200 μm isopropyl-β-d-1-thiogalactoside at an OD₆₀₀ between 0.5 and 0.6 and continued for 4 h at 30°C. Cells were harvested by centrifugation (4,000g, 5 min, 4°C), washed once with phosphate-buffered saline buffer (140 mm NaCl, 3 mm KCl, 10 mm Na₂HPO₄, and 2 mm KH₂PO₄, pH 7.4), and disrupted in lysis buffer (42 mm Na₂HPO₄, 8 mm NaH₂PO₄, 300 mm NaCl, 10 mm imidazole, 10% [v/v] glycerol, 0.5% [v/v] Triton X-100, 10 mm β-mercaptoethanol, 1 mg mL⁻¹ lysozyme, and 1× protease inhibitor cocktail [EDTA-free; Roche]) by sonication for 8 min on ice, which was applied sequentially with 10-s pulses and 10-s breaks in between. The debris was removed by centrifugation (16,000g, 30 min, 4°C); the supernatant was filtered through a 0.22-µm membrane and applied to a nickel affinity column (HisTrap HP; GE Healthcare) using FPLC at 4°C. The column was washed with two column volumes of Ni-NTA buffer (42 mm Na₂HPO₄, 8 mm NaH₂PO₄, 200 mm NaCl, 10% [v/v] glycerol, and 10 mm β-mercaptoethanol) containing 10 mm imidazole, followed by eluting gradually up to 500 mm imidazole. The eluate (eluted by 250 mm imidazole) was collected. Finally, a buffer exchange (42 mm Na₂HPO₄, 8 mm NaH₂PO₄, and 10% [v/v] glycerol) was performed with centrifugal filters (Merck Millipore; Amicon Ultra-4) of 50-kD cutoff to concentrate the protein.

Similar procedures were developed to overexpress and purify the His-tagged 60-kD Musashi except for some changes: codon-optimized DNA of His-tagged 60-kD Musashi (Supplemental Fig. S4) was synthesized and cloned into the NcoI and HindIII restriction sites of vector pET28a(+), resulting in expression vector pWS22; protein expression was induced with 1 mm isopropyl-β-d-1-thiogalactoside; the eluate (eluted in 175 mm imidazole) was collected; and buffer exchange was performed with centrifugal filters of 30-kD cutoff size to concentrate the protein.

The purified proteins were stored at −80°C and used for the pull-down assays. For antibody production, 1 mg of purified His-tagged 77-kD Musashi was used, and anti-Musashi antibodies were generated by Pineda Antikörper Service.

Dynabeads Pull-Down Assay Using Purified His-Tagged Proteins

For one assay, crude extracts from C. reinhardtii were prepared from a wild-type strain (SAG73.72) grown at 18°C and harvested at LD14. The proteasome inhibitor MG132 was added 1 h before harvesting (LD13) to a final concentration of 10 μm. Cold pull-down buffer (3.25 mm sodium phosphate buffer, pH 7.4, 70 mm NaCl, 5% [v/v] glycerol, and 1× protein inhibitor cocktail [EDTA-free; Roche]) was used for crude extract preparation. Cells were vortexed with glass beads (diameter of 0.45–0.5 mm; Sigma) five times at 4°C for 1 min; in between, there were 1-min breaks on ice. To collect the supernatant that contained the soluble proteins of the crude extract, a centrifugation step for 30 min with 16,000g at 4°C was done. Meanwhile, Dynabeads (cobalt-coated magnetic beads; Life Technologies) were vortexed in the bottle for 30 s, and 50 µL of the beads was transferred into a 1.5-mL Eppendorf tube, which was then placed for 2 min on a magnet (Promega); the supernatant was discarded, and the beads were washed three times with 300 µL of binding/wash buffer (50 mm sodium phosphate buffer, pH 8, 300 mm NaCl, 0.01% [v/v] Tween 20, and 1× protein inhibitor cocktail [EDTA-free; Roche]). Then, 80 µg of purified His-tagged proteins in 700 µL of binding/wash buffer was added to the washed beads, and they were rotated slightly for 10 min at 4°C to let the His-tagged proteins bind fully to the Dynabeads. Afterward, the beads were washed twice with 300 µL of binding/wash buffer followed by two washing steps with 300 µL of buffer 2 (50 mm sodium phosphate buffer, pH 8, 70 mm NaCl, 5% [v/v] glycerol, and 1× protein inhibitor cocktail [EDTA-free; Roche]). Next, 700 µL of crude extract from C. reinhardtii (about 2.5 mg of proteins) was added to the washed beads, and the mixture was rotated for 3 h at 4°C. The beads then were washed four times with 300 µL of buffer 2 for each step; for the last washing steps, 150 µL of supernatant was taken as W4. Finally, the beads were resuspended in 120 µL of elution buffer (300 mm imidazole, 50 mm sodium phosphate, pH 8, 300 mm NaCl, and 0.01% [v/v] Tween 20) and rotated for 15 min at 4°C; the whole supernatant was eluted.

In Vitro RNA Pull-Down Assay

This assay was performed according to Zhao et al. (2004) with some modifications. Crude extracts from C. reinhardtii were prepared as described above for the pull-down assay. At first, 1,000 μL of streptavidin-coated paramagnetic beads was washed four times with 500 μL of buffer B (1 m NaCl, 5 mm Tris-HCl, pH 7.5, and 0.5 mm EDTA, pH 8) by using a magnet. Then, 20 μg of biotinylated transcripts was added to the beads in a total volume of 200 μL with buffer B. After 1 h of incubation at room temperature, the nonbound transcripts were removed by 10 washing steps using 500 μL of buffer B for each step. Next, the beads were equilibrated eight times with 500 μL of buffer E (80 mm NaCl, 10 mm Tris-HCl, pH 7.5, 1 mm EDTA, pH 8, 5% [v/v] glycerol, 2 mm DTT, 0.5% [v/v] Igepal CA-630, 25 μg mL⁻¹ yeast tRNA [Sigma], and 1× protein inhibitor cocktail [EDTA-free; Roche]) for each step. Meanwhile, 1,000 μL of proteins from crude extracts (about 4 mg of proteins), which were isolated using extraction buffer (80 mm NaCl, 10 mm Tris-HCl, pH 7.5, 1 mm EDTA, pH 8, 5% [v/v] glycerol, 2 mm DTT, and 1× protein inhibitor cocktail [EDTA-free; Roche]), was incubated with 1 mg of polyguanylic acid (Sigma) as a nonspecific competitor for 20 min on ice; this reaction mixture then was added to the equilibrated streptavidin beads and incubated for 2 h at 4°C under slight agitation. The nonbound proteins were removed by five washing steps with buffer E (one time with 200 μL and four times with 500 μL). Afterward, the beads were washed another six times with 500 μL of buffer E− (lacking yeast tRNA) for each step; 300 µL of the supernatant was taken as W11 during the last washing step. Finally, the bound proteins were eluted with 210 μL of elution buffer (3 m NaCl, 10 mm Tris, pH 7.5, 1 mm EDTA, pH 8, and 5% [v/v] glycerol) by incubation under slight agitation for 45 min at 4°C; the whole supernatant was taken as eluate.

Crude Extracts and Immunoblots

C. reinhardtii soluble protein extracts were isolated, and immunoblotting for the detection of C3 and Musashi was performed as described previously (Zhao et al., 2004) with some modifications. Incubation of the nitrocellulose membranes with anti-C3 antibodies or anti-Musashi antibodies was performed overnight at an antibody dilution of 1:5,000 at 4°C. Horseradish peroxidase-conjugated anti-rabbit IgGs (Sigma) with a dilution of 1:6,666 were used as a secondary antibody, and peroxidase activity was detected by a chemiluminescence assay. The nitrocellulose membranes were stained with Ponceau S for loading controls. Immunoblotting to detect XRN1 was performed as described by Dathe et al. (2012), and the used polyvinylidene difluoride membranes were stained with Coomassie Brilliant Blue R 250 for the loading controls. Protein levels were quantified using ImageJ version 1.48v (National Institutes of Health).

Backcrossing of the xrn1_mut Strain to SAG73.72

The xrn1_mut strain was generated in the background of wild-type 4A+ (Dathe et al., 2012). It was first crossed with wild-type 21Gr two times, and an mt⁻ progeny was selected for further crosses with SAG73.72 (mt⁺) three times. Zygote germination and tetrad separation were completed according to the procedures of Jiang and Stern (2009).

Knockdown of XRN1 by amiRNA

Knockdown of XRN1 was achieved by usage of amiRNA following the protocol of Molnar et al. (2009). The design and off-target check of the oligonucleotides were conducted by the Web MicroRNA Designer version 3.1 from the Max Planck Institute for Developmental Biology (http://wmd3.weigelworld.org). 5ʹ-ctagtATGCCTTTGCCTAAAGACGAAtctcgctgatcggcaccatgggggtggtggtgatcagcgctaTTCGACTTTAGGCAAAGGCATg-3ʹ (amiRNA sequences are in uppercase letters) and 5ʹ-ctagcATGCCTTTGCCTAAAGTCGAAtagcgctgatcaccaccacccccatggtgccgatcagcgagaTTCGTCTTTAGGCAAAGGCATa-3ʹ oligonucleotides were aligned, phosphorylated, and ligated into the digested (with SpeI) and dephosphorylated vector pChlamyRNA3int (Chlamydomonas Resource Center), resulting in pJE1. Orientation of the insert was checked by gel electrophoresis, and sequence integrity was verified by sequencing. pChlamiRNA3int carrying the AphVIII gene was used as a control vector. Plasmid linearization was done by ScaI and used for transformation in C. reinhardtii. Linearized plasmid DNA (1 and 3 µg) was used for the transformation of C. reinhardtii wild-type strain SAG73.72 as described (Iliev et al., 2006). Paromomycin-resistant clones were grown on TAP agar plates with 50 mg L⁻¹ paromomycin sulfate salt (Sigma).

Overexpression of 60-kD Musashi in C. reinhardtii

Full Musashi DNA encoding the 60-kD form (Supplemental Fig. S7A) was synthesized by Life Technology and cloned into the XhoI and XbaI restriction sites of pSK1, which carries the tandem promoter HSP70A/RBCS2 followed by the first intron of RBCS2; the resulting vector was named pWS18. Then, pWS18 was digested with XbaI and KpnI and ligated with pHD-AEQ2, which had been digested with the same enzymes to give the final vector pWS19. pHD-AEQ2 carries the 3ʹ UTR of RACK1 and the Aph7ʹʹ gene cassette (Berthold et al., 2002). The schematic view of the overexpression vector of 60-kD Musashi is shown in Supplemental Figure S7B. Transformation of the linearized pWS19 (digested with SapI and PsiI) in C. reinhardtii was carried out as described above. Hygromycin-resistant clones were grown on TAP agar plates with 10 mg L⁻¹ hygromycin B (Sigma). The Musashi expression level in the transgenic lines was checked by immunoblots.

Temperature Assays

Different C. reinhardtii strains were grown in 50 mL of liquid TAP medium at 23°C under constant stirring at 250 rpm and under an LD12:12 with a light intensity of 75 μmol m⁻² s⁻¹. During the third cycle at LD11, the cell number was determined with a hemocytometer (Marienfeld-Superior) and adjusted to 1.5 × 10⁶ cells mL⁻¹ with liquid TAP medium; 1 mL of the cells was prepared in 2-mL microcentrifuge tubes. At LD12, cells were placed at 18°C, 28°C, 37°C, and 42°C and incubated for 12 h in darkness. The following day at LD0, the differentially incubated cells were diluted 1:20,000 and spread evenly onto TAP agar plates; three biological replicates and three technical replicates for each biological replicate were performed. Cells were then grown at 23°C under an LD12:12 with a light intensity of 75 μmol m⁻² s⁻¹ for 6 d, and the number of colonies for each plate was determined. For each assay, colony-forming units of wild-type cells incubated at 18°C were set to 100% (n = 9), and statistically significant differences were calculated by Student’s t test.

Musashi Transcript Analysis by RT-qPCR

Total RNA was isolated from C. reinhardtii strain SAG73.72 using the RNeasy Plant Mini Kit (Qiagen) and on-column digestion of DNA during purification with RNase-free DNase (Qiagen). An amount of 330 ng of total RNA was used per reaction in one-step RT-qPCR with primers for Musashi (forward, 5ʹ-gcgctgaagaacatctttgc-3ʹ; reverse, 5ʹ-aggttgttgttcatcatgctc-3ʹ, shown in Fig. 1, in pink) and for RACK1 (Mus et al., 2007) as an internal reference gene using QuantiTect SYBR Green RT-PCR (Qiagen) in the Mx3005P (Agilent Technologies) real-time PCR system. Relative transcript abundances for Musashi were calculated according to the 2^−ΔΔCT method (Livak and Schmittgen, 2001).

Biological Replicates

Biological replicates were done with different cultures side by side, unless stated otherwise.

Accession Numbers

Musashi sequence data from this article can be found in the GenBank/EMBL data libraries under https://trace.ncbi.nlm.nih.gov/Traces/sra/?study=SRP061735: identifiers are SRP061735 (Sequence Read Archive) and GSE71469 (Gene Expression Omnibus). The 77-kD Musashi (Cre16.g662702.t1.1) as well as C3 (Cre03.g177200.t1.1) and XRN1 (Cre06.g280050.t1.1) can be found in the C. reinhardtii v5.5 database hosted by the Phytozome v12 Web site (https://phytozome.jgi.doe.gov/pz/portal.html).

Supplemental Data

The following supplemental materials are available.

• Supplemental Figure S1. Sequence of codon-adapted His-tagged 77-kD Musashi and characterization of heterologously expressed 77-kD Musashi from E. coli.

• Supplemental Figure S2. Comparison of the 77-kD Musashi and Musashi variant 24.

• Supplemental Figure S3. Protein sequence of the 60-kD Musashi variant 12 and DNA sequences of cDNA clones (AV639968 and AV641618) compared with Cre16.g662702.t1.1.

• Supplemental Figure S4. Sequence of codon-adapted His-tagged 60-kD Musashi variant 12.

• Supplemental Figure S5. Total Musashi expression level based on RNAseq analysis (Zones et al., 2015).

• Supplemental Figure S6. Verification of vector controls for the temperature assays in comparison with the wild type.

• Supplemental Figure S7. Sequence of the 60-kD Musashi ORF used for overexpression in C. reinhardtii and schematic view of the vector for overexpression of 60-kD Musashi variant 12.

• Supplemental Figure S8. Protein amount quantification and cell viability assays of a 60-kD Musashi variant 12 overexpression strain.

• Supplemental Figure S9. Protein sequence alignments of different Musashi variants (related to Fig. 1).

• Supplemental Figure S10. Alignment of the sequences used for the phylogenetic tree of Musashi proteins (related to Fig. 3D).

• Supplemental Table S1. Potential interaction partners of XRN1 in addition to those given in Table 1.

• Supplemental Table S2. List of proteins and their identified peptides interacting with XRN1.

• Supplemental Table S3. LC-ESI-MS/MS analysis of five bands of heterologously expressed 77-kD Musashi with a C-terminal 6×His tag (related to Supplemental Fig. S1C.)

• Supplemental Table S4. Abundance of different Musashi splicing variants over a diurnal cycle LD12:12.

• Supplemental Table S5. LC-ESI-MS/MS analysis of proteins from the eluate of the coimmunoprecipitation using the protein sequences of the 25 Musashi variants.

• Supplemental Table S6. LC-ESI-MS/MS analysis of proteins from the eluate of the RNA pull down using the protein sequences of the 25 Musashi variants.

• Supplemental Text File S1. Convert GTF into GenBank.

• Supplemental Text File S2. Generate expression reports.