Abstract
mRNA transport determines spatiotemporal protein expression. Transport units are higher-order ribonucleoprotein complexes containing cargo mRNAs, RNA-binding proteins and accessory proteins. Endosomal mRNA transport in fungal hyphae belongs to the best-studied translocation mechanisms. Although several factors are known, additional core components are missing. Here, we describe the 232 kDa protein Upa2 containing multiple PAM2 motifs (poly[A]-binding protein [Pab1] associated motif 2) as a novel core component. Loss of Upa2 disturbs transport of cargo mRNAs and associated Pab1. Upa2 is present on almost all transport endosomes in an mRNA dependent-manner. Surprisingly, all four PAM2 motifs are dispensable for function during unipolar hyphal growth. Instead, Upa2 harbours a novel N-terminal effector domain as important functional determinant as well as a C-terminal GWW motif for specific endosomal localisation. In essence, Upa2 meets all the criteria of a novel core component of endosomal mRNA transport and appears to carry out crucial scaffolding functions.
Introduction
Active transport of mRNAs determines when and where proteins are synthesised. Such trafficking events are important for a wide variety of different cellular processes like asymmetric cell division, polar growth, embryonic development and neuronal activity (Eliscovich & Singer, 2017, Martin & Ephrussi, 2009). Several mRNA translocation mechanisms have been described. During cytokinesis of Saccharomyces cerevisiae, for example, the actin-dependent transport of ASH1 mRNA is mediated by the concerted binding of the RNA-binding proteins (RBP) She2p and She3p. These RBPs connect the cargo mRNA to the myosin motor Myo4 for transport towards the daughter cell (Edelmann et al., 2017, Niessing et al., 2018). In highly polarised cells such as fungal hyphae and neurons, mRNAs are transported along microtubules over long distances. Here, molecular motors such as kinesins and dynein are involved. Among the best studied examples of microtubule-dependent translocation is the mRNA transport on shuttling endosomes during polar hyphal growth in the fungus Ustilago maydis (Mofatteh & Bullock, 2017).
Upon infection of corn, U. maydis switches from budding to hyphal growth (Haag et al., 2015, Vollmeister et al., 2012). The resulting infectious hyphae grow with a defined axis of polarity: they expand at the apical pole and insert septa at the basal pole resulting in the formation of characteristic sections devoid of cytoplasm. In this growth mode, hyphae depend on active transport along microtubules. Loss of long-distance transport results in aberrant hyphal growth. Characteristic of this defect is the formation of bipolarly growing cells. Important carriers are Rab5a-positive endosomes that shuttle along microtubules by the concerted action of the plus-end directed Kinesin-3 type motor Kin3 and the minus-end directed cytoplasmic dynein Dyn1/2 (Baumann et al., 2012, Schuster et al., 2011). These endosomes carry characteristic markers of early endosomes involved in endocytosis (Haag et al., 2017). However, during polar growth they also function as transport endosomes moving organelles, like peroxisomes, and mRNAs with associated ribosomes attached to their cytoplasmic surface (Baumann et al., 2014, Higuchi et al., 2014).
The key RNA-binding protein for mRNA transport is Rrm4, an RRM (RNA recognition motif) protein containing tandem N-terminal RRMs separated by a spacer from a third RRM domain (Figure 1A). The mRNAs of all four septins were identified to be important cargo mRNAs, which are most likely translationally active while being transported on endosomes (Baumann et al., 2014, König et al., 2009, Zander et al., 2016). Consequently, the translation products Cdc3, Cdc10, Cdc11 and Cdc12 form heteromeric complexes on the cytoplasmic surface of endosomes in an Rrm4-dependent manner. These septin complexes are delivered to the hyphal growth pole to form a longitudinal gradient of filaments (Baumann et al., 2014, Zander et al., 2016). A recent transcriptome-wide view of endosomal mRNA transport revealed that Rrm4 binds thousands of mRNAs preferentially in their 3’ UTR in close proximity to the small glycine rich protein Grp1. This extensive mRNA transport is most likely needed for the distribution of mRNAs within the fast-growing hypha as well as for the transport of specific mRNAs encoding e.g. septins for heterooligomerisation (Olgeiser et al. in revision).
At its C-terminus, Rrm4 carries two MademoiseLLE (MLLE) domains that function as peptide binding pockets for specific interaction with the PAM2-like motifs of Upa1. Upa1 links Rrm4-containing mRNPs to transport endosomes using a FYVE zinc finger domain to recognise phosphatidylinositol-3-phosphate lipids of early endosomes. (Pohlmann et al., 2015). In addition, Upa1 also carries a classical PAM2 motif for interaction with the MLLE domain of the poly(A)-binding protein Pab1, an additional component of endosomal mRNPs (Becht et al., 2006, Haag et al., 2015). Although we already identified a number of components of endosomal mRNA transport, key factors might still be missing. Here, we unravelled and characterized one such factor, the protein Upa2, a second Ustilago PAM2 motif protein that had previously been identified by bioinformatic prediction of PAM2-containing proteins (Pohlmann et al., 2015).
Results
The PAM2-containing protein Upa2 interacts with Pab1
Upa2 (Ustilago PAM2 protein 2; UMAG_10350) is a 2121 amino acid (aa) protein with a conserved coiled coil domain of unknown function at its C-terminus, as well as four PAM2 motifs for potential interaction with the MLLE domain of Pab1. One PAM2 motif is situated at the immediate N-terminus and three in the central region of the protein (Figure 1A-B).
In order to validate the predicted PAM2 motifs we performed yeast two-hybrid studies that have already been successfully applied to demonstrate an interaction between Pab1 and the PAM2 motif of Upa1 (Pohlmann et al., 2015). Full-length Upa2 shows weak interaction with Pab1 but did not interact with Rrm4 (Figure 1C, Supplementary Figure S1A). The latter is consistent with the observation that PAM2-like motifs for Rrm4 interaction are missing in Upa2 (Pohlmann et al., 2015). Upon mapping the interaction domain of Upa2 with Pab1 we observed that the PAM2-containing N-terminal part of Upa2 (aa 1-1216) but not the C-terminal part (aa 1217-2121) interacted with Pab1 (Figure 1C). The interaction of Upa2 with Pab1 was mediated by the MLLE domain of Pab1, since this domain was necessary and sufficient for reporter gene expression in the yeast two-hybrid system (Figure 1C, Supplementary Figure S1A). Mutational analysis of the PAM2 motifs in Upa2 revealed that a single PAM2 motif was sufficient for interaction with the MLLE domain of Pab1 or full length Pab1 (Supplementary Figure S1A). Mutating all PAM2 motifs resulted in loss of interaction of Upa2 with the MLLE domain of Pab1 (Figure 1C) and strongly reduced interaction with Upa2 and full length Pab1 (Supplementary Figure S1A).
To verify this binding behaviour, we tested the interaction of the central PAM2 triplet of Upa2 (aa 834-1216) with the MLLE domains of Pab1 by GST-pulldown experiments using variants expressed in Escherichia coli. These experiments confirmed the specific interaction of the PAM2 motifs of Upa2 with the MLLE domain of Pab1. The interaction strength was dependent on the number of functional PAM2 motifs (Figure 1D). As expected, PAM2 motifs of Upa2 did not interact with Rrm4 (Supplementary Figure S1B-C).
Since coiled coil domains have been described as dimerization regions, we tested this in a yeast two-hybrid assay. When appended to both BD and AD reporter proteins the predicted C-terminal coiled coil domain of Upa2 supported yeast growth on selective medium suggesting that this region indeed functions as a dimerization domain (Figure 1E). Taken together, Upa2 contains multiple functional PAM2 motifs for interaction with Pab1 and a C-terminal dimerization domain. The interaction with Pab1 is the first hind that Upa2 might function in endosomal mRNA transport.
Upa2 is essential for efficient unipolar hyphal growth
If Upa2 is indeed important for endosomal mRNA transport, loss of Upa2 should exhibit phenotypes similar to mutations in the previously identified transport components Rrm4 and Upa1 (Pohlmann et al., 2015). To test this assumption, we generated upa2 deletion mutants in the genetic background of laboratory strain AB33. In this strain, hyphal growth can be elicited synchronously and reproducibly by switching the nitrogen source in the medium (Brachmann et al., 2001). Hyphae grow with a defined axis of polarity: they expand at the apical tip and insert septa at their base leading to the formation of characteristic empty sections (Figure 2A). While loss of Upa2 did not affect growth or alter shape of yeast cells (Supplementary Figure S2A-B) upa2Δ strains exhibited a bipolar growth phenotype and the formation of empty sections was delayed. This aberrant growth mode is typical for defects in microtubule-dependent transport and has been described for loss of the key proteins of endosomal mRNA transport like Rrm4 and Upa1 (Figure 2A-B). Analysing the upa1Δ/upa2Δ double mutant revealed that the number of hyphae exhibiting aberrant bipolar hyphal growth was not additive (Supplementary Figure S2), providing genetic evidence that both PAM2-containing proteins function in the same cellular process.
A second read-out for defects in the endosomal mRNA transport machinery is the unconventional secretion of chitinase Cts1 (Koepke et al., 2011, Pohlmann et al., 2015). Loss of Upa2 resulted in reduced extracellular Cts1 activity specifically during hyphal growth. This is reminiscent of defective Cts1 secretion upon deletion of rrm4 and upa1 (Supplementary Figure S2D). Hence, Upa2 might function in concert with Rrm4 and Upa1 during endosomal mRNA transport.
This hypothesis was supported by a phylogenetic analysis showing that orthologs of Upa2 were found in several basidiomycete fungi with comparable spacing of the predicted PAM2 motifs (Figure 2C). Consistent with the notion that Rrm4, Upa1 and Upa2 function together, these fungi also had orthologs for Rrm4 and Upa1. Notably, Upa2 was absent in fungi, which presumably have lost the endosomal mRNA transport machinery, like Malassezia globosa and Cryptococcus neoformans var. neoformans. The latter only possesses diverged orthologs of Rrm4 and Upa1 lacking key functional domains, while M. globosa is even lacking a clear ortholog of Rrm4 (Figure 2C). In essence, the evolutionarily conserved Upa2 is a novel factor essential for efficient unipolar growth that might function in endosomal mRNA transport.
Upa2 shuttles on Rrm4-positive transport endosomes
upa2 and rrm4 appear to be genetically and phyogenetically linked. We therefore examined whether the protein can also be found on transport endosomes by expressing a functional version of Upa2 with C-terminally fused Gfp (eGfp, Clontech; Figure 2A-B; Supplementary Figure S2D). Fluorescence microscopy revealed that in hyphae Upa2-Gfp localised exclusively in bidirectionally moving units (Figure 3A, Supplementary Video V1). Movement was inhibited by benomyl indicating that it was microtubule-dependent (Figure 3A, Supplementary Video V1). Importantly, Upa2-Gfp shuttling resembled the movement of Rrm4-Gfp and Pab1-Gfp (Figure 3A, Supplementary Video V1). This holds true for the amount of processive signals as well as for their velocity (Figure 3B-C). Hence, Upa2 appeared to shuttle on Rrm4-positive transport endosomes (Baumann et al., 2012, Pohlmann et al., 2015).
For experimental verification of endosomal shuttling we stained processive endosomes with the lipophilic dye FM4-64 (Haag et al., 2017). Using dual-colour dynamic live imaging, we detected extensive co-localisation of Upa2-Gfp and FM4-64-stained, processively moving endosomes (Figure 3D-E). Furthermore, we tested strains co-expressing Upa2-Gfp and Rrm4-mCherry or Pab1-mCherry (functional C-terminal fusions with the monomeric red fluorescent protein mCherry; Baumann et al., 2014, König et al., 2009). In this instant we also observed extensive co-localisation of Upa2 and Rrm4 (Figure 3D-E, Supplementary Video V2) indicating that Upa2 was present on almost all shuttling endosomes (Baumann et al., 2012, Pohlmann et al., 2015). Upa2-Gfp also co-localised with processively moving Pab1-mCherry (Figure 3D-E). However, unlike Rrm4, Pab1-mCherry additionally localised in the cytoplasm resulting in diffuse and static signals that are easily detectable around the region of the nucleus (Figure 3A, D; Baumann et al., 2014, König et al., 2009). Notably, Upa2 did not exhibit this cytoplasmic staining indicating that it is not present in cytoplasmic Pab1-containing complexes, but specifically interacts with endosome-associated Pab1 (Figure 3A, D, E). In essence, Upa2 shuttles exclusively on almost all Rrm4-positive transport endosomes supporting the notion that it is a new component of endosomal mRNPs.
Endosomal localisation of Upa2 depends on the RNA-binding capacity of Rrm4
Since Upa2 proved to be an endosomal protein, we investigated the functional relationship between Upa2 and Rrm4 by studying the subcellular localisation of endosomal mRNP components in rrm4Δ strains. As previously described, the endosomal localisation of Upa1-Gfp is not affected in rrm4Δ strains, because its interaction is mRNA-independent and mediated by its FYVE domain (Figure 4A-B, Supplementary Video V3; Pohlmann et al., 2015). In contrast, Pab1-Gfp is no longer detectable on shuttling endosomes in rrm4Δ strains indicating that without Rrm4 no mRNAs are transported on endosomes (Figure 4A-B, Supplementary Video V4; König et al., 2009). Interestingly, the endosomal localisation of Upa2 was also no longer detectable in rrm4Δ strains (Figure 4A-B, Supplementary Video V5) suggesting that the association of Upa2 depends on direct interaction with endosomal mRNP components.
To investigate a possible mRNA-dependent interaction of Upa2 with shuttling endosomes we studied the effect of the allele Rrm4mR1-Rfp (Rrm4 variant with C-terminally fused monomeric Rfp). Previously, it was shown that Rrm4mR1 carrying a loss of function mutation in the first RRM domain resulted in drastically reduced RNA binding of Rrm4 (Becht et al., 2006). Notably, Upa2 is no longer detectable on endosomes in a Rrm4mR1-Rfp background, whereas for Pab1 we observed a strong reduction in movement (Figure 4C-D, Supplementary Videos V6-7). This indicates that the recruitment of Upa2 to motile transport endosomes is dependent on the RNA-binding capacity of Rrm4. Taken together, although loss of the endosomal components Upa1 and Upa2 results in similar defects during hyphal growth, the mode of endosomal localisation is clearly different. Contrary to Upa1, endosomal localisation of Upa2 is mRNA-dependent, suggesting an interaction with mRNA either directly or indirectly through a protein component of mRNPs such as Pab1 that Upa2 binds via its PAM2 motifs.
Upa2 carries two functionally important regions, a N-terminal effector domain and a C-terminal GWW motif
Since the mRNA-dependent localisation could be explained by the direct PAM2-dependent interaction with Pab1, we tested the role of the identified PAM2 motifs of Upa2 with regards to function and endosomal localisation. Therefore, we expressed Upa2-Gfp variants in laboratory strain AB33 carrying mutations in one, in multiple and in all identified PAM2 motifs. Surprisingly, none of the PAM2 mutations affected the hyphal growth of U. maydis indicating that the interaction of Upa2 with Pab1 is not essential for its role during hyphal growth (Figure 5A-B). Furthermore, endosomal shuttling was not affected by mutating all four PAM2s indicating that the interaction with Pab1 is dispensable for endosomal localisation (Supplementary Figure S3A). Hence, Upa2 must contain so far unknown domains critical for endosomal transport functions.
To map these domains, we studied N-terminal truncations carrying C-terminal Gfp fusions (Figure 5C). Respective alleles were integrated at the homologous locus. Removal of the first 338 amino acids did not interfere with the function of the protein (Upa2-339-2121; Figure 5C-D). However, additional truncations resulted in increase of bipolar hyphae indicating reduced (Upa2-399-2121) or loss of function (Upa2-599-2121; Figure 5C-D). Importantly, endosomal shuttling of these versions and protein amounts were not drastically affected (Figure 5E; Supplementary Figure 5B). Hence, a currently unknown effector domain important for endosomal mRNA transport is located between amino acid position 339 and 599. Notably, the non-functional variant Upa2-599-2121 still contained three PAM2 motifs, consistent with the finding that the interaction with Pab1 is not sufficient for function.
For mapping of the domain that mediates endosomal localisation of Upa2, we assayed additional N-terminal truncations. All variants still shuttled indicating that the last 163 amino acids were sufficient for endosomal shuttling (Figure 5C, F). Upa2-1217-2121 did no longer contain any PAM2 region for interaction with Pab1 (Figure 5C). This was consistent with the finding that the PAM2 motifs of Upa2 were not essential for endosomal localisation. Closer inspection by sequence comparison revealed a conserved GWW sequence at the very C-terminus (Figure 5G), which was shown in other proteins to function in protein/protein interaction (Hardwick et al., 2011, Hardwick et al., 2012). Mutating this short sequence resulted in loss of shuttling without drastic changes in protein amounts (Figure 5F, Supplementary Figure S3B). This holds true when testing the mutation in the context of the full length protein (Figure 5H, Supplementary Video V8). Importantly, we also observed a loss of function phenotype for this mutation, i.e. an increased number of bipolar hyphae (Figure 5H, Supplementary Figure S3C). This was not due to altered protein amounts (Supplementary Figure 3B). Thus, the conserved C-terminal GWW motif is essential for endosomal localisation of Upa2 and endosomal localisation is important for the function of the protein. In essence, besides the PAM2 motifs, Upa2 contains a functionally important N-terminal effector domain and a C-terminal GWW motif for interaction with an endosomal mRNP component (see Discussion).
Loss of Upa2 causes defects in the formation of endosomal mRNPs
Finally, we studied the role of Upa2 during endosomal mRNA transport in closer detail. An important function of endosomal mRNA transport is the local assembly of septin heteromeric subunits that are transported to hyphal growth poles to form filaments with a gradient emanating from this pole (Baumann et al., 2014, Zander et al., 2016). Using a functional septin fusion protein Cdc3-Gfp as read-out (Cdc3 N-terminally fused to eGfp; Baumann et al., 2014) we observed that loss of Upa2 abolished endosomal shuttling of Cdc3-Gfp and as an expected consequence also the Cdc3-Gfp containing gradient at hyphal tips was no longer detectable (Figure 6A-C, Supplementary Video V9).
Studying septin mRNA transport with RNA live imaging (Baumann et al., 2015; Supplementary Figure S4A) revealed that the number of processively transported cdc3B16 mRNAs was reduced to about 50% (Figure 6D-E). While the velocity of the cdc3B16 containing mRNPs was not affected (Supplementary Figure S4B), the range of movement was reduced (Figure 6F) suggesting that mRNPs might fall off endosomes with higher frequency.
To address whether the defect is specific to the cdc3 mRNA we tested shuttling of other endosomal components. In case of Rrm4-Gfp we observed that Rrm4-Gfp still shuttles in upa2Δ strains, although a substantial fraction of the protein also stained structures that resembled microtubule bundles (Figure 6G; Supplementary Figure 4C). Consistently, in benomyl treated hyphae bundle-like Rrm4-Gfp signals were no longer detected (Supplementary Figure S4D-E). This suggests that Rrm4 localisation is disturbed in the absence of Upa2.
Observation of endosomal shuttling of Rrm4-Gfp, Pab1-Gfp, Upa1-Gfp and Rab5a-Gfp revealed that loss of Upa2 caused a slight increase in the velocity and amount of shuttling signals (Figure 6G-I, Supplementary Videos V10-13). There was only one exception: the amount of Pab1-containing endosomes was reduced to about 50%, suggesting that Upa2 is needed for a stable interaction of Pab1-containing mRNAs on endosomes. This is consistent with the reduced shuttling range of cdc3B16 mRNA as described above and the observation that Cdc3-Gfp protein is no longer present on endosomes in the absence of Upa2 (Zander et al., 2016; see Discussion). Importantly, Upa2 is essential for the correct association of mRNAs, Pab1, as well as Rrm4, on endosomes. Thus, we identified an endosome-specific and functionally important factor that functions as novel core component of endosomal mRNA transport.
Discussion
Functional units of mRNA transport are higher-order mRNPs consisting of cargo mRNAs, RNA-binding proteins and additional interacting proteins (Mofatteh & Bullock, 2017). A crucial task is to differentiate between core components that are essential to orchestrate transport, and accessory components. The latter might mediate for example translational regulation of cargo mRNAs. Here, we identified Upa2 as a novel core component of endosomal mRNA transport. This interactor of the poly(A)-binding protein appears to stabilise higher-order endosomal mRNPs during transport suggesting that it functions as a scaffold protein (Figure 7).
A novel multi PAM2 protein for endosomal mRNA transport
Upa2 is a large protein of 232 kDa that contains four functional PAM2 motifs for interaction with the C-terminal MLLE domain of Pab1. It is known from earlier structural studies that the C-terminal MLLE domain of human PABC1 forms a defined peptide pocket to accommodate various PAM2 sequences (Kozlov et al., 2001). Currently, only three MLLE-containing proteins are described: the poly(A)-binding protein, the E3 ubiquitin ligase UBR5 and mRNA transporter Rrm4. UBR5 is a HECT-type (homologous to the E6AP C terminus) ligase that functions in translational regulation and microRNA-mediated gene silencing (Su et al., 2011).
The PAM2 motif is present in various PABC1-interaction partners such as eRF3, GW182 and PAN3 functioning in various steps of posttranscriptional control such as translation, miRISC assembly and deadenylation, respectively (Xie et al., 2014). The vast majority of proteins harbour a single PAM2 motif, only eRF3 and Tob contain two overlapping PAM2 motifs for MLLE interaction (Kozlov & Gehring, 2010, Ezzeddine et al., 2007). Therefore, it is exceptional to find four PAM2 motifs in Upa2 and due to the potential dimerisation via its C-terminal coiled coil region Upa2 offers an extensive interaction platform for Pab1 (Figure 7).
Experimental evidence confirmed that the PAM2 motifs of Upa2 interact with Pab1. However, it seems to be an utter paradox that these evolutionarily conserved PAM2 motifs are not important for the function of Upa2 during endosomal mRNA transport. The same holds true for Upa1, where the PAM2 motif is also dispensable for function (Pohlmann et al., 2015). A possible explanation is a high level of redundancy. This is supported by the fact that in endosomal mRNPs there are verified PAM2 motifs in Upa1 and Upa2 as well as additional cryptic versions in Rrm4. It is conceivable that all PAM2 sequences and their variants must be mutated to observe defects in mRNA transport and consequently in hyphal growth. In fact, redundancy was already observed during the study of PAM2-like motifs. Only deletion of both PAM2-like sequences in Upa1 resulted in loss of function (Pohlmann et al., 2015). Alternatively, we might not be able to detect the functional significance of the PAM2 / Pab1 interaction under optimal laboratory conditions. The interaction could be of specific importance under certain stress conditions. Comparably, a function of the small Glycine rich protein Grp1 was only observed during hyphal growth under cell wall stress conditions (Olgeiser et al. in revision).
Besides PAM2 motifs and a coiled coil region for dimerization, we identified two additional functionally important regions. Firstly, we succeeded in mapping a novel N-terminal effector domain. Interestingly, this domain is embedded in a part of the protein that is predicted to be rich in intrinsically disordered regions (IDRs; Supplementary Figure S5). In the future, we need to carry out an extensive fine mapping to identify clear boundaries of this domain and identify the respective interaction partner. Secondly, we discovered a C-terminal GWW motif that is essential for the endosomal localisation of Upa2. Since the presence of Upa2 depends on the RNA-binding capacity of Rrm4, we assume that the GWW motif does not bind endosomal lipids directly. Most likely, it interacts with a protein component of endosomal mRNPs (Figure 7). This assumption is supported by the fact that an evolutionarily conserved GWW motif at the C-terminus of bacterial endoribonuclease RNase E interacts with a peptide-binding pocket of a PNPase (exoribonuclease polynucleotide phosphorylase; Hardwick et al., 2011, Hardwick et al., 2012). Furthermore, GWW motifs have been implicated in intramolecular interactions of two SH3 domains of the NADPH oxidase component p47phox (Groemping et al., 2003). Related dispersed GW motifs in the posttranslational regulator GW182 mediate recruitment of CCR4-NOT deadenylation components during miRNA-mediated repression (Chekulaeva et al., 2011). In essence, we identified at least four different types of interfaces for protein/protein interactions in Upa2 supporting the hypothesis that it serves as a scaffold protein.
The function of Upa2 during endosomal transport
In order to study the function of Upa2 in vivo we combined genetic with cell biological approaches. Loss of Upa2 causes the same aberrant growth phenotype and reduced Cts1 secretion as observed in rrm4Δ and upa1Δ strains (König et al., 2009, Pohlmann et al., 2015), suggesting that these proteins function in the same pathway. This is supported by a phylogenetic analysis showing the co-appearance of all three proteins.
The bipolar hyphal growth phenotype can in part be explained by the defects in forming the gradient of septin filaments at the growth pole, since septins are important during the initial phase of unipolar hyphal growth (Baumann et al., 2014, Zander et al., 2016). The formation of the septin gradient depends on the endosomal transport of septin heteromers and in turn heteromer assembly depends on endosomal transport of the septin encoding mRNAs (Baumann et al., 2014, Zander et al., 2016). Thus, the most rational explanation for the observed defects in hyphal growth in upa2Δ strains is the reduced transport of septin mRNAs resulting in fewer septin proteins on endosomes. Since the presence of the different septins on transport endosomes are interdependent (Zander et al., 2016), the reduced amount might not be able to support endosomal assembly and localisation of septin proteins causing the observed defects in septin gradients. The hypothesis of local translation on the endosomal surface for septin assembly (Baumann et al., 2014, Zander et al., 2016) has recently been supported by the finding that most heteromeric protein complexes are co-translationally assembled in eukaryotes (Shiber et al., 2018).
Loss of Upa2 also causes a slight increase in the amount and velocity of transport endosomes suggesting that the absence of Upa2 and the reduced associated mRNA cargos has an influence on transport endosomes in general. However, the most profound differences were recognized studying the mRNP components Rrm4 and Pab1. We observed an aberrantly strong formation of Rrm4 on microtubules and reduced endosomal localisation of Pab1. Hence, Upa2 appears to be crucial for the stable association of mRNPs on the surface of endosomes during transport. This is consistent with important key findings of this study that Upa2, like Rrm4, specifically localises to transport endosomes and that it is present an almost all transport endosomes. In conjunction with the aforementioned scaffolding function, Upa2 fulfils the necessary criteria to function as a novel core component of endosomal mRNP transport.
Conclusions
In recent years the identification of components of the endosomal mRNA transport machinery in U. maydis has advanced significantly (Figure 7). Importantly, it is the first system, where a transcriptome-wide binding landscape of the key RBP is available. Comparing bound RNAs of Rrm4 with the accessory component Grp1 revealed that Rrm4 orchestrates a tailored transport strategy for distinct sets of cargo mRNAs (Olgeiser et al., in revision). Here, we add the evolutionarily conserved Upa2 as an important novel piece to our jigsaw puzzle of endosomal mRNA transport.
Key concepts appear to be conserved throughout evolution and might also be applicable to higher eukaryotes. In fungi, for example, Rrm4, Upa1 and Upa2 are conserved throughout Basidiomycetes suggesting that endosomal mRNA transport is more wide-spread than currently anticipated (J. Müller, T. Pohlmann and M. Feldbrügge, unpublished). Consistently, in animal systems, endosomal components have been implicated in mRNA transport during axonal growth (Falk et al., 2014, Konopacki et al., 2016) and mRNAs associated with Rab5-positive endosomes were recently discovered in plants (Yang et al., 2018). Finally, discovering the key RNA-binding protein Rrm4 that is linked intensively and intimately with endosomal membranes fits to the new emerging concept that RNA and membrane biology are tightly intertwined. Membrane-associated RBPs (memRBPs) are most likely important at all intracellular membranes to orchestrate spatio-temporal expression (Béthune et al. in revision). These findings stress the vital role of U. maydis as a model for RNA biology (Haag et al., 2017, Niessing et al., 2018).
Materials and methods
Plasmids, strains and growth conditions
For cloning of plasmids and GST pulldown experiments, E. coli Top10 cells (Life Technologies, Carlsbad, CA, USA) and E. coli Rosetta2 pLysS (Merck 71403) were used, respectively. Transformation, cultivation and plasmid isolation were conducted using standard techniques. All U. maydis strains are derivatives of AB33, in which hyphal growth can be induced (Brachmann et al., 2001). Yeast-like cells were incubated in complete medium (CM) supplemented with 1% glucose, whereas hyphal growth was induced by changing to nitrate minimal medium (NM) supplemented with 1% glucose, both at 28°C (Brachmann et al., 2001). Detailed growth conditions and cloning strategies for U. maydis are described elsewhere (Baumann et al., 2012, Brachmann et al., 2004, Terfrüchte et al., 2014). All plasmids were verified by sequencing. Strains were generated by transforming progenitor strains with linearised plasmids. Successful integration of constructs was verified by diagnostic PCR and by Southern blot analysis (Brachmann et al., 2004). For ectopic integration, plasmids were linearised with SspI and targeted to the ipS locus (Loubradou et al., 2001). Wild-type strain UM521 genomic DNA was used as a template for PCR amplifications of ORFs, unless otherwise stated. Yeast two-hybrid tests were carried out using S. cerevisiae strain AH109 (Clontech Laboratories Inc., Mountain View, CA, USA). A detailed description of all plasmids and strains is given in Supplementary Tables T1 to T6. Sequences are available upon request.
Microscopy, image processing and image analysis
Laser-based epifluorescence-microscopy was performed on a Zeiss Axio Observer.Z1 equipped with CoolSNAP HQ2 CCD (Photometrics, Tuscon, AZ, USA) and ORCA-Flash4.0 V2+ CMOS (Hamamatsu Photonics Deutschland GmbH, Geldern, Germany) cameras. For excitation we used a VS-LMS4 Laser-Merge-System (Visitron Systems, Puchheim, Germany) that combines solid state lasers for excitation of Gfp (488 nm at 50 or 100 mW) and Rfp/mCherry (561 nm at 50 or 150 mW).
For the quantification of unipolar hyphal growth, cells were grown in 20 ml cultures to an OD600 of 0.5 and hyphal growth was induced. After 6 hours, more than 100 cells per experiment were imaged and analysed for growth behaviour. Cells were scored for unipolar and bipolar growth as well for formation of empty sections. At least three independent experiments were conducted.
For analysis of signal number and velocity, we recorded videos with an exposure time of 150 ms and 150 frames taken. All videos and images were processed and analysed using Metamorph (Version 7.7.0.0, Molecular Devices, Seattle, IL, USA). Kymographs were generated using a built-in plugin and processively moving particles (covered distance of hypha per 22.5 s < 50 µm) were counted manually. The average velocity was determined by quantifying processive signals (movement > 5 µm). Data points represent means from three independent experiments (n = 3) with mean of means (red line) and s.e.m. For each experiment at least 30 signals per hypha were analysed out of 10 hyphae per strain.
For quantification of Cdc3 fluorescence intensity line-scans were conducted in a region of 10 µm from hyphal tips. Relative fluorescence intensities of < 28 hyphal tips were averaged.
Colocalization studies of dynamic processes were carried out by using a two-channel imaging system (DV2, Photometrics, Tucson, AZ, USA; Baumann et al., 2016, Pohlmann et al., 2015).
RNA live imaging, FM4-64 staining and benomyl treatment
RNA imaging in living cells was conducted by using the improved λN*-based green-RNA method described previously (Pohlmann et al., 2015). For RNA visualization a modified λN (λN*) was fused to three copies of enhanced Gfp as well as to a nuclear localization signal (NLS) and 16 copies of the boxB loop were inserted in the cdc3 mRNA. Both constructs were under the control of the constitutive active promoter Potef. For each experiment, >18 hyphae were analysed per strain. Statistical tests were performed using Graphpad Prism5 (version 5.00; Graphpad Software, La Jolla, CA, USA). A detailed protocol for subsequent data analysis was described elsewhere (Zander et al., 2016). For staining of cells with FM4-64, a 1 ml sample of hyphal cells was labelled with 0.8 µM FM4-64 (Thermo Fisher, Waltham, MA, USA). After incubation for 1 min at room temperature, the labelled cells were analysed by fluorescence microscopy. For benomyl treatment, a 1 ml sample of hyphal cells was treated with 20 µM benomyl (Sigma-Aldrich, Taufkirchen, Germany). After incubation for 30 min at room temperature with agitation samples were analysed by microscopy.
Fluorimetric measurement of chitinolytic activity
Chitinolytic activity measurements of U. maydis cells were carried out as described elsewhere (Koepke et al., 2011, Pohlmann et al., 2015). Briefly, U. maydis cell suspensions were grown to an OD600 of 0.5. The culture was divided in half, yeast-like growing cells were measured directly while activity of hyphae was measured 6 h after induction of hyphal growth. 30 µl of the culture were mixed with 70 µl 0.25 µM 4-Methylumbelliferyl-β-D-N,N‘,N“-triacetylchitotrioside (MUC, Sigma-Aldrich, Taufkirchen, Germany), a fluorogenic substrate for chitinolytic activity. After incubation for 1 h, the reaction was stopped by addition of 200 µl 1M Na2CO3, followed by detection of the fluorescent product with the fluorescence spectrometer Infinite M200 (Tecan Group Ltd., Männedorf, Switzerland) using an excitation and emission wavelength of 360 nm and 450 nm, respectively. Chitinase activity was set and reported in relation to AB33 (wt) activity. Five independent biological experiments were performed with three technical replicates per strain.
Yeast two-hybrid analysis
The yeast two-hybrid analyses were carried out as described elsewhere (Pohlmann et al., 2015). Briefly, using the two-hybrid system Matchmaker 3 from Clontech, strain AH109 was co-transformed with derivates of pGBKT7-DS and pGADT7-Sfi (Supplementary Table T4) and cells were grown on synthetic dropout (SD) plates without leucine and tryptophan at 28° C for 4 days. Subsequently, colonies were patched on SD plates without leucine and tryptophan (control) or on SD plates without leucine, tryptophan, histidine and adenine (selection). Plates were incubated at 28°C for 3 days to test for growth under selection condition. For qualitative plate assays cells were cultivated in SD without leucine and tryptophan to OD600 of 0.5 and successively diluted with sterile water in four steps at 1:5 each. 4 µl were spotted on control as well as selection plates and incubated at 28° C for 3 days. Colony growth was documented with a LAS 4000 imaging system (GE Healthcare Life Sciences, Little Chalfont, UK).
Protein extracts and Western blot analysis
U. maydis hyphae were harvested 6 hours post induction (h.p.i.) by centrifugation (7546 x g, 10 minutes) and resuspended in 2 ml l-arginine rich buffer (0.4 M sorbitol; 5 % glycerol; 50 mM Tris/HCl pH7.4; 300 mM NaCl; 1 mM EDTA; 0.5% Nonidet P-40; 0.1% SDS; 72.5 mM l-arginine; 1 tablet of complete protease inhibitor per 25 ml, Roche, Mannheim, Germany; 1 mM DTT; 0.1 M PMSF; 0.5 M benzamidine). Cells were lysed in a Retsch ball mill (MM400; Retsch, Haan, Germany) while keeping samples constantly frozen using liquid nitrogen. 2 ml cell suspension per grinding jar with two grinding balls (d = 12 mm) were agitated for 2 x 10 minutes at 30 Hertz. Protein concentrations were measured by Bradford assay (Bio-Rad, Munich, Germany). For Western Blot analysis, 80 µg of protein were heated to 60°C for 10 minutes, resolved by 8% SDS-PAGE and transferred to a nitrocellulose membrane (GE Healthcare, Munich, Germany) by semi-dry blotting. Membranes were probed with α-Gfp (Roche, Freiburg, Germany) and α-actin (MP Biomedicals, Eschwege, Germany) antibodies. As secondary antibody a mouse IgG HRP conjugate was used (Promega, Madison, WI, USA). Detection was carried out by using AceGlow (VWR Peqlab, Erlangen, Germany).
GST pull down experiments
Derivatives of plasmids pGEX and pET15B (Supplementary Table T5) were transformed into E. coli Rosetta2 pLysS (Merck 71403). Overnight cultures were diluted 1:50 in a final volume of 50 ml. Protein expression was induced with 1mM IPTG for 4 h at 28°C. Cells were pelleted, resuspended in 10 ml lysis buffer and lysed by sonication. Cell lysate was centrifuged at 16,000 g for 15 minutes and the supernatant was transferred to the fresh microcentrifuge tubes. 50 µl glutathione beads (GE Healthcare) were transferred to a new 1.5 ml microcentrifuge tube and washed 3 times with 1 ml lysis buffer (20 mM Tris-Cl, pH 7.5; 200 mM NaCl; 1mM EDTA, pH 8.0; 0.5 % Nonidet P-40; 1 tablet complete protease inhibitor per 50ml, Roche, Mannheim, Germany). For each pulldown, 1 ml of supernatant of GST-tagged protein was added to the washed beads, incubated for 1 h at 4° C and subsequently washed 5 times with 1 ml lysis buffer. 1 ml supernatant containing His-tagged Upa2 variant was added directly to the washed GST bead and incubated for 1 h at 4° C and subsequently washed 5 times with lysis buffer. The beads were boiled for 6 min at 99° C in 100 µl of 1x Laemmli buffer. 10 µl of each GST-pulldown fraction was analysed by SDS-PAGE and Coomassie blue (CBB R250) staining. 1 ml supernatant containing Upa2 variant was added directly to the washed Ni-NTA (Macherey-Nagel) bead, incubated for 1 h at 4° C and subsequently washed 5 times with 1 ml lysis buffer. 20 µl of each Upa2 variants were loaded on control lanes in SDS PAGE as an input. Note, that the input fraction “H” shows Ni-NTA precipitated Upa2, which could not be visualized before enrichment.
For Western blotting protein samples were resolved by 10% SDS-PAGE and transferred to a PVDF membrane (GE Healthcare) by semi-dry blotting. Western blot analysis was performed with α-GST (Sigma G7781), α-His (Sigma H1029). α-Rabbit IgG HRP conjugate (Promega W4011), α-mouse IgG HRP conjugate (Promega W4021) were used as secondary antibodies. Activity was detected using the AceGlow blotting detection system (VWR Peqlab, Erlangen, Germany).
Phylogenetic analysis and bioinformatics
Sequence data for U. maydis genes was retrieved from the PEDANT database (http://pedant.gsf.de/). Accession numbers of U. maydis genes used in this study: upa2 (UMAG_10350), rrm4 (UMAG_10836), pab1 (UMAG_03494), upa1 (UMAG_12183), rab5a (UMAG_10615) and cdc3 (UMAG_10503). Orthologs were identified using fungiDB (Basenko et al., 2018, Stajich et al., 2012), Ensembl Fungi (Kersey et al., 2018) and NCBI blastp tool (Altschul et al., 1990). Sequences alignments were performed with ClustalX (version 2.0.12; Larkin et al., 2007). Domains were predicted using SMART (Letunic & Bork, 2018, Schultz et al., 1998), conserved domain database from the NCBI (CDD; Marchler-Bauer et al., 2017) and active search using ScanProsite (de Castro et al., 2006). The coiled coil dimerisation domain was predicted using the Interpro COILS program (Lupas et al., 1991). The phylogenetic tree is based on the NCBI taxonomy and was created using phyloT online tool (phylot.biobyte.de). The lengths of the lines do not represent evolutionary distance. The accession numbers for the proteins can be found in Supplementary Table T7).
Intrinsically disordered regions were predicted using the PONDR VL3-BA algorithm (www.pondr.com, Molecular Kinetics, Inc., IN, USA). VL3-BA is a feedforward neural network predictor generating outputs between 0 and 1 which are smoothed over a sliding window of 9 amino acids. Regions with values of 0.5 or above are considered disordered and are marked in red, while peptide regions with values lower than 0.5 are considered ordered and are marked in blue.
Conflict of interest
The authors declare that they have no conflict of interest.
Supplementary Tables T1-T7
Legends to Supplementary Videos
Supplementary Video V1
Composite video showing Upa2-Gfp in comparison with other components of endosomal transport mRNPs. (1st line) Upa2-Gfp is moving bidirectionally in hyphae. (2nd line) Upa2-Gfp movement is abolished by treatment with microtubule inhibitor benomyl. (3rd line) Rrm4-Gfp is moving bidirectionally in hyphae. (4th line) Pab1-Gfp is moving bidirectionally in hyphae. Additionally, Pab1-Gfp shows a prominent localisation in the cytoplasm (size bar, 10 µm; timescale in seconds, 150 ms exposure time, 150 frames, 6 frames/s display rate; MP4 format, 2373 kb; corresponds to Figure 3A).
Supplementary Video V2
Composite video showing co-localisation of Upa2-Gfp and Rrm4-mCherry (upper and lower part, respectively). Both proteins co-localise in shuttling units in hyphae. Videos were recorded simultaneously using dual-colour detection (size bar, 10 µm; timescale in seconds, 150 ms exposure time, 150 frames, 6 frames/s display rate; MP4 format, 2387 kb; corresponds to Figure 3D).
Supplementary Video V3
Composite video showing movement of Upa1-Gfp in hyphae with or without Rrm4 (top and bottom, respectively). Upa1-Gfp is moving bidirectionally in hyphae. Loss of Rrm4 results in bipolar growing cell, but does not affect Upa1-Gfp shuttling (size bar, 10 µm, timescale in seconds, 150 ms exposure time, 150 frames, 6 frames/s display rate; MP4 format, 2371 kb; corresponds to Figure 4A, top).
Supplementary Video V4
Composite video showing movement of Pab1-Gfp in hyphae with and without Rrm4 (upper and lower part, respectively). Pab1-Gfp is moving bidirectionally in the hyphal cell and shows a prominent cytoplasmic localization. Loss of Rrm4 results in bipolar growing cell, and abolishes Pab1-Gfp shuttling (size bar, 10 µm; timescale in seconds, 150 ms exposure time, 150 frames, 6 frames/s display rate; MP4 format, 506 kb corresponds to Figure 4A center).
Supplementary Video V5
Composite video showing movement of Upa2-Gfp in hyphae with and without Rrm4 (upper and lower part, respectively). Upa2-Gfp is moving bidirectionally in hyphae. Loss of Rrm4 results in bipolar growing cells, and abolishes Upa2-Gfp shuttling (size bar, 10 µm; timescale in seconds, 150 ms exposure time, 150 frames, 6 frames/s display rate; MP4 format, 2364 kb corresponds to Figure 4A, bottom).
Supplementary Video V6
Composite video showing co-localisation of Upa2-Gfp and Rrm4mR1-mRfp (upper and lower part, respectively) in hyphae. While Rrm4mR1-Rfp is present in bidirectionally moving units, Upa2-Gfp does not show processive movement. Videos were recorded simultaneously using dual-colour detection (size bar, 10 µm; timescale in seconds, 150 ms exposure time, 150 frames, 6 frames/s display rate; MP4 format, 2300 kb; corresponds to Figure 4C, left).
Supplementary Video V7
Composite video showing co-localisation of Pab1-Gfp and Rrm4mR1-mRfp (upper and lower part, respectively) in hyphae. Rrm4mR1-Rfp is present in bidirectionally moving units. Pab1-Gfp shows strongly reduced shuttling. Videos were recorded simultaneously using dual-colour detection (size bar, 10 µm; timescale in seconds, 150 ms exposure time, 150 frames, 6 frames/s display rate; MP4 format, 2364 kb; corresponds to Figure 4C right).
Supplementary Video V8
Composite video showing movement of Upa2-Gfp and Upa2mGWW-Gfp in hyphae (upper and lower part, respectively). While Upa2-Gfp is moving bidirectionally, Upa2mGWW-Gfp shuttling is no longer detectable (size bar, 10 µm; timescale in seconds, 150 ms exposure time, 150 frames, 6 frames/s display rate; MP4 format, 2367 kb; corresponds to Figure 5H).
Supplementary Video V9
Composite video showing movement of Cdc3-Gfp with and without Upa2 in hyphae (upper and lower part, respectively). While Upa2 is present, Cdc3-Gfp is moving bidirectionally and forms a gradient starting from the hyphal tip, loss of Upa2 results in absence of Cdc3-Gfp movement as well as gradient formation (size bar, 10 µm, timescale in seconds, 150 ms exposure time, 150 frames, 6 frames/s display rate; MP4 format, 2247 kb; corresponds to Figure 6A).
Supplementary Video V10
Composite video showing movement of Pab1-Gfp with and without Upa2 in hyphae (upper and lower part, respectively). While Pab1-Gfp is moving bidirectionally in the presence of Upa2, loss of Upa2 results in reduced movement of Pab1-Gfp (size bar, 10 µm; timescale in seconds, 150 ms exposure time, 150 frames, 6 frames/s display rate; MP4 format, 945 kb; corresponds to Figure 6G).
Supplementary Video V11
Composite video showing movement of Rab5a-Gfp with and without Upa2 in hyphae (upper and lower part, respectively). While Rab5a-Gfp is moving bidirectionally in the presence of Upa2, loss of Upa2 results in bipolar growing cells, but does not drastically affect movement of Rab5a-Gfp (size bar, 10 µm; timescale in seconds, 150 ms exposure time, 150 frames, 6 frames/s display rate; MP4 format; 1429 kb, corresponds to Figure 6G).
Supplementary Video V12
Composite video showing movement of Rrm4-Gfp with and without Upa2 in hyphae (upper and lower part, respectively). While Rrm4-Gfp is moving bidirectionally in the presence of Upa2, loss of Upa2 results in bipolar growing cells, but does not drastically affect movement of Rrm4-Gfp. Note, the aberrant staining of a microtubule bundle (size bar, 10 µm, timescale in seconds, 150 ms exposure time, 150 frames, 6 frames/s display rate; MP4 format; 948 kb corresponds to Figure 6G).
Supplementary Video V13
Composite video showing movement of Upa1-Gfp with and without Upa2 in hyphae (upper and lower part, respectively). While Upa1-Gfp is moving bidirectionally in the presence of Upa2, loss of Upa2 results in bipolar growing cells, but does not drastically affect movement of Upa1-Gfp (size bar, 10 µm; timescale in seconds, 150 ms exposure time, 150 frames, 6 frames/s display rate; MP4 format; 2357 kb corresponds to Figure 6G).
Acknowledgements
We acknowledge Drs. A. Brachmann, J. Béthune, K. Zarnack and lab members for discussion and reading of the manuscript. We are grateful to U. Gengenbacher for excellent technical assistance. The work was funded by grants from the Deutsche Forschungsgemeinschaft to MF (FE448/5-2 DFG-FOR1334; FE448/10-1 DFG-FOR2333; DFG-CRC1208 and CEPLAS EXC1028).
Footnotes
↵# present address: Cell and Developmental Biology, Centre for Genomic Regulation (CRG), Barcelona, Spain