The HAC1 mRNA is unconventionally spliced in V. dahliae

The Ensembl Fungi gene annotation of the potential HAC1 homolog in V. dahliae predicted only one mRNA splice variant with a 53 nt intron, resulting in a spliced 1581 nt mRNA. Since unconventional splicing of the HAC1 mRNA was described for known homologs preferentially under ER stress inducing conditions, it was tested, whether different mRNA splice variants could be obtained for V. dahliae HAC1 from liquid cultures in presence or absence of dithiothreitol (DTT). DTT is a reducing agent which prevents the formation of disulfide bonds and thereby induces misfolding of proteins in the ER (Guillemette et al., 2011).The HAC1 cDNA sequence amplified from un-induced growth was in accordance with the database predicted splice variant, referred to as HAC1^u (Figure 16A). Additionally, a second splice variant could be obtained from these growth conditions, which was also amplified from ER stress inducing growth conditions, named HAC1ⁱ (Sequence given in S10). The HAC1ⁱ mRNA results from splicing of a second intron with 20 nt in size (Figure 16A). Splicing of the 20 nt intron alters the ORF, leading to a shorter variant of 1254 nt.

Figure 16: The V. dahliae HAC1 mRNA is unconventionally spliced. (A) mRNA and protein variants of V. dahliae HAC1. The 2634 nt HAC1 pre-mRNA contains a 53 nt conventional (black) and a 20 nt unconventional (blue) intron. V. dahliae produces two HAC1 mRNA variants: splicing of the conventional intron results in HAC1^u mRNA (1581 nt) for the potential Hac1^u protein (526 aa), additional unconventional splicing results in a shorter induced HAC1ⁱ ORF (1254 nt) encoding the Hac1protein (417 aa). N-termini (268 aa) of both proteins are identical, containing a basic leucine zipper domain (bZIP, grey, PS50217; 107-164 aa) and a nuclear localization signal (NLS, red, 94-105 aa), whereas Hac1^u and Hac1 possess unique C-termini (dark or light green). (B) Alignment of the 5´- and 3´-splice sequences of the unconventionally spliced introns shows high conversation of V. dahliae HAC1 (VDAG_JR2_Chr2g09780a; blue), A. fumigatus hacA (XM_743634), A. nidulans hacA (AN9397), T. reesei QM6a hac1 (M419DRAFT_128619), A. brassicicola HacA (Joubert et al., 2011), N. crassa hac-1 (NCU01856), U. maydis cib1 (UMAG_11782), S. cerevisiae S288C hac1 (NC_001138.5), H. sapiens XBP1 (NM_005080.3). The consensus sequence of Ire1 splice sites (Hooks & Griffiths-Jones, 2011) is indicated by CNG’CNGN. Ire1 splice sites are indicated by arrows, intron sequences as lowercase characters, splice sequences as capital letters. Numbers of nucleotides for those not shown are given. (C) Twin stem-loop secondary structures of 5´- and 3´-splice sequences of V. dahliae and S. cerevisiae unconventional HAC1 introns. Ire1 splice sites are indicated by arrows, intron sequences as lowercase characters, splice sequences as capital letters, discontinuation of intron sequence as //.

RNAfold was used for folding prediction (http://rna.tbi.univie.ac.at/cgi-bin/RNAfold.cgi).

In yeast and filamentous ascomycetes, translation of the uninduced mRNA variant into Hac1^u was described to be blocked (Chapman & Walter, 1997; Rüegsegger et al., 2001;

Saloheimo et al., 2003; Mulder et al., 2004; Joubert et al., 2011; Heimel, 2015).

Differently, translation of the uninduced protein variant, acting as negative regulator of the UPR, was observed in higher eukaryotes (Yoshida et al., 2006, 2009). Hints to protein translation were given in the basidiomycete U. maydis (Heimel et al., 2013). The alternative splice variant HAC1ⁱ encodes a 417 aa Hac1protein (aa sequence shown in Figure S11) with a molecular weight of 44 kDa. For both hypothetical Hac1 protein variants the N-terminal 268 aa are identical with a nuclear localization signal (NLS) and bZIP domain. The bZIP protein domains are specifically found in transcription factors and are formed by a basic loop responsible for DNA binding and the hydrophobic leucine zipper required for homo- or heterodimerization of two DNA binding proteins (Hurst, 1995). The NLS proposes protein import into the nucleus. Due to the frame shift after unconventional splicing, the C-terminus of the predicted Hac1^u and Hac1 proteins are unique with 258 unique amino acids in Hac1^u versus 149 in Hac1.

The mRNA of HAC1 homologs from several fungal species up to humans was described to be cleaved by the cytosolic endoribonuclease domain of Ire1 by recognition of the conserved consensus splice sequence 5´-CNG’CNGN-3´ which is specific for unconventional splicing (Yoshida et al., 2001; Saloheimo et al., 2003; Mulder et al., 2004;

Wimalasena et al., 2008; Richie et al., 2009; Cheon et al., 2011; Hooks & Griffiths-Jones, 2011; Joubert et al., 2011; Heimel et al., 2013; Montenegro-Montero et al., 2015).

Alignment of the 5´- and 3´-intron-exon-borders of both introns from V. dahliae HAC1 to the consensus sequence displayed conservation of the 20 nt intron, which was found to be spliced under ER stress induced growth conditions (Figure 16B). The 53 nt intron spliced in both, HAC1^u and HAC1ⁱ, did not match the consensus sequence.

Comparison of the 20 nt intron from V. dahliae HAC1 to homologous sequences in other organisms displays the same intron size in A. brassicicola, A. fumigatus, A. nidulans, and T. reesei. N. crassa (23 nt) and human (26 nt) homologs possess similar intron sizes.

The unconventionally spliced intron of the homolog Cib1 in U. maydis shows intermediate length (65 nt), whereas the yeast S. cerevisiae homolog is much longer with a size of 252 nt (Figure 16B).

The mRNA of HAC1 homologs was described to form a twin stem-loop structure in several organisms, which is targeted by the endonuclease domain of Ire1 homologs (Yoshida et al., 2001; Saloheimo et al., 2003; Mulder et al., 2004; Wimalasena et al., 2008; Richie et al., 2009; Cheon et al., 2011; Hooks & Griffiths-Jones, 2011; Joubert et al., 2011; Heimel et al., 2013; Montenegro-Montero et al., 2015). In order to allow predictions about the splice mechanism for V. dahliae HAC1, the secondary structure of

the unconventionally spliced intron was analyzed. Figure 16C shows the twin stem-loop secondary structure of the 20 nt intron from V. dahliae HAC1 in comparison to the model organism S. cerevisiae, calculated by the RNAfold web server with the lowest free energy. The predicted Ire1 cleavage sites are located in the formed seven nucleotides loops.

Similarities between the V. dahliae Hac1 protein encoded by the unconventionally spliced mRNA HAC1ⁱ to described UPR regulatory proteins in other fungi and human XBP1 (X-box binding protein) were assessed (Figure 17). The highest similarity was found to T. reesei HACI with 55% aa sequence identity. The corresponding A. nidulans and A. fumigatus proteins display a higher amino acid similarity to the V. dahliae protein with 36% than the S. cerevisiae homolog with 28%. The similarities to U. maydis cib1 and H. sapiens XBP1 was below 20% (Figure 17A). The proteins from V. dahliae and T. reesei cluster in one subclade in the phylogenetic tree (Figure 17B). Considered homologs from ascomycetes form one clade with lower amino acid similarities to the homologs of the basidiomycete U. maydis and the H. sapiens homolog XBP1.

In summary, V. dahliae HAC1 shows high similarity to homologs in related fungi. It encodes an unconventionally spliced intron with a proposed twin stem-loop structure, which is presumably recognized by the endoribonuclease Ire1 and displays similarities in the deduced protein sequence and protein domain predictions to UPR regulatory proteins from related fungi.

Figure 17: Similarities of Hac1 proteins from different organisms. (A) Protein sequence identity matrix of Hac1-like proteins (ClustalW algorithm, sequence identifiers given in B were used for the calculation). (B) Phylogenetic tree of Hac1-like proteins with V. dahliae Hac1 (Figure S11), U. maydis cib1^s (XP_011390112.1; isolate 521), S. cerevisiae HAC1ⁱ (NP_116622.1; isolate S288C), T. reesei HACIⁱ (XP_006964054.1; isolate QM6a), A. nidulans HacAⁱ (Q8TFU8.2; isolate FGSC A4), A. fumigatus HacAⁱ (ACJ61678.1; isolate H237), H. sapiens XBP1 (NP_001073007.1) sequences. ClustalW algorithm in MEGA6.0 was used with the Maximum likelihood method. The scale bar represents the average number of amino acid substitutions per site.

3.3.3 The unconventionally spliced mRNA variant HAC1ⁱ is translated into the