In silico analysis of putative PHY4 interactors

Sequence data was further analysed by BLAST searches against the current version of the Physcomitrella nucleotide database on both the cosmoss and the NCBI database server. Alignments of the sequencing data with the CDS identified from BLAST searches revealed that 62 % of the analysed sequences were indeed represented as full-length sequences within the cDNA library, whereas the majority of the remaining 38 % missed parts of the 5 ´end.

cDNA hits were then translated into protein sequences and verified by protein BLAST against the non-redundant protein database of Physcomitrella. To identify bona fide phytochrome interacting/signaling partners it was one criterion whether or not isolated proteins share homologs in Arabidopsis thaliana and/or other plants.

Isolated proteins were further analysed by BLAST searches of the non redundant Arabidopsis protein database. Identified protein sequences were further analysed regarding conserved protein domains or motifs and their predicted functions using SMART (http://smart.embl-heidelberg.de) and NCBI CDD (conserved domains database) search. The cDNA hits as well as the identified conserved domains and their related/predicted function are listed in table 10.

62 Table 10: In silico analysis of sequences isolated from Y2H screen.

Hits of outstanding interest and chosen for further analysis are marked in orange, hits of minor interest are marked in yellow. TM: transmembrane domain.

As expected, a considerable number of identified sequences were related to plant housekeeping genes like 40 and 60 S ribosomal proteins, proteins related to chlorophyll synthesis or photosynthesis or membrane pores and are likely false positives. Since annotation of the Physcomitrella genome is still an ongoing process, a large amount of genes have not been annotated yet. If possible, such sequences were further classified according to conserved domain or motifs contained in their sequence. Following cDNA and protein identification and in silico analysis, protein sequences were analysed regarding their predicted subcellular localisation. Putative interacting partners may be excluded from further analysis due to localisation into compartments phytochromes do not access. Literature provides examples for predicted phytochrome interacting partners of such controversial nature (186, 196, 197). Prediction of subcellular localisation was carried out using WoLF PSORT (http://wolfpsort.org). Results are presented in table 11.

63 Table 11: Prediction of subcellular protein localisation of putative PHY4 interactors.

Cp: chloroplasts, Cyto: cytoplasm, Cytosk: cytoskeleton, ER: endoplasmatic reticulum, Mito:

mitochondrion, Nuc: nucleus, PM: plasma membrane, Vac: vacuole. Colour code as in table 4.

Although exact prediction of subcellular protein localisation is difficult, especially for organellar targeted proteins, proteins predicted to localise in other compartments as either the cytoplasm or the nucleus were formally excluded from further analysis.

The following sections present putative candidates chosen from analysed Y2H screening data.

4.4.4.1. #16.1 - predicted protein containing a p-loop motif (PLP)

The sequence obtained from yeast colony #16 contained cDNA of a yet unknown

“predicted” protein and was the only sequence chosen for further analysis, which could not be unequivocally identified by in silico analysis. The reason for it being of interest as a probable phy4 interacting partner is based on its conserved functional domain. When initially analysed in 2009, CDD search identified a C3HC4 RING zinc finger motif. Since phytochrome interacting partners containing RING type zinc fingers have been described in Arabidopsis, this protein appeared to make a good candidate for further analysis. As was realised, prediction algorithms changed by the beginning of 2011 (198). Using current algorithms of

64 CDD search and SMART-prediction both searches are to date unable to identify a RING-type zinc finger motif. Instead, a transmembrane domain located at the N-terminus and a sulfotransferase domain containing a phosphate binding loop, a so-called Walker-motif (or p-loop motif) were predicted. p-loops are made up by a sequence of GXXXXGK(T/S) resulting in a structure containing a β-sheet followed by a glycine rich loop and a α-helix (199). This motif is capable of binding ATP, or less common, also GTP at the β-γ-phosphate moiety (200) and is commonly found in nucleotide binding proteins, but also in ATP-synthases, helicases or kinases (201). Thus, the “predicted protein” identified from Y2H apparently is grouped into the superfamily of p-loop containing nucleoside triphophate hydrolyses. Plant sulfotransferases are either soluble or transmembrane proteins and transfer a sulfate group to proteins or glycosaminoglycans (202).

Fig. 20: Alignment of Physcomitrella PLP and its putative Physcomitrella and Arabidopsis homologs.

Transmembrane domains in yellow, sulfotransferase domains in orange. Scale bar as indicated.

Protein BLAST identified one close homolog in Physcomitrella. This protein lacks the transmembrane domain and was presumably listed due to its high degree of similarity originating from the sulfotransferase domain.

PLP contains three homologs in Arabidopsis with the closest homolog being designated as a p-loop containing nucleoside triphosphate hydrolyse family protein – like protein, the two remaining homologs are annotated as nodulation proteins. ClustalW2-alignment of both the Physcomitrella and Arabidopsis proteins yielded 58.5 % consensus with 42.9 % sequence identity. Additional homologs are found in several other species from lower to higher plants and theirs phylogenetic relation are depicted in suppl. figure 21.

65 Fig. 21: Alignment (ClustalW2) of PLP isolated from a Physcomitrella cNDA library and its Physcomitrella and Arabidopsis homologs.

Predicted transmembrane domains (yellow) and p-loop containing sulfotransferase domains (red) are highlighted.

4.4.4.2. #33.7 - pleiotropic regulator locus (PRL)

Although no annotation for the Physcomitrella protein was available at the time of BLAST searches this sequence could be assigned to the pleiotropic regulator locus 1 (PRL1) on the basis of its high homology to the respective Arabidopsis protein. PRL1 appears to be a single copy gene in Physcomitrella whereas Arabidopsis carries two PRL homologs, PRL1 and PRL2. CLUSTALW2 alignment of Pp.PRL with At.PRL1 and At.PRL2 yielded 64.8 % consensus with 27.3 % sequence identity.

66 Fig. 22 Alignment (ClustalW2) of PRL isolated from a Physcomitrella cNDA library and its Arabidopsis homologs PRL1 and PRL2.

WD40 repeats are highlighted in red.

SMART analysis of both the Physcomitrella and the Arabidopsis protein sequence revealed a 7 x repeat of a WD40 motif, structurally forming a circular 7 sheeted beta-propeller (203). Commonly WD40-domain proteins function as platforms in multi-protein-complex assembly, with the propeller serving as a rigid scaffold for protein-protein-interactions, as in the case of the β-subunit of heterotrimeric G-proteins or E3-ubiqitin ligases. As such, WD40-domain G-proteins mostly serve in signal transduction or transcriptional regulation (204, 205). Other than the WD40-motifs, no particular domains or motifs were identified.

Protein BLAST identified PRL-homologs in many different plant species (see suppl.

fig. 5); all of which were single copy genes with the exception of Arabidopsis thaliana, which bears two copies.

67 Fig. 23: Schematic of Physcomitrella PRL with its Arabidopsis homologs PRL1 and PRL2.

WD40 motifs in red. Scale bar as indicted.

4.4.4.3. #54.1 - elongation factor 1 α (EF1α)

This clone was assigned to the elongation factor 1 α (EF1α). EF1α is a ubiquitous protein from bacteria to the animal kingdom and exhibits a high degree of conservation even on protein level throughout all species.

Fig. 24 Alignment (ClustalW2) of EF1α isolated from a Physcomitrella cNDA library and its closest Arabidopsis homolog.

EFTU domains (EF-TU1, EF-TU_D2 and EF-TU_D3) are highlighted in purple.

EF1α proteins belong to the family of Ras-like GTPases and thus comprise of several GTP binding sites, as well as binding sites for other interacting proteins of the EF-family like EF-Tu proteins. The binding motif of the β-γ-phosphate moiety of an ATP or GTP is comprised within the first domain of EF1α. GTP_EFTU_D2 structurally forms a β-barrel and is responsible for binding of charged tRNAs and probably also for cytoskeletal association of the protein. GTP_EFTU_D3 domain generally exhibits GTP hydrolysis capacity.

68 Fig. 25: Schematic of EF1α protein’s domain organisation in Physcomitrella and Arabidopsis.

Conserved domains in purple. Scale bar as indicated.

In Physcomitrella 6 copies of EF1α proteins were detected by BLAST search;

Arabidopsis posses 4 homologous proteins of EF1α, EF1A1-EF1A4 (206). As such, EF1α has to be considered a ubiquitous protein although the number of protein homologs in other plant species may vary in a great range (e.g. more than 10 homologs in Zea mays) (207).

4.4.4.4. # 61.4 - pirin-like protein (Pirin)

This sequence was assigned to a pirin-like protein in Physcomitrella by its high degree of homology to the Arabidopsis pirin-like protein. Arabidopsis carries 3 homologous proteins of the Pirin-family: Pirin, a putative Pirin lacking approx 30 amino acids of the N-terminus and a Pirin-like protein that carries an N-terminal extension of about 20 amino acids. The protein identified in Physcomitrella exhibits highest homology to the pirin-like protein of Arabidopsis with 63 % consensus positions and 54 % sequence identity.

Fig. 26 Alignment (ClustalW2) of Pirin isolated from a Physcomitrella cNDA library and its putative Arabidopsis homologs.

The cupin domain (light blue) and the C-terminal domain of pirin-proteins (dark blue) are highlighted.

69 SMART identified a conserved Cupin-domain in all of the proteins. Cupin domains were named after their conserved structure, forming a barrel out of 6 β-strands.

The C-terminus is a conserved part of all pirin-proteins.

Fig. 27: Schematic of pirin proteins domain organisation in Physcomitrella and Arabidopsis.

Cupin domains in light blue, conserved C-terminal domains in green. Scale bar as indicated.

Pirin is a single copy gene in Physcomitrella, whereas Arabidopsisposses3 related sequences. Pirin proteins were identified in many other plant species, but not within tobacco or colza.

4.5. Light dependent interaction of phy4 with its putative interactors in yeast