Mistletoes (European mistletoe; Viscum album), grown on an apple tree (Malus sp.) on the campus of Leibniz University Hannover were used as starting material. Various organs (leaves, stems, flower buds) of male and female plants were harvested in summer and in winter, shock-frozen using liquid nitrogen, and stored at –80°C until use. mRNA isolation and quality evaluation were carried out as described previously (Schröder et al., 2022b).

mRNA fractions were reverse transcribed into cDNA. The summer and winter fractions were analyzed separately using Illumina PE150 (paired-end read) sequencing. cDNA libraries containing 250~300 bp inserts were constructed and sequenced. Quantitative data as well as quality evaluation data of the libraries are given in Table 1 .

The raw Illumina RNA-seq reads were first trimmed for low-quality and adaptor regions using Trimmomatic (Bolger et al., 2014) (version 0.36, ILLUMINACLIP : ADAPTER.fa:2:20:7 SLIDINGWINDOW:4:15 MINLEN:50). The trimmed reads were assembled using Trinity (Haas et al., 2013) (version 2.11.0) including in silico normalization with a target coverage of 50 x. The PacBio IsoSeq reads were included in the assembly using the “–long_reads” option of Trinity. The assembled transcripts were screened for coding sequences using TransDecoder (Haas and Papanicolaou, n.d.). The potential protein sequences were merged with the previously created protein sequences (Schröder et al., 2022b) and redundant sequences were filtered using CD-HIT (Li and Godzik, 2006) (version 4.7, -c 0.95 -aS 0.99). The functional annotation was computed using the best blast hit method on the UniProt/Swiss-Prot (Boutet et al., 2007) and UniProt/TrEMBL (The UniProt Consortium et al., 2023) databases. The TPM values were computed with Salmon (Patro et al., 2017) (version 1.9.0). The completeness of the filtered transcripts was estimated with BUSCO (Seppey et al., 2019) (version 5.0.0) using the viridiplantae database.

The final V. album Gene Space II database was created by combining PacBio and Illumina sequencing. It includes 90,039 distinct proteins. The database is accessible at https://viscumalbum.pflanzenproteomik.de/. Furthermore, all sequences were submitted to NCBI.

The complexome profiling approach is introduced in Schröder et al., 2022c. Primary complexome profiling data were taken from Senkler et al., 2018. Re-evaluation of the mass-spectrometry (MS) data and annotation of proteins were carried out with MaxQuant (version 2.1.4.0) using the novel VaGs II database. The settings were the same as in Senkler et al., 2018. For heatmap generation, the abundance profiles (based on iBAQ values; Schwanhäusser et al., 2011; calculated by MaxQuant) of all identified proteins were used. These profiles were aligned according to similarity by hierarchical clustering using the NOVA software (version: 0.5.8; Giese et al., 2015). The Complexome Map of the re-evaluated V. album mitochondria fractions is available at the ComplexomeMap portal at https://complexomemap.de/75.

The datasets were processed as described in the Materials and Methods section and assembled using the Trinity software package version 2.11.0 (Haas et al., 2013). The PacBio IsoSeq reads (Schröder et al., 2022b) were included to the assembly using the “–long_reads” option of Trinity. The Trinity assembly produced 650,594 transcript contigs. TransDecoder predicted potential coding sequences on 144,517 contigs. The clustering with CD-HIT produced 104,405 unique sequences, which were used for further analyses.

Functional annotation of the 104,405 transcripts was computed using the best blast hit method on the UniProt/Swiss-Prot (Boutet et al., 2007) and UniProt/TrEMBL (The UniProt Consortium et al., 2023) databases. As a result, 51% of the sequences could be assigned a function. As expected, in most cases the greatest similarity is with sequences from the phylum Viridiplantae (green algae and the land plants). However, there are also numerous transcripts which show highest similarity to transcripts from other phyla, especially fungi. No functions could be assigned to many other sequences, which can likely be explained by the fact that no genome sequences of plants more closely related to V. album are yet available.

V. album organs used for transcript isolation were from plants living in the field. V. album leaves are known to be colonized by several endophytic fungi (Peršoh et al., 2010; Kotan et al., 2013, Ariantari et al., 2019, reviewed in Krasylenko et al., 2020). To investigate the presence of fungal sequences in our database, we evaluated the functional annotation of Illumina transcript sequences according to Phyla for the winter and summer samples ( Figure 1 ). 37.614 of the V. album transcripts assigned to Viridiplantae. 91% of them (34.159 transcripts) likewise were present in both, the winter and the summer sample. Furthermore, 11.830 transcripts of our V. album samples were assigned to the fungi Phylum ( Figure 1 ). Almost all fungal sequences are from the winter sample, which is probably due to the increased age of the mistletoe leaves (new leaves appear in the spring and are kept in the winter). Only 11% of the fungal sequences occur likewise in the summer and winter samples. Few of the transcripts of our V. album samples were assigned to other Phyla than Viridiplantae or fungi ( Figure 1 ). Transcripts that could not be functionally annotated were labeled “unknown” ( Figure 1 ). The number of transcripts of this category is similar in the winter and summer samples; also, they show quite a high intersection in terms of their occurrence in the two seasons (76%), similar to the transcripts assigned to Viridiplantae. We conclude that these sequences should be predominantly V. album transcripts. The relatively high proportion of transcripts of unknown function likely is due to the lack of comparative sequences from related plant species, but also due to the fact that V. album has a special way of life, which requires numerous proteins that do not occur in other phyla.

To further analyze the origin of sequences in our V. album samples, we quantified all transcripts identified by our Illumina sequencing approach according to their assignment to Phyla ( Figure 2 ). Transcripts per million (TPM) values of transcripts in the summer and winter samples were examined separately. As expected, TMP values are highest for transcripts assigned to Viridiplantae and clearly lower in those assigned to the fungi Phylum. This is particularly visible in the sample harvested in summer. TMP values of unassigned transcripts more resemble the values obtained for transcripts assigned to Viridiplantae, which is especially evident in the summer sample. This reconfirms that the transcripts in this category are derived from V. album rather than from fungi colonizing V. album. The absolute summed up TPM values (summer and winter sample combined) for the four categories are 1,096,579 for Viridiplantae (73%), 385,858 for “unknown” (25%), 11,497 for fungi (0.8%) and 17,789 for “others” (1.1%). Thus, about 1% of the transcripts of our V. album fractions can be classified as fungal. This should realistically reflect the conditions in the “ecological niche” V. album.

The previous V. album Gene Space database, which was created based on Single Molecule Real-Time (SMRT) sequencing (PacBio sequencing), includes 39,092 entries encoding 32,064 distinct proteins (Schröder et al., 2022b). We now designate this database VaGs I. BUSCO analyses had revealed a completeness of about 78%. Through implementation of the new Illumina sequencing data and improved transcript annotation, we hereby present a novel V. album Gene Space database, VaGs II. We decided to remove transcripts that are unambiguously assigned to the Phyla Fungi, Animals, Bacteria and Viruses based on sequence comparisons. VaGs II includes all transcripts assigned to the categories “Viridiplantae” and “unknown”. The totel number of distinct transcripts is 90,039. The BUSCO score of VaGs II is 93% (91% complete, 2% fragmented; Figure 3 ).

A database has been developed that allows all sequences to be viewed ( Figure 4 ). Sequences can be downloaded, proteins can be searched by ID, name, and sequence, and instructions for BLAST searches are given. Furthermore, all V. album transcript have been uploaded at the European Nucleotide Archive (ENA) (https://www.ebi.ac.uk/ena/browser/home).

To test the quality of the new VaGs II databank, a published proteomic dataset on V. album mitochondria (Senkler et al., 2018) was re-evaluated. This is a complexome dataset which was created as follows: First, mitochondria were isolated from V. album and a mild detergent was used to dissolve the mitochondrial membranes. Protein complexes in the resulting solution were separated by Blue-native polyacrylamide gel electrophoresis. A lane of the native gel was then cut from top to bottom into 54 small gel slices, which were individually subjected to label-free quantitative shotgun mass spectrometry for protein identifications. Finally, abundance profiles along the blue-native gel lane were calculated for all identified proteins. The results were visualized as a heatmap and the profiles of individual proteins were aligned according to similarity. On the resulting heatmap, proteins belonging to the same protein complex form characteristic clusters. It thus is possible to systematically investigate the protein complexes of a biochemical fraction, the ‘complexome’. The original data of this experiment still had to be analyzed using a protein database for the model plant Arabidopsis thaliana (Senkler et al., 2018), since a V. album database was not available. This allowed the identification of 477 proteins in total. Re-evaluation of the same dataset using the VaGs II database now allowed the identification of 1392 proteins ( Figure 5 ). The newly evaluated mitochondrial complexome of V. album is presented as a supplement of this publication ( Supplementary Table 1 ) and is also accessible at the ComplexomeMap portal at https://complexomemap.de/75. It offers novel insights into the molecular biology of V. album mitochondria. Also, the number of peptides identified in the complexome fractions (gel slices) significantly increased based on VaGs II evaluation, resulting in a better coverage of proteins by peptides.

To further analyze the mitochondrial complexome from V. album, as revealed by VaGs II evaluation, we took a closer look at the protein clusters corresponding to respiratory chain complexes III and IV ( Figure 6 ). Complex III is an ubiquinol:cytochrome-c oxidoreductase and complex IV a cytochrome-c:O₂ oxidoreductase. These two protein complexes catalyze the last two steps of the mitochondrial respiratory chain and were found to form two exceptionally stable supercomplexes in V. album mitochondria (Senkler et al., 2018). According to current knowledge (Maldonado et al., 2021), complexes III and IV in plants each consist of 10 different subunits (the 10 subunits of complex III, however, all are present in duplicate, as complex III occurs as a functional dimer). In V. album, this dimer (III₂) associates with one or two copies of complex IV; the corresponding supercomplexes are designated III₂IV₁ and III₂IV₂. Of the total 20 different proteins present in these supercomplexes, all 20 proteins were found in our complexome dataset upon evaluation using VaGs II ( Figure 6 ; compared to only 8 proteins based on the original evaluation, Senkler et al., 2018; see Supplementary Figure S1 ). 18 of the identified subunits form part of the two supercomplexes, whereas two (QCR6 of complex III₂ and Cox6b from complex IV) are partly detached and migrate in the low-molecular-mass region of the blue native gel, probably because they became detached during membrane solubilization. For some of the proteins, different isoforms are present in V. album (e.g. the Cox5b and the α- and β-MPP subunits, Supplementary Figures S2 , S3 ). The summed-up abundance profiles of the subunits of complex III and of complex IV reveal that the amounts of the two supercomplexes are similar; furthermore, the stoichiometry of the two complexes within the supercomplexes is correctly revealed ( Figure 6 , bottom).