Evidence for plastome loss in the holoparasitic Mystropetalaceae

By definition, holoparasitic plants have lost their ability to photosynthesize, yet nearly all of them have retained a plastome, albeit reduced size and gene content, compared with autotrophic plants (Wicke & Naumann, 2018; Sanchez-Puerta et al., 2023). The holoparasitic plant Pilostyles aethiopica (Apodanthaceae) has the smallest published plastome of 11 348 bp, containing only five plastid genes unrelated to photosynthesis (Bellot & Renner, 2015). To date, only two holoparasitic angiosperm lineages in land plants have been reported to have lost their plastomes: Rafflesia and Sapria in Rafflesiaceae (Molina et al., 2014; Cai et al., 2021) and section Subulatae of Cuscuta in Convolvulaceae (Braukmann et al., 2013; Banerjee & Stefanović, 2023). The latter still needs to be verified with more data, especially from the nuclear genome. Another reported case of plastome loss in green plants is from the non-photosynthetic green alga Polytomella (Smith & Lee, 2014). The sandalwood order Santalales makes up about half of all parasitic angiosperm species (Nickrent, 2020), and two of its families, Balanophoraceae and Mystropetalaceae, consist only of holoparasites. Known plastomes of Balanophoraceae are highly reduced in size (14.6–28.4 kb), extremely AT-rich (78.8–88.4%), and contain the only two genetic code changes recorded in land plants (Ceriotti et al., 2025). Mystropetalaceae, sister to the mistletoe family Loranthaceae (Su et al., 2015), consists of three monotypic genera, namely Dactylanthus, Hachettea, and Mystropetalon (Fig. 1), which are endemic to New Zealand, New Caledonia, and South Africa, respectively. To date, no plastid sequences for this family have been deposited in public repositories like GenBank (accessed on 1 July 2025) or are available in the literature. However, there have been a very limited number of mitochondrial (cox3 and matR) and nuclear genes (18S rRNA, 26S rRNA, and rpb2) available for this family. By analyzing Illumina DNA sequencing (DNA-seq) data for all three genera and Illumina RNA sequencing (RNA-seq) data of Hachettea, here we provide compelling evidence for plastome loss in the family. Sampling details for Dactylanthus taylorii, Hachettea austrocaledonica, and Mystropetalon thomii are provided in Supporting Information Table S1. Tissues from one inflorescence of each species were collected. DNA isolation of the three samples and RNA isolation of the H. austrocaledonica sample, library construction, and Illumina sequencing were conducted following the methods of Yu et al. (2022). Sequencing was performed on the Illumina Novaseq 6000 platform (v.1.5 Reagent Kit), yielding paired-end reads of 150 bp. Sequencing information is provided in Table S2. The raw Illumina reads were filtered using Trimmomatic v.0.39 (Bolger et al., 2014) with default parameters, before further analyses. We used the program GetOrganelle v.1.7.1 (Jin et al., 2020) to assemble the mitogenomes of the three species with parameters ‘-F embplant_mt -R 30 -k 107’. The mitogenomes of H. austrocaledonica and M. thomii were successfully assembled into circular-mapping contigs, whereas that of D. taylorii was assembled into a complex network. To assess the assembly continuity and sequencing depth, for H. austrocaledonica and M. thomii, we mapped their Illumina DNA-seq reads onto their mitogenomes using Bowtie2 v.2.5.3 (Langmead et al., 2018) with parameters ‘--end-to-end --no-mixed --no-discordant’. For D. taylorii, we used 52 long gene-containing contigs to calculate its sequencing depth to avoid inaccuracy in sequencing depth assessment due to too many short contigs. We used two approaches for plastome assembly. First, we used the reference-based program GetOrganelle to assemble their plastomes with the plastomes of 12 Santalales species (Table S3) as the ‘seed’ and other parameters ‘-F embplant_pt -k 57, 77, 97 -R 30’. The 12 plastomes included two extremely AT-rich Balanophoraceae plastomes. This was done in case the Mystropetalaceae plastomes have the same highly biased nucleotide composition, which could cause read mapping failure in the first step of the assembly process. Second, we used the de novo assembly program SPAdes v.3.13.1 (Prjibelski et al., 2020) to assemble draft genomes, with the parameter -k set to 57, 77, and 97. For both approaches, only contigs > 200 bp in length were used for subsequent analyses. Nuclear rRNA operons of the three species were assembled using GetOrganelle with parameters ‘-F embplant_nr -R 30 -k 57, 77, 97’. To identify potential plastome contigs, we first mapped the Illumina DNA-seq reads onto the contigs assembled by GetOrganelle and SPAdes, using Bowtie2 with default parameters except for ‘--end-to-end’. For H. austrocaledonica, we also mapped its RNA-seq reads onto its contigs assembled by SPAdes using Hisat2 v.2.2.1 (Kim et al., 2019) with default parameters. The mapping results were supplied to SAMtools v.1.19.2 (Li et al., 2009) to calculate the average sequencing depth of each contig, with the filtering option -q set to 30. We then identified potential mitochondrial and nuclear rRNA operon contigs by mapping these assembled contigs to their own mitogenomes and nuclear rRNA operons. Contigs with a perfect match were considered as mitochondrial and nuclear rRNA operon contigs. Finally, for the remaining contigs > 200 bp in size, we did a Blastn search using the 12 plastomes of Santalales as a query (Table S3), with default parameters except -word_size set to 7, and an HMM (hidden Markov model)-based search using Nhmmscan v.3.4 (Eddy, 2011) with the E-value threshold set to 1e-5. The HMMs were built based on sequence alignments of plastid protein coding and nuclear rRNA genes of the 12 plastomes of Santalales. To identify potential plastid trnE gene sequences in the three species, we conducted an additional HMM-based search against their own genome contigs > 200 bp using Nhmmscan. The HMM was built using plastid trnE genes of the 12 species of Santalales. We used the mitogenomes of Liriodendron tulipifera (Genbank accession no. NC_021152) and seven Santalales species (Table S4) as references to annotate the protein coding, rRNA, and tRNA genes in the Mystropetalaceae mitogenomes. To identify mitochondrial plastid sequences (MtPts) in the mitogenomes of H. austrocaledonica and M. thomii, we first masked all the annotated mitochondrial genes and then did a Blastn search against the masked mitogenome using the plastomes of 41 angiosperm species (Table S3) as a query, with default parameters except -word_size set to 7. We did not identify MtPts in the D. taylorii mitogenome due to its short contig lengths. To infer the origins of the identified MtPts, a phylogenetic analysis was carried out for each MtPt, alongside plastid counterparts of other species. Briefly, nucleotide sequences were aligned using Mafft v.7.505 with the L-INS-i strategy (Nakamura et al., 2018), and a maximum likelihood (ML) tree was inferred in RAxML-NG v.1.2.2 (Kozlov et al., 2019), with GTR + Gamma selected as the nucleotide substitution model and 1000 bootstrap replicates. For a MtPt, if H. austrocaledonica/M. thomii is nested within Santalales with bootstrap support > 50%, we consider it a potential intracellular gene transfer (IGT) event; if H. austrocaledonica/M. thomii is nested within a non-Santalales lineage with bootstrap support > 50%, we consider it a potential horizontal gene transfer (HGT) event. A total of 24 species, including 12 holoparasitic or non-photosynthetic plants, four hemiparasitic plants, and eight autotrophic plants (Table S5), were used to investigate the loss of nuclear genes encoding organelle-targeted proteins. Although whether the two Cuscuta species are hemi- or holoparasites remains debatable, we tentatively regarded them as holoparasites, following the treatments of Cai et al. (2021) and Chen et al. (2023). For species without available genome sequences (Table S6), we de novo assembled their transcriptomes using Illumina RNA-seq reads with Trinity v.2.5.1 (Grabherr et al., 2011) with default parameters. Cd-Hit v.4.8.1 (Fu et al., 2012) with the parameter ‘-c 0.9’ was used to remove redundancy for the assembled transcripts, and then, TransDecoder v.5.7.1 (https://github.com/TransDecoder/) was employed to predict open reading frames. To eliminate bacterial contamination, contigs were searched against the NCBI bacterial nonredundant database. Contigs with over 80% of aligned length and over 90% identity to bacterial sequences were discarded. The assembly quality was evaluated using the predicted proteins against the eukaryota_odb10 database in Busco v.5.5.0 (Simão et al., 2015) (Table S7). We used NHMMSCAN to search all assembled transcripts against the HMMs of plastid genes of the 12 Santalales mentioned above, with the E-value threshold set to 1e-5. To determine the sequencing depth for each transcript, we mapped the Illumina DNA-seq and RNA-seq reads using Bowtie2, setting the parameters ‘--local’ for the former and ‘--end-to-end --no-mixed --no-discordant’ for the latter. Transcripts with at least one NHMMSCAN hit and both DNA-seq depth and RNA-seq depth ≥ 1 were further searched against sequences of the mitogenome and nuclear rRNA genes of H. austrocaledonica and GenBank NT database using Blastn with default parameters. The nuclear Benchmarking Universal Single-Copy Orthologs (BUSCO) genes of H. austrocaledonica were first identified from its assembled transcripts using the Busco assessment tool v.3.0.2 (Simão et al., 2015) with the ‘protein’ mode, based on the eukaryota_odb10 database (255 genes). The corresponding DNA contigs of these BUSCO genes were determined by searching these transcripts against the SPAdes assembly using Blastn with default parameters. To identify the DNA contigs of D. taylorii and M. thomii containing BUSCO genes, we searched the transcripts of the H. austrocaledonica BUSCO genes against the SPAdes assemblies of the two species using Tblastn with default parameters, keeping only hits > 30 amino acids and > 60% identity in amino acid sequence. Sequencing depths of these DNA contigs corresponding to the BUSCO genes of the three species were then directly extracted from the results calculated above. Of the 255 conserved single-copy genes in the eukaryota_odb10 database in Busco, 211 are commonly present in at least 17 species examined in this study. If a species has two or more copies for a gene, we randomly selected one for phylogenetic analysis. Protein sequences of the 211 genes were aligned using Mafft with the L-INS-i strategy. Poorly aligned regions were manually trimmed. All aligned sequences were then concatenated into a supermatrix. The best-fitting amino acid sequence substitution model (JTT + I + G4) was estimated using ModelTest-NG v.0.1.7 (Darriba et al., 2019), and the ML tree was constructed using RAxML-NG with 1000 nonparametric bootstrap replicates. We used Orthofinder v.2.5.5 (Emms & Kelly, 2019) with default parameters to construct orthogroups (OGs) for all the protein sequences of the 24 species, which generated a total of 56 765 OGs. The largest 100 OGs were removed. When an OG contains proteins of at least eight hemiparasitic or autotrophic plants, it is defined as a conserved OG. Based on this criterion, 8795 were considered conserved OGs and used for subsequent analyses. We used Localizer v.1.0.5 (Sperschneider et al., 2017) with the ‘-p’ option, TargetP2 (Armenteros et al., 2019) and DeepLoc v.2.0 (Thumuluri et al., 2022), with default parameters to predict the localization of these proteins in the conserved OGs. For each protein, we considered it to be either plastid- or mitochondrion-targeted when there were at least two programs indicating the presence of plastid or mitochondrial signaling peptides with scores or probabilities ≥ 0.5. When one OG contained predicted plastid- or mitochondrion-targeted proteins from at least eight hemiparasitic or autotrophic plants, it was treated as a conserved plastid- or mitochondrion-targeted OG. If a conserved OG did not contain proteins of any examined species from a lineage, this OG was considered lost in that lineage. We examined the status of conserved OGs involved in certain organellar functions, including organellar DNA transcription, translation, and protein degradation (Table S8). Proteins for these functions and the localizations of these proteins were extracted from the literature (Tiller & Bock, 2014; Börner et al., 2015; Nishimura & van Wijk, 2015; Woodson et al., 2016; Richardson & Schnell, 2019). To infer whether plastids are present in the cells of H. austrocaledonica, we also examined the OGs related to plastid protein transport (TOC–TIC import complex) and heme biosynthesis (Table S9). PROTmp genes of Scurrula parasitica var. graciliflora and Ipomoea batatas, which were not annotated in their gene annotation files, were identified using Tblastn with default parameters by searching the PROTmp sequence of Arabidopsis thaliana (AT5G15700) against their genomes (Table S10). When an OG includes paralogs derived from gene duplication, we constructed a ML tree with the same pipeline for species tree construction to distinguish the paralogs. To understand the function of OGs lost in certain holoparasitic or we carried out a analysis. Briefly, annotation of these OGs was performed based on the of thaliana the or the The et al., 2021) was used to determine based on a threshold of with a using 8795 conserved OGs as paired-end reads from Illumina DNA we assembled the mitogenomes of Dactylanthus taylorii Hachettea austrocaledonica and Mystropetalon thomii The contigs of the latter two species were assembled into with their and bp, and average sequencing depths of and The mitogenome of D. taylorii was assembled into a very complex of contigs, the presence of (Fig. total size is bp with an average sequencing depth of The three mitogenomes have and and genes of other plants, except for a in H. austrocaledonica that to due to an (Table Blastn of the mitogenomes of H. austrocaledonica and M. thomii against the plastome of an in that regions are highly to its plastome (Fig. Table analysis that plastid regions could from intracellular DNA from the plastome or from other plants that their own plastid DNA sequences Table assembly using the program GetOrganelle (Jin et al., 2020) with a Santalales plastome database (Table S3) as the generated and contigs for D. taylorii, H. austrocaledonica, and M. thomii, (Table However, for D. taylorii and M. thomii, contigs at least one Blastn or Nhmmscan hit to a database containing 12 plastomes of Santalales were all identified as either mitochondrial sequences or nuclear For H. austrocaledonica, the assembled contigs with at least one hit mitochondrial sequences or nuclear with DNA-seq and RNA-seq and nuclear contigs with DNA-seq depths and extremely to RNA-seq depths (Table novo draft genome assembly using SPAdes (Prjibelski et al., 2020) generated 12 and contigs for D. taylorii, H. austrocaledonica, and M. thomii, (Table which and contigs have a Blastn or Nhmmscan hit to the 12 plastomes of Santalales. However, in to mitochondrial or nuclear rRNA sequences other contigs are to be nuclear based on their DNA-seq depths for of D. taylorii and H. austrocaledonica, and for all of M. which are to DNA-seq depths of their own nuclear BUSCO genes Of these and rRNA contigs of D. taylorii and H. austrocaledonica and have a > DNA-seq search of the 24 contigs with a parameter that contain either short AT-rich regions to the plastid of other plants short coding regions of plastid genes with or no hits sequences of their nuclear with the of mitochondrial and nuclear rRNA gene contigs, all contigs of H. austrocaledonica very (Fig. with the that plant plastid genes very et al., be that the plastids of Mystropetalaceae have extremely Although this be it that holoparasitic plants from the related Balanophoraceae of plastid DNA compared with mitochondrial DNA (Table no plastome contigs were in the plastome and genome assemblies of the three Mystropetalaceae species. assembly of Hachettea generated transcripts (Table S6), of which have at least one NHMMSCAN hit to any of the 12 Santalales plastomes and both DNA-seq depth and RNA-seq depth ≥ search of the transcripts against sequences of the mitogenome and nuclear rRNA genes of Hachettea and the GenBank NT database that and are transcripts of its own mitochondrial genes and nuclear rRNA genes, and the remaining either are to nuclear rRNA genes nuclear genes mitochondrial DNA plastid DNA of other species or no hits (Table the eight transcripts to plastid DNA of other species, seven have both very DNA-seq depths and very RNA-seq depths The remaining is by a DNA-seq depth and a very short hit DNA-seq and RNA-seq depths for each of these eight contigs that are to be plastid gene transcripts nuclear sequences based on their DNA-seq depths to of BUSCO genes (Table no plastid gene transcripts were in plastome loss in Hachettea is further by loss of nuclear genes that plastid DNA transcription, translation, and other We identified OGs of nuclear genes using the data of 24 plants, including holoparasitic or non-photosynthetic lineages and their related autotrophic hemiparasitic plants The holoparasitic or non-photosynthetic lineages loss of conserved nuclear genes encoding proteins compared with autotrophic and hemiparasitic plants (Fig. the holoparasitic or non-photosynthetic and Hachettea have lost at least as many conserved genes encoding proteins as the two other holoparasitic lineages with plastomes and which includes and (Fig. of genes encoding mitochondrion-targeted proteins is also in the holoparasitic or non-photosynthetic the of loss is that of proteins (Fig. analysis of genes encoding proteins lost in the holoparasitic or non-photosynthetic lineages that the for commonly lost OGs in and Hachettea were for plastid translation, the three lineages and one or two other holoparasitic lineages were for photosynthesis and plastid (Fig. Table We further three of proteins related to plastid DNA transcription, plastid translation, and plastid protein and for the of we also their counterparts in the mitochondrion-targeted (Table S8). The first proteins for plastid DNA genes are by two of RNA in RNA and RNA mitochondrial genes are by only et al., land plants, there three namely mitochondrion-targeted and et al., 2015), in there is only mitochondrion-targeted of all plastid genes to et al., the three genes, both and were in all eight autotrophic and hemiparasitic plants and one holoparasitic lineage only was in all other holoparasitic plants except Hachettea, which both and Table S8). is that the examined holoparasitic plants with a plastome only for plastid DNA By the mitochondrion-targeted was in all 24 examined plants (Fig. Table S8). The of both and in Hachettea plastid DNA highly The proteins for plastid is performed by a that is of a and a of all proteins the and of all proteins the are by the nuclear the remaining are by the plastome & Bock, 2014). all nuclear genes encoding these were present in the examined plants with a plastome holoparasitic or very were in Polytomella Rafflesia Sapria and Hachettea (Fig. Table S8). However, mitochondrion-targeted protein genes were present in all these lineages (Fig. Table S8). The loss of all or nearly all nuclear genes encoding plastid proteins in these lineages makes plastid highly The the plastid which is for the of plastid proteins and the of or proteins & van Wijk, 2015). the plastid complex from one gene, and nuclear and & van Wijk, 2015). the of and Hachettea, in which and of nuclear genes encoding only proteins were not other species at least eight of these genes (Fig. Table S8). All examined species is to plastids and et al., loss of genes encoding the plastid complex cause of plastid proteins. By the four only to the and at least two are retained in all species (Fig. Table S8). The of any contig in the three genera of Mystropetalaceae, the of plastid gene transcripts and the loss of nuclear genes involved in plastid DNA transcription, translation, and protein degradation in Hachettea, provide compelling evidence for their plastome the three genera are as (Su et al., 2015), it is that plastome loss in the of the family. that there is a that Mystropetalaceae a plastome with limited loss of nuclear genes that proteins involved in plastid and have a on the of a However, that the loss of these nuclear genes in the of the three genera, it is that the plastome with limited has been lost over a long plastome loss in due to the same loss of related nuclear genes as identified in Sapria and Rafflesia (Fig. Table S8). all three genera of Rafflesiaceae have lost their plastome loss in their the of non-photosynthetic lineages in green plants, are extremely are plastomes retained in nearly all non-photosynthetic The tRNA that in translation, is for heme biosynthesis in plants, including non-photosynthetic et al., However, the trnE gene is reported as from the plastomes of four holoparasitic plant three from Balanophoraceae and et al., 2025) and one from & Renner, 2015), this We and copies of the gene from genome contigs of the three species of be to the nuclear or mitochondrial genomes (Table This could from from the plastomes before their still be present in cells of Mystropetalaceae, as was in Polytomella (Smith & Lee, 2014) and Rafflesia (Molina et al., 2014). This is by the of for the of genes in heme biosynthesis out of and encoding the import complex out of in Hachettea (Table S9). using be an to whether This was by the of to The Hachettea was and by the South of New We and for the of Mystropetalaceae species. We are also to three and for their on and and performed and and the and to this All the raw sequencing reads generated in this were deposited at the for Information the All the assembled mitogenomes or mitochondrial genes were to GenBank accession and The assembly and annotated coding sequences of assembled mitogenomes in this are also available in the database of Hachettea austrocaledonica and Mystropetalon of based on DNA-seq the mitogenome of Mystropetalon thomii, and DNA-seq and RNA-seq data that of Hachettea The assembly of the mitogenome of Dactylanthus phylogenetic analysis of plastid regions in the mitogenomes of Hachettea austrocaledonica and Mystropetalon and DNA sequencing depths for SPAdes contigs of Dactylanthus taylorii and Mystropetalon thomii with a hit to the plastomes of 12 Santalales species. A maximum likelihood tree of 24 plants based on protein sequences of organelle-targeted RNA Table Sampling details for three species of Table Sequencing data generated for three Mystropetalaceae species in this study. Table data used for plastome assembly and plastome contig search three species of Mystropetalaceae, and for phylogenetic analysis. Table of Santalales mitogenomes. Table or data of 24 plants used in this study. Table of assembled transcriptomes of five plants examined in this study. Table Benchmarking Universal Single-Copy Orthologs (BUSCO) assessment of nuclear genome or data of 24 Table of genes in orthogroups related to plastid RNA proteins and in 24 Table of genes in orthogroups related to plastid and heme biosynthesis in 24 Table of PROTmp from the genomes of Scurrula parasitica var. graciliflora and Ipomoea using with the sequence of Arabidopsis thaliana (AT5G15700) as a Table analysis for plastid regions (MtPts) in the mitogenomes of Hachettea austrocaledonica and Mystropetalon Table of plastome assembly using GetOrganelle and genome assembly using SPAdes for three species of Table of contigs of Dactylanthus taylorii assembled using Table of contigs of Mystropetalon thomii assembled using Table of contigs of Hachettea austrocaledonica assembled using Table of contigs of Dactylanthus taylorii assembled using Table of contigs of Hachettea austrocaledonica assembled using Table of contigs of Mystropetalon thomii assembled using Table DNA sequencing depth of BUSCO Universal Single-Copy genes of Dactylanthus Table DNA sequencing depth of BUSCO Universal Single-Copy genes of Hachettea Table DNA sequencing depth of BUSCO Universal Single-Copy genes of Mystropetalon Table Sequencing depths of and of three Balanophoraceae species. Table Information of transcripts of Hachettea austrocaledonica with a NHMMSCAN hit to 12 plastomes of Santalales. Table of genes encoding organelle-targeted proteins in 24 Table orthogroups for 24 Table and of organelle-targeted orthogroups in 24 Table of 24 plants examined in this study. Table mitochondrion-targeted orthogroups for 24 Table analysis for the conserved orthogroups that are commonly lost species Table of potential plastid trnE genes from contigs 200 of three Mystropetalaceae species assembled using is not for the or of any Supporting Information supplied by the be to the New The is not for the or of any information supplied by the be to the corresponding for the The New remains with to in and in any