The GC% landscape of the Nucleocytoviricota

Abstract

Genomic studies on sequence composition employ various approaches, such as calculating the proportion of guanine and cytosine within a given sequence (GC% content), which can shed light on various aspects of the organism’s biology. In this context, GC% can provide insights into virus-host relationships and evolution. Here, we present a comprehensive gene-by-gene analysis of 61 representatives belonging to the phylum Nucleocytoviricota, which comprises viruses with the largest genomes known in the virosphere. Parameters were evaluated not only based on the average GC% of a given viral species compared to the entire phylum but also considering gene position and phylogenetic history. Our results reveal that while some families exhibit similar GC% among their representatives (e.g., Marseilleviridae), others such as Poxviridae, Phycodnaviridae, and Mimiviridae have members with discrepant GC% values, likely reflecting adaptation to specific biological cycles and hosts. Interestingly, certain genes located at terminal regions or within specific genomic clusters show GC% values distinct from the average, suggesting recent acquisition or unique evolutionary pressures. Horizontal gene transfer and the presence of potential paralogs were also assessed in genes with the most discrepant GC% values, indicating multiple evolutionary histories. Taken together, to the best of our knowledge, this study represents the first global and gene-by-gene analysis of GC% distribution and profiles within genomes of Nucleocytoviricota members, highlighting their diversity and identifying potential new targets for future studies.

Supplementary Information

The online version contains supplementary material available at 10.1007/s42770-024-01496-7.

Introduction

The advent of DNA sequencing, pioneered by various researchers but made possible by the Sanger method in 1977, heralded a new era for biological studies [1]. Since then, a wide range of DNA sequencing technologies have been developed and enhanced, enabling exploration of diverse biological aspects in much greater depth [2]. Genomics is considered a Big Data science, demanding various data analysis methods, bioinformatic tools, and approaches [3]. Metrics utilized for genomic analysis are of significant importance, such as sequence composition analysis, which encompasses features like length, k-mer word frequency, codon usage, and nucleotide percentages. Specifically, the measured proportion of guanine (G) and cytosine (C) nucleotides (GC% content) within a target genomic sequence (RNA or DNA) is an easily calculated metric associated with a broad range of biological aspects, providing crucial information about genome structure and function as it affects the stability, flexibility, and interaction of nucleic acid molecules [4–6].

Used as a fundamental parameter of genome sequence variation, GC% content has been extensively explored, for example, in evolutionary and microbiological studies [7–12]. Additionally, GC% content has been identified as one of the genomic features associated with important biological events, such as horizontal genetic transfer (HGT), evolution, and acquisition of virulence factors [10, 13–17]. Similarly, various studies have demonstrated specific correlations between the GC% of pathogens and host organisms, suggesting that sequence composition plays a crucial role in the coevolution of different species [18, 19], including many viruses [20].

The known virosphere encompasses the most diverse biological entities described to date, highlighting the diversity of viruses and their genomes [21]. The phylum Nucleocytoviricota comprises a hypothetical monophyletic taxon of large and giant DNA viruses. Currently, there are eleven official families described in this phylum, according to the International Committee on Taxonomy of Viruses. Nucleocytoviruses are known pathogens of algae, protists, invertebrates, and vertebrates, including humans [22–26].

For instance, these viruses can reach sizes of up to 2.3 µm in particle size, with genome lengths as long as 2.8 Mb [27, 28]. Due to their large genomes, intergenic regions might be more abundant, which impacts their genome evolution [27, 28]. Another notable characteristic of nucleocytoviruses is the identification of unusual genes involved in canonical cellular metabolism, such as peptide synthesis factors, tRNAs, aminoacyl tRNA synthetases, and factors associated with tRNA/mRNA maturation and ribosome protein modification, bringing Nucleocytoviricota genomes closer in content to that of their hosts [27].

Therefore, considering the diversity of species, genomes, and host ranges, members of the Nucleocytoviricota are of significant interest for genomic and evolutionary studies. In this study, we describe a comprehensive characterization of Nucleocytoviricota genomes based on GC% content, focusing on gene/open reading frame (ORF) sequences. Our data suggest that gene-by-gene analyses of GC% content can provide insights into the evolution, genome expansion, and host-virus relationships of nucleocytoviruses.

Methods

Nucleocytoviruses selection and data acquisition

Representative viruses from different viral families within the phylum Nucleocytoviricota were selected to build the dataset used in this study. After extensive manual curation, we selected representative genomes from each genus within the families, based on the official taxonomy described by ICTV as [29, 30]. The primary criterion used was to select nucleocytoviruses previously considered prototype species of each genus or viral family. Considering that the term "prototype species" is no longer used and in some taxa, there is no prototype reference, we employed a second criterion for selection: viral representatives based on viral characterization (e.g., Eastern kangaroopox virus, genus Macropopoxvirus). Additionally, we purposely included more giant Nucleocytoviricota viruses since these viruses carry the largest genomes in the phylum, enriching our sequence composition analysis within the dataset. Lastly, publicly available complete genomes were downloaded from NCBI’s GenBank [31]. The genomic data were obtained in files generated by GenBank, which consisted of the complete FASTA sequences and descriptions of all predicted CDS/ORFs for each viral genome.

Virus-host GC% correlation analysis

To observe possible correlations between the GC% content of viruses and host organisms, we manually curated known hosts for each virus included in this study using the Virus-Host Database website [19]. In cases where neither a specific host was listed nor complete host genomes were available on GenBank (e.g., lymphocystis disease virus 1), representative host taxa were listed. Subsequently, genomic GC% data of hosts were obtained directly from reference genomes deposited on GenBank (Table 1).

Table 1.

GC% content average of viral genome, viral ORF, and associated host organism genome. Overall GC% content of Nucleocytoviricota viruses (Phycodnaviridae, Mimivirdae, Ascoviridae, Iridoviridae, Marseilleviridae, Asfarviridae, Extended Asfarviridae, Poxviridae, “Pithoviridae”, Yarviridae, "Pandoraviridae", and unclassified DNA viruses). Observed viral taxonomy at family level, viral species, GenBank accession for viral genome, genome size (kb), genomic GC% content, ORF GC% content of each virus, host organisms, and GC% content of host organisms are described

Family	Virus	GenBank Accession	Genome size (kb)	Viral genomic GC%	Viral ORF GC%	Host organism	Host genomic GC%
Phycodnaviridae	Paramecium bursaria Chlorella virus 1	NC_000852.5	330.6	40	40.6	Chlorella variabilis	65.5
	Emiliania huxleyi virus 86	NC_007346.1	407.3	40.2	40.6	Emiliania huxleyi	64.5
	Ectocarpus siliculosus virus 1	NC_002687.1	335.6	51.7	52.2	Ectocarpus siliculosus	53.49
	Heterosigma akashiwo virus 1	NC_038553.1	274.7	30.4	30.8	Heterosigma akashiwo	N/A
Mimiviridae	Cafeteria roenbergensis virus	NC_014637.1	617.4	23.3	23.4	Cafeteria roenbergensis	70.44
	Acanthamoeba polyphaga mimivirus	NC_014649.1	1,181.5	28	28.8	Acanthamoeba castellanii	58.35
						Acanthamoeba polyphaga	58.7
	Samba virus	KF959826.2	1,181.5	28	28.7	Acanthamoeba castellanii	58.35
	Megavirus chiliensis		1,246.1			Acanthamoeba castellanii	58.35
		NC_016072.1		25.3	26.3	Acanthamoeba polyphaga	58.7
						Acanthamoeba griffini	N/A
	Moumouvirus australiensis	MG807320.1	1,098	25.1	26	Acanthamoeba polyphaga	58.7
	Tupanvirus deep ocean	MF405918.2	1,439.5	29.4	30.5	Acanthamoeba castellanii	58.35
				29.4	30.5	Vermamoeba vermiformis	42.48
	Tupanvirus soda lake	KY523104.2	1,516.2	29.1	30.2	Acanthamoeba castellanii	58.35
						Vermamoeba vermiformis	42.48
	Aureococcus anophagefferens virus	NC_024697.1	370.9	28.7	29.3	Aureococcus anophagefferens	69.92
	Bodo Saltans Virus	MF782455.1	1,385.8	25.3	25.7	Bodo saltans	51.6
	Phaeocystis globosa virus group I	NC_021312.1	459.9	32	33.4	Phaeocystis globosa	N/A
	Chrysochromulina ericina virus	NC_028094.1	473.5	25.4	26	Haptolina ericina	N/A
	Tetraselmis viridis virus	KY322437.1	668	41.2	40.6	Tetraselmis viridis	N/A
	Cotonvirus japonicus	AP024483.1	1,476.5	25.3	26.6	Acanthamoeba castellanii	58.35
Ascoviridae	Spodoptera frugiperda ascovirus 1a	NC_008361.1	156.92	49.2	50.3	Spodoptera frugiperda	36.37
Iridoviridae	Lymphocystis disease virus 1	NC_001824.1	102.65	29.1	29.2	Osmeridae (Hypomesus transpacificus **)	44.5
						Percichthyidae (Maccullochella peelii **)	40.5
						Percidae (Etheostoma cragini **)	40.5
						Sparus aurata	41.94
						Centrarchidae	N/A
						Pleuronectidae	N/A
						Pleuronectoidei	N/A
						Soleidae (Brachirus orientalis **)	39
						Clupeidae (Alosa alosa**)	42.5
	Frog virus 3	NC_005946.1	105.9	55.1	57.1	Notophthalmus viridescens	N/A
						Lithobates pipiens	N/A
						Dryophytes versicolor	N/A
						Lithobates sylvaticus	N/A
						Oophaga pumilio	26.9
	Invertebrate iridescent virus 3	NC_008187	191.1	47.9	50.4	Mosquitos ( Aedes taeniorhyncus**)	N/A
	Decapod iridescent virus 1	MF599468.1	165.8	34.6	29.8	Penaeus vannamei	36.5
	Invertebrate iridescent virus 6	NC_003038.1		28.6	35	Acheta domesticus	38.5
						Gryllus bimaculatus	38.6
			212.4			Spodoptera frugiperda	36.37
						Choristoneura fumiferana	38.1
						Chilo suppressalis	35.7
Marseilleviridae	Marseillevirus marseillevirus	NC_013756.1	368.4	44.7	45	Acanthamoeba castellanii	58.35
						Acanthamoeba polyphaga	58.7
	Brazillian marseillevirus	KT752522	362.2	43.3	43.9	Acanthamoeba castellanii	58.35
	Golden marseillevirus	NC_031465.1	360.6	43.1	43.7	Acanthamoeba castellanii	58.35
						Limnoperna fortunei	33.6
	Lausannevirus	NC_015326.1	346.7	42.9	43	Acanthamoeba castellanii	58.35
	Tunisvirus	NC_038511.1	380	43	43.6	Acanthamoeba castellanii	58.35
Asfarviridae	African swine fever virus	NC_001659.2		38.6	38.7	Chlorocebus aethiops	40.9
			189.3			Sus scrofa	41.6
						Phacochoerus africanus	40.48
Extended Asfarviridae	Pacmanvirus S19	MZ440852.1	418.5	33.2	34.3	Acanthamoeba castellanii	58.35
	Faustovirus e12	KJ614390	465.9	37.7	36.9	Vermamoeba vermiformis	42.48
	Kaumoebavirus	MT334784.1	362.5	43.1	43.4	Vermamoeba vermiformis	42.48
Poxviridae	Fowlpox virus	NC_002188.1	291	31.2	31.3	Gallus gallus	42
						Meleagris gallopavo	41.1
	Sheeppox virus	NC_004002.1				Meleagris gallopavo	41.1
			149.9	25	25.3	Capra hircus	42.1
						Ovis aries	43.2
	Yokapox virus	NC_015960.1	175.7	25.6	26.2	Mus musculus	41.95
	Mule deerpox virus	AY689437.1	170.5	27	27.6	Odocoileus virginianus	41.5
	Nile crocodilepox virus					Crocodylus niloticus	N/A
		NC_008030.1	190	61.9	62.4	Crocodylus porosus	43.85
						Crocodylus johnsoni	N/A
	Myxoma virus	NC_001132.2	162.4	43.5	43.5	Oryctolagus cuniculus	43.97
	Eastern kangaroopox virus	MF467281.1	170.1	54	54.3	Macropus giganteus (eastern gray kangaroo*)	44.3
	Molluscum contagiosum virus	MH646551.1	192.1	64.3	63.9	Homo sapiens	40.4
	Sea otterpox virus	NC_037656.1	127.8	31.3	31.6	Enhydra lutris	N/A
	Vaccinia virus	NC_006998.1	182.5	33.4	34.6	Homo sapiens	40.4
						Bos taurus	41.92
	Cotia virus	KM595078.1	185.1	23.6	24.4	Chlorocebus aethiops	40.9
						Mus musculus	41.95
	Orf virus					Homo sapiens	40.4
		NC_005336.1	139.9	63.8	64.3	Capra hircus	42.1
						Ovis aries	43.2
	Pteropox virus	NC_030656.1	133.4	33.8	34	Pteropus scapulatus	N/A
	Salmon gillpox virus	NC_027707.1	241.5	37.5	37.1	Salmo salar	N/A
	Squirrelpox virus	NC_022563.1	148.8	66.7	67	Sciurus vulgaris	39.26
	Swinepox virus	NC_003389.1	146.4	27.4	27.8	Sus scrofa	41.6
	Eptesipox virus	NC_035460.1	176.6	23.6	23.9	Eptesicus fuscus	43.5
						Chlorocebus aethiops aethiops	N/A
	Yaba monkey tumor virus					Erythrocebus patas	41.05
		NC_005179.1	134.7	29.8	30.1	Papio hamadryas	40.9
						Homo sapiens	40.4
	Anomala cuprea entomopoxvirus	NC_023426.1	245.7	20	20.5	Anomala cuprea	N/A
	Amsacta moorei entomopoxvirus	NC_002520.1	232.3	17.8	18.2	Lymantria dispar	38.55
	Melanoplus sanguinipes entomopoxvirus	NC_001993.1	236.1	18.3	19	Locusta migratoria	41
						Schistocerca gregaria	42.55
	Diachasmimorpha longicaudata entomopoxvirus	KR095315.1	252.9	30.1	31	Diachasmimorpha longicaudata	N/A
“Pithoviridae”	Pithovirus sibericum	NC_023423.1	610	35.8	40.2	Acanthamoeba castellanii	58.35
	Cedratvirus A11	NC_032108.1	589	42.7	43	Acanthamoeba castellanii	58.35
Pithoviridae-like	Orpheovirus	NC_036594.1	1,473.5	25	28.1	Vermamoeba vermiformis	42.48
Yaraviridae	Yaravirus brasiliensis	MT293574.1	44.9	58	58	Acanthamoeba castellanii	58.35
"Pandoraviridae"	Pandoravirus quercus	NC_037667.1	2,077.2	60.7	64.4	Acanthamoeba castellanii	58.35
Unclassified DNA virus	Medusavirus	AP018495.1	381.2	61.7	61.5	Acanthamoeba castellanii	58.35
	Mollivirus sibericum	NC_027867.1	651.5	60.1	60.2	Acanthamoeba castellanii	58.35

GC% content calculation

After establishing the database containing members of the phylum Nucleocytoviricota, the GC% content calculation of CDS/ORFs was performed. We employed the publicly available InfoSeq tool developed by EMBL’s European Bioinformatics Institute, EMBOSS (available at https://www.ebi.ac.uk/Tools/emboss/) [32]. The InfoSeq outputs provided a list of each coding sequence within the viral genome, as well as its length (kb) and calculated GC% content, respectively.

Data analysis and plotting

Kruskal–Wallis’ test (p < 0.05) was used to determine the variance of the GC% content of each CDS/ORF from each nucleocytovirus isolate genome. Dunn’s multiple comparison test quoted comparison between two isolates against each other within a family, subfamily, or group. To better understand the dimension and profile of the variation in GC% content, gene by gene, along the studied viral genomes, scatter plots were created using Microsoft's Power BI tool. GC% values are represented on the y-axis, and relative CDS/ORF positions along viral genome extension are represented on the x-axis. A tendency line (red) and GC% mean values (dashed blue) were included. Similarly, to visualize ORF GC% ranges within taxa/subgroups, considering overall minimum, maximum, and mean GC% of each, a heatmap-like graphic analysis was performed using Microsoft’s Excel. Conditional formatting was used to color categorize GC% value variation associated with each individual CDS/ORF within the genomes. Blue corresponds to lower GC% values, yellow corresponds to mean GC% value, and red corresponds to higher GC% values observed on each individual taxa/subgroup analyzed. Considering that different nucleocytoviruses had great variability among CDS/ORF quantity (not necessarily corresponding to genome size), we normalized the heatmap-like analysis with the number of CDS/ORF represented on the scale.

Analysis of potential gene duplication, phylogenetic, and HGT inferences

To test the hypothesis that CDS/ORFs of discrepant GC% could be associated with possible gene duplication events or HGT between host-viruses, we constructed a new dataset of target sequences. Considering outliers’ variations from GC% sequences along viral genome extension, and many CDS/ORFs to be analyzed, we purposefully decided to evaluate the three highest and three lowest GC% values observed on each individual viral genome. Therefore, based on gene sequences of minimum and maximum GC% of each viral genome, targeted sequences were selected for further analysis.

Following this approach, a targeted CDS/ORF dataset was built and used for phylogenetic and HGT analysis, as well as gene duplication event search using the NCBI BLASTp tool [33] (Supplementary Table 1). For the establishment of possible gene duplication events, BLASTp analysis of target sequences against original viral genomes and manual curation of data were performed, considering a threshold of 40% coverage and 40% identity. Subsequently, amino acid sequences from NCBI’s databank were compared to the proposed dataset employing the BLASTp tool (e-value threshold of 0.05) [33]. Upon these local alignments, viral targets that presented homologous sequences outside the virosphere were selected. BLASTp was also used against each viral host specific databank. Genes usually associated with cellular metabolism were favored as targeted sequences, as their identification on viral genomes suggests potential content acquisition from another organism. Lastly, phylogenetic analyses were performed for BLASTp hits considering a coverage threshold of 40% and an identity threshold of 30%. The availability of sequence hits of other taxa (non-viral) was also considered whilst determining the final dataset for phylogenetic analysis. For each dataset, the sequences were aligned using the MAFFT software, and phylogenetic analyses were conducted using the maximum likelihood method with the IQ-TREE 2 software [34, 35]. The amino acid substitution model automatically determined the best fit for each dataset. Bootstrap analysis with 1000 replicates was performed. Tree topologies were analyzed to identify potential horizontal gene transfer events accordingly to the methodology described by Irwin et al. (2022) [36].

Results and discussion

Nucleocytoviricota dataset preparation: open reading frame (ORF) vs intergenic regions’ GC% content

In this study, we carefully selected viral genomes to construct our main dataset, including only those with complete genomes available on GenBank (NCBI). Representative genomes from the main viral families of Nucleocytoviricota were chosen, totaling sixty-one complete viral genomes: Phycodnaviridae (n = 4), Mimiviridae (n = 13), Ascoviridae (n = 1), Iridoviridae (n = 5), Marseilleviridae (n = 5), Asfarviridae (n = 1), Poxviridae (n = 23), “Pithoviridae” (n = 2; and 1 pithoviridae-like), “Pandoraviridae” (n = 1), as well as extended Asfarviridae (n = 3), and unclassified DNA viruses (n = 2), according to the proposed ICTV taxonomy in 2020 (Walker et al., 2020). Additionally, we included the recently discovered Yaravirus (Yaraviridae) in our analysis, despite not being classified within Nucleocytoviricota, due to phylogenomic analyses indicating a close relationship between Yaravirus and nucleocytoviruses [37].

We manually curated data from the Virus-Host Database, indicating host organisms for all viruses within the dataset (see Table 1), and extracted the genomic GC% content of host organisms. For lymphocystis disease virus 1, which has multiple probable hosts within fish species, single representatives with available complete genomes were selected based on NCBI’s taxonomy. However, the genomic GC% content from host organisms for certain viruses was not available, including Heterosigma akashiwo virus 1, Phaeocystis globosa virus group 1, Chrysochromulina ericina virus, Tetraselmis viridis virus, invertebrate iridescent virus 3, sea otterpox virus, pteropox virus, salmon gillpox virus, Anomala cuprea entomopoxvirus, and Diachasmimorpha longicaudata entomopoxvirus.

As described in various models, sequence composition may vary along genome extensions, particularly between intergenic and intragenic regions [6, 38–40]. In our study, we found that the GC% content of coding sequences and intragenic regions tended to present similar sequence composition, as the whole genome GC% values were comparable to the ORF mean GC% (see Table 1). Contrary to expectations, this suggests there may be no differential selective pressure over these distinct parts of the viral genome, or at least none associated with GC%. However, some exceptions were noted, including decapod iridescent virus 1, invertebrate iridescent virus 6, and pithovirus sibericum, which exhibited differences between the total genome GC% mean and the ORF GC% mean (ranging from 4.4% to 6.4% variation). Additionally, codon-usage analysis was performed and compared to %GC content profiles, revealing a clear relationship between %GC and the use of codons rich in C and G.

Nucleocytoviricota coding sequence (CDS)/ORF GC% variation

The GC% content variation profile of nucleocytoviruses coding sequence (CDS)/ORF assessed has a notable range with a minimum value of 8.13% (Amsacta moorei entomopoxvirus; CDS NP_065034.1), and a maximum value of 83.91% (Orf virus; CDS NP_957782.1), both described as hypothetical proteins of representatives of the Poxviridae family (Fig. 1 and supplementary Table 1). These aspects cohesively demonstrate the Poxviridae family as the family with the greatest GC% range within Nucleocytoviricota, thus presenting the most influence over the phylum’s GC% profile.

Fig. 1.

Nucleocytoviricota heatmap-like GC% profile. Lowest GC% values categorized in blue, highest GC% values categorized in red. Minimum, mean and maximum GC% of the entire taxon dataset are used for categorizing ORF values and scale. Respective identified viruses: Paramecium bursaria Chlorella virus 1 (PbCv1), Emiliania huxleyi virus 86 (Ehv86), Ectocarpus siliculosus virus 1 (Esv1), Heterosigma akashiwo virus 1 (Hav1) (Phycodnaviridae 1–4); Cafeteria roenbergensis virus (Crov), Acanthamoeba polyphaga mimivirus (Apmv), Samba virus (Sbv), Megavirus chiliensis (Mvch), Moumouvirus australiensis (Mvau), Tupanvirus deep ocean (Tvdo), Tupanvirus soda lake (Tvsl), Aureococcus anophagefferens virus (Auanv), Bodo Saltans Virus (Bsv), Phaeocystis globosa virus group I (Phgvg1), Chrysochromulina ericina virus (Chrev), Tetraselmis viridis virus (Tvv), Cotonvirus japonicus (Cvj) (Mimiviridae 5–17); Spodoptera frugiperda ascovirus 1a (Sfa1a) (Ascoviridae 18); Lymphocystis disease virus 1 (Ldv1), Frog virus 3 (Fv3), Invertebrate iridescent virus 3 (Iniv3), Decapod iridescent virus 1 (Div1), Invertebrate iridescent virus 6 (Iniv6) (Iridoviridae, 19–23); Marseillevirus marseillevirus (Mvmv), Brasillian marseillevirus (Bramv), Golden marseillevirus (Gmv), Lausannevirus (Lauv), Tunisvirus (Tuv) (Marseilleviridae 24–28); African swine fever virus strain BA71V (Asfv), Pacmanvirus S19 (PvS19), Faustovirus e12 (Fve12), Kaumoebavirus (Kauv) (Asfarviridae and extended asfarviridae (29–32); Fowlpox virus (Fpoxv), Sheeppox virus 17077–99 (Shpoxv), Yokapox virus (Ykpoxv), Mule deerpox virus (Mdpoxv), Nile crocodilepox virus (Ncpoxv), Myxoma virus (Myxv), Eastern kangaroopox virus (Ekpoxv), Molluscum contagiosum virus (Mcv), Sea otterpox virus (Sopoxv), Vaccinia virus (Vacv), Cotia virus (Cov), Pteropox virus (Ptpoxv), Salmon gillpox virus (Sgpoxv), Orf virus (Orfv), Squirrelpox virus (Spoxv), Swinepox virus (Swpoxv), Eptesipox virus (Eptpoxv), Yaba monkey tumor virus (Ymtv), Anomala cuprea entomopoxvirus (AcEpoxv), Amsacta moorei entomopoxvirus (AmEpoxv), Melanoplus sanguinipes entomopoxvirus (MsEpoxv), Diachasmimorpha longicaudata entomopoxvirus (DlEpoxv) (Poxviridae 33–54); Pithovirus sibericum (PvS), Cedratvirus A11 (CdvA11), Orpheovirus (Orphv) (“Pithoviridae” and pithoviridae-like 55–57); Yaravirus brasiliensis (Yvbra) (Yaraviridae 58), Pandoravirus quercus (Pvq) (“Pandoraviridae”, 59), Medusavirus (Medv), and Mollivirus sibericum (MolS) (Unclassified DNA viruses, 60–61)

In terms of CDS/ORF GC% variation and mean within viral families, the following ranges were observed: (i) From 18.61% to 64.63% in Phycodnaviridae, with Ectocarpus siliculosus virus 1 and Heterosigma akashiwo virus 1 presenting an overall maximum and minimum GC% content, respectively; (ii) From 15.15% to 69.09% in Iridoviridae, with both frog virus 3 and invertebrate iridescent virus 3 presenting significantly higher GC% content in comparison to other family members; (iii) From 27.88% to 57.36% in Marseilleviridae, with no specific representatives; (iv) From 19.66% to 62.96% in Asfarviridae (and extended Asfarviridae), with kaumoebavirus and pacmanvirus S19 presenting an overall maximum and minimum GC% content, respectively; (v) From 9.18% to 57.83% in “Pithoviridae” (and pithoviridae-like), with orpheovirus presenting notably lower GC% content; (vi) From 9.68% to 62.55% in Mimiviridae, with Tetraselmis viridis virus and Cafeteria roenbergensis virus presenting an overall maximum and minimum GC% content, respectively; and (vii) From 8.13% to 83.91% in Poxviridae, with lower GC% represented in Entomopoxvirinae (Figs. 2 and 3).

Fig. 2.

Heatmap-like GC% profile of nucleocytoviruses families. Lowest GC% values (blue) and highest GC% values (red) are represented. Minimum, mean, and maximum GC% of the entire taxon dataset are used for categorization of ORF values and scale. Represented families: Phycodnaviridae, Iridoviridae, Marseilleviridae, Asfarviridae* (Asfarviridae and extended asfarviridae), and “Pithoviridae”** (“Pithoviridae” and pithoviridae-like)

Fig. 3.

Heatmap-like GC% profile of nucleocytoviruses families. Lowest GC% values (blue) and highest GC% values (red) are represented. Minimum, mean, and maximum GC% of the entire taxon dataset are used for categorization of ORF values and scale. Represented families: Mimiviridae and Poxviridae

When looking specifically into CDS/ORF GC% variation within individual viral genomes, we observed how the GC% varies in coding sequences along the extension of genomes, which could allow for the identification of possible hotspots for HGT or duplication events (Fig. 4). Interestingly, we also observed an absence of specific patterns for GC% variation among coding sequences of the viral families of the nucleocytoviruses assessed (supplementary Figs. 1–7). Thus, representative genomes of each family were selected for demonstrating the main aspects of ORF GC% distribution along genome extension (Fig. 4).

Fig. 4.

Scatter plot of ORF GC% variation along genome extension. Dots represent GC% of specific ORF in its position within the virus genome, according to annotation. Tendency line (red), ORF GC% mean (dashed blue), and outliers (red triangles) are represented. Selected representative viruses of each family: A Orfv, B Ldv1, C Fv3, D Vacv, E Asfv, F Mvmv, G PvS, H Chrev, I Ehv86, J Tvdo

When considering how our data might have presented statistical difference amongst analyzed groups, we performed Kruskal–Wallis’ test (p < 0.05) followed by Dunn’s multiple comparison test for the quoted comparison between two isolates against each other within a family, subfamily, or group. Our statistical evaluation of the dataset proposed and explored herein demonstrated that there are significant differences among isolates within viral families, subfamilies, and groups (supplementary Fig. 8). This was a consistent result even when comparing isolates of the Marseilleviridae which had the least GC% variation among all nucleocytoviruses’ families assessed, pointing out that the five isolates' genomes analyzed are significantly different (p < 0.0001) in terms of GC% content (supplementary Fig. 8).

Virus-host GC% similarities and HGT analysis

Virus-host coevolution plays a pivotal role in sharing genomic features between viruses and their hosts. Studies have demonstrated how sequence composition and nucleotide frequency correlation between virus-host pairs can be associated with viral adaptability dynamics, or even used as reliable metrics for inferring virus-host linkages [18, 41, 42]. When comparing the viral GC% content profile of nucleocytoviruses to those of host organisms included in this study, contrary to what one might expect, values were not similar in most cases (Table 1). A few exceptions were noted: Ectocarpus siliculosus virus 1 (51.7%) and its host Ectocarpus siliculosus (53.49%); decapod iridescent virus 1 (34.6%) and its host Penaeus vannamei (36.5%); invertebrate iridescent virus 6 (35% ORF GC mean) and its hosts Spodoptera frugiperda (36.37%) and Chilo suppressalis (35.7%); African swine fever virus (38.6%) and its hosts Chlorocebus aethiops (40.9%), Sus scrofa (41.6%), and Phacochoerus africanus (40.48%); kaumoebavirus (43.1%) and its host Vermamoeba vermiformis (42.48%); myxoma virus (43.5%) and its host Oryctolagus cuniculus (43.97%); and Yaravirus (58%) and its host Acanthamoeba castellanii (58.35%). Although a few patterns of GC% similarity between viruses and hosts were observed in this dataset, it is important to consider that many of these viruses can infect more than one host organism, implying different selective pressures on sequence composition. Additionally, many host organisms did not have complete genome sequences available for GC% calculation, making it impossible to compare viral GC% content at the moment.

Although limited evidence of direct correlation was observed between the GC% content of viruses and hosts analyzed in this work, the influence of host genomic characteristics was not excluded from the possibilities to further explain CDS/ORF GC% variation observed in viral genomes. Genome expansion through the acquisition of genes from host organisms by horizontal gene transfer (HGT) has been widely debated regarding nucleocytoviruses, and the literature has shown that different viral families among Nucleocytoviricota present different tendencies of HGT events [43]. Overall, Nucleocytoviricota viruses are considered to have some propensity to acquire host genes by HGT, and such events may play an important role in composing the content diversity and size of these viruses’ genomes [43]. It has even been hypothesized that NCLDVs encode homologs of conserved genes commonly found among the domains of Bacteria, Archaea, and Eukarya, suggesting the phylum could be considered as a fourth domain of life. However, this hypothesis has not been supported by multiple phylogenetic analyses, indicating instead multiple independent acquisitions from various cellular lineages [44–47].

Considering this, HGT was hypothesized as one of the possible causes of gene GC% variation in different nucleocytovirus genomes [15], especially in cases where proximal gene groups presented GC% content distant from the overall viral GC% mean. For instance, this was observed for Emiliania huxleyi virus 86 (Ehv86) (Fig. 4I), where a cluster of CDS/ORFs with discrepant GC% values (positions 290 to 315) was identified. Although events of HGT of entire metabolic pathways have been previously described for Ehv86 and its host Emiliania huxleyi [48], no hits for any of the CDS/ORFs were found when aligned to the host genome, leaving the discussion open regarding why these sequences differ in GC% composition and how they originated. Similarly, a discrepant GC% CDS/ORF cluster was identified within the genome of Chrysochromulina ericina virus (ChreV) (positions 422 to 442) (Fig. 4H), which may be linked to ChreV’s remarkable genomic characteristics, such as abundant mobile genetic elements, complex gene evolution, and host gene acquisition among other features [49].

After extensive data curation, we carefully selected 30 genes from our target CDS/ORF dataset based on the three maximum and three minimum GC% values observed for each viral genome (Supplementary Table 1). We then proceeded to assess phylogeny and evaluate HGT events according to the methodology described by Irwin et al. (2022) [36]. Among all analyzed targets, 14 potential HGT events were identified, such as the “Histone H2B/H2A fusion protein” (AMQ10945.1) of the Brazilian marseillevirus (Supplementary Figs. 9–35; Fig. 5A). Interestingly, a probable HGT from virus to host was observed (Fig. 5B) when evaluating the targeted sequence of “Papain-like cysteine peptidase” (YP_009310305.1) of Golden marseillevirus. Moreover, these findings are in accordance with previously described HGT events in Marseilleviridae members [50, 51].

Fig. 5.

Phylogenetic and HGT inference. Representative targets: A HGT from host to virus, target “Histone H2B/H2A fusion protein” (AMQ10945.1) of brazilian marseillevirus (KT752522), and lausannevirus (NC_015326.1); B HGT from virus to host, target “Papain-like cysteine peptidase” (YP_009310305.1) of golden marseillevirus (NC_031465.1); and (C) no evidence of HGT, target “CD47-like protein” (NP_659700.1) of sheeppox virus (NC_004002.1). Cyan: viral clusters; orange: target viral sequence; black: non-viral cluster. Analysis performed with 1000 replicates

Finally, other analyzed sequences were considered inconclusive or did not indicate potential HGT events, such as the targets “CD47-like protein” (NP_659700.1) of Sheeppox virus (Fig. 5C), “Putative replication factor and/or DNA binding/packing protein” (NP_078747.1) of Lymphocystis disease virus 1, and “NAD-dependent DNA ligase” (ATE87064.1) of Decapod iridescent virus 1 (both inconclusive regarding HGT inference). This does not exclude the possibility that these events could potentially be observed more frequently when accessing all the identified outliers for GC% content variation among CDS/ORFs of viral genomes and other viral isolates in future studies. Therefore, events of HGT remain a possible cause of GC% variation within viral genomes, yet to be further explored.

Potential paralogs and gene duplication events

Another hypothesis for the gene GC% variation observed in this study was the presence of duplicated ORFs. In this scenario, if a given ORF had one or more copies along the virus genome, it is hypothesized that one copy would remain conserved, while other copies would potentially evolve under different selective pressure conditions [52–56], likely resulting in GC% content variation. For instance, gene duplication is one of the known mechanisms by which genome evolution can be accelerated, allowing for the emergence of new genes with different functions [52, 55]. Moreover, gene duplication has already been described as a major component involved in genome expansion and genetic diversity among giant virus genomes [57].

Another piece of evidence suggesting that gene duplication could be associated with GC% variation lies within the Poxviridae family, specifically in the inverted terminal repetition sequences (ITR). ITRs consist of genome terminal regions containing inverted duplicated sequences, known to harbor most of the variable genetic content of poxvirus genomes, while conserved genes are primarily observed within the central genome region [58–60]. This aligns with the majority of GC% variation in poxvirus genes observed along the genome extension in the present work (supplementary Fig. 6).

Among all selected ORFs of maximum and minimum GC% analyzed (n = 336), we identified 60 genes with at least one potential duplication, using a threshold of 40% coverage and 40% identity. Of these ORFs, 46 presented two potential copies, five presented three potential copies, and nine presented four or more potential copies. The maximum quantity of probable copies identified for an ORF was 17, regarding the targeted “Putative ankyrin repeat protein” (AMK61738.1) of Samba virus. However, the majority (70%) of the identified ORFs are characterized as hypothetical proteins, while the remaining 30% are miscellaneous. Underrepresented miscellaneous groups of probable paralogs included “MHC-like TNF binding protein” (3.3%), “EFc gene family protein” (3.3%), “Putative ankyrin repeat protein” (3.3%), “Chemokine-binding protein” (1.7%), and “Collagen and repeat containing protein” (1.7%) (Supplementary Table 1).

Still considering the 60 identified ORFs, 55% are represented in the Poxviridae family, followed by Mimiviridae (13.3%), Phycodnaviridae (8.3%), Iridoviridae, and “Pithoviridae” (both with 6.7%), and Marseilleviridae and Asfarviridae (both with 5%) (supplementary Table 1). It is worth noting that a higher observation of duplicated genes is expected for poxvirus genomes due to ITRs. Moreover, the identification of probable gene duplication events based on GC% variation among genomes has led to interesting targets for future ortholog studies. Further investigation regarding gene duplication and orthology in nucleocytovirus genomes should be conducted to better understand these events.

Perspectives in sequence composition studies of viral genomes

When studying genomes and sequence composition, various approaches can be considered, one of which is the evaluation of GC% content. However, the ratio of guanines and cytosines in a DNA/RNA sequence is just one of the many nucleotide ratios that can be measured. A different metric, although still based on the G + C ratio, is the calculation of CpG dinucleotides, which differs from GC% content as it specifically measures bonded cytosines to guanines (Cytosine-Phosphate-Guanine). CpG dinucleotide is known to be associated with gene regulation through methylation, cancer-inducing factors, and even virulence augmentation, since certain antiviral proteins specifically bind to CpG [61–64].

In addition to CpG, other dinucleotide compositions in genomic sequences can provide important information about viruses and host organisms. In fact, different dinucleotide relative ratios can reflect the chemistry of dinucleotide stacking energies and base-step conformational tendencies of an organism, as well as species-specific properties of DNA modification, replication, and repair mechanisms [65]. Regarding other nucleotide composition evaluations, trinucleotide and tetranucleotide compositions should also be considered as relevant metrics for retrieving biological information from genome sequences of viruses and hosts [66, 67].

Furthermore, considering how assessing the GC% content profile of the phylum Nucleocytoviricota has provided an insightful view of large viral genomes, we consider other metrics of sequence composition analysis to be promising—especially since there remains a plethora of information yet to be unveiled from nucleocytoviruses’ sequence composition features.

Supplementary Information

Below is the link to the electronic supplementary material.

Authors contributions

All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by Amanda Stéphanie Arantes Witt, João Victor Rodrigues Pessoa Carvalho, Mateus Sá Magalhães Serafim. Jônatas Abrahão designed and supervised the study. The first draft of the manuscript was written by Amanda Stéphanie Arantes Witt and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript. This text has been revised by artificial intelligence.

Funding

We acknowledge financial support from Rede Vírus—Ministério da Ciência, Tecnologia e Inovações (MCTI), Câmara Pox—405249/2022–5. We thank Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq), Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), grant number 88882.348380/2010–1, Fundação de Amparo à Pesquisa do estado de Minas Gerais (FAPEMIG), Programas Institutos Nacionais de Ciência e Tecnologia (INCT), grant number 406441/2022–7, chamada 58/2022, and Pró-Reitorias de Pesquisa e Pós-Graduação of UFMG. J.S.A. is a CNPq researcher.

Data availability

The datasets generated during and/or analysed during the current study are available from the corresponding author.

Declarations

Ethics approval

This work is registered at SISGEN – Ministério do Meio Ambiente – number A2291C9.

Competing interests

The authors have no relevant financial or non-financial interests to disclose.