Understanding the origins and evolution of genes remains one of the central challenges in evolutionary biology. A fundamental question is how new genetic material arises, becomes integrated into existing cellular networks, and contributes to cell fitness and diversity. While ancient genes underpin essential cellular functions, new genes can introduce novel traits and adaptive advantages. Exploring the mechanisms and dynamics of gene emergence and loss is therefore key to understanding the evolution of genomes over time. The oldest genes present in the extant genomes are the ones there were also present in the last universal common ancestor (LUCA; Mushegian, 2008). Newer genes are the ones that emerged later. Besides a variety of duplication-based mechanisms (Tautz and Domazet-Lošo, 2011), new genes could emerge from non-coding DNA sequences through random mutations (Neme and Tautz, 2014). While it has been proposed that this type of gene emergence is common in all species, the frequency of this phenomenon probably varies a lot, and is difficult to assess quantitatively (Light et al., 2014). Recently, Iyengar and Bornberg-Bauer (2023) developed a mathematical model that suggests that genes are lost much more rapidly than they emerge, and that new genes preferentially arise in transcribed regions. This concept is well-illustrated by a recent study of human proto genes (Grandchamp et al., 2022), suggesting that new genes tend to “capture” regulatory sequences in their vicinity, such as introns, promoter motifs, enhancers. Proto genes that survive the initial selection tend to grow by gaining new functional domains. While comparative genomics has provided strong evidence for the phenomenon of de novo gene emergence in various organisms, experimental validation of gene emergence is obviously challenging. To experimentally illustrate the selection process following “gene birth,” Knopp et al. (2019) constructed plasmid libraries with short randomly generated open reading frames (ORFs) and expressed them in Escherichia coli. They demonstrated that short peptides offering a fitness benefit during exposure to aminoglycoside antibiotics could be effectively selected by bacterial cells.

Genomic phylostratigraphy is a computational method for studying genome evolution based on estimating evolutionary age of individual genes (Domazet-Lošo et al., 2007). For any species to be analyzed, the first step is to create a consensus phylogeny tree. In this species-centered tree, each node is named a phylostratum (PS). Each PS is then populated with individual genes whose founders emerged at that specific node. From the whole genome perspective, this means that all genes get distributed in phylostrata corresponding to their evolutionary age. In the early evolutionary history, the first few PS correspond to deep, shared ancestry (e.g., LUCA, Bacteria) and are therefore common across different bacterial species such as E. coli and B. subtilis. Subsequent PS become increasingly species-specific as bacterial lineages diverge. Phylostratigraphy has proven to be a powerful method for explaining macro-evolutionary phenomena. Certain complex functions that emerged at some point in time tend to involve large sets of genes, which then cluster to a specific PS. Hence, phylostratigraphy was used to demonstrate that cancer-related genes could be coupled to the emergence of multicellularity (Domazet-Lošo and Tautz, 2010a), and that the expression of genes during ontogenic development of metazoa follows an age-specific pattern (Domazet-Lošo and Tautz, 2010a). The latter study has been particularly noted for providing direct evidence for the hourglass theory of development (Casci, 2011). There has been some debate about the reliability of sequence similarity search algorithms underlying phylostratigraphic approach (Moyers and Zhang, 2015, 2016; Domazet-Lošo et al., 2017, 2024). However, the statistical mapping of well-studied functional data on phylogenies repeatedly demonstrates that classical sequence similarity searches accurately recover macroevolutionary information (Domazet-Lošo et al., 2017, 2024; Xia et al., 2025).

It is known that bacterial genomes undergo extensive horizontal gene transfer (Arnold et al., 2022). This evolutionary process is particularly prominent in mixed bacterial communities (Brito, 2021). Mobile genetic elements, such as plasmids and phages, have been recognized as the main driver of horizontal gene transfer (Lang et al., 2017). Since phylostratigraphy was initially developed for metazoa, i.e., species with predominantly vertical mode of evolution, it was not initially clear how well the method would perform in analyzing bacterial genome evolution. However, it turned out to work remarkably well. Using genomic phylostratigraphy, Futo et al. (2021) demonstrated that the development of Bacillus subtilis biofilms recapitulates phylogeny at the expression level. This finding suggested that the ontogeny of bacterial biofilms is a developmental process similar to that of e.g., metazoans or plants (Koska et al., 2024). Similarly, phylostratigraphy has been used to characterize the development of different morphotypes of Borreliella (syn. Borrelia) burgdorferi (Corak et al., 2023). Based on the assumption that bacterial sporulation is also a true developmental phenomenon, phylostratigraphy has been successfully used to predict new sporulation genes, among uncharacterized genes in sporulation-related phylostrata (Shi et al., 2020). This validated the earlier proposition that phylostratigraphy could be a useful tool in genome mining pipelines, as previously suggested (Mijakovic, 2020).

Bacterial genomes are large circular DNA molecules, in which the origin of replication (and terminus of replication, at 180° with respect to origin) plays a major organizational role (Duigou and Boccard, 2017). Each chromosome arm between the origin of replication and replication terminus consists of two regions, a non-structured region and a so-called macrodomain (Valens et al., 2004). The “Right” and “Left” macrodomains are defined by their incapacity to interact with each other, a phenomenon linked to high level structural organization of the bacterial genome. The position of genes with respect to the bacterial origin of replication has a deep significance. Genes near the replication origin tend to be more highly expressed (Ying et al., 2014; Kosmidis et al., 2020; Lato and Golding, 2020). Gene essentiality is also high near the origin of replication (Kosmidis et al., 2020; Lato and Golding, 2020). Genes farther away from the origin are more prone to molecular changes, such as substitutions, recombination events and genomic rearrangements (Lato and Golding, 2021). In B. subtilis, during asymmetric division leading up to spore formation, chromosomal location governs the timing of expression of sporulation genes (Zupancic et al., 2001).

In this report, we used genomic phylostratigraphy to examine the relationship between gene age and chromosomal organization in two model bacterial species, B. subtilis and E. coli. We hypothesized that differences in lifestyle and ecology between B. subtilis (predominantly soil-dwelling, sporulating bacterium) and E. coli (a facultative anaerobe associated with animal hosts) may have influenced their evolutionary paths and their propensity to acquire new genes. We further hypothesized that bacterial genome dynamics, including rate of gene acquisition, genomic placement of new genes, their transcriptional regulation, and their lateral mobility, might exhibit distinct patterns compared to those observed in eukaryotes.

The bacterial species studied were selected based on their differing habitats, with B. subtilis being a predominantly soil-dwelling, sporulating bacterium, and E. coli a facultative anaerobe associated with animal hosts. Additionally, they were chosen because they are among the best-characterized bacterial models, with early selection driven by their biological properties (such as sporulation and rapid growth, respectively), followed by extensive development of genetic tools and comprehensive genomic resources. For E. coli, we specifically focused on the non-pathogenic laboratory K-12 strain, which serves as a standard reference genome and minimizes confounding variation associated with pathogenicity.

What can be learned from looking at bacterial genome organization from the perspective of evolutionary age of genes? The bulk of bacterial genes are dedicated to core functions, including housekeeping, and these are naturally encoded by the oldest genes, situated in PS1–5 (Figures 1A, B, Supplementary Data Sheets 1, 2). A majority of the proteins in both species belonged to PS1 indicating that many of the core functions in present day bacteria also existed in the last universal common ancestor (LUCA). This contrasts with the situation in Eukarya, were the fraction of PS1 proteins is much lower, indicating that many more new genes appeared in this domain during the course of evolution. Developmental phenomena in bacteria, such as biofilm formation (Futo et al., 2021) and sporulation (Shi et al., 2020) are known to be more recent inventions, and as suggested by our study, typically involve newer genes (Figures 5C–F). This is in line with bacterial developmental processes being cited as models for ongoing social evolution in bacteria (Boyle et al., 2013). Since bacterial social interactions in biofilms are highly dynamic and evolvable, their intense evolution can be directly observed in adaptive evolution experiments (Martin et al., 2016). B. subtilis is often cited for its rich repertoire of developmental and community-based genetic programs, involving biofilm and pellicle formation, sporulation, cannibalism and genetic competence (Ricci-Tam et al., 2023). By contrast, E. coli has a less diverse lifestyle, with biofilm formation being the only developmental process it can do. This is very accurately reflected in the phylostratigraphy analysis, with only 5% of E. coli genes in PS6–11 (Figure 1B), and B. subtilis with 20% of all genes in PS6–15. Does this mean that E. coli evolved fewer de novo genes than B. subtilis during its evolution? Not necessarily. This question just brings us to horizontal gene transfer, recognized as a major generator of novelty in bacterial genomes (Arnold et al., 2022). The origins of novel bacterial genes likely involve a combination of evolutionary mechanisms. These include horizontal gene transfer from other bacteria or phages, gene duplication followed by divergence, frameshift-based innovation, and de novo emergence from previously non-coding sequences (Lang et al., 2017; Tautz and Domazet-Lošo, 2011; Neme and Tautz, 2014; Xia et al., 2025). While prophage regions represent identifiable hotspots for gene acquisition in both E. coli and B. subtilis, they account for only a fraction of the younger genes. Additional new genes may arise gradually within transcriptionally active regions, as proposed for proto-gene evolution in eukaryotes (Grandchamp et al., 2022), although the relative importance of these mechanisms in bacteria remains difficult to quantify. In many cases, phages represent a large fraction of the strain-specific DNA sequences (Brüssow et al., 2003). Our results suggest that the prophage regions in E. coli and B. subtilis genomes represent hot spots for more recent genes (Figure 3). Gene transfer is particularly intense in bacterial communities (Brito, 2021). This means that any de novo genes that evolve in bacteria became more or less immediately available to all community members by means of horizontal gene transfer. Then, the genetic makeup of an individual bacterial species gets defined by the preferred niche and environmental challenges. The soil dwelling B. subtilis, having to cope with vary variable and adverse environmental challenges, picked up more of the “novelty” tools for its adaptation toolbox. E. coli, adapting to a less challenging and more constant environment of a symbiont, does not require such a diverse developmental toolkit, hence the reduced proportion of novel, development-related genes. However, E. coli is fully capable of taking up additional functions when switching from commensal to pathogen lifestyle (Dobrindt et al., 2010), a phenomenon that has been described as the “unexhausted potential” of E. coli (Blount, 2015). Overall, it would appear that de novo genes are relatively accessible to bacteria, that are far less siloed than e.g., plants or metazoans, with strictly vertical evolution patterns.

Younger genes identified in bacteria follow the same pattern as those in more complex life forms (Neme and Tautz, 2014), they tend to be short, non-essential, and expressed to a low level or only under certain specific conditions. This description, in conjunction with the known association of essential and highly expressed genes to the bacterial origin of replication (Ying et al., 2014; Kosmidis et al., 2020; Lato and Golding, 2020), would suggest that the more recent genes should be expected to cluster closer to the replication terminus. Surprisingly, our analysis revealed a weak (with prophages included) or no correlation (with prophages excluded) between location of more recent genes and the replication origin (Figure 2). One possible explanation for this finding is linked to constraints on placement of new genes in the genome. As noted by Grandchamp et al. (2022), de novo genes in humans are more likely to arise in transcribed regions, coupled to existing elements of transcriptional regulation. This makes sense from the probabilistic perspective, since there are fewer constraints for creating an open reading frame from a non-coding sequence, than for creating a promoter region and other regulatory sequences required for ensuring gene expression in each organism. While bacteria do not have complex transcription regulation based on chromatin de-condensation to enable transcription, they do possess transcriptional regulatory units known as operons. We report an analogous observation that in bacteria new genes get predominantly inserted in preexisting operons (Figures 4A, B) and thus get placed under control of available transcription regulators (Figures 4C–F). As a housekeeping process, transcription control is generally ensured by old genes, and we found very few examples on new transcription regulators, such as ComK, Rok, GerR and SwrAA/1 from B. subtilis. Evolving of new developmental phenomena seldom involves establishing completely novel operons. Only 11 such operons were found in B. subtilis and 13 in E. coli (Tables 1, 2). Rather, new developmental programs in bacteria seem to arise as a combination of new genes, and extensive repurposing of older genes and operons (Figures 5C–F). This is supported by our observation that many of the genes involved in processes such as sporulation, competence, and biofilm formation originate from the oldest PS, particularly PS1, and have likely been incorporated into newer regulatory frameworks. This aligns with the view that bacterial developmental systems can evolve as modular additions, but also indicates that these modules are, at least in part, constructed from pre-existing components whose original functions were retained or adapted to new roles. In this light, bacterial development reflects both evolutionary innovation and the reorganization of existing genetic material, rather than purely the acquisition of entirely novel gene modules.

Our analysis of the two model bacteria suggests that there are important lessons to be learned from looking at bacterial genome evolution and genome organization from the perspective of evolutionary age of new genes. More recent bacterial genes tend to be short, non-essential, and their level of expression is generally low. Despite these features, newer genes are surprisingly not preferentially located far from the origin of replication. Their genomic location is rather uniform, and they are only in some instances enriched in areas containing mobile genetic elements, such as prophages. While most bacterial transcription regulators belong to the oldest phylostrata, they were found to regulate expression of both older and more recent genes alike. This suggests that many newer genes get inserted in the existing operons under the control of conserved regulatory elements. In particular, E. coli contains significantly fewer novel genes than B. subtilis, and this is mirrored by its more limited repertoire of developmental processes. By contrast, B. subtilis exhibits a more eventful evolutionary past in terms of acquisition of new genes, and particularly in this species, which possesses highly elaborate developmental regulation (e.g., during sporulation), it is evident that new regulatory systems have also evolved. The observation that many genetic operons contain genes from different phylostrata indicates a layered evolutionary history, in which new genes are incorporated into both pre-existing and newly formed frameworks for gene expression.

Genomic features described above suggest that phylostratigraphy approaches could assist in genome mining efforts (Mijakovic, 2020), not by predicting individual gene function directly, but by identifying phylostrata enriched in specific biological processes. For example, in the case of sporulation in B. subtilis, genes involved in this process were overrepresented in certain phylostrata, and a higher fraction of uncharacterized proteins from these phylostrata were experimentally shown to be involved in sporulation (Shi et al., 2020), demonstrating how phylostratigraphy can highlight candidate genes for targeted validation, particularly in biological processes involving functionally related gene sets that emerged during the same evolutionary interval.