Helitrons are one of the three groups of eukaryotic class 2 transposable elements (TEs) so far described. They are the eukaryotic rolling-circle transposable elements which are hypothesized to transpose by a rolling circle replication mechanism via a single-stranded DNA intermediate. They were first discovered in plants (Arabidopsis thaliana and Oryza sativa) and in the nematode Caenorhabditis elegans, and now they have been identified in a diverse range of species, from protists to mammals. Helitrons make up a substantial fraction of many genomes where non-autonomous elements frequently outnumber the putative autonomous partner. Helitrons seem to have a major role in the evolution of host genomes. They frequently capture diverse host genes, some of which can evolve into novel host genes or become essential for Helitron transposition.

Helitrons were the first group of TEs to be discovered by computational analysis of whole genome sequences. The first Helitrons described were called Aie, AthE1, Atrep and Basho which are Non-autonomous Helitrons found in the genome of Arabidopsis thaliana, a small flowering plant. Despite these discoveries, the classification of Helitrons was unknown until 2001 when the discovery of protein coding-elements which were predicted to be the autonomous partners. Kapitonov and Jurka investigated the coding capacity of Helitrons in A. thaliana, Oryza sativa, and Caenorhabditis elegans using in silico studies of repetitive DNA of these organisms, computational analysis and Monte Carlo simulation. They described the structure and coding potential of canonical Helitrons and proposed the rolling-circle mechanism of transposition as well as the possibility that some of the encoded genes captured from the host are now used for replication. Their survey of the genome of these organisms showed that Helitron activity could contribute to a significant fraction (~ 2%) of the plant and invertebrate genomes where they were found, but the extent of their distribution elsewhere was not clear.

In 2003, a group of investigators studied the structure of proteins related to Helitrons and the different coding domains within them by looking for Helitron-like elements in vertebrates, specifically zebra fish, Danio rerio and a puffer fish, Sphoeroides nephelus. The Rep/Helicase proteins were predicted to be 500 to 700 amino acids longer because of a C-terminal fusion of a domain with homology to apurinic-apyrimidinic (AP) endonuclease. Previous phylogenetic studies showed that the AP endonuclease is nested within the Chicken Repeat 1 (CR1) clade of non-long terminal repeat (non-LTR) retrotransposons. This relationship suggested that AP endonuclease originated from a retrotransposon insertion either nearby or within a Helitron. These investigators were not able to identify the ends of the Rep/Helicase/Endonuclease unit of Helitrons.

In recent years, Helitrons have been identified in all eukaryotic kingdoms but their genomic copy numbers are highly variable, even among closely related species. They make up 1–5% of the genomic DNA in different fruit flies, 0–3% in mammals, >0.5% in the frog. In most mammals Helitron's presence is negligible and limited to remnants of old transposons, with the exception of bat genomes, which are populated by numerous young elements. However, many years after the description autonomous Helitrons, no mechanistic studies have been published and therefore the rolling-circle mechanism of transposition remains a well-supported but not yet tested hypothesis.

Helitrons are structurally asymmetric and are the only class of eukaryotic DNA transposons that do not generate duplications of target sites during transposition. Canonical Helitrons typically begin with a 5′ T (C/T) and terminate with the nucleotides CTRR (most frequently CTAG, but occasionally variation has been noted) but do not contain terminal inverted repeats. In addition, they frequently have a short palindromic sequence (16 to 20 nucleotides) hairpin about 11 bp from the 3′ end. They integrate between an AT host dinucleotide. Some families of Helitrons also carry tandem repeats, like microsatellites and minisatellites which are generally highly mutable sequences.

Most Helitrons are non-autonomous elements and share common termini and other structural hallmarks with autonomous Helitrons, but they do not encode any complete set of proteins encoded by the autonomous elements. The main enzymatic hallmarks of Helitrons are the rolling-circle (RC) replication initiator (Rep) and DNA helicase (Hel) domains, which are present in a protein comprising 1000–3000 amino acids (aa) (Rep/Hel) encoded by all autonomous Helitron elements. The Rep/Helicase protein includes zinc finger motifs, the Rep domain (which is a ~100-aa and has HUH endonuclease activity), and an eight-domain PiF1 family helicase (SuperFamily1) which are universally conserved in Helitrons. The zinc finger-like-motifs have been associated with DNA binding. The ~400-aa Hel domain is classified as a 5' to 3' DNA Hel which is involved in the breaking and joining of single-stranded DNA and are characterized by both the presence of the HUH motif (two histidine residues separated by a hydrophobic residue) and the Y motif (one or two tyrosine residues that are separated by several amino acids). The PiF1 family of helicases (Hel) has 5′ to 3′ unwinding activity which for many rolling-circle entities this activity is host encoded. Plant Helitrons also encode an open reading frame with homology to single-stranded DNA-binding proteins (RPA). Typically, the RPA proteins in Helitrons are 150 – 500-aa long and are encoded by several exons. In all Helitrons, the Rep domain precedes the Hel domain.

The three-dimensional structure of Helitron transposase covalently bound to the left transposon end has been recently determined by cryoEM.

Helitrons are proposed to transpose by a mechanism similar to rolling-circle replication via a single-stranded DNA intermediate. Two models are proposed for the transposition mechanism: the concerted and the sequential. In the concerted model, the donor strand cleavage and ligation occurs simultaneously while in the sequential model they occur in a stepwise fashion. The concerted model does not require a circular intermediate although they could occur if a step fails or is bypassed during transposition. The sequential model differs in that a circular intermediate is a required step of transposition and because, until very recently, circular intermediates were not known for Helitrons, the concerted model was adapted to explain transposition.