1. Characteristics of ITS2 Secondary Structure

The first acknowledged ITS2 secondary structure was determined in S. cerevisiae based on chemical and enzymatic probing together with minimum free-energy calculation of the in vitro synthesized 37S pre-rRNA^[1]. In this structure, ITS2 sequence folds extensively through long-range interaction adjacent to the 5′ end of 25S and the 3′ end of 5.8S, yielding five tightly base-paired helixes, which was later known as the "hairpin model" (Figure 1c). Subsequent verification tests in vivo showed that site-directed mutations or deletions on different sites of this structure could completely or partially block the maturation of 25S rRNA^[2][3][4]. These findings supported the ITS2 "hairpin model" and showed its distinct functional components.

Figure 1. Overview of the ITS2 (internal transcribed spacer 2) processing and folding in the ribosome biogenesis of the yeast S. cerevisiae. (a) ITS2 location and processing in pre-rRNA from genome to transcriptome. ITS2 and ITS1 are intercalated in the 18S–5.8S–25S tandem arrays, separating the elements of the pre-rRNA. The endonucleolytic cleavage sites are labelled A–E on the pre-rRNA. ITS2 region is highlighted in dark red in each processing state. (b) ITS2 location in the secondary structure of the 5.8S/25S pre-RNA. The ITS2 and 5.8S are displayed in dark red and bright green, respectively; the six typical domains of 25S are labelled I–VI in distinct colors. (c) The three proposed secondary structure models for ITS2. The C₂ cleavage site is highlighted in red in stem III of the ring-pin model. The secondary structure scheme is taken and modified from S. cerevisiae LSU (http://apollo.chemistry.gatech.edu/RibosomeGallery/).

Several experimental methods in vivo, such as cryo-EM, site-directed mutagenesis and chemical and structural probing are reliable to infer RNA secondary structures, but they are laborious and are not suitable for large dataset analyses. Alternatively, the most conventional method is use of thermodynamic energy optimization to predict secondary structures and then use phylogenetic comparative analysis to examine common structures from closely related lineages. Specifically, as more and more ITS2 structural features become available in databases^[5], ITS2 secondary-structure predictions will be based not only on the nucleotide sequence but also on the templates for homology modeling, in which subtle structural motifs that rarely automatically fold by standard minimum free energy (MFE) approach are preset^[6].

Unexpectedly, in silico prediction from a broad range of eukaryotic diversity has yielded two more ITS2 secondary-structure models, which are distinct from the "hairpin model" being found in yeast and a few other species. One is the “ring model”, which is detected mainly in vertebrates^[2][7]. In this model structure, the 5′ to 3′ long-range interactions in the "hairpin model" disappear, wherein each strand folds by itself into two separate stems instead. In addition, the longest stem always shares a common short base-stem with another stems, thereby characterized as “giant stem with lateral branches” (Figure 1c). Collectively, four or five extended stems in this structure radiate from an open central core or ring.

The third ITS2 secondary-structure model termed "four-helix model"^[8] or "ring-pin model"^[9], is widely observed throughout the eukaryote^[10][11][12]. In this structure, the longest stem of the ring model and its two neighboring stems merge together into a much longer stem. Therefore, the “ring-pin model” is typically recognizable as a much longer helix together with three relatively short helices radiating from an open central ring. Notably, not all the eukaryotes have the fixed four helices because the helix I and helix IV are most variable and may disappear in some organisms (Figure 1c). In contrast, the longest helix III and its neighboring helix II are more stable and common to all eukaryotes. The basal pairing of the helices I and helix II are usually base-paired conserved and serve as a scaffold for shaping the structure. The most conserved component of helix III at the tip of the helix, and in land plants a hallmark motif UGGU, exist in this region^[8][10].

2. Dynamic Model of ITS2 Secondary Structure

Cote et al. first examined the occurrence and functional significance of the distinct ITS2 structures between the "hairpin model" and "ring model" using functional genetic assay^[13]. Unexpectedly, they found both model structures are necessary in vivo. Given their different effects on downstream processing, Cote et al. hypothesized that the "ring model" structure promotes AFs binding and is required in earlier pre-rRNA complex assembly, and is then followed by an induced “zipping up” transition to the “hairpin model” structure that facilitates formation of the ITS2 proximal stem. This idea is supported by the recent high-resolution cryo-EM structures of pre-60S particles which give convincing evidence that ITS2 function through conformational changes^[14][15]. This process is also coordinated with stepwise binding or separating of AFs, by which the transient structure complex can be stabilized or flexible to another change^[16]. In contrast, Burlacu et al. monitored rRNA restructuring processing events from 35S to 27SB using a high-throughput RNA structure probing method (ChemModSeq15), by which they quantified the nucleotide flexibilities of ITS2 structure^[9]. The result showed that ITS2 generally maintains a conserved structure, which agrees best with the “ring-pin model”. Thus, it remains largely elusive whether ITS2 functions with a fixed or dynamic conformational model.

Although each model has been confirmed by alternative methods, the current three ITS2 secondary-structure models are more or less derived from minimum free energy folding (MFE). Actually, ITS2 does not fold and function by itself but is mutually interdependent with dozens of AFs that function as chaperones and scaffolds to direct and stabilize ITS2 folding^[9][17][18]. It is reasonable to speculate that the final ITS2 folding shape does not necessarily follow the law of thermodynamics. It is not surprising that the proposed structure in vivo is rarely automatically yielded by standard MFE folding. Despite recent advances in cryo-EM that have provided detailed portions of ITS2 secondary-structure information^[19], it is still challenging to get a comprehensive ITS2 secondary structure from the snapshots of a number of assembly intermediates.