Human ribosomal RNA gene arrays display a broad range of palindromic structures

Abstract

The standard model of eukaryotic ribosomal RNA (rRNA) genes involves tandem arrays with hundreds of units in clusters, the nucleolus organizer regions (NORs). A first genomic overview for human cells is reported here for these regions, which have never been sequenced in their totality, by using molecular combing. The rRNA-coding regions are examined by fluorescence on single molecules of DNA with two specific probes that cover their entire length. The standard organization assumed for rDNA units is a transcribed region followed by a nontranscribed spacer. While we confirmed this arrangement in many cases, unorthodox patterns were also observed in normal individuals, with one-third of the rDNA units rearranged to form apparently palindromic structures (noncanonical units) independent of the age of the donors. In cells from individuals with a deficiency in the WRN RecQ helicase (Werner syndrome), the proportion of palindromes increased to one-half. These findings, supported by Southern blot analyses, show that rRNA genes are a mosaic of canonical and (presumably nonfunctional) palindromic units that may be altered by factors associated with genomic instability and pathology.

Figure 1.

Structural analysis of the human rDNA locus by molecular combing. (A) Schematic representation of two canonical rDNA units. Restriction with EcoRI (sites E) yields four distinct fragments spanning the transcribed region (thick line) and the intergenic spacer or IGS (thin line). The orientation of the rRNA transcript and the positions of the ribosomal genes (black boxes) are shown. (B) Two-color hybridization on combed human DNA. The red probe is the 5′ EcoRI fragment B detected with Texas Red. The green probe is the 3′ fragment A detected with FITC. The image displays 10 canonical rDNA units in tandem, each composed of a dual fluorescent signal and the adjacent nonhybridizing spacer segments. (C) Hybridizations of the probes on human DNA that illustrates noncanonical units. The image displays a region containing two canonical units (left) followed by seven palindromic units, with each half joined by its 3′ region (3′–3′ palindromes) and separated by short IGS segments. (D) Hybridizations illustrating successive inverted units with gaps. The image displays five canonical units, followed by a series of seven palindromic units separated by short gaps. (E) Hybridizations illustrating 5′–5′ palindromic units. The image displays six canonical units (left), followed by two gapless palindromic units joined at their 5′ extremities and two canonical units (separated by a short IGS) in an inverted orientation with respect to the canonical units on the left. (F) Hybridizations illustrating 5′–5′ palindromic units with gaps. The image displays six canonical units, followed by a single 3′ fragment and three 5′–5′ palindromes with short central gaps separated by short IGS segments. (G) Hybridization illustrating complex recombinant structures. The image displays closely spaced, inverted units (left), followed by two complex units with alternating 5′ and 3′ coding sequences, separated by short IGS sequences, and one canonical unit (right). Scale bar, 10 kb, applicable to all images.

Figure 2.

Variability in rDNA spacer length. (A) A series of hybridization signals depicting the variability in the length for 3′–3′ palindromes. (B) Signals for 5′–5′ palindromes. (C) Bar graphs of the distribution of palindromic signals with respect to the total length between the outer probes, for 5′–5′ palindromes in red and 3′–3′ palindromes in green. The data represent a compilation of measurement on 306 individual palindromes.

Figure 3.

Noncanonical units in human cell lines. A bar chart showing the total percentage of noncanonical rDNA units in control cell lines in dark gray (with the total number of counted rDNA units for each cell line in parenthesis): (1) AG12657 (1948); (2) AG11561 (1554); (3) AG05283 (1640); (4) AG13077 (1549); (5) AG07898 (998); (6) AG07897 (2522); (7) D1; (8) MRCV (3649); (9) IMR90 (6357); and in Werner syndrome cell lines in light gray: (10) AG07896 (1269); (11) AG03141 (2421); (12) AG03829 (2206); (13) AG12797 (1275); (14) AG06300 (1275); (15) AG11395 (1381); (16) WV1 (1003). Horizontal bars are set at the average percentage for control and WS cells. Error bars are shown for the mean of three experiments; the other values are each the mean of two experiments.

Figure 4.

Southern blotting of human DNA hybridized with the 5.9-kb probe (fragment B, Fig. 1). DNA was extracted from a control (D1) and two Werner (AG03829 and AG07896) cell lines, in addition to two primary fetal cells (IMR90 and MRC5). (A) EcoRI digestion. The two major expected bands are apparent, at 5.9 and 20 kb, after a complete and partial enzymatic digestion due to a polymorphism of an EcoRI site (Wellauer and Dawid 1979). The arrows indicate the supplemental bands. On the right of the gels, a schematic representation presents the genomic structures that can account for the different bands after EcoRI digestion. Red lines represent the 5.9-kb probe, whereas green bars indicate the 7.1-kb probe (Fig. 1), black bars correspond to the DNA fiber. Gray bars show the regions recognized by the radioactive probe. Vertical black dotted lines show the restriction sites. The vertical gray dotted lines signify a position of digestion that can be lost after the rearrangement. (B) HindIII digestion. The band at ∼14.5 kb is expected from the digestion of canonical units (at the top of the scheme). Additional bands are visible, between 5.9 and 14.5 kb. The scheme on the right shows the noncanonical units that could generate those bands upon HindIII digestion. (C) I-PpoI digestion, run on a PFEG. The control and Werner sample (D1 and AG03829, respectively) show the expected band at 43 kb and several additional bands. One band of the Werner sample shows a different size, suggesting that a type of rearrangement is more represented in this cell type. The scheme explains the canonical and noncanonical units that could generate the supplementary bands. Numbers on the right of the gels indicate the approximate size of the bands, expressed in base pairs.

Figure 5.

Competing processes underlying palindrome formation, variation, expansion, and elimination. The pathway of formation on the left and the pathway of elimination on the right. (A) Schematic representation of canonical rDNA genes (5′ in red, 3′ in green) separated by IGS sequences. (B) One of many possible mechanisms, inversions are presented based on recombination between Alu repeats (represented by black arrows) of opposite orientation present in the IGS sequences. (C) One of the two products of recombination with crossover in B producing inverted rDNA genes. (D) Single-strand loop formation (possibly with cleavage at the base). (E) A 3′–3′ palindrome generated by the preceding reactions. (F) Expansion of the palindromic sequence. (G) Competing reaction of elimination of the palindromic sequence, analogous to those presented in D or DNA replication slippage, shown by the arrow coming from the right.