To be responsive to developmental requirements, rRNA synthesis is strictly regulated (Moss, 2004; Grummt, 2003; Moss and Stefanovsky 2002), reflecting specific needs of individual cell-types, developmental stages, and physiological conditions. The rRNA transcription requirement for most cellular activities remains unknown. Traditionally, the rate of transcription has been measured by metabolic labeling such as the nuclear run-on assay, which can be carried out only in isolated or cultured cells and, thus, is severely limited in its scope of application. In order to investigate rRNA transcription regulation in different cell types in vivo, we developed an in situ hybridization technique, which detects the quantity of the full-length transcript of rRNA genes, i.e., the 47S precursor ribosomal RNA (pre-rRNA) and rRNA processing intermediates. We demonstrated that the quantity of 47S pre-rRNA detected by this technique correlated positively with rRNA transcription rate in vivo (Cui and Tseng, 2004), and showed that 47S pre-rRNA was a far more sensitive indicator for perturbation in rRNA synthesis than the steady-state level of mature 18S or 28S rRNAs.
The mammalian eye is a highly specialized organ containing multiple types of tissues, each performing a distinctive function that contributes to vision. Within these tissues, cells of different developmental origins are in various stages of quiescence, proliferation, and differentiation. This abundance of cell types and physiological states in conjunction with its compact size make the eye an ideal organ to study cell type–specific regulation of rRNA transcription by in situ hybridization. Earlier studies on rRNA synthesis and stability in a limited number of ocular tissues revolved around in vivo or in vitro incorporation of radioactively labeled precursor RNA (Bok, 1970; Erickson and Fisher, 1990; Koenig, 1971; Kong and Nagata, 1995; Kong et al., 1992; Schmidt, 1983). These investigations yielded information on the rate of precursor incorporation, which relates to the intensity of synthesis, but no specific sequence could be correlated directly with the incorporation.
Regulation of rRNA transcription has been studied for more than 30 years, and a wealth of detailed molecular knowledge regarding the composition and modulation of Pol I transcription apparatus is now available. However, a systematic investigation of rRNA transcription regulation during development and other cellular processes has not been attempted. In contrast, the expression pattern and dynamics of many Pol II transcripts have been studied in the embryonic and neonatal development, as well as in adults and in diseases using in situ hybridization and microarray technology. To study the cell type–specific rRNA transcription in ocular tissues, we establish, with single cell resolution, a map of the 47S pre-rRNA level of the entire eye. This map suggests that the rRNA transcription rate varies widely in different ocular tissues. Corneal and lens epithelial cells, which are either proliferative or synthesizing protein at a high rate, have high levels of 47S pre-rRNA, rRNA processing intermediates, and mature rRNAs. Surprisingly, quiescent retinal neurons contain high levels of pre-rRNA without accumulating the mature rRNAs. These findings provide a technique and baseline for future studies of cell type–specific regulation of rRNA synthesis in ocular and other tissues.
The in situ hybridization condition was optimized to maximize signal intensity while preserving morphological details. Among the parameters examined (fixation, hybridization temperature, wash stringency, etc.), fixatives were the most critical in yielding better ocular tissue morphology as well as in situ hybridization signal. Paraformaldehyde (4%), the most commonly used fixative, caused the corneal epithelium and retina cell layers to contract (shrink) excessively (Fig. 1B,D), leading to a much weaker in situ hybridization signal in these tissues (Fig. 1F,H), whereas this fixative had a relatively minor influence on the tissue morphology and in situ hybridization signal in the corneal stroma (Fig. 1A,B) or the muscle layers posterior to the retina (Fig. 1C,D). Alcoholic formalin (65% ethanol 10% formaldehyde, by volume) yielded overall the highest signal intensity and best morphology (Fig. 1A,C), especially in the corneal epithelium and the inner nuclear layer of the retina (Fig. 1E,G as compared with F,H).
Mapping 47S pre-rRNA Level
47S pre-rRNA levels varied significantly in the mouse ocular tissues. Intense 47S pre-rRNA signals were observed in four regions: the basal layers of corneal epithelium (CE in Fig. 2), the lens transitional zone (LTZ) and the ganglion cell- (GCL) and inner nuclear layers (INL) of the retina (ret). In cornea, the highest level of 47S pre-rRNA was concentrated in the basal and immediate suprabasal epithelial cells, weak signals were detected in limbal cells (Fig. 2, lim), where corneal epithelial stem cells reside (Cotsarelis et al., 1989). In retina, the majority of the ganglion cells contained large nucleoli, which hybridized intensely with the probe (Fig. 2). In the retinal inner nuclear layer (INL), a majority of cells also contained a high level of 47S pre-rRNA, with the strongest signal in the inner (i.e., amacrine cells) and the weakest in the outer (i.e., horizontal cells) edges. The 47S rRNA level increased gradually in the lens epithelial cells as they moved toward the LTZ (Fig. 2). However, just prior to their entry into the lens and within the range of a few cells, the content of this RNA was sharply elevated and maintained as the cells transited through the LTZ (Fig. 2; see also Fig. 5I–L).
To quantify the in situ hybridization signal, we labeled the RNA probe with digoxigenin, which produced a better-defined signal area than autoradiography and allowed us to use the signal area (approximately the size of nucleoli) as a means of quantification. The signal area was shown to be highly consistent and reproducible in the same cell type in consecutive tissue sections as well as between experiments (Fig. 3A,B). The lens equator cells (LTZ) had the largest average signal area (20 pixels), followed by that of retina GCL (13.7) and the INL (6.6). Lens epithelial cells near the equator had an average of 9.8 pixels, and that in basal corneal epithelium, 8.3 pixels. This result, in conjunction with the fact that in some ocular cells the level of 47S pre-rRNA was below detection, suggests that the range of rRNA synthesis in ocular tissues is unexpectedly broad (Fig. 3C,D) (see Discussion section).
The 47S pre-rRNA level did not always correlate with various downstream rRNA processing intermediates and the mature 28S rRNA (Fig. 4). The epithelia of cornea and lens and GCL in the retina had the best correlation (Fig. 4A–E, K–O, and F–J). Curiously in the retinal INL cells, a correlation was observed with the 5′-ETS (externally transcribed spacer) and 3′-ETS probes (Fig. 4F to G), which detected early processing steps, but the correlation broke down when ITS (internally transcribed spacer) and 28S probes were used, which detect a later processing step and the final product (Fig. 4I and J). The reciprocal was seen in the photoreceptor cells, whose nuclei (in the ONL) contained very little 47S pre-rRNA, yet their protein synthesis region, the inner segment (IS) (Young, 1976), contained more ITS and mature 28S rRNA (Fig. 4F–J). The most striking breakdown of correlation was seen in the lens transition zone, where cells showed a strong presence of the 47S probe but no detectable processing intermediates, suggesting the enhanced level of pre-rRNA was a result of reduced processing. This is a surprise because lens cells in early differentiation seemed to have a higher level of protein synthesis (Piatigorsky, 1981; Ozaki et al., 1985; Lieska et al., 1992; Russell et al., 1996), yet they apparently stopped processing pre-rRNA. Also unexpected was the cytoplasmic location of some of the 3′ ETS and ITS hybridization signals, which suggest that at least some of the rRNA processing intermediates were exported from the nucleolus.
This distribution of 47S pre-rRNA suggests that ocular tissues have a wide range of rRNA transcription activities and a complex rRNA processing regulation.
Developmental Regulation of the 47S Pre-rRNA Synthesis
The rRNA transcription pattern in adult eye was the result of a highly coordinated regulation during neonatal development. In corneal epithelium, the cellular 47S pre-rRNA level increased gradually from postnatal day 3 (P3) to P13, and the number of cells containing this RNA also increased significantly during this period (Fig. 5A–C). While the 47S pre-rRNA was up-regulated dramatically in epithelial cells during neonate development, it remained virtually unchanged in corneal stroma and endothelial cells. Neonatal rRNA transcription regulation in retina GCL and INL was similar to that of the corneal epithelium (Fig. 5E–H). On P13, while numerous cells in the GCL and INL contained the 47S pre-rRNA, it was not seen in ONL (Fig. 5G). The pattern of 47S pre-rRNA level on P13 was maintained throughout the rest of neonatal development and adulthood (Figs. 2, 5D,H). Cells in the lens epithelium and the transitional zone showed distinctive 47S pre-rRNA patterns (Fig. 5I–L). In the epithelium, 47S pre-rRNA level appeared similarly regulated as that of the cornea epithelium, where transcription activity was up-regulated between P3-P13 to reach a peak prior to eye opening (Fig. 5I–K). In contrast, at birth, the 47S pre-rRNA pattern in the lens transition zone already reached a steady state, which was maintained during neonatal development and adulthood (Fig. 5I).
Age-related decline in 47S pre-rRNA synthesis was observed in lens, but not in corneal epithelium or retina GCL and INL. The 47S pre-rRNA levels were analyzed in eyes of four age groups (1, 6, 12, and 22 months, and each group consisted of 2 males and 2 females). In the 22-month-old mice, there was none apparent in corneal epithelia (Fig. 6A–D) and retina (Fig. 6E–H), despite evident thinning of corneal epithelium and disorganization of retinal tissue structure and fewer cells. In contrast, our results showed a clear absence of this pre-RNA in lens epithelium of 22-month-old eye (Fig. 6L). Furthermore, a drop of 47S pre-rRNA level in fiber cells was noted at the internal equator as early as 6 months of age, presumably a sign of maturity, not aging (Fig. 6M).
The dynamics of 47S pre-rRNA expression suggests that it is precisely regulated spatially and temporally during neonatal eye development. The expression of 47S pre-rRNA, therefore, can serve as a marker of developmental stages. Our results also indicate that when used as an indicator for rRNA synthesis rate, the level of rRNA processing intermediates should also be examined to verify that a higher level of 47S pre-rRNA is not due to reduced processing.
47S Pre-rRNA and Ocular Cell Proliferation
It has been well established that cell proliferation requires enhanced rRNA transcription, presumably because ribosomes and other cellular components are diluted in each cell division, and a high level of protein synthesis is necessary to make new cells. To verify this concept in ocular tissues and to identify other cellular processes that may require a high level of rRNA transcription, the cellular content of 47S pre-rRNA was compared with the presence of Ki67 antigen, which is a cell-cycle marker. In adult corneal and lens epithelia, the presence of Ki67 correlated well with a high level of 47S pre-rRNA (see Fig. 7A–C for corneal epithelium, and Fig. 7G–I, for lens epithelium). In the lens, however, the high level of the 47S pre-rRNA appeared as the result of the lens cells' exit from the cell cycle and a dramatic reduction in pre-rRNA processing (Fig. 4). Cells of the inner nuclear layer (INL) in retina were an example of high levels of pre-rRNA and its processing intermediates without cell proliferation. Throughout the adult life, the cells in the INL maintained a very high level of 47S pre-rRNA (Fig. 7F) but did not express any Ki67 (Fig. 7D), which is consistent with the notion that these cells would never enter the cell cycle. These retinal INL cells do proliferate in neonatal eyes (Toriyama, 1995). We thus examined the proliferative status of these cells with BrdU incorporation. An intense proliferative activity of the INL cells occurred in P2 and P3 retina (Fig. 8A and C), which quickly subsided, first in the central retina, then the periphery (Fig. 8E). By P6, very little, if any, BrdU incorporation was seen (Fig. 8G). Surprisingly, this proliferative activity was not accompanied by an increased level of 47S pre-rRNA, which was at a very low level when proliferation was at a peak on P2 (Fig. 8A and B). The 47S pre-rRNA level in INL increased belatedly on P3 (Fig. 8D) but kept its momentum despite a decline in proliferation. By P6, the adult patterns of the 47S pre-rRNA and Ki67 were established (Fig. 8G and H).
We present here a map of the content of rRNA precursor and processing intermediates in mouse ocular cells. The purposes of this map are to (1) characterize the rRNA transcription activity in each ocular tissue; (2) compare directly the rRNA transcription activity in different cell types, which relates to the dynamic range of rRNA transcription regulation; (3) correlate the levels of 47S pre-rRNA with that of rRNA processing intermediates and mature rRNAs (e.g., 28S, 18S); and (4) correlate rRNA transcription with cellular functions.
Two features of this map distinguish it from the previous studies on rRNA regulation. First, this map is constructed by employing in situ hybridization; therefore it reveals rRNA synthetic activity in vivo within single cells, which differs from most transcription-rate studies that are carried out in vitro (i.e., in cultured or isolated cells). Second, because the 47S pre-rRNA level correlates with rRNA transcription rate (Cui and Tseng, 2004), this map along with the levels of rRNA processing intermediates is an approximation of the rRNA transcription activity (rate), which differs from most in situ hybridization studies that detect the steady-state level of a mature transcript. Combined, these features make this map unique in assessing rRNA transcription rate in individual cells in vivo. Furthermore, mapping of rRNA transcription of the entire eye affords us the first opportunity to compare directly the levels of rRNA transcription in different tissues (within the same specimen), thus revealing the developmental dynamics of this essential gene.
We demonstrate that the 47S pre-rRNA level varies significantly in different ocular tissues as well as in different developmental and differentiation stages within the same ocular tissue. Using the signal area as a means of quantification, the transcription rate varies within 2–3-fold in the tissues most actively transcribing rRNA. The regulatory range, however, was much greater because in many tissue/cells, the 47S pre-rRNA level was not detectable, which suggested a very low level of rRNA transcription. Therefore, the rate of rRNA transcription between the lowest and the highest activities can be much larger (i.e., 50- to 100-fold). Such variation in the rRNA transcription rate must reflect a cellular need, which is specific to cell-type, developmental stage, and physiological conditions of a cell.
Although the rRNA transcription requirement for most cellular activities remains unknown, the requirement of a high level of rRNA transcription has been well documented in cell proliferation (e.g., in renewing tissues and cancer), cell volume expansion, and protein/ribosome accumulation and storage (e.g., during oogenesis) (Shultz, 1993). Regulation of rRNA synthesis during the corneal and lens epithelia neonatal development is consistent with the notion that proliferative cells require more rRNA. The dramatic change in rRNA transcription during the neonatal period also highlights its critical importance in eye development, as various tissues are completing their final differentiation in preparation for eye opening. Our results suggest that in ocular tissues, rRNA transcription is not significantly affected by aging. This result is in agreement with a recent report on rRNA synthesis in fibroblasts from Werner syndrome patients, whose rRNA level remained within normal range despite accelerated rDNA methylation (Machwe et al., 2000).
Interestingly, while some of our findings confirm the existing notions of rRNA transcription regulation, others cannot be fully explained by the current concepts of rRNA transcription in relation to ribosomal biogenesis. For example, at least two types of retinal neurons, the ganglion cells (GCL) and the amacrine cells (most likely also bipolar cells in INL), contain a high level of 47S pre-rRNA throughout adulthood. These cells are involved in transmitting and processing visual signals from retina to visual cortex. In adult mouse eyes, these cells are post-mitotic as shown by our Ki67 immunostaining and BrdU labeling studies (Fig. 7D), as well as by others (Kong et al., 1992; Toriyama, 1995). These cells, however, do proliferate briefly during neonatal development (Fig. 8A and C), but they do not up-regulate rRNA synthesis during the neonatal proliferative period (Fig. 8B and D). Their high level of rRNA synthesis commences after the cells become quiescent (Fig. 8G and H) and is maintained thereafter, suggesting this synthesis serves a purpose other than proliferation. None of these cells are considered to have an unusually large cytoplasmic volume or storage function. It is well known, however, that localized protein synthesis is critical for synaptic plasticity and maintenance (Sutton and Schuman, 2005; Sutton et al., 2004; Wiersma-Meems et al., 2005). Furthermore, neuronal signal transmission may require a high level of protein synthesis; however, a requirement for more ribosomes per se does not completely explain the life-long sustained high level of pre-rRNA synthesis. For example, the adjacent photoreceptor cells renew their content continuously, which requires a high level of protein synthesis. The protein synthesis rate of photoreceptor cells was shown to be four times higher than that of the ganglion cells (Hollyfield and Anderson, 1982). For this purpose, the photoreceptor cells accumulate large amounts of ribosomes in the myoid portion of the inner segment (IS) (Young, 1976). Our finding of an intense hybridization of 28S mature rRNA in the IS region (Fig. 4J) is consistent with this notion but also demonstrates that to amass and maintain this amount of ribosome does not require a detectable presence of 47S pre-rRNA (Fig. 4F, ONL), presumably due to the long half-life of ribosomes, e.g., 5–12 days in rats (Goodlad and Ma, 1975; Lifshits et al., 1976). Conversely, the high level of 47S pre-rRNA does not result in a significant accumulation of 28S rRNA in amacrine, bipolar, and ganglion cells, suggesting that an up-regulated degradation process kept its steady-state level relatively low. This implies that in these cells, the half-life of mature rRNA is very short and a corollary question is why retinal neurons turn over rRNA at a high rate?
In summary, we present here a map of rRNA transcription activity in the mouse eye. Our results confirm existing notions about the regulation of rRNA synthesis; they also suggest the presence of cellular processes, which have yet-understood rRNA transcription requirements (i.e., retinal neurons). Conceivably, techniques described in the present study can be employed to create similar maps of other tissue/organs or the entire mouse, in various developmental and pathological states. Such surveys will be of great value not only to studying rRNA transcription regulation, but also for identifying and understanding other cellular processes.
Animals and Sample Preparation
All animals were handled in accordance with the ARVO Resolution on the Use of Animals in Research. Normal adult eyes were harvested from CF1 mice purchased from Charles River Laboratories (Wilmington, MA). Neonatal mice were obtained by crossing CF1 with B6SJLF1/J (Jackson Laboratories, Bar Harbor, ME). For the aging experiment, 1-, 6-, 12-, and 22-month-old BALB/C mice were purchased from the National Institute on Aging (NIA, Bethesda, MD). Mice were housed in a climate-controlled facility with a 12-hr dark/light cycle. After CO2 asphyxiation, eyes were removed and immersed immediately in 4% paraformaldehyde/PBS (PFA/PBS) or alcoholic formalin (65% ethanol and 10% formaldehyde by volume) for 16 hr. Five-micrometer paraffin sections were mounted on X-tra slides (Surgipath Medical Industries Inc., Richmond, IL).
In Situ Hybridization
The templates for the sense and antisense riboprobes were prepared by PCR from mouse genomic DNA (CF1) with one of the primers containing a T7 promoter (5′-GGGTAATAGGACTCA CTATAGGGCGA). The primers for the 47S leader region were: forward, 5′-GCCTGTCACTTTCCTCCCTG; reverse, 5′-GCCGAAATAAGGTGGCCCTC; for the 5′ETS: forward, 5′TGTCTTGCCCCGCGTGTAAG; reverse, 5′-CGCTTACAAGAAACAGCGCG; for the ITS1: forward, 5′-GTGGAGCGAGGTGTCTGGAG;reverse, 5′-AACGCGACAGCTAGGTACCC; for the 28S rRNA: forward, 5′-GGAAACTCTGGTGGAGGTCC; reverse, 5′-CCTTAGCGGATTCCGACTTC; and for the 3′ETS: forward, 5′-CCCGAGGACGGTTCGTTTCTCTTTC; reverse, 5′-TCCACACCGCGACATCCTCC. PCR conditions were: 5 min at 94°C; 35 cycles (94°C for 1 min, 60°C for 1 min, and 72°C for 1 min) and at 72°C for 7 min. The PCR products were purified and checked by direct sequencing. Digoxigenin (DIG)-labeled probes were generated by in vitro transcription from the above PCR templates using DIG RNA labeling Kit (Roche Diagnostics, Indianapolis, IN). The reaction product was treated by DNaseI, purified on a ProbeQuant G-50 Micro Column (Amersham Biosciences, Arlington Heights, IL), precipitated in ethanol and suspended in DEPC-H2O. The riboprobe yield was estimated by gel electrophoresis and spot test.
In situ hybridization was performed as previously described (Cui and Tseng, 2004), except that both prehybridization and the hybridization was at 62°C. After washing, the hybridization signal was visualized using an alkaline phosphatase-conjugated antidigoxigenin antibody (Roche Applied Science, Indianapolis, IN) and nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate (NBT/BCIP). Consecutive sections were hybridized with sense riboprobes as controls.
Quantification of the in situ hybridization was based on measuring the size of the hybridization signal area by hand or by the Analyze Particles function of ImageJ (1.32j), a software package available through NIH (W. Rasband, NIH, http://rsb.info.nih.fov/ij). Digital image files of microphotographs (200×), at a resolution of 300 dpi, were cropped in Photoshop CS to obtain the region of interest and imported to ImageJ. The threshold was set by equating the mask area to the hybridization signal and the mask area was then measured (with the minimal area set at one pixel and Bins set at a value of 20) to obtain the number of particles (cells) analyzed, the total signal area, the average signal area, as well as the area distribution. For each tissue type, three local areas in a photograph were analyzed to obtain an average and several different experiments were analyzed to assure reproducibility. We did not use signal intensity to quantify because it was saturated under the color-development condition employed. The quantification was, therefore, only an approximation.
Ki67 was detected in paraffin-embedded eye sections. Sections were deparaffinized, rehydrated, blocked with 1% BSA in PBS, and incubated with the antibody (NCL-L-Ki67-MM1; Novocastra Laboratories Ltd, Newcastle upon Tyne NE12 8EW, UK) at a dilution of 1:50, for 1 hr at room temperature. Control slides were incubated without antibody. After washing, the primary antibody was reacted with a biotin-conjugated secondary antibody at room temperature for 1 hr. The biotin-labeled antibody was visualized by a Cy3-conjugated streptavidin (1:450; Jackson ImmunoResearch Laboratories, Inc., West Grove, PA). Some immunostained tissues were counterstained with DAPI.
BrdU Labeling of Ocular Tissues
Mice were injected intraperitoneally with BrdU at a dose of 100 μg/g body weight and euthanized 90 min later. Eyes were processed for 5-μ paraffin sections. BrdU detection was performed according to the manufacturer's instruction (5-Bromo-2′-deoxy-uridine Labeling and Detection Kit II AP; Roche Applied Science, Indianapolis, IN), except that sections were heated in 0.01 M citrate buffer (pH 6.0) by microwave for 10 min to unmask the antigen before the antibody incubation.
The authors thank the reviewers for their insightful suggestions and Noga Vardi for comments. This work was supported by grants from NIH (EY13637 to H.T., EY06769 to R.M.L.).