Superresolution imaging of transcription units on newt lampbrush chromosomes
We have examined transcription loops on lampbrush chromosomes of the newt Notophthalmus by superresolution microscopy. Because of the favorable, essentially two-dimensional morphology of these loops, an average optical resolution in the x–y plane of about 50 nm was achieved. We analyzed the distribution of the multifunctional RNA-binding protein CELF1 on specific loops. CELF1 distribution is consistent with a model in which individual transcripts are tightly folded and hence closely packed against the loop axis.
KeywordsRNA polymerase II CELF1 oocyte localization microscopy
Charge coupled device camera
CUG-BP1 and ETR-3 Like Factor 1
CUG Binding Protein 1
Embryonic lethal abnormal vision Type RNA-binding protein 3
digital optical reconstruction microscopy
Ground state depletion imaging microscopy
- Pol II
RNA polymerase 2
Spectral position determination microscopy
single stranded RNA
The physical organization of chromatin at intermediate levels of resolution is poorly understood—that is, between the level of the nucleosome and what can be resolved by conventional light microscopy. Because of the extreme level of chromatin compaction in mitotic chromosomes and interphase nuclei, giant chromosomes with their more extended organization have provided much useful information, particularly the polytene chromosomes of Diptera and the lampbrush chromosomes (LBCs) of various vertebrates. LBCs are the largest known chromosomes, reaching lengths of up to 1 mm in oocytes of some amphibians. Their major structural features have been known for a long time and are well summarized in Callan’s classic monograph (Callan 1986). In brief, the longitudinal axis of each chromosome consists of multiple regions of condensed inactive chromatin, the chromomeres, from which paired loops of active chromatin extend laterally (Gaginskaya et al. 2009; Gall et al. 2004; Morgan 2002). Each loop consists of one or more transcription units (TUs) readily visible by conventional light microscopy (reviewed in Gall et al. 1983). LBCs can be centrifuged onto a standard microscope slide or coverslip, which reduces their complex three-dimensional organization to two dimensions.
To gain new insight into the organization of the TUs, we used spectral position determination microscopy (SPDM) (Lemmer et al. 2008). SPDM is a method of localization microscopy (Betzig et al. 2006; Hess et al. 2006; Rust et al. 2006) that uses standard fluorophores; it is also termed direct optical reconstruction microscopy (Heilemann et al. 2008) or ground-state depletion imaging microscopy (Fölling et al. 2008). SPDM is based on the precise determination of the position of each individually detected fluorescent molecule. From this information, an image of the underlying biological structure can be reconstructed with an effective resolution far below the diffraction limit of light. Localization microscopy and other superresolution methods, in combination with fluorescent immunostaining and in situ hybridization, have already given new insights into several aspects of nuclear structure and gene activity, including details of the nuclear envelope, the organization of chromatin domains, and the relationship between transcription and chromatin structure (Bohn et al. 2010; Markaki et al. 2010; Schermelleh et al. 2008; Tykocinski et al. 2010)
For the analysis shown below, we made use of both the high structural resolution (~50 nm) and the single-molecule information provided by the detection of individual fluorophores. This unique combination, which is currently provided only by localization microscopy, allowed us to gain insight into the folding of mRNA on LBC loops with a previously unmatched level of detail.
Materials and methods
Setup for SPDM
Superresolution microscopy measurements were performed with a wide-field fluorescence microscopy setup according to Lemmer et al. (2008). This technique detects single molecules using a light-induced, long-lived dark state of the fluorophore (Dickson et al. 1997; McAnaney et al. 2005; Peterman et al. 1999; Schuster et al. 2005; Sinnecker et al. 2005). Illumination of the sample with an excitation intensity in the 10 kW/cm² range pushes the fluorophores into a long-lived dark state within a few seconds. The stochastic recovery of the molecules to the fluorescent state (followed by their transition to another long-lived dark state) is then used for their detection.
For image acquisition, we used diode-pumped solid-state lasers with wavelengths of 488 and 568 nm (Sapphire HP 488 and Sapphire 568, Coherent, Dieburg, Germany). No additional laser was needed for the switching/bleaching of the molecules. Fluorescent light was detected with a 1.4 NA oil-immersion objective (HCX PL APO, 63X, Leica, Wetzlar, Germany) and a high quantum efficiency CCD camera (SensiCam QE, PCO Imaging, Kehlheim, Germany). Bandpass filters (XF3003 520DF40, Laser Components GmbH, Olching, Germany; F37-609, AHF, Analysentechnik, Tübingen, Germany) were used to select the emitted fluorescence.
For the present experiments, data stacks consisting of several thousand images were recorded with an exposure time of 150 ms each. Measurements for the two different channels were performed sequentially. Determination of the fluorophore positions was performed as described previously (Gunkel et al. 2009; Kaufmann et al. 2009; Lemmer et al. 2008). The chromatic shift between the images from the two different channels was globally corrected via cross-correlation. Local shifts were considerably smaller than the achieved structural resolution and therefore have been neglected in further analyses.
Adult female newts, Notophthalmus viridescens, were purchased from Charles D. Sullivan, Nashville TN and maintained in aquaria at ~8 °C with weekly feeding. Ovaries were removed and LBC preparations made as described (Gall and Wu 2010) with the following modifications. LBCs were centrifuged onto 22 mm coverslips instead of microscope slides. They were stained with two antibodies: mouse mAb 3B1 (IgG) against CELF1 (CUG-BP1) from Abcam and mouse mAb H14 (IgM) against RNA polymerase II phosphorylated on serine 5 of the C-terminal domain, a gift from Jeffry Corden (Patturajan et al. 1998). After immunostaining, the coverslips were mounted on standard 3″ × 1″ microscope slides with Prolong Gold antifade reagent (Invitrogen, Grand Island, NY). TetraSpeck Fluorescent Microspheres (0.5 μm; Invitrogen) were included for microscope calibration.
Distance analysis of CELF1 molecules
Superresolution microscopy of CELF1 and polymerase II
The results reported here extend observations originally made by Morgan on LBCs of the axolotl, Ambystoma mexicanum (Morgan 2007). Morgan analyzed the distribution of the multifunctional RNA-binding protein CELF1 (Barreau et al. 2006) on specific LBC loops, showing three features clearly. First, CELF1 protein is strongly concentrated on a small subset of transcription loops. Second, it is limited to a restricted region of the TUs on which it occurs. And third, within that region, it has a characteristic distribution: proximally (toward the start of the TU) it is closely associated with pol II on the loop axis, whereas distally (away from the start) it becomes more and more displaced from the axis. Morgan interpreted these features as follows. Because CELF1 is an RNA-binding protein, it binds to the nascent transcript when pol II transcribes a specific binding sequence, and it remains attached to this sequence as the nascent transcript elongates. Consequently, the binding sequence and its attached CELF1 become progressively displaced laterally from the loop axis (Fig. 1; Morgan’s Fig. 6).
Distance analysis of CELF1 molecules along the DNA strand
Fang et al. (2011) describe a model for the folding of single-stranded RNA (ssRNA) based on the formation of double-stranded duplexes by contiguous base pairs. According to their simulations, the lengths of a folded ssRNA molecule and its fully extended chain are related by a power law. Based on their analysis, we fitted model functions of the type c + Nα to the experimental data, where N is the length of the chain. The yellow curves in the diagrams in Fig. 4 represent the best fit to the data. The curve shapes in Fig. 4a, b are very similar. The simulations of Fang et al. yielded a value of ~0.7, whereas the fit to our experimental data gives a value closer to ~0.5. This corresponds to a secondary structure of the mRNA that is more condensed relative to the ssRNA than predicted by the theoretical simulation.
Several features make LBCs particularly favorable objects for the study of chromosome organization. First is their extraordinary size, which results from extreme decondensation at the molecular level. Second, they are easy to isolate freehand from the giant germinal vesicle, thereby physically separating them from each other and from other nuclear components. Third, they are transcriptionally active, permitting analysis not only of general chromosome structure but also of individual TUs. Finally, they can be centrifuged onto a microscope slide or coverslip, thus reducing the three-dimensional complexity of the intact nucleus to a two-dimensional problem. These features have been used to their advantage here and in numerous earlier studies.
Our superresolution analysis of CELF1 staining provides quantitative data about the TUs in LBC loops. The generally accepted model of LBC structure assumes that each loop represents one or more TUs. In the simplest case, a loop corresponds to a single TU with transcription starting at or near the “thin” end of the loop and terminating at the “thick” end. This model is strongly supported by electron microscopic images of GV contents prepared as “Miller spreads.” Such images show closely spaced nascent transcripts extending laterally from a linear axis (Hamkalo and Miller 1973; Miller and Hamkalo 1972). The width of a LBC loop as viewed by phase contrast or confocal microscopy, however, is very much less than the length of the attached nascent transcripts seen in these electron micrographs. Furthermore, the increase in width from thin to thick end is not proportional to the distance from the thin end. These two features are reflected in the data for CELF1 staining shown in Fig. 4a, b.
We presume that folding of the nascent RNA chains is the major factor responsible for the small width of a loop relative to the length of its attached RNA fibers. In addition to intrinsic folding of the RNA molecule, additional compaction probably results from association of RNA binding proteins with the nascent transcripts. An indication of this tighter folding is the smaller value of our measured α compared with the value provided by the theoretical simulations of Fang et al. (2011). In their model, folding of the ssRNA depends only on the development of local double strands. Splicing is a third factor that could influence loop morphology. The splicing machinery definitely associates with nascent LBC loops (Patel and Bellini 2008; Patel et al. 2007), but it is not known whether spliced introns are removed from the loops during transcription or after release of the transcripts into the nucleoplasm. Early release of introns would presumably influence the width of a loop, preventing the width from increasing in proportion to the distance from transcription initiation.
Superresolution images should help to clarify additional aspects of loop structure, especially if combined with a variety of molecular markers. Particularly useful will be analysis of additional splicing components, the exon-junction complex, and hnRNP proteins that may be involved in targeting transcripts to specific pathways or locations in the cell.
CC and RK thank the Frontier Project of Heidelberg University and the Deutsche Forschungsgemeinschaft (DFG) for support. Research reported in this publication was supported by the National Institute of General Medical Sciences of the National Institutes of Health under award number R01GM33397 to JGG. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. JGG is an American Cancer Society Professor of Developmental Genetics.
- Gall JG, Diaz MO, Stephenson EC, Mahon KA (1983) The transcription unit of lampbrush chromosomes. In: Subtelny S, Kafatos F (eds) Gene structure and regulation in development, vol. 41. Alan R. Liss, New York, pp 137–146, Symp Soc Dev BiolGoogle Scholar
- Kaufmann R, Lemmer P, Gunkel M et al (2009) SPDM: single molecule superresolution of cellular nanostructures. Proc Soc Photo-Opt Instr Eng 7185:1–19Google Scholar