Visualization of single RNA transcripts in situ

Science

Volumn 280, Issue 5363, 1998, Pages 585-590

  Femino, Andrea M ^a   Fay, Fredric S ^b   Fogarty, Kevin ^b   Singer, Robert H ^a  

^a ALBERT EINSTEIN COLLEGE OF MEDICINE   (United States)

^b UNIVERSITY OF MASSACHUSETTS MEDICAL SCHOOL   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

BETA ACTIN; MESSENGER RNA;

ARTICLE; BINDING SITE; DIGITAL FILTERING; DNA PROBE; FLUORESCENCE IN SITU HYBRIDIZATION; INFRARED RADIATION; PRIORITY JOURNAL; RNA ANALYSIS; RNA PROCESSING; TRANSCRIPTION REGULATION;

ACTINS; ANIMALS; CELL LINE; FLUORESCEIN-5-ISOTHIOCYANATE; IN SITU HYBRIDIZATION, FLUORESCENCE; KINETICS; OLIGONUCLEOTIDE PROBES; RATS; RNA PROCESSING, POST-TRANSCRIPTIONAL; RNA, MESSENGER; TRANSCRIPTION, GENETIC;

EID: 0032562720     PISSN: 00368075     EISSN: None     Source Type: Journal    
DOI: 10.1126/science.280.5363.585     Document Type: Article

References (33)

1. J. B. Lawrence, R. H. Singer, L. M. Marselle, Cell 57, 493 (1989); C. L. Sundell and R. H. Singer, Science 253, 1275 (1991); G. Zhang, K. L. Taneja, R. H. Singer, M. R. Green, Nature 372, 809 (1994); R. M. Long et al., Science 277, 383 (1997).
2. J. B. Lawrence, R. H. Singer, L. M. Marselle, Cell 57, 493 (1989); C. L. Sundell and R. H. Singer, Science 253, 1275 (1991); G. Zhang, K. L. Taneja, R. H. Singer, M. R. Green, Nature 372, 809 (1994); R. M. Long et al., Science 277, 383 (1997).
3. J. B. Lawrence, R. H. Singer, L. M. Marselle, Cell 57, 493 (1989); C. L. Sundell and R. H. Singer, Science 253, 1275 (1991); G. Zhang, K. L. Taneja, R. H. Singer, M. R. Green, Nature 372, 809 (1994); R. M. Long et al., Science 277, 383 (1997).
4. J. B. Lawrence, R. H. Singer, L. M. Marselle, Cell 57, 493 (1989); C. L. Sundell and R. H. Singer, Science 253, 1275 (1991); G. Zhang, K. L. Taneja, R. H. Singer, M. R. Green, Nature 372, 809 (1994); R. M. Long et al., Science 277, 383 (1997).
5. P. K. Elder, L. J. Schmidt, T. Ono, M. J. Getz, Proc. Natl. Acad. Sci. U.S.A. 81, 7476 (1984).
6. M. E. Greenberg, A. L. Hermanowski, E. B. Ziff, Mol. Cell. Biol. 6, 1050 (1986).
7. J. B. Lawrence and R. H. Singer, Nucleic Acids Res. 13, 1777 (1985); R. J. Schwartz and K. N. Rothblum, Biochemistry 20, 412 (1981).
8. J. B. Lawrence and R. H. Singer, Nucleic Acids Res. 13, 1777 (1985); R. J. Schwartz and K. N. Rothblum, Biochemistry 20, 412 (1981).
9. C. S. Thummel, Science 255, 39 (1992); A. W. Shermoen and P. Z. O'Farrell, Cell 67, 303 (1991).
10. C. S. Thummel, Science 255, 39 (1992); A. W. Shermoen and P. Z. O'Farrell, Cell 67, 303 (1991).
11. A. M. Femino, F. S. Fay, K. Fogarty, R. H. Singer, data not shown.
12. R. Ashfield et al., EMBO J. 13, 5656 (1994); R. Ashfield, P. Enriquez-Harris, N. Proudfoot, ibid. 10, 4197 (1991).
13. R. Ashfield et al., EMBO J. 13, 5656 (1994); R. Ashfield, P. Enriquez-Harris, N. Proudfoot, ibid. 10, 4197 (1991).
14. Y. N. Osheim, O. L. Miller, Jr., A. L. Beyer, Cell 43, 143 (1985).
15. J.-C. Dantonel, K. G. K. Murthy, J. M. Manley, L. Tora, Nature 389, 399 (1997).
16. D. K. Hawley and R. G. Roeder, J. Biol. Chem. 262, 3452 (1987).
17. S. Huang and D. L. Spector, Genes Dev. 5, 2288 (1991).
18. R. W. Dirks, K. C. Daniel, A. K. Raap, J. Cell Sci. 108, 2565 (1995); Y. Xing, C. V. Johnson, P. T. Moen Jr., J. A. McNeil, J. Lawrence, J. Cell. Biol. 131, 1635 (1995).
19. R. W. Dirks, K. C. Daniel, A. K. Raap, J. Cell Sci. 108, 2565 (1995); Y. Xing, C. V. Johnson, P. T. Moen Jr., J. A. McNeil, J. Lawrence, J. Cell. Biol. 131, 1635 (1995).
20. For details of the digital imaging microscopy procedures, see F. S. Fay, K. E. Fogarty, J. M. Coggins, Soc. Gen. Physiol. Ser. 40, 51 (1986); F. S. Fay, W. A. Carrington, K. E. Fogarty, J. Microsc. 153, 133 (1989). Images were obtained with an inverted Nikon Diaphot 200 epifluorescence microscope equipped with a 100-W mercury lamp and modified to capture images under computer control. Between 20 to 35 optical sections were acquired at 250-nm z-intervals and at an effective pixel size of 187 nm by 187 nm with a thermoelectrically cooled (-45.O°C) charge-coupled device (CCD) (model 220; back-thinned RCACCD chip; 50 kHz; Photometrics, Tucson, AZ). A high quantum efficiency (0.8 at 500 nm) and a low noise (50 photons equivalent) were required for imaging at the low light levels.
21. For details of the digital imaging microscopy procedures, see F. S. Fay, K. E. Fogarty, J. M. Coggins, Soc. Gen. Physiol. Ser. 40, 51 (1986); F. S. Fay, W. A. Carrington, K. E. Fogarty, J. Microsc. 153, 133 (1989). Images were obtained with an inverted Nikon Diaphot 200 epifluorescence microscope equipped with a 100-W mercury lamp and modified to capture images under computer control. Between 20 to 35 optical sections were acquired at 250-nm z-intervals and at an effective pixel size of 187 nm by 187 nm with a thermoelectrically cooled (-45.O°C) charge-coupled device (CCD) (model 220; back-thinned RCACCD chip; 50 kHz; Photometrics, Tucson, AZ). A high quantum efficiency (0.8 at 500 nm) and a low noise (50 photons equivalent) were required for imaging at the low light levels.
22. Oligonucleotide probes were synthesized, purified, and labeled as described [E. H. Kislauskis, Z. Li, R. H. Singer, K. L. Taneja, J. Cell Biol. 123, 165 (1993)] (23). Below are listed the probes used in this work (nucleotide number start, length in nucleotides). The sequence is for β-actin unless specified (accession J00691). For Fig. 1, B and D through H, the antisense probes to the 3′-UTR were 3133, 51; 3312, 52; 3434, 53; 3488, 52; and 3542, 52. For Fig. 2, same as Fig. 1 and the antisense probes to the γ-actin (accession X52815) 3′-UTR were 1151, 50; 1282, 61; 1409, 52; 1470, 54; and 1535, 56. For Fig. 2C, same as Fig. 1 and the antisense probes to the splice junctions were (exon nucleotides only) 284, 52; 1340, 52; 1667, 52; 2567, 56; and 2839, 53. For Fig. 2D, the antisense probe to the 5′-UTR was 255, 50; to the coding region, 2839, 53 and 3034, 52; and to the 3′-UTR, 3542, 52. For Fig. 3A, the sense probes to the introns were 630, 63; 998, 63; 1069, 63; 1369, 63; 1691, 61; 1889, 64; 1960, 63; 2040, 63; 2590, 63; and 2864, 62. For Fig. 3B the antisense probe to the 5′-UTR was 225, 53; to the splice junctions, the same as Fig. 2; and to the 3′-UTR, the same as Fig. 1. For Fig. 4, A through F, the antisense probe to the 5′-UTR was 225, 53 and the antisense probes to the 3′-UTR were 3133, 51; 3312, 52; 3434, 53; 3488, 52; and 3542, 52. For Fig. 4I, the antisense probe to the 5′-UTR was 255, 50, and the antisense probes to the 3′-UTR were 3133, 51 and 3542, 52.
23. 3]. The pixel size was calibrated for the optical setup of the microscope (x = y = 93 or 187 nm with a ×60 magnification, 1.4 numerical aperture objective, and ×5 or ×2.5 camera eyepiece, respectively). The TFI is dependent on the light flux at the sample; therefore, a calibration curve was calculated for each imaging session. The slope is equal to the TFI per fluorochrome under imaging conditions (CY3: m = 65 ± 1.3, R = 0.992; FITC: m = 20.6 ± 0.8, R = 0.985; CY5: m = 56.3 ± 1.4; correlation coefficient R = 0.992). Exposure times for CY3, FITC, and CY5 were 3, 4, and 15 s, respectively. (iv) The TFI value per probe obtained from solution was adjusted to that for a restored image and multiplied by a factor (50) equal to the number of optical sections in the point spread function (PSF) used for deconvolution of the experimental image. The TFI for one probe molecule is calculated and normalized to that for a restored image (CY3, 16250 ± 975; FITC, 5150 ± 600; and CY5, 14075 ± 1050). For exhaustive photon reassignment (EPR), the sampled PSF defines the depth (number of optical sections) over which light is integrated by the deconvolution process.
24. Fluorescent probes diluted in mounting media (2 pg/ μl) were allowed to adsorb to the surface of a cover slip and then optically sectioned. The restored images revealed the distribution of intensities attributed to individual probes. The empirical measurements of the TFI for one probe either in solution or immobilized on glass cover slips were compared with a theoretical estimate of the number of photons emitted by one dye molecule by using the molar extinction coefficient, the quantum yield, the measured light flux of the microscope, and the quantum efficiency of the CCD camera. This estimate came within 80% of the actual measurement obtained.
25. 4) after serum stimulation, washed, and stored in PBS. The FISH protocol was modified from K. L. Taneja and R. H. Singer [J. Cell. Biochem. 44, 241 (1990)]. Cells were hybridized for 3.5 hours at 37°C. Cover slips were washed and mounted on slides with phenylenediamine in 90% glycerol with PBS. Multicolored (FITC, tetramethyl rhodamine isothiocyanate), 0.099-μm-diameter latex beads (Molecular Probes) were included in the mounting media and used as fiduciary markers to align images.
26. -6 and a convergence value of 0.001. The procedure has been detailed previously [W. A. Carrington et al., Science 268, 1483 (1995); K. L. Taneja, L. M. Lifshitz, F. S. Fay, R. H. Singer, J. Cell Biol. 119, 1245 (1992)].
27. -6 and a convergence value of 0.001. The procedure has been detailed previously [W. A. Carrington et al., Science 268, 1483 (1995); K. L. Taneja, L. M. Lifshitz, F. S. Fay, R. H. Singer, J. Cell Biol. 119, 1245 (1992)].
28. Background in a restored image comprises point sources of low-level autofluorescence, light scattering, and random noise contributed by the imaging electronics. A threshold value was established and applied to the restored images to eliminate these low-level point sources of light by analyzing the background TFI frequency distribution of nonzero pixels in a cell subjected to a mock hybridization. The mean TFI per nonzero voxel +3 SD was chosen as the threshold value; all pixels less than the threshold value were set to zero, whereas those greater were retained at their original value. This eliminated >95% of the tow-TFI background pixels. The threshold value was <10% of that predicted for one probe molecule (five CY3 fluorochromes) and when applied to the experimental image did not adversely affect the value from one hybridized probe. Brighter background pixels that eluded the thresholding filter but whose TFI value was less than that for one hybridized probe were excluded during the mapping analysis. The result was a filtered image where one hybridized CY3 labeled probe could be distinguished.
29. Data Analysis and Visualization Environment (DAVE); copyright 1995 by Lawrence M. Lifshitz and the University of Massachusetts Medical School.
30. The reassignment of light to a point source in a restoration results in one or two brighter voxels at the location of the point source in addition to a number of associated contiguous voxels. A point source is thus described as an object comprising a group of contiguous nonzero voxels in a restored image after thresholding. All the nuclear and cytoplasmic point sources therefore had a finite size and were treated as three-dimensional objects.
31. Additional evidence that every discrete signal is a single RNA molecule is that, when each of the five probes was hybridized individually, analysis showed that >95% of the detected sites had one probe hybridized. When two probes to the 3′-UTR were used, a population of single and double probes was detected (6). As an additional control, a set of five probes to an mRNA not expressed in these cells (α-actin) gave about 20 single probe signals per cell (2% of the β-actin single probe signal), but no multiple probe signals when processed by identical methods.
32. Supplementary material is available at www. sciencemag.org/feature/data/975399.shl. General information is available at www.singerlab.aecom.yu. edu.
33. We thank members of the Biomedical Imaging Facility, L. M. Lifshitz and J. Collins for their assistance with the DAVE software, R. A. Tuft for carrying out the light-flux measurements of the microscope, Y.-L. Wang for NRK cells, and K. Taneja for oligonucleotide probe synthesis. Supported by NIH grant GM 54887 to R.H.S. We are deeply saddened by the loss of our colleague and close friend Fredric S. Fay during the preparation of this manuscript.