High-Throughput DNA FISH (hiFISH)

Abstract

High-throughput DNA fluorescence in situ hybridization (hiFISH) combines multicolor combinatorial DNA FISH staining with automated image acquisition and analysis to visualize and localize tens to hundreds of genomic loci in up to millions of cells. hiFISH can be used to measure physical distances between pairs of genomic loci, radial distances from genomic loci to the nuclear edge or center, and distances between genomic loci and nuclear structures defined by protein or RNA markers. The resulting large datasets of 3D spatial distances can be used to study cellular heterogeneity in genome architecture and the molecular mechanisms underlying this phenomenon in a variety of cellular systems. In this chapter we provide detailed protocols for hiFISH to measure distances between genomic loci, including all steps involved in DNA FISH probe design and preparation, cell culture, DNA FISH staining in 384-well imaging plates, automated image acquisition and analysis, and, finally, statistical analysis.

Keywords: 3D genome; DNA FISH; High-throughput imaging; Nuclear architecture.

Fig. 1). FISH probe preparation and cell culture workflows.

A) Flow chart for the preparation of BAC or Fosmid FISH probes. The image of the agarose gel represents the quality control step to ensure the success of the nick-translation labelling reaction. A successful reaction should appear as a smear between 100 bp and 1,000 bp on the gel (Green star symbols). An unsuccessful reaction due to failed digestion will run as the original full-size BAC or Fosmid (Red star symbol). An unsuccessful reaction due to excessive digestion will run as a small smear at very short lengths (Blue star symbols). B) Flow chart for the oligoFISH probe amplification reactions. The picture of the agarose gel represents the quality control/yield step obtained by loading a small sample volume (2 – 5 μL) of the completed reaction. A successful pooled library of oligos should run as a single band at approximately 150 – 200 bp. C) Flow chart for cell culture in 384-well imaging and fixation. As shown in the image, users should use a brightfield microscope to check that the confluency at the time of fixation is approximately 70–80%.

Fig. 2). FISH Staining workflow.

A) Flow chart for FISH staining of PFA-fixed 384-well plates. B) 3D maximally projected images of ~70% – 80% confluent healthy human primary fibroblasts successfully stained with the FISH protocol. CH02, CH03 and CH04 images represent the channels assigned to the FISH probes. As expected, since they are diploid, most cells contain two FISH signals in each channel. Scale bar: 10 μm. C) Same as B), except that in this case the cells are either too confluent or clumped while they were seeded, making it difficult or impossible to automatically segment nuclei with automated image analysis pipelines (See Note 14). This clumping also causes cell stress, resulting in the enlarged/fused nucleoli visible in DAPI staining and cellular debris visible in all channels. D) Same as B), except that in this case the FISH probes used in the experiment did not efficiently bind to their genomic targets, and/or the FISH staining conditions were not properly optimized (See Notes 13, 27 and 28).

Fig. 3). Image and statistical analysis workflows.

A) Flow chart for the automated image analysis pipeline used to identify single cells and localize FISH signals in X, Y and Z (Optional). The pseudocolored nucleus mask image shows a representative output of the automated nuclear segmentation step. The yellow box indicates an oversegmentation error. The cyan box represents a merge error. The teal box represents segmentation of a nucleus in mitosis. All these segmentation events need to be filtered out from the final analysis by applying a combination of filters set based on morphology, fluorescence intensity, and/or fluorescence texture features calculated from the DAPI channel. Scale bar: 20 μm. The FISH/Spot Localization images show detection of FISH signals (Pseudocolored circles), as output by the image analysis software. The white masks represent the border of the nucleus masks where the spots are detected. The question mark represents questionable or missed FISH signal detections, which are particularly common in situations when the FISH channel images show low fluorescence signal to background ratio (S/B). Scale bar: 15 μm. B) Flow chart for the automated statistical analysis pipeline. The histogram represents the distribution of the number of detected FISH spots signals per cell in one channel for diploid human primary fibroblasts for a successful FISH probe experiment (bottom), or a failed FISH experiment (top), respectively. The two pictograms represent the nucleus in gray, the detected FISH spot centers as pseudocolored circles (A, B and C for FISH probes in different colors), and all the Euclidean distances between FISH spots in different channels as dashed colored lines, for the whole or minimum sets of distances per nucleus, respectively. The boxplot represents the single chromosome level distribution of distances between different FISH probes labelled with two different colors (Pink), or the positive co-localization control where cells are FISH stained with a mix of the same probe labelled with two different colors (Teal).