Remember Current Protocols in Molecular Biology? Those red binder-encased manuals, as common in biology labs as coffee cups and whiteboards, detail everything from nucleic acid extraction and basic cell culture to DNA microarrays and metabolomics.
Current Protocols is divided into chapters called “units.” In Unit 4.9, the instructions for the Northern blot—one of the first methods for studying gene expression—can be found. The RNA equivalent of a DNA Southern blot, a Northern blot uses labeled probes to detect the expression of a particular sequence in gel-fractionated RNA. The recommended amount of RNA per lane is 0.5 to 10 µg—the equivalent of 50,000 to 1 million cells.
Northern blots produce a population profile, often interpreted as a transcriptional description of an “average” cell. But in reality, says Daniel Larson, head of the Systems Biology of Gene Expression Section at the National Cancer Institute, “There is no average cell. Every cell is different.”
Where one cell may have 100 copies of a given RNA transcript, its immediate neighbor may have only 1 or 2, even in theoretically homogeneous populations such as cultured yeast or mammalian cells—a fact that often is overlooked in population-based analyses. The problem is further compounded in tissues, tumors, and other clearly heterogeneous samples.
It's not just steady-state RNA abundances that differ; the kinetics of gene expression itself also vary from gene to gene and cell to cell. Gauging such variability requires examining one cell at a time and reconsidering some assumptions about gene expression— a considerable technical and computational challenge for today's researchers.Fishing for transcripts
Single-cell transcription studies fall largely into two categories: imaging and sequencing.
The oldest imaging-based method is a variation of fluorescence in situ hybridization (FISH), a fixed-cell technique that uses fluorescently labeled oligonucleotide probes to detect complementary sequences, in either DNA or RNA, at the single-cell level. FISH hails from the 1980s, but its use at the single-transcript level only dates back to 1998, when Robert Singer of Albert Einstein College of Medicine and his colleagues used sets of 5 oligos, each at least 50 nucleotides long and studded with 5 fluorophores apiece, to resolve individual β-actin transcripts in rat kidney cells via epifluorescence microscopy.
“It was, I think, the first demonstration of single-molecule detection in a cell,” Singer says.
Single-molecule FISH works by effectively reducing each transcript to a single diffraction-limited volume in the cell: the smallest resolvable fluorescent point. By then counting those spots and tracking their appearance over time, Singer's team and others have slowly chipped away at the question of how genes are regulated. Their results imply a level of randomness in gene expression that few researchers, if any, could have imagined.
The traditional view of gene regulation is that transcription regulatory complexes assemble on gene regulatory regions, either activating or repressing RNA polymerase activity. That understanding is codified in the pretty diagrams with looping arrows so often found in transcriptional studies, all of which imply an orderly progression of transcriptional events. It's a foundational tenet of molecular methods such as footprinting, which are predicated on the idea that these complexes are stable.
But FISH tells a different story. “Many of these events that occur are what we've come to realize as ‘stochastic’—that is, they're random. When a gene gets transcribed by a polymerase is subject to the laws of diffusion,” explains Singer.
It turns out that transcription is not a smooth, continuous process; instead, it often occurs in bursts separated by extended periods of inactivity, meaning that apparent differences in transcript abundance from cell to cell might simply be due to when the researcher happens to look.
In 2008, Singer's team used FISH to explore the transcriptional properties of “housekeeping genes,” workaday genes that are always on and whose abundance, theoretically, should never change. Housekeeping genes are routinely used as controls in gene expression studies. Singer's team measured the abundance of transcripts from three essential genes, MDN1, KAP104, and DOA1, from yeast that were believed to harbor exactly one copy of each RNA per cell. The average number was actually between 3 and 6 transcripts per cell, with the absolute number varying from 1 to 15 copies. A small fraction of cells—8% in the case of DOA1—contained no copies of the transcript at all.