The 100 colours of the Brainbow

Researchers from Harvard University have developed a remarkable genetic technique that enabled them to visualize complete neuronal circuits in unprecedented detail, by using multiple distinct colours to label individual neurons.

The technique, called Brainbow, works in much the same way as a television uses the three primary colours to generate all the colour hues. With multiple combinations of up to four differently coloured fluorescent proteins, a palette of approximately 100 labels has been produced. 

brainbow hippocampus

Hippocampus. Tamily Weissman, Harvard University

To develop Brainbow, the researchers used the Cre/loxP site-specific recombination system, a sophisticated method which is commonly used to generate mutant (or “knockout“) mice lacking a specific gene.

The cre gene encodes a viral enzyme that recognizes a specific 34-base pair DNA sequence called loxP, which is usually present in pairs that are located close together on the chromosome. When a Cre molecule binds to it’s target sequence, one of two things can happen. The length of DNA in between the two loxP sites can be snipped out, inverted and then re-inserted into the chromosome, or removed altogether. In both of these cases, the DNA in the construct has been snipped and ‘recombined’, hence, the process is called recombination.

The Cre/loxP system can be used to disrupt the coding sequence of a given gene, and mutant mice lacking that gene in a specific part of the body can be created. The method can be used, for example, to create mice that have a receptor missing from the brain, or from a specified region of the brain.The mice can even be engineered so that the gene of interest is only disrupted at a specified time during embryonic development.


The method is complex and time-consuming – it involves breeding at least two different strains of mutant mice created from embryonic stem cells transfected with specially designed DNA constructs. One strain is derived from stem cells containing the cre gene under the control of a regulatory DNA sequence (the promoter) which drives expression in a specific tissue or cell type. The other is derived from cells in which the gene of interest has been replaced with a construct containing the same gene flanked by loxP sites. The two mouse strains are then mated, and recombination occurs in some of the offspring but not others.

The Harvard team generated various constructs, with the simplest one (shown above) consisting of the genes encoding the red and cyan fluorescent proteins (RFP and M-CFP, respectively) flanked by loxP sequences. Note that the sequence of the M-CFP gene is the wrong way around. So, because DNA is translated into a protein sequence in one direction only, a cell containing this construct will express RFP but not M-CFP.

Cre recombination inverts the DNA sequence, so that the M-CFP gene is facing the right way. Cells containing the recombined construct will therefore express the M-CFP gene. But the recombination event occurs randomly, so that some cells will express RFP and emit a red fluorescence, while others will express M-CFP and emit a cyan fluorescence.


Cerebral cortex. Tamily Weissman, Harvard University

Another construct contained the genes encoding green, red, yellow and cyan fluorescent proteins, with two invertable sequences (as in the construct above) arranged in tandem. With this construct there are five possible recombination events because of the presence of two additional loxP sites. Three different inversions and two different excisions (iv and v), and a choice of inversion and/or excision, can lead to expression of any of one of the four genes.

The beautiful results shown here were obtained because multiple copies of the constructs can be integrated into a stem cell chromosome, thus enabling the four genes to be expressed in multiple, random combinations. If, for example, a mouse was generated from stem cells containing 3 copies of a construct that had all recombined differently, the neurons would be labelled with a mixture of at least 10 different colours. Transfecting the same with a combination of constructs would therefore generate more colour combinations.

Using confocal microscopy to generate three-dimensional reconstructions, this multi-colour palette was used trace complete neural circuitry in various regions of the brain, such as the the dentate gyrus of the hippocampus (above) and the inner granule layer of the cerebellum (below). In some cases, different kinds of glial cells were also labelled.


Cerebellum. Tamily Weissman, Harvard University

Thus, established techniques have been used to develop a system which has been used to label large numbers of neurons, as well as the connections formed by them. Refinements to the technique will enable the researchers to add even more colours to the palette, and to somehow limit the distribution of each colour to specific populations of cells or regions of the brain.

Brainbow has the potential to be an extremely powerful tool for researchers. Developmental neurobiologists may, for example, soon have at their disposal a method with which they can label distinct populations of neural progenitor cells with different colours. This would enable them to trace the fates of the progeny of each population, and to gain a better understanding of how a small number of identical cells gives rise to the extraordinarily complex mature brain.

Reference: Livet, J. et al. (2007). Transgenic strategies for combinatorial expression of fluorescent proteins in the nervous system. Nature 450, 56-62. [Abstract]


4 thoughts on “The 100 colours of the Brainbow

  1. That, and the photos are frickin’ beautiful. Every neuroscientist in the world is going to be dying to have this technique so they can put these photos up on a screen during their talks. It’s going to make the MRI guys jealous.

Comments are closed.