What do thoughts look like? A fireworks show of neurons in action


The human brain is composed of 86 billion cells1. To put that number in perspective, 86 billion is still more than the current population of the earth times 10 (7 billion x 10= 70 billion). Proposed by the Spanish anatomist Santiago Ramon y Cajal 100 years ago, the neuron theory is the fundamental idea that neurons are the discrete functional units of the nervous system2. Despite the gargantuan numbers, neurons somehow work together to orchestrate day-to-day events, from coordinating muscle movement to analyzing sensory information such as pain. We may see nothing special in these daily routines, but behind the scenes, neurons are hard at work communicating within complex networks. Understanding how neurons talk to each other has been an area of active research for about 60 years. While we are still decades away from a complete understanding of how the brain works, recent developments in neural recording techniques have provided scientists with next generation tools to elucidate mysteries of the nervous system.

A massive mystery begs for a massive effort. Rafael Yuste, a neuroscientist from Columbia University, proposed a large-scale public effort to understand fundamental neural networking. Dubbing the ambitious effort the Brain Activity Map project, Yuste hopes to unite scientists from varying disciplines in order to elucidate the mystery of brain function3.

The Obama administration was quick in its move to support neuroscience research. “Now, as humans, we can identify galaxies light years away. We can study particles smaller than an atom, but we still haven’t unlocked the mystery of the three pounds of matter that sits between our ears,” said Obama in a speech unveiling the BRAIN (Brain Research through Advancing Innovative Neurotechnologies) Initiative earlier this year4. Beginning in fiscal year 2014, the BRAIN Initiative will give 100 million dollars to the National Institute of Health (NIH), Defense Advanced Research Projects Agency (DARPA), and the National Science Foundation (NSF). Through research funding, the ambitious effort aims to foster development in technologies that will allow scientists to image complex neural circuits in the brain. To crack a modern neuroscience puzzle, researchers need newer and better tools to observe live neurons in action.

Before a research idea is carried out in humans, it must first demonstrate efficacy in animal models, due to practical and ethical reasons. The mouse is one of the most well-known animals in scientific research, owing to its striking similarity to humans, including a complex nervous system. Current trans-cranial imaging techniques can be used to look at neurons in an intact mouse5. Since the brain is protected by the skull, the bone must be shaved into a thin transparent layer. If the neurons are labeled with a fluorescent protein, they can be imaged in real time by a powerful two-photon microscope. With its skull still intact, the mouse returns to its cage and recovers from the procedure for subsequent neural recordings. While this method allows for high-resolution imaging of brain cells, the microscope cannot penetrate past the outside layer of the brain. In reality, only about 0.1% of the brain can be viewed. There are numerous other neurons deep inside the brain that the microscope cannot reach.

The humble zebrafish emerges as the great hero of brain research. Size proves not to matter, as the the tiny zebrafish is the key to unlocking a vast body of knowledge about the nervous system. For the first few days of life, the entire zebrafish is entirely transparent. Due to this developmental quirk, the zebrafish does not face the mouse’s problem: powerful fluorescent microscopes can easily penetrate through the zebrafish’s optically clear tissues. Thus, the whole brain can be imaged, as opposed to a tiny fraction from the mouse.

Additionally, the zebrafish nervous system has two major advantages over those of mice and invertebrate models. First, the zebrafish has a more complex system than invertebrate organisms such as C. Elegans. Second, the zebrafish has fewer neurons than the mouse, making it much easier to characterize neuronal loops. As a vertebrate animal, the zebrafish has numerous genes that are evolutionarily conserved across many species, including humans. Therefore, parallels can be drawn between zebrafish and humans, making zebrafish an important tool for studying neural circuits and development.

Figure 1: A) A zebrafish embryo at about 30 hours post fertilization. Image taken from Biocircuits Institute. B) c-Fos labeling of neuronal activity in sections of 5-day old zebrafish brain. Axons are labeled with green fluorescence, while red fluorescence labels C-fos protein, a marker of brain activity. Image taken from Lauderdale lab.

Neurons send messages to each other through synaptic transmission. When the nerve impulse (called action potential) enters the terminal of the axon, a neurotransmitter is released. When the neurotransmitter binds to the receptor on the receiving neuron, the neuron becomes activated. In 1987, it was discovered that c-Fos protein is expressed by spinal cord neurons as a response to stimulation from peripheral afferent neurons, suggesting the possibility of using c-Fos as a marker for mapping neuronal activity6. Interestingly, c-Fos is a proto-oncogene that is implicated in cervical cancer invasiveness7 c-Fos protein levels in the neuron can be detected by immunohistochemistry staining, or c-Fos mRNA transcripts can be by in-situ hybridization (an antisense can bind to RNA through complimentary base pairing). Activated neurons will show c-Fos labeling, allowing scientists to partially trace neural pathways. As a result, scientists have utilized c-Fos to study neural activation in various cognitive functions such as learning and memory8.

There are limitations with using c-Fos for tracing neuronal activities. Neuronal signaling happens in milliseconds, whereas c-Fos levels cannot be detected until 30 minutes have elapsed since the initial neuronal activation. In addition, the animal must be euthanized and sectioned before viewing. The c-Fos method is not ideal for real-time recording of neurons, and lacks sufficient temporal resolution to precisely delineate neural circuitry that occurs in a blink of an eye. Using c-Fos to take a snapshot of brain activity is like looking at photographs of a soccer game that happened 30 minutes ago. Like sport aficionados who would rather watch live soccer games, neuroscientists would prefer to watch neuronal action in real-time.

Intracellular recording techniques do not face the same temporal challenges as c-fos labeling. In 1939, scientists inserted a microelectrode inside a squid axon to study neuronal action potentials9. This technique allowed for accurate reading of the voltages generated by neurons in real time. However, the technique was invasive and could be used for only single neurons. Considering the fact that neurons work together in large populations, observing an individual cell in a complex circuit is comparable to looking at a few pixels on an HDTV10.

Multielectrode arrays can record activities of a large population of neurons at one time. Building on previous single microelectrode technology, multielectrode arrays involve the use of up to 10,000 electrodes for in-vitro recordings and over a hundred in in-vivo recordings11. Other techniques to study large populations of neurons are electroencephalography (EEG) and functional magnetic resonance imaging (fMRI). In an EEG, electrodes are placed all over the scalp, picking up electrical activities of numerous neurons in the superficial layer of the brain. fMRI maps out neuronal activation by measuring changes in cerebral blood flow, as blood flow correlates to neuronal activity. The ability to simultaneously map out activities of neurons on a large scale, however, comes with a sacrifice: single-cell resolution was not possible. Without individual-neuron specificity, intricate neural loops requiring activation of precise neurons remain elusive.

In the field of neuroimaging, there is an inverse relationship between the number of neurons that can be captured and how clear the image is: would I rather have a blurry picture of a bunch of neurons, or a crisp picture of only one neuron?

GCamp is the new school answer to an old school problem. Belonging to a family of genetically encoded calcium indicators, GCamp is used to indicate fluctuations of intracellular calcium ions as a consequence of neuronal activation. First developed by a group of Japanese researchers in 2001, the design of GCamp involves a circularly permuted fluorescent protein (cpFP), calcium-binding calmodulin (CaM), and M13 fragment of the myosin light chain kinase12. The C-terminus of cpFP is attached to CaM, while the N-terminus of cpFP binds to M13. Normally, the fluorescent protein will emit a photo signal following excitation by a fluorescent microscope. However, when calcium ions bind to CaM, the cpFP changes structure, emitting a much higher intensity fluorescence. Since calcium ions enter neurons immediately following activation, brain activity can be imaged as the neurons fire (real-time), as opposed to previous methods using c-fos that have delay periods. In addition, GCamp allows for imaging activities of many different neurons at once without sacrificing resolution.


Figure 2: Structure of GCamp. The binding of calcium ion to CaM facilitates a conformational change that increases fluorescence intensity. Image taken from Knopfel, 2013.

Taking advantage of the zebrafish’s optical transparency, GCamp can be introduced into zebrafish embryos and larvae for real-time imaging of neuronal activity, without killing the animal. Several notable studies have utilized GCamp for answering questions pertaining to how the brain perceives the outer world. A team of Japanese researchers approached this mystery by studying how a zebrafish thinks as it watches its food, the paramecium. The researchers genetically engineered a zebrafish to make it express GCamp in the optic tectum, or the region of the brain that processes visual information. Thus, as an influx of calcium moves inside an activated neuron, GCamp fluoresces to label active neurons. To look at how the brain behaves as a response to visual stimulus, a 3.5 inch LCD screen was placed next to the zebrafish. The LCD screen displayed a circular spot that can be turned on/off and moved around to different locations. The researchers watched in real-time as neuronal activity rippled across the brain in response to the movement of the spot on the screen. In another experiment, a live paramecium was substituted for the screen. As the zebrafish watched its prey move around, its thought was represented by flashes of fluorescent signals from the GCamp. When the paramecium moved from left to right, a wave of brain activity in the zebrafish moved in the opposite direction13. The videos recorded by the researchers show what thoughts look like: different regions light up as dots of fluorescence zipped across the brain. You can certainly get a kick out of watching these videos of zebrafish brain activity, as they resemble a fireworks show more than anything!

Figure 3: A) and B) Brain activity of larval zebrafish (5-day-old) captured by GCamp. Images taken from Muto et al, 2013 and Ahrens et al, 2013

The Japanese researchers were not the only ones to make use of GCamp. Other groups have jumped on the GCamp bandwagon. Another report published in Nature Methods also utilized the power of GCamp to look at zebrafish brain activity. By putting GCamp under the control of a pan-neuronal promoter, 80% of all neurons in the zebrafish brain could be seen. Most of all, resolution was not sacrificed, as activity could be traced down to a single neuron, making this a way to record activities of numerous neurons at single-cell resolution. Every 1.39 second, a snapshot was taken and then the pictures were put together in a cohesive video. From this method, the scientists identified two populations of neurons with correlated activity patterns14.

These studies involving GCamp have managed to jump over a major hurdle: large amounts of neurons could be visualized without sacrificing resolution, thanks to the optical transparency of the young zebrafish. While GCamp empowers scientists with the ability to look at brain activity patterns in zebrafish embryos and larvae, can we take a step further to image adult brain? As the zebrafish matures, it loses its optical transparency due to development of pigment cells. The adult zebrafish is no longer see-through, making it challenging to image nervous system in real-time. However, the Casper zebrafish developed five years ago may alleviate this problem. Named after Casper the Friendly Ghost for its ghost-like appearance, the Casper zebrafish has mutations in two pigmentation genes that makes it almost see-through15 Without the pigmentation, it may be easier penetrate through the outside tissue, paving the way for the next generation of technology to study the adult brain.



Figure 4: The Casper zebrafish has mutations in two pigmentation genes, nacre and roy. Image taken from White et al., 2008.

GCamp technology goes further than simply allowing us to visualize neuronal activity (even though it is fun to watch neurons fire). The brain can be thought of as an information processing center. Input (visual information about the paramecium) from the sensory neurons feed into the brain. After this information is processed, the brain orchestrates an output (moving towards the paramecium) as a response to the input. As we watch the zebrafish think through GCamp, we begin to understand the relationship between patterns of brain activity and how the brain manages input/output traffic. What is the underlying process in the brain that governs function, from simpler tasks such as visual processing to more intricate processes such as making complex decisions? How does structure of neural circuits govern their function? How do molecules and development determine the structure? Answering these questions will give us a fundamental understanding of how the brain works. From this solid groundwork, we can apply our knowledge to mental health conditions, in which the brain does not work correctly. The cure to these rising disorders such as Alzheimer’s and Parkinson’s rests upon correcting the malfunctioning brain, returning it to the normal functioning state. GCamp technology is humanity’s next step towards figuring out how the brain works, which still remains the greatest mystery in neuroscience research.


  1. Azevedo, Carvalho, Grinberg, Farfel, Ferreti, & Herculano-Houzel. (2009). Equal numbers of neuronal and nonneuronal cells make the human brain an isometrically scaled-up primate brain. Journal of Comparative Neurology []
  2. Lopez-Munos, Boya, & Alamo. (2006). Neuron theory, the cornerstone of neuroscience, on the centenary of the Nobel Prize award to Santiago Ramón y Cajal. Brain Research Bulletin. []
  3. Alivisatos, Chun, Church, Greenspan, Roukes, & Yuste. (2013). The Brain Activity Map Project and the Challenge of Functional Connectomics. Neuron. []
  4. President Obama Calls For A BRAIN Initiative. (2013, April 5). []
  5. Grutzender, J., Yang, G., Pan, F., Parkhurst, C., & Gan, W. (2011). Transcranial two-photon imaging of the living mouse brain. Cold Spring Harbor Protocols. []
  6. Hunt, Pini, & Evan. (1987). Induction of c-Fos-like protein in spinal cord neurons following sensory stimulation. Nature. []
  7. Cheung, Leung, Chung, Lam, To, & Wong. (1997). c-fos overexpression is associated with the pathogenesis of invasive cervical cancer. Gynecologic and Obstetric Investigation. []
  8. Herrera, & Robertson. (1996). Activation of c-fos in the brain. Progress in Neurobiology. []
  9. Hodgkin, & Huxley. (1939). Action potentials recorded from inside a nerve fibre. Nature. []
  10. Alivisatos, Chun, Church, Greenspan, Roukes, & Yuste. (2013). The Brain Activity Map Project and the Challenge of Functional Connectomics. Neuron. []
  11. Spira, M., & Hai, A. (2013). Multi-electrode array technologies for neuroscience and cardiology. Nature Nanotechnology. []
  12. Nakai, J., Ohkura, M., & Imoto, K. (2001). A high signal-to-noise Ca2+ probe composed of a single green fluorescent protein. Nature Biotechnology. []
  13. Muto, Ohkura, Abe, Nakai, & Kawakami. (2013). Real-Time Visualization of Neuronal Activity during Perception. Cell. []
  14. Ahrens, Orger, Robson, Li, & Keller. (2013). Whole-brain functional imaging at cellular resolution using light-sheet microscopy. Nature Methods. []
  15. White, R., Sessa, A., Burke, C., Bowman, T., & Leblanc, J. (2008). Transparent adult zebrafish as a tool for in vivo transplantation analysis. Cell Stem Cell. []


  1. Wrong. Human Brain = 100 BILLION neurons. Cells, many many more – you’re forgetting the important glia – astrocytes, microglia, oligodendrocytes and Schwann cells. GET YOUR FACTS RIGHT!!!

    • Yes you are right. There are countless support cells in the CNS, but for simplicity’s sake in the introduction, I did not choose to specify. I’m a neuroscientist. I can’t forget about the glia.

      Thank you for the comment,
      Duy Phan

    • and in fact if you closely look at the paper I cited (Azevedo et al. 2009), the number of neuronal cells are actually quite similar to the number of non-neuronal cells.


Please enter your comment!
Please enter your name here