Optogenetic fMRI

OF all the techniques used by neuroscientists, none has captured the imagination of the general public more than functional magnetic resonance imaging (fMRI). The technique, which is also referred to as functional neuroimaging and, more commonly, "brain scanning", enables us to peer into the human brain non-invasively, to observe its workings and correlate specific thought processes or stimuli to activity in particular regions. fMRI data affect the way in which people perceive scientific results: colourful images of the brain have persuasive power, making the accompanying data seem more credible.

Functional neuroimaging is used widely by researchers, too, with tens of thousands of research papers describing fMRI studies being published in the past decade. Yet, a big question mark has been hanging over the validity of the technique for over a year and, furthermore, the way in which fMRI data are interpreted has also been called into question. Using a novel combination of fMRI and a recently developed state-of-the-art technique called optogenetics, researchers now provide the first direct evidence that the fMRI signal is a valid measure of brain activity.

fMRI measures brain activity indirectly, using signals generated by the flow of blood around the brain. It is based on the assumption that increased blood flow to a particular region of the brain is related to activity in that region, because the cells within it require a supply of oxygen to generate nervous impulses. But several recent studies challenged this assumption. The most important of these, by Yevgeniy Sirotin and Aniruddha Das of Columbia University, showed that the brain pre-empts itself by increasing the flow of blood to regions that might become active in the near future, but that this anticipatory blood flow is not always needed. This prompted the question of whether or not fMRI actually measures what researchers have always claimed it measures.

The new study, published online in the journal Nature, finally answers this question, by confirming that the fMRI signal is indeed closely correlated with increased brain activity. Senior author Karl Deisseroth  published one of the first papers describing the optogenetics technique, back in 2005. That paper showed that light-sensitive proteins called channelrhodopsins, which had been isolated from algae several years earlier, can make the neurons sensitive to light when shuttled into them. As a result, light pulses of the appropriate can be used to control the activity of individual, specified cells, on a millisecond timescale. The technique was quickly applied to controlling simple behaviours in small organisms such as nematode worms and fruit flies; more recently, it has been used to successfully control complex functions, such as reward behaviors and memory, in mammals; it has even been used to restore vision in blind mice.  

This time, Deisseroth and his colleagues injected a virus carrying the channelrhodopsin gene into the primary motor cortex of mice. The gene was engineered so that it would be expressed specifically in the large pyramidal cells that send axons down into the spinal cord and control movement. The mice were then placed into a custom-made MRI-compatible cradle fitted with a stereotaxic frame to keep their heads still. With the animals inside the scanner, the researchers delivered light pulses to the virus injection site, using an optical fibre inserted through a hole in the skull. In this way, they could activate specified subsets of cells in the motor cortex and simultaneously monitor the fMRI signals generated by them.

optogenetic fMRI.JPG The researchers observed fMRI signals (above right) in the motor cortex 3-6 seconds after delivering pulses of light to the motor cortex. The signals originated in the area into which the light pulses were delivered (indicated above by the asterisk) and spread away from it, lasting for about 20 seconds before returning to a baseline level. By contrast, no fMRI response was seen in control mice injected with a salt water solution instead of the ChR2 gene (left). Activity in a part of the brain called the thalamus, which receives connections from the motor cortex, was also observed. This downstream activity was initially weaker than that seen in the cortex, but after a 5-second delay showed a very similar pattern, and reliably followed the cortical activity by several hundredths of a second.

The activity pattern observed in the thalamus again demonstrates that fMRI can accurately measure the activity of groups of neurons. It also shows that the combination of optogenetics and fMRI can be used to investigate how activity in one region of the brain alters activity in distant regions via long-range connections - the 5-second delay is consistent with network activity that modulates the output of the cortex and activity in the thalamus. By introducing ChR2 into the thalamus and performing simultaneous optical stimulation and neuroimaging, the researchers also revealed hitherto unknown details about the pathways connecting the thalamus and motor cortex. 

The thalamus is thought of as a 'relay station', which receives sensory information en route to the cortex, and provides the cortex with feedback. The feedback pathways are mostly ipsilateral, that is, they project to the sensory cortical areas on the same side of the brain. The thalamus also sends and receives information to and from the motor cortex, but the former pathways are thought to project to motor areas on both sides of the brain, because control movement involves co-ordination between the two sides of the body. The experiments confirmed this, by showing that optical stimulation of one side of the thalamus evoked activity in the motor cortical areas on both sides.

Previous attempts to investigate the relationship between blood flow and neuronal activity have involved using microelectrodes to stimulate small groups of cells whilst simultaneously performing functional neuroimaging. This method has limitations, because electrical stimulation activates not only those cells targeted by the electrodes, but also distant cells whose axons traverse the stimulated brain region. In these neurons, the electrodes elicit nervous impulses that travel backwards along the axon, causing the cell body to generate further impulses. The resulting activity can therefore potentially confound the fMRI signal, because it is unrelated to that of the targeted cells.

Optogenetic fMRI largely overcomes this problem. Optical stimulation will activate ChR2-expressing axons in the targeted area, and the resultant nervous impulses will be back-propagated. So although some unwanted activity might be observed, it will be related to the cells of interest, because only those cells will be sensitive to light. As well as confirming the validity of fMRI data, this initial description of optogenetic fMRI shows that it is a very powerful technique for investigating neuronal activity at both the local and the global level. The new study does not, however, rule out the possibility that fMRI also detects other unrelated signals.  Many questions still remain about exactly how the fMRI signal is generated but, using optogenetics, researchers may soon begin to answer them.


Lee, J., et al (2010). Global and local fMRI signals driven by neurons defined optogenetically by type and wiring. Nature DOI: 10.1038/nature09108. [PDF]

Boyden, E., et al. (2005). Millisecond-timescale, genetically targeted optical control of neuronal activity. Nat. Neurosci. 8: 1263-1268. [PDF]

McCabe, D. P. & Castel, A. D. (2008). Seeing is believing: The effect of brain images on judgments of scientific reasoning. Cognition: 107: 343-52. [PDF


More like this

What disappointed me, however, is that Dr. Bernardine Healy, former Director of the NIH, was a guest and that she didn't slap Maher down hard for his idiotic statements about vaccines and Pasteur.

Ok....I'm just wondering if it is just me or if there is a more widespread thought that subjecting the brain to these kinds of bombardments may have long term effects that have not been sufficiently considered. You know...there are people now who can see the change in energy coloration without the use of machines? Perhaps giving these gifted ones the opportunity to be useful...testing to see if they see what the machine sees...would provide interesting data. Just as those who are on the aspergers to autistic spectrum are able to read the unspoken thoughts of others...and could be used to assist in situations where the patient is unable to speak/communicate....those with other gifts could also be utilized to diagnose and heal.

Do you actually believe that there are people who "can see the change in energy coloration"? And are you really expecting a serious answer?

This is very nice work, and an excellent technique, but no, no, no, no, no it does not show that the fMRI BOLD signal can always be used as a surrogate for neuronal activity (and I appreciate that you are not saying that it can be, but you do say it is âconfirming the validity of fMRI dataâ, it isn't). It is only sometimes that the fMRI BOLD signal corresponds to neuronal activity. That âsometimesâ may be 99% of the time, but it is still only sometimes. Sirotin and Das showed that quite clearly.


Pretending that a technique is more reliable than it actually is may falsely inflate the purported value of the research results; it actually decreases their value. Results with fMRI BOLD need to be understood to be not primary measures of neuronal activation. Even when a volume element appears to be highly activated by fMRI (or by any technique), only a small minority of the neurons in that volume element are actually activated simultaneously.

This is a very large issue in my work on basal nitric oxide. The vasodilatation observed in fMRI is produced by NO, by neurogenic NO that causes local vasodilatation. What the BOLD fMRI technique is showing, is where the local NO level is high enough to activate sGC, form cGMP and produce vasodilatation.

Changing the basal NO level changes the amount of neurogenic NO that is needed to raise the local NO level to the threshold that activates sGC and so causes vasodilatation. There is no threshold for changes in basal NO levels to change the volumes activated because differences in NO are already used for regulating the volumes activated.

There are many researchers who don't want to appreciate that the fMRI BOLD technique does not actually measure neuronal activity because it complicates the analysis of their data. I have no sympathy for such researchers. If you are going to use a technique, you need to know and understand the limitations of that technique. It is not being a âpuristâ to demand rigor in how techniques are used and described.

Although the study clarifies previous studies' results on the validity of fMRI by removing the artifacts of electrode stimulation (while creating a cool new method for use in animal work), you're implying that this validates the BOLD signal for all applications of fMRI.

I'd thought that for certain areas fMRI has not been found to have a good correlation with neural firing and that this correlation can switch between tasks. For instance, fMRI is often used to look at the temporal lobe for memory and learning studies. However, fMRI may not be related to actual firing there, but rather to phase-locking to certain brain waves that may correspond better with general dendritic integration rather than actual action potential firing. This phase-locking might vary if the task is a memory encoding task, a memory retrieval task, or a navigation task.

There's also thought that this could be because areas like the motor, visual, and auditory cortices exhibit population coding while the hippocampus and other temporal lobe regions may exhibit sparse coding.

The paper ends with the statement that ofMRI a tool to further probe the controversy - not a solution to the question of whether the BOLD signal is valid.

It's a very interesting study, but I don't see how it validates fMRI as a technique. This study shows: if there's neural activity, then there's an fMRI signal. Most research is based on the assumption: if there's an fMRI signal, then there's neural activity. These are different statements; one may be true and one may be false.

fMRI is valid as a technique to look at vasodilatation.

Vasodilatation is mediated through neurogenic NO. Neuronal activity is mediated through propagation of action potentials.

The coupling between those two things is not one-for-one.

Well..is that true that fMRI signals based on elevated levels of oxygenated blood in specific parts of the brain are caused by an increase in the excitation of specific kinds of brain cells?!?