Menu

Close

Functional MRI of large scale activity in behaving mice

Madalena S. Fonseca, Mattia G. Bergomi, Zachary F. Mainen, Noam Shemesh

Preprint posted on April 18, 2020 https://www.biorxiv.org/content/10.1101/2020.04.16.044941v2

Into the brain: Exploring the big picture of complex neural activity in awake mice.

Selected by Mariana De Niz

Background

Adaptive behaviour requires dynamic interactions between neurons and neural circuits across many brain regions. Access to study neuronal circuits in rodents has provided important insights into how local neuronal activity relates to behaviour. While studies focusing on specific regions of the brain at any one time have given us key insights into the function in such regions and the connections involves, a brain-wide view is key to understand how interactions betweenregions gives rise to behaviour, for example, how interactions between the motor cortex, the basal ganglia, the cerebellum, and the thalamus result in motor learning and execution. A brain-wide approach would also help in guiding, in an unbiased way, the selection of brain regions to be studied at higher detail. In their preprint, Fonseca et al (1) present a novel method based on functional magnetic resonance imaging (fMRI), to study interconnections between brain areas to answer outstanding and previously inaccessible behavioural and neurological questions. Moreover, this technique allows for longitudinal and non-invasive studies, and can be used in rodent models and in humans, bridging findings across species.

 

Key findings and developments

Tool development

The authors used fibre optics, pressure sensors, and 3D printed parts to generate an fMRI setup that includes a head-fixation system, multiple sensory delivery systems for olfactory and visual stimuli (input), and three separate behavioural measures (licking and right and left lever press) (output) (Figure 1).

Figure 1. Schematic of the scanner and coil used for MRI imaging. ​Schematic of the behavioural setup, including a head-fixation system, a water delivery tube, a detector for licking behaviour, a detector for right (red) and left (green) forepaw lever presses, an odour delivery tube and two optic fibres delivering blue (left) and yellow (right) light stimuli. (Adapted from Ref1).

 

Biological findings

            Using this setup, mice were trained in a classical conditioning task, whereby they learned to associate different odours (conditioned stimuli), with biologically relevant (unconditioned stimuli) outcomes, such as the presence or absence of water rewards. This was chosen to allow identifying brain activity as related to odour identity and reward association. An identified challenge, and possibly a confounder, is the stress experienced by mice undergoing MRI. To minimize stress, the authors implemented 4 steps: 1) to train the mice to voluntarily enter the head-fixation apparatus; 2) gradual exposure to MRI sounds only once the mice were proficient in the tasks, as stress is known to influence learning; 3) maintaining sounds at the highest level in the study for at least 7 days to allow for acclimation; 4) habituating the mice to the vibrations of the scanner. While measuring brain-wide activity by recording blood-oxygenation-level-dependent fMRI signals across the brain, the authors detected that despite minimal brain movement, large amplitude changes in brain signals were temporally coupled with lick events -which require jaw and tongue movements. The authors aimed to correct these artefacts, not at the behavioural level, but at the analysis level. They therefore analysed information present in the muscle tissue to predict and remove the artefacts observed in the brain, and found muscle-based pre-processing was effective at correcting the artefacts.

Key findings in this study using the fMRI approach were, first, confirmation that there are distinct spatial activity patterns for different odours in the olfactory bulb –odourant receptors are arrayed into a spatial map of glomeruli, such that each odour activates a unique spatial pattern of activity. Second, the mapping of both motor and reward correlates in the odour guided classical conditioning task. Activity was detected during the odour period, in the primary olfactory (piriform) cortex, while in the outcome period, multiple cortical and subcortical areas were engaged. This included motor-related regions as well as various areas involved in memory , autonomic regulation, and aversion and appetite responses.

Third, in a separate task, they explored the voluntary preparatory action (lever pressing), unlinked to the reflexive action of licking. This offers the possibility to study the readout for cognitive phenomena such as decision-making, and the neural correlates in this process. Press-related brain activity showed sequential activation of the anterior cingulate area, and the primary motor cortex, followed by the secondary motor cortex, the dorsal striatum and the upper limb region of the somatosensory cortex (SSp(UL)). These were finally followed by activity in the ipsilateral SSp(UL).

Fourth, they explored neural correlates of preparatory (lever pressing) and consummatory behaviour (licking). They found that right lever presses preferentially engaged the upper region of the somatosensory cortex in the contralateral hemisphere, while licking correlated best with the mouth region, and was more bilateral. Lever-pressing and licking also recruited various cortical and subcortical regions. Because light was used as stimulus in the task, the visual cortex and the superior colliculus were also engaged.

Decoding approaches further allowed investigating task-related activity. This included comparison of valid and invalid responses and their relation to reward contingencies. The areas identified as discriminating between the two conditions, were regions previously implicated in associative learning, decision-making, motivation, action planning, as well as initiation of consummatory actions, reward processing, and memory. Among their findings, the authors propose for the first time, that the piriform cortex may play a role in non-olfactory associative processes.

Behaviour depends not only on intra-area local processing but critically on the interaction and coordination of information processing across multiple areas. The authors studied functional coupling between brain regions upon performance of the different tasks. Among the interesting findings was the identification of the retrosplenial (RSP) cortex as being a key active area during lever pressing. The RSP is at the intersection of areas that encode visual information, motor feedback, higher-order decision-making, and hippocampal formation. The findings in this work are consistent with suggestions that the RSP may specifically contribute to cognitive processing, by integrating visual and motivational cues, with information generated by self-motion (2).

Altogether, the authors explored fMRI to gain information about functional coupling and activation across brain regions during behaviour. They envisage that this information can be used to define areas to explore using complementary invasive circuit techniques, and propose other possible types of analysis can be performed with fMRI beyond the one explored here, including how functional coupling changes over time, and how it relates to functional connectivity.

 

What I like about this preprint

            I like the rationale of using a technique that allows bridging findings across species – this is explored and discussed as a key point in this paper. Equally, I like work that has in mind the big picture. I think conventionally, research tends to be very specific, and although this of course has led to a lot of discoveries, it has its shortcomings.  I see as a positive thing that we are entering a different time, where the interconnectivity of systems (be this systems, circuits, organs, host-pathogen interactions, etc) is seen as equally important to understand. And I think since we started heading in this direction, we have discovered things we couldn’t have imagined before. I like this paper because a) it goes in this direction of exploring a full organ during the performance of specific tasks, and because b) the authors developed the instrument/setup that allows such research. Tool development is what often precedes the questions we can ask, and this will be key for the advances we can achieve.

 

Open questions

  1. I think you did a great job addressing the fact that when an animal is stressed, this affects its ability to learn and perform tasks – you addressed this by helping the animals acclimatize gradually to the setup. Is there any way in which in the future, this setup can be further optimized to allow measurements in a free-moving animal?
  2. Are there studies in humans already, of a similar nature? fMRI has been used in humans already. Do you know if and to what level your findings correlate with the behavior and cognition in the human brain?
  3. You investigated various complex processes beyond motor activity and reward association, namely, decision-making, motivation, and memory. If you can correlate this to findings in humans, many diseases affecting the central nervous system could be studied in rodent models. Are you interested in any specific ones for which there are rodent models?
  4. Beyond studies in the brain, fMRI can be used to study for instance, the spinal cord. Do you envisage further developing your method to allow investigating activity beyond the brain? And therefore covering diseases that affect both the central and peripheral nervous system.
  5. Considering the global picture, what is the main limitation you found in the work you performed (be this the setup, or the findings), and how would you address it/them in the future?

 

References

  1. Fonseca MS, Bergomi MG, Mainen ZF, Shmesh N, Functional MRI of large scale activity in behaving mice, bioRxiv, 2020
  2. Cooper BG, and Mizumori SJY, Temporary inactivation of the retrosplenial cortex causes a transient reorganization of spatial coding in the hippocampus, J Neurosci, 21(11), 2001

 

Posted on: 18th May 2020

doi: https://doi.org/10.1242/prelights.20673

Read preprint (No Ratings Yet)




  • Author's response

    Madalena Fonseca shared

    Open questions 

    1. I think you did a great job addressing the fact that when an animal is stressed, this affects its ability to learn and perform tasks – you addressed this by helping the animals acclimatize gradually to the setup. Is there any way in which in the future, this setup can be further optimized to allow measurements in a freely-moving animal?

    With other techniques (e.g. microscopy), miniaturized equipment has been mounted on the head of the animals to achieve freely moving behaviour. Unfortunately, to produce the magnetic fields that we need for MRI, we need to use large machines, therefore it is currently impossible to achieve these measurements in freely moving animals. Inside the MRI scanner, we further need the head to be steadily positioned to achieve reliable imaging, this is why we need animals to be head-fixed. However, while the brain needs to be stable, the setup can certainly be optimized to allow further body mobility. For instance, we envision adding a treadmill to the setup. This would allow mice to run and further interact with the setup, while allowing us to further quantify their behaviour and correlate it with the neural signals.

    2.Are there studies in humans already, of a similar nature? fMRI has been used in humans already. Do you know if and to what level your findings correlate with the behavior and cognition in the human brain?

    There is a very extensive literature in humans using fMRI to measure brain activity during a variety of behavioural tasks. While the tasks that we used in this study are not explicitly designed as analogues to tasks in humans, they do engage processes that have been studied in both species (e.g. forelimb movements, olfactory and visual processing, reward processing) and, in general, many of the areas that we identified in our study are consistent with studies in humans. However, more explicit and fine-grain comparisons will require the explicit use of comparable tasks in both species. There are already some human tasks with rodent analogues (and vice-versa). However, it has not been easy to compare their neural correlates as rodents and human studies typically measure brain activity at drastically different scales (whole-brain in humans vs. single region in rodents) and record different kinds of signals (hemodynamic signals in humans vs. invasive measures of neural activity in rodents). This is one of the main reasons we believe fMRI in behaving rodents can be so powerful, allowing researchers to obtain directly comparable measures of neural activity during similar behaviours. We hope this will facilitate the study of similarities as well as differences between species.

    3.You investigated various complex processes beyond motor activity and reward association, namely, decision-making, motivation, and memory. If you can correlate this to findings in humans, many diseases affecting the central nervous system could be studied in rodent models. Are you interested in any specific ones for which there are rodent models?

    We are very interested in decision-making and in impulsivity, which are very relevant to both healthy and clinical human populations, and to which several rodent models already exist. We are particularly interested in the role played by serotonin in these processes. Serotonin is a neuromodulator in the brain and one of the main targets of psychiatric drugs. Despite being widely used in the clinic, we still know fairly little about how these drugs work and how serotonin influences brain activity and behaviour. Using the powerful invasive tools that we have available in rodents, we can now manipulate the neurons that produce serotonin with remarkable specificity and temporal precision – something that we cannot do in humans but that is essential to understand serotonin’s function. By combining these manipulations with whole-brain fMRI during behavioural tasks that are comparable to those used humans, we can start to understand how serotonin influences whole-brain activity, as commonly measured in humans, and how this relates to the pharmacological manipulations used in the clinic. This is what we are focusing on now.

    4.Beyond studies in the brain, fMRI can be used to study for instance, the spinal cord. Do you envisage further developing your method to allow investigating activity beyond the brain? And therefore covering diseases that affect both the central and peripheral nervous system.

    We have focused on the brain, but the approach could be extended to the spinal cord with some additional hardware developments. The major technical challenge that we anticipate is motion as maintaining the spinal cord immobile in a behaving animal will likely be challenging. Solutions could involve a combination of body restrainment, behavioural training, and the use of novel imaging methods that are less susceptible to motion artefacts.

    5.Considering the global picture, what is the main limitation you found in the work you performed (be this the setup, or the findings), and how would you address it/them in the future?

    One important limitation that restricts the kind of tasks that can be used, and thus the questions that can be currently tackled, is the hardware. Although our setup is compatible with a wide range of tasks, it is not compatible with all. For instance, recent years have seen the emergence of virtual reality setups for mice where they can navigate in virtual environments. We have not yet found a way to fit screens in our setup that would allow for virtual environments. This is mainly due to spatial constraints. We are hopeful that with some additional creativity and perhaps collaborative work with the scanner manufactures, we and others can progressively increase the complexity of the behavioural hardware compatible with MR imaging. Another important limitation is the interpretation of the signal we are measuring. With fMRI we are measuring neural activity indirectly, and the precise relationship between neural activity and fMRI signals is still debated. This is both a limitation of the technique we are using but also a reason to be very excited about rodent fMRI. In rodents, we can combine a variety of cell-specific invasive measures of neural activity with fMRI measures and thus further study the nature of the fMRI signals themselves. This will inform not only rodent fMRI studies but the human fMRI literature, helping us better interpret fMRI readouts.

    Have your say

    Your email address will not be published. Required fields are marked *

    This site uses Akismet to reduce spam. Learn how your comment data is processed.

    Sign up to customise the site to your preferences and to receive alerts

    Register here
    Close