This protocol describes the steps and data analysis required to successfully perform optogenetic functional magnetic resonance imaging (ofMRI). ofMRI is a novel technique that combines high-field fMRI readout with optogenetic stimulation, allowing for cell type-specific mapping of functional neural circuits and their dynamics across the whole living brain.
The investigation of the functional connectivity of precise neural circuits across the entire intact brain can be achieved through optogenetic functional magnetic resonance imaging (ofMRI), which is a novel technique that combines the relatively high spatial resolution of high-field fMRI with the precision of optogenetic stimulation. Fiber optics that enable delivery of specific wavelengths of light deep into the brain in vivo are implanted into regions of interest in order to specifically stimulate targeted cell types that have been genetically induced to express light-sensitive trans-membrane conductance channels, called opsins. fMRI is used to provide a non-invasive method of determining the brain’s global dynamic response to optogenetic stimulation of specific neural circuits through measurement of the blood-oxygen-level-dependent (BOLD) signal, which provides an indirect measurement of neuronal activity. This protocol describes the construction of fiber optic implants, the implantation surgeries, the imaging with photostimulation and the data analysis required to successfully perform ofMRI. In summary, the precise stimulation and whole-brain monitoring ability of ofMRI are crucial factors in making ofMRI a powerful tool for the study of the connectomics of the brain in both healthy and diseased states.
Optogenetic functional magnetic resonance imaging (ofMRI) is a novel technique that combines the spatial resolution of high-field fMRI with the precision of optogenetic stimulation1-11,38, enabling cell type-specific mapping of functional neural circuits and their dynamics across the whole brain. Optogenetics allows for specific cell types to be targeted for stimulation by the introduction of light-sensitive trans-membrane conductance channels, called opsins. Specific elements of neural circuits are genetically modified to express these channels, enabling millisecond-timescale modulation of activity in the intact brain1-15. fMRI provides a non-invasive method of determining the brain's global dynamic response to optogenetic stimulation of specific neural circuits through measurement of the blood-oxygen-level-dependent (BOLD) signal16-18, which provides an indirect measurement of neuronal activity.
The combination of these two techniques, termed optogenetic functional magnetic resonance imaging (ofMRI), is advantageous over other methods of recording brain activity during stimulation such as electrophysiology because it can provide a view of the entire brain at relatively high spatial resolution. This enables the detection of neuronal activity in response to targeted stimulation at great distances from the site of stimulation without the need for implantation of invasive recording electrodes1-11. ofMRI is advantageous over the more traditional method of performing electrical stimulation during fMRI, which can recruit different cell types near the electrode and thus confound the causal influence of each population19. In addition, the electrodes used for electrical stimulation and the generated current can produce artifacts during MR imaging20. Indeed, ofMRI enables the observation of the influence on global brain activity from the specific modulation of a wide variety of cell types through the use of advanced genetic targeting techniques such as the Cre-Lox system in transgenic animals or the use of promoters. Combinatorial optical control with whole-brain monitoring is possible with ofMRI through the use of both NpHR to inhibit and ChR2 to excite specific cell types. The optogenetic toolkit available for use in ofMRI is also rapidly improving over time with the introduction of opsins with increased light-sensitivity or improved kinetics, of stabilized step function opsins (SSFOs) or of red-shifted opsins that may negate the requirement for implanted fiber optics, enabling non-invasive stimulation during imaging21. These possibilities are not available with electrical stimulation.
However, signal artifacts resulting from tissue heating due to light delivery in the brain have been reported22, where temperature-induced modification of relaxation times has been shown to produce pseudo activation. Researchers performing ofMRI should therefore be aware of this potential confound. With the proper setup and controls, the issue can be addressed. Additionally, relatively low temporal resolution of measuring the hemodynamic response in fMRI may be a limiting factor for certain applications of this technique.
This protocol first describes the construction of the fiber optic implants that enable delivery of specific wavelengths of light deep into the brain in vivo. The protocol then describes the delivery of the opsin-encoding viral vector to a precise brain region using stereotactic surgery. Next the protocol describes the process of whole-brain functional MRI during simultaneous light stimulation. Finally, the protocol outlines basic data analysis of the acquired data.
Of note, the optogenetics described here require a chronic implant for light delivery. However, the fiber optic implants are stable and bio-compatible, allowing for longitudinal scanning and investigation of neural circuitry over a period of months23,24.
In summary, the precise stimulation and whole-brain monitoring ability of ofMRI are crucial factors in making ofMRI a powerful tool for the study of the connectomics of the brain. In addition, it can provide novel insight into the mechanisms of neurological diseases25 when coupled with different animal models. Indeed, ofMRI has been used to elucidate the network activity of distinct hippocampal subregions associated with seizures8. Therefore, laboratories interested in answering systems-level neuroscience questions will find this technique of importance.
Ethics Statement: Experimental procedures here have been approved by the Stanford University Institutional Animal Care and Use Committee (IACUC).
1. Preparing Patch Cables and Ferrule Implants
Note: Although patch cables and ferrule implants are available commercially, producing these in-house enables specialty designs and will cost less.
2. Stereotaxic Implantation Surgery and Virus Injection
3. Optogenetic Functional MRI
CAUTION: Exercise caution around the permanent magnetic field of an MRI scanner. Secure equipment, including the function generator, light source, ventilator, capnograph and gas tanks, sufficiently far away (at least beyond the 5 Gauss limit).
4. ofMRI Data Analysis
Note: The following steps are performed in MATLAB as described in a publication on high-throughput ofMRI27.
Figure 1 and Figure 2 show representative data resulting from 20 Hz (15 msec pulse width, 473 nm, 30% duty cycle) optogenetic stimulation of the motor cortex. A stimulation paradigm of 30 seconds of baseline followed by 20 sec on/40 sec off for six min was used. Previous studies have shown that this paradigm produces robust BOLD signal from optogenetic stimulation1,8. Figure 1 shows activated voxels detected both at the local site of stimulation (motor cortex) and in the thalamus, as a result of the long-range synaptic connections between these regions. Figure 2 shows that temporal information can be gleaned from HRFs, as the thalamic response is delayed (lower initial slope) compared to the motor cortex response after optogenetic stimulation.
Figure 1. Activation Map of BOLD Signal Induced by Optogenetic Stimulation of CaMKIIa-expressing Cells in Motor Cortex. Coherence values of active voxels, identified as those significantly synchronized to repeated stimulations, are shown overlaid on a T2-weighted coronal anatomical slice. Data collected over a six min, 30 sec period (initial 30 sec baseline and six stimulation cycles of 20 sec on/40 sec off with 473 nm light, 20 Hz, 15 msec pulse width) are condensed into one activation map. Sequential slices are 0.5 mm apart and the location of the fiber optic implant is denoted by the triangle. Please click here to view a larger version of this figure.
Figure 2. Hemodynamic Response Function. (Left) The percent modulation of the BOLD signal relative to baseline is shown for active voxels in motor cortex and thalamus during optogenetic stimulation of motor cortex (six stimulation cycles of 20 sec on/40 sec off with 473 nm light, 20 Hz, 15 msec pulse width). Shaded grey error bars denote standard error across activated voxels within ROI. (Right) Time-averaged responses are given by the hemodynamic response functions (HRF). The thalamus HRF shows a delayed response relative to the stimulated motor cortex. Blue bars indicate periods of 473 nm photostimulation. Shaded grey error bars denote standard error across six cycles. Please click here to view a larger version of this figure.
Motion of the subject during imaging is a significant source of artifact that can lead to data corruption. Appropriately securing the animal on the imaging cradle can minimize such artifacts as will maintaining appropriate anesthesia levels. Here, we used isoflurane but alternative anesthetics, such as medetomidine or ketamine and xylazine, should also be considered. However, the levels and choice of anesthetic can influence many parameters in the brain, including the BOLD response28. Isoflurane can cause changes in neuronal excitability29. Other anesthetics can also affect GABA synaptic inhibition30. Thus, the choice of anesthesia is important when performing ofMRI given its ability to affect neuronal activity. ofMRI in the absence of anesthesia is possible but can be challenging with increased motion from the animal, which can be reduced if the animal is habituated; such awake ofMRI studies have previously been performed and would avoid the confounding effect of anesthesia on the brain9,10. Post-processing motion correction algorithms can be used to greatly mitigate the effects of motion. Several of these methods exist, including the inverse Gauss-Newton algorithm employed in this protocol, which minimizes the sum of squares cost function of the reference image and image under correction. The algorithm is useful because it enables fast and robust motion correction, using a GPU parallel platform design to reduce processing times27.
For data reconstruction in this protocol, custom written software in a MATLAB environment was used for two-dimensional gridding reconstruction, where spiral samples are reconstructed in k-space into gridded images31-33. Time series data were generated by calculating the percent modulation of the BOLD signal of each voxel relative to the baseline period collected prior to stimulation. Voxels whose time series were synchronized to blocks of optogenetic stimulation with a coherence value of 0.35 or greater were defined as activated voxels; this coherence value corresponds to a less than 10-9 P value8. Coherence values were calculated as the magnitude of the Fourier transform at the frequency of repeated stimulation cycles divided by the sum-of-squares of all frequency components8,27. Familywise error can be controlled using the Bonferroni correction for multiple comparisons. Alternative methods of analysis can be used, including parametric statistical tests such as the general linear models (GLMs). The coherence method requires less prior knowledge of the HRF compared to the conventional general linear model. Therefore, it is advantageous when exploring data using ofMRI. However, the coherence method can only be used for data with block designs or selected event-related designs with a fixed interstimulus interval and may not be used in ofMRI data with other event-related designs or mixed designs. Subsequently, dynamic causal modeling (DCM) can be used to analyze interactions between brain regions identified through ofMRI. DCM is a Bayesian statistical technique developed for analysis of functional connectivity from system responses to experimental inputs during fMRI34.
Additional technical concerns for ofMRI are discussed here. Implants can be damaged or fall off, leading to the removal of the affected animal from the study. Re-implantation surgeries are not recommended due to the additional uncertainty of targeting the same ROI as in the original implantation surgery and due to animal welfare issues. Because of the significant amount of time and resources invested into each animal subject, consideration of the strength of the material is a significant concern when choosing a suitable dental cement for use in ofMRI studies. The implantation surgery is a critical factor in maximizing the longevity of the implant and animal subject. For example, ensuring that the skull is dry before applying the dental cement and placing an adequate amount of cement around the ceramic ferrule implant can ensure stability over the potential months-long timeline of the animal during the study. Additionally, alternative cage designs can be explored and discussed with the local animal care facility to avoid cages with wire tops holding the food and water that often protrude into the cage and provide opportunities for the animal to damage the implant. Importantly, the dental cement must be chosen carefully to reduce artifacts that affect imaging and alternative cements can be tested by application onto a phantom and imaging in a scanner before use in animal experiments. Trial and error with various dental cements has shown that the cement used in this protocol gives relatively few artifacts. Another technical challenge in performing ofMRI is the accuracy of fiber optic placement at the intended ROI, given the extremely small distances that can exist between nuclei in the brain35. After completing the implantation surgeries, T2-weighted anatomical scans can be used to determine correct placement by overlaying onto a brain atlas. The skill of the surgeon and practice performing these surgeries can improve correct placement rates. The specificity and expression of the opsin at the intended ROI can be verified at the conclusion of the study by perfusing the animal and fixing the brain, using immunohistochemistry or the endogenous fluorescence of a reporter-protein tagged to the opsin for visualization. These reporter proteins can also be colocalized with other proteins to ensure that the opsin is expressed in the desired neural cell types1,8,15,25. As mentioned previously, artifacts can arise when performing ofMRI due to tissue heating from light delivery22. The tissue heating causes modification of relaxation times, resulting in false BOLD signal. To ensure that activation resulting from light stimulation during ofMRI is not due to this artifact, opsin-negative controls should be performed in which either saline injected animals or animals injected with control fluorophore vectors (such as AAV-CaMKIIa-EYFP) undergo ofMRI. Additionally, only well-constructed fiber optic implants with good light transmission efficiency should be used to remove the need to use high laser powers. ofMRI studies have been performed in which false activation due to tissue heating has not been an issue1,6-8,10,11.
Regarding the choice of vector to introduce the required optogenetic genes into neurons for expression, AAVs are not known to cause disease in humans and are therefore a convenient option, given the lower biosafety level required to use these agents (BSL-1). In addition, a multitude of vector cores carry AAVs packaged with various optogenetic genes in stock and with multiple serotypes. The serotype of AAV must be chosen based on the intended cell population target to ensure optimal expression levels36,37. Lentiviruses can also be used but require a higher biosafety level. The time period required for sufficient expression of the optogenetic genes is variable depending on the specific animal model used, on the particular AAV used and on the specific experimental paradigm. In this protocol, Sprague Dawley rats at 11 weeks old are used and optogenetic studies begin four to six weeks after virus injection. Transgenic mice can also be used in optogenetic studies. It is necessary to perform pilot experiments to determine the specific amount of time required for sufficient expression of the opsins. Stimulation paradigms can vary depending on the specific opsin used. In this protocol, AAV5-CaMKIIa-hChR2(H134R)-EYFP is used and the stimulation paradigm is 20 sec on/40 sec off. If using an SSFO, the stimulation paradigm will vary because the SSFO requires only a brief pulse of light to be activated and then a brief pulse of light at another wavelength to be terminated.
An additional critical concern when performing ofMRI is preventing light leakage from the ferrule implant interface with the fiber optic patch cable during optogenetic stimulation to prevent a confounding brain signal originating from visual stimulation, even when the animal is anesthetized. Cones of black electrical tape can be used to block the light from the ferrules and to cover the eyes of the animal. Importantly, physiological values including expiratory CO2 and body temperature of the subject must be properly maintained throughout the duration of the imaging. Expiratory CO2 should be kept between 3 – 4% and body temperature at 37 °C. In addition, the shimming sequences to reduce as much inhomogeneity as possible in the magnetic field prior to starting ofMRI scans greatly determines the quality of the resulting BOLD data. Control of these factors is critical in producing reliable ofMRI data. In this protocol, DPSS lasers are used as the light source for optogenetic stimulation. Because laser light is coherent, more than enough power can be easily supplied through the fiber optic. LED light sources coupled to fiber optics are available from commercial vendors, but have the disadvantage of decreased power of light transmission. The laser light source does require alignment to each particular fiber optic patch cable, but with practice, the alignment can be accomplished within seconds to minutes.
Future applications of ofMRI include the use of next-generation opsins such as red-shifted opsins to enable non-invasive stimulation during imaging. Additionally, the implantation of MRI-compatible EEG or similar recording electrodes along with the fiber optic implant could allow for the acquisition of high temporal resolution data in addition to the high spatial resolution data of MRI. ofMRI with electrophysiological recording could provide extensive information on the functional connectivity of the brain. In summary, the power of ofMRI to monitor the entire brain in response to the stimulation of specific cell populations defined by genetic or anatomical identity makes ofMRI a critical tool to use in the study of neurological diseases and of the connectomics of the healthy brain.
The authors have nothing to disclose.
This work was supported through funding from the NIH/NIBIB R00 Award (4R00EB008738), Okawa Foundation Research Grant Award, NIH Director’s New Innovator Award (1DP2OD007265), the NSF CAREER Award (1056008), and the Alfred P. Sloan Foundation Research Fellowship. J.H.L. would like to thank Karl Deisseroth for providing the DNA plasmids used for the optogenetic experiments. The authors would also like to thank Andrew Weitz and Mankin Choy for editing the manuscript and all the Lee Lab members for their assistance with the ofMRI experiments.
7 Tesla scanner | Agilent Technologies | Discovery MR901 System | |
Sprague Dawley rats | Charles River | Crl:SD | 11 weeks old |
fiber cleaver | Fujikura | CT-05 | |
multimode optical fiber | Thor Labs | AFS105/125Y | |
fiber stripper | Thor Labs | T08S13 | |
ceramic split sleeve | Precision Fiber Products | SM-CS1140S | |
epoxy glue | Thor Labs | G14250 | |
cotton-tipped applicators | Stoelting Co. | 50975 | |
multimode ceramic zirconia ferrules | Precision Fiber Products | MM-FER2002 | |
FC/PC multimode connector | Thor Labs | 30128C3 | |
fiber optic polishing disk | Precision Fiber Products | M1-80754 | |
aluminum oxide lapping sheet, 0.3 µm | Thor Labs | LFG03P | |
aluminum oxide lapping sheet, 1 µm | Thor Labs | LFG1P | |
aluminum oxide lapping sheet, 3 µm | Thor Labs | LFG3P | |
binocular biological microscope 40X-1000X | Amscope | B100 | |
laser safety glasses | Kentek | KXL-62W01 | |
473 nm DPSS laser | Laserglow | LRS-0473 | |
594 nm DPSS laser | Laserglow | LRS-0594 | |
Allen hex wrench set | 2.0 mm (5/64") for alignment of fiber tip to focal point of coupler in the laser | ||
power meter, Si Sensor, 400-1100 nm | Thor Labs | PM121D | |
Isoflurane (Isothesia) | Henry Schein | 50033 | |
isoflurane vaporizer with induction chamber | VetEquip | 901806 | |
NanoFil 100uL syringe | World Precision Instruments | NANOFIL-100 | |
UltraMicroPump with SYS-Micro4 Controller | World Precision Instruments | UMP3-1 | |
function generator | A.M.P.I. | Master-8 | |
small animal stereotax | David Kopf Instruments | Model 940 | |
Model 683 small animal ventilator | Harvard Apparatus | 550000 | |
Type 340 capnograph | Harvard Apparatus | 733809 | |
dental drill (rotary micromotor kit) | Foredom Electric Co. | K.1070 | |
ophthalmic ointment (Artificial Tears) | Rugby | 00536-6550-91 | |
instrument sterilizer | CellPoint Scientific | Germinator 500 | glass bead sterilizer |
antibiotic powder | Pfizer | NEO-PREDEF | neomycin sulfate, isoflupredone acetate and tetracaine hydrochloride |
buprenorphine painkiller | Hospira | NDC:0409-2012 | schedule III controlled substance , 0.3 mg/mL stock |