A Wireless miniScope for Deep Brain Imaging in Freely Moving Mice

Abstract

Background:

The increasing interest in the study of neuronal activities at the microcircuit level is motivating neuroscientists and engineers to push the limits in developing miniature in vivo imaging systems. This inter-disciplinary effort led to an increasingly widespread use of wearable miniature microscopes, constantly improving in size, cost, spatial and temporal resolutions, and signal to noise ratio.

New Method:

Here we developed a miniature wireless fluorescence microscope (miniScope) that allows recording of brain neural activities at single cell resolution. The wireless miniScope has onboard field-programmable gate array (FPGA) and Micro SD Card storage, and is powered by a battery backpack.

Results:

Using this wireless miniScope, we simultaneously recorded activities from hundreds of medium spiny neurons (MSNs) in the dorsal striatum of two freely moving mice interacting with each other in an open field, with excellent spatial and temporal resolutions.

Comparison with Existing Methods:

Existing miniaturized microscope systems have connecting cables between the microscope sensor and the data acquisition system, consequently limiting the recording to one animal at a time. The wireless miniScope allows simultaneous recording of multiple mice in a group, and could also be applied to freely behaving small primates in the future.

Conclusion:

The wireless miniScope expands the realm of possible behavioral experiments, both by minimizing the repercussions of the cable from the imaging device on the rodent’s behavior and by enabling simultaneous in vivo imaging from multiple animals.

1. Introduction

The study of neuronal activities at the microcircuit level is crucial for the understanding of fundamental neurological processes and their association with specific behaviors (Ji et al., 2016). The miniaturization of light microscopy systems has a major impact on the way neuroscientists study the complex cognitive processes such as learning, reward and addiction, social interaction, decision making or mouse models for neurological diseases such as Parkinson disease (Barbera et al., 2016; Cai et al., 2016; Ghosh et al., 2011). Recently, portable fluorescence microscopy is becoming an excellent choice for probing mammalian brain activity in freely behaving animals, by leveraging key advantages over more traditional techniques, particularly the fine detail in both spatial and temporal resolution. Current state-of-the-art miniature imaging system allows tracking the activity of the same neuronal ensemble for over a month (Ziv et al., 2013), and this is paving the way for new studies aiming at answering crucial questions about long term neurological processes such as learning and memory and neural encoding mechanisms of animal behavior control.

However, one of the key limitations of the existing miniScope systems that still hinder the study of more complex behaviors, is the tethering of the miniScope mounted on the animal’s head to the data acquisition system, which may either influence the behavioral outcome of the experiment, or even prevent one to carry out the experiment itself, such as simultaneous recording of multiple animals in a group housing settings. There have been attempts to create miniature wireless calcium imaging systems (Liberti et al., 2017), however due to the size and weight of the transmitter and battery, this system cannot be used wirelessly in small rodents or birds. To address this issue, we developed a wireless miniScope, capable of recording activities of hundreds of individual neurons in vivo. We tested capabilities of the wireless miniScope by simultaneously imaging the dorsal striatum of two mice interacting with each other.

2. Methods

2.1. Microscope design

The wireless miniScope imaging system comprised a 3D printed miniature epifluorescence microscope body (Barbera et al., 2016), an image sensor board hosting the CMOS image sensor (MT9V022, ON Semiconductor), a daughter board hosting the field-programmable gate array (FPGA, AGL400V5, Microsemi Corporation), a LED driver, a Micro SD Card and a communication interface with host computer (Fig. 1a and 1b). Frames from the image sensor (10 fps, 200 × 200 pixels, corresponding to a 500 μm × 500 μm field of view) were saved directly to the Micro SD Card as raw data, including a header with timestamp and frame number to facilitate image extraction. The daughter board also hosted an IR sensor that was used as a trigger for initiation of recording and for synchronization with other wireless imaging microscopes. Two operation modes were available: a streaming mode and a recording mode. In the streaming mode, the system was tethered and streamed live images to the host computer through a custom data acquisition board and a USB cable. The streaming mode was useful for positioning of the microscope and focus adjustments. The recording mode was used during the experiments, and the wireless system stores all frames into the Micro SD Card.
The entire system was tested using a single 3.8 g rechargeable LiPo battery (180 mAh), which could provide 40 minutes of continuous recording.

Figure 1. Wireless miniScope imaging system.

A. An illustration of a mouse wearing the wireless miniScope with a battery backpack. B. An illustration of mounting of the wireless miniScope on mouse head and coupled to a GRIN lens for in vivo deep brain imaging. C. A photo of an actual wireless miniScope side by side with a Quarter. D. Overall design of wireless miniScope, scale bar 4 mm. E. System architecture overview. F. A representative standard deviation projection image of GCaMP6s fluorescence images captured by the wireless miniScope imaging system. The D2 neurons detected by the automatic cell identification algorithm are overlaid in green. G. Representative calcium transient traces from 25 representative identified D2 neurons; the red arrows mark the onset of the detected calcium transients. H. Representative mouse locomotion trajectories of two mice both wearing a wireless miniScope during a 3-minute open field session.

A brief summary of the miniScope’s main optical parameters was listed below:

Number	Name	Weight	Specs and vendor
1	Miniscope Housing	1.7g	Custom design, 3D printed with SLArmor Nickel-NanoTool, Protolabs
2	LED		XLamp XP-E blue LED, Cree
3	Collimating lens		4-mm diameter, 2.73-mm focal length aspherical lens, #83–605, Edmund optics
4	Excitation filter		3 mm × 3 mm×1mm, ET470/40×, Chroma Technology
5	Dichroic mirror		5 mm × 5 mm×1mm, FF495-Di02, Semrock
6	Emission filter		3 mm × 3 mm×1mm, ET525/50m, Chroma Technology
7	Objective lens		4-mm diameter, 2.73-mm focal length aspherical lens, #83–605, Edmund optics
8	Imaging lens		4-mm diameter 6-mm focal length achromatic doublet lens, #63–690, Edmund optics
9	Imaging sensor PCB including SD card	2.2g	Custom design
10	Polymer Lithium Ion Battery	See below	See below

Total weight of wireless miniScope without battery: 3.9 grams

Dimension of wireless miniScope: 16 mm × 18 mm × 30 mm

Below are three different models of battery that we tested with wireless miniScope.

Type	Model	Size	Weight	Expected runtime
100 mAh	GM261534	2.5mm × 15mm × 34 mm	2.2 g	16 min
180 mAh	GM052025	5mm × 20mm × 25mm	3.8 g (Used in exp.)	28 min
240 mAh	GM502030-PCB	5mm × 20mm × 30mm	5.0 g	38 min

All the design files and codes can be found at the following GitHub link: https://github.com/giovannibarbera/wireless_v1.0

2.2. Animal surgery

All surgical procedures were performed in accordance with the guidelines of Institutional Animal Care and Use Committee, the Intramural Research Program, National Institute on Drug Abuse, National Institutes of Health. Transgenic mice (minimum 30 g of body weight) expressing Cre recombinase under the control of the dopamine D1 receptor promoter (D1-Cre, FK150 line, C57BL/6J congenic, Gensat, RRID: MMRRC_036916-UCD) or dopamine D2 receptor promoter (D2-Cre, ER44 line, C57BL/6J congenic, Gensat, RRID: MMRRC_032108-UCD) were used in the experiment. Adeno-Associated Virus (AAV) expressing calcium indicator Gcamp6s, AAV1.CAG.Flex.GCaMP6s.WPRE.SV40 (University of Pennsylvania Vector Core), was injected into the dorsal striatum of mice using the stereotactic coordinates (A/P: −0.93 mm, M/L: +1.8 mm, D/V: −3.46 mm, with 30° angle shift to caudate). Ten days after AAV injection, a 1mm diameter GRIN lens was implanted into the mouse brain right above the dorsal striatum under ketamine/xylazine anesthesia (ketamine:100 mg/kg, xylazine:15 mg/kg), as previously described (Barbera et al., 2016). Three weeks after GRIN lens implantation, miniScope base was mounted on the mouse head. Data acquisition was performed 4 weeks after GRIN lens implantation.

2.3. Data acquisition and analysis

The open field behavior tests were performed during the light cycle. Mouse were under light isoflurane anesthesia when the miniScope was mounted on the mouse head. Mouse was then allowed to recover from anesthesia in home cage for 30 minutes. We conducted 3 sessions (5 min per session) open field test for the two mice simultaneously in a 34 cm × 40 cm × 20 cm (length × width × height) chamber with 5 min interval between imaging sessions. The GCaMP6s fluorescent signal and the video of mice behavior were recorded simultaneously. The calcium images and behavior video were processed and analyzed using custom scripts in MATLAB.

3. Results

3.1. In vivo recording of striatal MSNs

To test the performance of the wireless miniScope imaging system, we simultaneously recorded neural activities from two freely behaving mice (Supplementary Video). We imaged the activity of dorsal striatal D1-MSNs using one D1-Cre mouse and D2-MSNs using one D2-Cre mouse. Three consecutive imaging sessions were performed using this mice pair. A representative standard deviation projection image from a sample recording was shown in Fig. 1c. Representative calcium traces were shown in Fig. 1d. The two mice were placed in an open field and let free to explore the environment and interact with each other. The trajectory of both mice were shown in Fig. 1e. We were able to identify 129 active D1-MSNs from the D1-Cre mouse and 166 active D2-MSNs from the D2-Cre mouse (Fig. 2a), using the automatic cell identification algorithm adopted in (Barbera et al., 2016). In both D1- and D2- MSNs, we observed an increase in neural activity during active behavioral states (e.g. motion initiation or speed increase), in line with the theory that concerted activity of direct and indirect pathway MSNs underlies locomotor activity (Barbera et al., 2016; Cui et al., 2013). In both D1- and D2- MSNs, we identified a wide range of levels of correlation between neural activity and mouse locomotion velocity (Fig. 2b and 2c). Some MSNs showed positive correlation with mouse locomotion velocities, whereas others displayed negative correlation with mouse locomotion velocities (Fig. 2b and 2c). On the population level, both D1- and D2- MSN population in general showed positive correlation with mouse locomotion velocities (Fig. 2d, left panel), consistent with recent findings on D1- and D2- MSN activity and action initiation or locomotion (Barbera et al., 2016; Cui et al., 2013). However, upon closer look, a small fraction of D1- or D2- MSNs showed negative correlation with mouse locomotion velocities (Fig. 2d, middle panel), while more D1- and D2- MSNs showed positive correlation with mouse locomotion velocities (Fig. 2d, right panel). This functional heterogeneity among neighboring neurons could only be captured through the study of neural circuits at a single cell level, consistent with our recent findings using the tethered version of miniScope (Barbera et al., 2016). In sum, the wireless miniScope offers excellent spatial and temporal resolution and performance is comparable to the tethered version of miniScope.

Figure 2. Simultaneous imaging of two mice using wireless miniScope in open field test.

A. Rasterplot of the normalized calcium traces (top) and time plot of the average population neural activity and locomotion velocity of the mouse (bottom). Both population activities of D1- and D2- MSNs show clear correlation with locomotor activity. B. Neuronal map of the two mice; color code indicates the average correlation coefficient between each neuron and the mouse locomotor speed. C. Distribution of the average correlation coefficient between neural activity and mouse speed, compared against the average correlation coefficient for calcium traces randomly shuffled 1000 times (gray bars). D. Average neuronal activity as a function of mouse speed for all neurons (left), some neurons are negatively correlated with speed (middle), and some neurons are positively correlated with speed (right). The spatial location of these neurons for the D1 (left map) and D2 (right map) mouse are shown in the inset maps (scale bar 100 μm).

4. Discussion

In this manuscript, we developed a wireless miniScope that offer similar performance compared to our wired version of miniScope. We showed that we could simultaneously record two mice under freely social interaction settings. The advantage of the wireless miniScope over the wired version are several folds. Firstly, it is apparent from our recording that the wireless miniScope allows simultaneous imaging of multiple animals in group housing conditions, allowing study of more complex behavior under group house conditions, whereas the tethered version of miniScope would only allow imaging one animal at a time, due to potential tangling of cables connecting each miniScope sensor to FPGA module. Secondly, the wireless miniScope would facilitate other animal behavior studies requiring connecting tubing from the behavior box to the animal, such as intravenous cocaine self-administration (IVSA) studies, in which a cocaine infusion tubing need to be connected to the animal jugular vein. It would be extremely difficult to perform in vivo imaging of neural activities while animal perform IVSA in the behavior box, again due to the fact that the infusion tubing would tangle with cable connecting miniScope sensor to the FPGA module. The wireless miniScope would allow such studies by eliminating the cable for the miniScope. Finally, the wireless miniScope could be used towards in vivo imaging of deep brain neural activities in freely behaving small primates in the future. Freely behaving small primates would likely be able to reach to their head and disconnect miniScope from the FPGA module if there is a cable connecting the two. A wireless miniScope with a well protective shield would prevent this from happening, therefore allowing such studies to be carried out.

While our current wireless miniScope prototype are promising, there are several improvements could be implemented in future generations. For example, the current version of wireless miniScope uses a CMOS sensor that is power hungry, we therefore have to use a larger battery backpack to power the entire device. New CMOS sensors with up to 90% less power consumption is now available. Incorporating new CMOS sensors into our wireless miniScope system would allow us to use much smaller battery for the system, or significantly prolog the recording time using the same battery backpack. Future generation of wireless miniScope will also consider wireless powering options. We recently developed a wireless optogenetic stimulation device using radio frequency power transfer (Lee et al., 2015), similar wireless powering option will be explored, once the more energy efficient CMOS sensor has been incorporated into the wireless miniScope.

Finally, the current version of the system saves locally the recorded data on an onboard SD-Card. The alternative of streaming out the data in real time to a host computer, although appealing, poses some challenges: a Bluetooth interface would hardly provide enough bandwidth to stream video data (even the maximum 2Mbps of the BLE Bluetooth 5.0 is not enough to transfer the recorded images which, at 10 Hz, require a sustained transfer speed of 26Mbps). On the other hand, a WiFi interface for data transfer will be considered for a future generation of wireless miniScope specifically designed for small primates, since the increase in PCB surface and power required would be impractical for use with small rodents. Incorporating onboard battery with wireless charging and WiFi streaming option will render the wireless miniScope extremely attractive for freely behaving small primate studies.

Highlights:

• Wireless miniature fluorescent microscope

• Simultaneously imaging of multiple animals