A chat with Na Ji

Like Comment
Read the paper

Na Ji, from the University of California Berkeley, talks about the role of curiosity in her work, why she likes taking knowledge from one discipline to another, and she describes imaging systems that she and her team developed to observe brain activity in vivo.

In two papers in Nature Methods, Na Ji presented ways of imaging the brain in vivo in behaving mice. One method is Kilohertz two-photon fluorescence microscopy imaging of neural activity in vivo. The other method is: Rapid mesoscale volumetric imaging of neural activity with synaptic resolution. 

I published a profile of her here. And this podcast has more from the conversation with her. 

Note: These podcasts are produced to be heard. If you can, please tune in. Transcripts are generated using speech recognition software and there’s a human editor. But a transcript may contain errors. Please check the corresponding audio before quoting.

Transcript of podcast : A conversation with Na Ji

Hi, this is Conversations with scientists, I'm Vivien Marx. This podcast is about scientists, about some of the work they do, about why they do what they do, about the questions and puzzles they pursue. It's about their approach to science, their background, how they were mentored and how they mentor and much else. Sometimes the podcast is based on a conversation with one scientist, and sometimes it involves several researchers.

Na Ji [00:39]:  I'm very simply a microscopist who wants to be a neurobiologist. When I say that I people will say but you are already a neurobiologist, and that makes me very, very happy.

Vivien: That's Dr Na Ji, a researcher at the University of California, Berkeley. She's a physicist and has a PhD in chemistry. She has a split appointment. She is on the faculty of UC Berkeley's physics department and in molecular cell biology. She is part of a pan-departmental research institute. She's involved in the graduate programs in biophysics and another in neuroscience. She's busy.At Berkeley, she teaches an introductory class in physics. She calls it physics for modern citizens.

Na Ji:  Now I call it physics for modern citizens. I am teaching general physics concepts, energy, atoms, all to universe to people who never had physics before. It's my first time teaching this course, so yeah, lots of time to prepare, I love it, students really respond to it. And I feel like I'm making the kind of impact beyond  what papers can. 

Vivien: Scientists have impact with their research and the papers they publish about the research. Na Ji Has a sense that her teaching has a wide impact beyond her publications in scientific journals. She was a researcher at Janelia Research Campus in Ashburn, Virginia, where she did not teach, and then she moved to join the faculty at UC Berkeley.

Na Ji: That's a nice thing I like about Berkeley is that I feel like I'm making more impact. Janelia was a wonderful place, we had all the resources we need, but  the impact you can make is basically through your papers. Here  I don't just talk to people in my life and I'm a graduate student mentor for physics students, I'm teaching undergraduate courses. I just feel like I make an impact about to communicate science and life, what can you do with a graduate education.

Vivien:  Na Ji works on ways to study the brain. And she wants to make sure the ways to do so are ones that neurobiologists will use, not just physicists.  She likes to bridge-build between different scientific disciplines.Her approach is shaped by her background.She started out in chemical physics, which is about using physics tools to understand chemical systems. She received her PhD degree in chemistry at UC Berkeley. Looking back, she realizes she has always enjoyed taking something from one field to address questions in another. After her PhD, he thought she would try something different for her postdoctoral fellowship.

Na Ji:  My adviser always encouraged us to do something different. So there was never expectation. Well, you know, I was doing that for my PhD and I will be doing something similar for my postdoc or for my independent career. 

Vivien: She had always been interested in biology, and it was her first choice for college.

Na Ji: But I didn't get into the biology department because I didn't do well enough in my college entrance  exam. During those years, biology was an extremely popular major in China at that time. So I always had this biology  bug,  I want to study biology.

Vivien: The biology bug made her think broadly about a postdoctoral fellowship in biology. After her PhD, her PhD advisor was  Yuen Ron Shen, now a physicist emeritus at UC Berkeley, and she liked that. He encouraged trainees to branch out. Na Ji  mentors her students the same way she encourages them to be brave and try something different after completing their PhD.

Na Ji [04:28] :  I think that's when you go to do postdoc, you should something different, because. Because this is really a chance for you to even broaden, because right now everything is so interdisciplinary with the type of things that we do. When you are dong PhD, that's the place where you are where you are really interesting, your depth of understanding of a subject. So I suppose that you really should broaden it out. But by going beyond that, what I want my students to appreciate is that you don't have to be a research scientist after you finish.You don't have to be you know, you don't have to have academics as your research goal.  But I do think that all those fields  we learn in graduate school is very useful for anything.

Vivien: She is open minded about her mentee's career choices, they might want to become academics, but they might, for example, join a company or start a business. They might work for the government or in law. When she finished her PhD, as she looked around, she realized how captivating neurobiology is to her.

Na Ji: In a way, it's almost like, it's a puzzle on this more metaphysical philosophical, I was doing chemical physics for along time I have  a very deep understanding of how molecules, how atoms work, how they vibrate, they rotate, I know them so well and I use my brain. But I know nothing. I knew nothing about my brain , I guess I still don't know that much, I know a little bit more now. But I just feel like this is really like the ultimate mystery of who we are.

Vivien: Neuroscience was it, she decided she wanted to do neuroscience and be a neurobiologist, but she is glad that she had studied and trained in other fields.

Na Ji: Even if I had known that I'm going to be doing neurobiology like looking back given all my knowledge now, I probably still will do the same thing because they just made a great foundation for me in terms of just generally understanding the world, the inanimate part of the world,   but also the training I received from my graduate advisor. It was just awesome. And I think, you know, in terms of research, we have basically everything in all the skills you learn by doing chemical physics research can be applied in any other type of research. Those are all great experiences.

Vivien: In her research, Na Ji i develops methods to help neurobiologists with experimental setups that you do not need a physics PhD to operate.

Na Ji: So I understand because I do neurobiology in my own lab and I understand the challenges that neurobiologists already facing just to just to use standard techniques, it already very hard, you need to train animals, you need to get the surgery done. So what you really want to have is something that is really robust and then also also it has to be something that can fundamentally change the type of question you could answer, because if it's just something that is a little bit better, you know, it's just not worth it for them to adopt your method.

Vivien: In Nature Methods she recently published two new ways to help neurobiologists study the brain. And these are approaches she is using in her lab too. One of the two methods lets scientists image a mouse brain along a narrow path down to a depth of six hundred microns in a live mouse. I say depth, but it's important to keep in mind this is imaging to a depth of about 200ths of an inch. But it's deep enough to capture what is happening along the entire length of a neuron which has a lot of branches and a lot of interactions with other neurons. To use this method the mouse has been prepped with a small hole in its skull, a brief laser pulse will make sensors in the neuron light up, and those signals can be captured with the help of a microscope. Researchers can watch the entire dendritic tree of a neuron, get a picture of the activity of this entire neuron. This is unlike more traditional approaches that let a researcher only look at part of this tree at one time.

Na Ji: So it's really very difficult to say how, if you want to ask the question, what is the input, what are the inputs does a single neuron receive. And how does it input across the dendritic, synaptic and  eventually somatic level computation? To answer that kind of question, you really need to be able to look at all the inputs into a single neuron at the same time.

Vivien: Typically, a lab will label cells in the brain and watch their activity in vivo through the window in the skull of a mouse, and by using two-photon microscopy and then the image, a small area inside the brain, she and her team made a system that uses a so-called Bessel beam to work with a microscope called a mesoscope. This mesoscope was developed at Janelia Research Campus in the lab of Karel Svoboda. What's special about the mesoscope is that it has a large field of view. Instead of the more traditional, less than one millimeter field of view, he and his team made a microscope the mesoscope with a five millimeter field of view that can resolve single neurons. There are other large field of view microscopes, but labs can't use those to resolve what is happening in the individual neurons. Five millimeters might not sound like much, but in the brain of a mouse, that's a lot of brain area you can watch.

Na Ji: Now, you could actually ask the questions, which we couldn't ask before. Basically a single-neuron computation.

Vivien: The system uses a Bessel beam and if a lab wanted to set it up, it is not hard to do, says Na Ji. She and her team added a Bessel module to the existing mesoscope.

Na Ji: So this is the thing I love about Bessel beam, is that it's actually fairly easy to set up, onto an existing system. So this is something you can add to your existing two photon system, because if you need a physics PhD to maintain the microscope, nobody's going to use it. Except people in your group. It's is actually very, very simple to add onto the mesoscope. But once we have the add-on, we were able to, for example, do the single cell computation type of example that we just talk about, which adds one ft by one foot little piece we bolted onto the mesoscope. We made it to allow it to generate all this data that we had.

Vivien: Labs can set up both this module and the mesoscope, these components are available through a company called Thor Labs for people to set up and use on their own. When they use the system labs can switch between the Bessel mode and the conventional Gaussian. When labs use a conventional two-photon microscope, they are looking at neurons and a single focal plane.

You basically scan at different depths to get an idea of what is happening in the three dimensions of a neural circuit that leads to an image stack. A lot of images. Scanning with the Bessel beam avoids that big stack and gives you information from a volume of the brain and what is going on there? You can cover what is called a whole dendritic tree, Na Ji explains.

Na Ji: The Bessel beam, the focal volume is still very narrow, it has really good resolution in the XY direction, so you can see spines and synapses. But in the Z direction, instead of being 1-2 microns long , we make it 100 microns long. 100 microns long that means all the structures that are within 100 micron depth can be excited and give you signal. Now you're retrieving information from a volume to a depth that is the length of the Bessel beam,  100. Instead of taking 300 images along 600 micron long cell, we we only take six images. So we scan the Bessel focus at 100 micron steps. Six images. Now we can cover this whole dendritic tree. 

Vivien: Using this system, The team was also able to look at thousands of GABA-ergic neurons in a mouse that was awake and resting. GABA stands for gamma aminobutyric acid. GABA-ergic neurons are a kind of neuron. They are inhibitory neurons. In mammals, and humans are mammals, as are mice. The GABA-ergic system is involved in muscle activity, but also things such as anxiety and stress response. Inhibitory neurons play different roles depending on where they are in the brain.

Na Ji:  Inhibitory neurons are not a uniform population. The neurons at different depths can have different roles different functions.

Vivien: Using a mesoscope alone without the Bessel beam might take the gaze of a researcher to around 160 microns. But the Bessel beam got the team to be able to see to a depth of 600 microns, and they got a wide field of view so they could look at what was happening in several brain regions

Na Ji [14:06]: So the mesoscope is basically  based on a two-photon microscope with a very large field of view, you can image a very large area. So now with that,  why do people want to image in multiple brain regions at the same time. Traditionally in a standard two-photon system that you buy, non-mesoscope usually image one to two millimeter diameter area, which is really just a single cortical brain region of a mouse. But if you really look at the behaviors. We can do lower resolution imaging experiments, look at the whole dorsal surface as the whole top surface of the cortex of a mouse that is engaged in a behavior, what you will see is that all brain regions become very active.

And you see activity goes back and forth, the waves, it's just mesmerizing. So it's it's really, it's great to  study a simple region and try to understand that that kind of  research is very, very valuable. But ultimately, when you go more and more towards a system-level understanding or cognitive level understanding of the brain, you would want to be able to image multiple brain areas at the same time, because we already know that all those areas  are engaged when we try to do any kind of behavior. In this particular example for four- brain regions, the mouse was not engaged in anything.  It was just staying in the dark, just kind of daydreaming.

Vivien: The mouse was daydreaming and, yeah, the scientists can't know what it was daydreaming about, but it was sitting in the dark and the scientists got a glimpse of its brain activity even when it is not scrambling around or tussling with others, multiple brain regions are active. The scientists were able to image 9,000 neurons across four areas of the brain, which seems like a neat accomplishment. Na Ji is happy with the work, but she sees plenty of one uppance  going around in neuroscience. This idea of 'I imaged more neurons than you did'.

Na Ji: Yah, we did 9,000 neurons, great. But is that is only better than 8000 neurons or is that worse than twenty thousand neurons. It's that it's not something that, you know, because bigger is not necessarily better. 

Vivien: When she was looking around for a postdoctoral fellowship, her mentor told her to go talk to a neurobiologist. One of the things one neurobiologist told her was that becoming a good neurobiologist was contingent on the questions you ask in physics. The question scientists ask are about quantity, she says.

Na Ji: It's all about value, quantity, but in biology it's really about the questions you ask  which is also what intrigued about biology. I feel like there's a lot more thinking and intellectual effort going into asking a neurobiology question. For technology you can be simple-minded: larger field of view, faster speed, that's it. 

Vivien: What she found interesting is that they could observe 9000 neurons across four areas of the brain.

Na Ji: The information really kept propagating within the brain. So this is why the meso-scale is really important. And to be able to look at the whole propagation.

Vivien With the mesoscope, they watched calcium concentration change inside neurons. That's one way of watching neuronal activity. What this tracks is that when a neuron fires, there is an electrical signal that propagates down the length of the neuron. Calcium flows into the neuron. The calcium signal can last for one or two seconds, but the actual electrical signal is much faster. It's a little like thunder and lightning. With calcium imaging, you're catching the thunder, not the lightning. And then neurons do a lot more than fire or not fire. Signals arrive to a neuron that lead to what is called sub-threshold activity.

It's not like the neuron is holding its breath, but the neuron is collecting signals. The signals haven't reached the threshold at which the neuron is ready to fire, Na Ji  explains subthreshold signaling with an example from parenting.

Na Ji: They may be unhappy about something, but they're not telling me. They don't vocalized it. All of a sudden, like a little thing pushes them over, like people will say the straw that breaks the elephan't's back. So when that happens they suddenly vocalizing it. They complain. You think it's because they didn't get a cookie, that's making them upset. But it's really because they had a bad day at school, they're fighting with their friends, all those subthreshold events. are really interesting and important to know

Vivien A parent will want to know about subthreshold events, all the events that led up to the explosion of unhappiness about something. In brain research scientists want to capture these subthreshold events, these undercurrents. There are voltage sensors they can use to do that. Voltage imaging captures different information than calcium imaging. But to do voltage imaging  one needs a system that can image superfast image at 1000 frames per second

Na Ji  An action potential lasts around one millisecond. That means that you have image your sample one thousand times per second. I was telling you  with two-photon, you scan your focus in two dimensions, to take one image, make one frame. With a camera you click, take  picture, take everything in parallel. With two photon, you have to scan your focus  serially in time in order to cover a frame. With the best commercial microscope you can buy, you will have a typical frame rate of 30 Hertz, depending. 30 frames per second. Now we need to scan much faster. At 1000 and we demonstrated that we can image at three thousand frames per second.

Vivien The team built a system able to image those kinds of speeds, it is based on a system called FACED, which stands for Freespace Angular Chirp Enhanced Delay, developed by Kevin Tsia, an electrical engineer from the University of Hong Kong.

He and his team built the system to watch cells as they flow in a microfluidic device, a kind of device that is built out of a polymer with narrow channels in which you can sort cells, expose them to drugs and see what happens. It's used, for example, when working on diagnostic tests where you want to have high throughput of cells passing through quickly. His FACED system is fast. He could get 80 million frames per second Na JI at Kevin Tsia at a conference, and his system gave her an idea for a system that could be combined with two-photon microscopy.

Na Ji: I look at it,  well, we're going to have to use it for two photon, but it's not straightforward. We can't  transport microscope make it work work. We need to for example think about the nanosecond delay which wasn't a problem for what he was doing, because he wasn't looking at fluorescence, But it is a problem for us.

Vivien: They adapted FACED for fluorescence microscopy so that it could image at 3000 frames per second and capture voltage signals happening in the mouse's brain down to 345 microns deep. The module they use is a little like the Infinity Room by the Japanese artist Yayoi Kusama. When you walk into this room she created, light is bounced all around and it's a continuous series of nested reflections. If you've ever seen an Infinity Room, it's a little dizzying, at least to me. But it's cool

Na Ji: in the mirror parallel mirror. You know that the images further away from you.  those images form the picture than the images than those that are closer to you. 

Vivien The researchers made a system that excites the sensors with a laser, but each spot is hit by light with a two nanosecond delay. That's because it takes a moment for the sensor to react and then release the signal that is then captured by the microscope. Next, another point lights up. This approach lets the team tell the signals apart. So then they had a system able to capture electrical signals in neurons in such a way that they could explore subthreshold events.

Na Ji: We use that difference to tell  tell apart the fluorescence signal that is generated . The problem is when you shine light it takes a few nanoseconds before the fluorescent signal comes out.  We also need to have the voltage sensor that will work with two photon excitation. There are quite a bit of voltage sensors is actually quite a bit of sensors on the market. Many of them will not not report voltage signal reporting, when you use two photon excitation.Single photon is ok two photon is not. The probe ASAP3 actually changes with two photon excitation. It as a great collaboration with him because he hasn't published it when we started a project.

The paper was just coming out in Cell describing I think late last year. We got this unpublished construct, we expressed it in the mouse brain, we used it,  What we got,  remember calcium it's too slow now we can see individual action potentials, speed is not a problem. But that's a really exciting thing, those subthreshold the things that you wouldn't be able to know. If you look at the traces you see bumps, but it doesn't have a little spike on top of it.Those are the subthreshold responses. which are traditionally only measured with electrophysiology, now we could actually see all of that. And that would really allow people to ask more questions they weren't able to ask before using imaging methods.  

Vivien: Asking questions one could not ask before is something many labs need. Imaging methods helps neuroscience and it helps other disciplines, too.One nice thing, 

Na Ji: One nice thing, those imaging methods can be applied to many other different samples, even materials, going back to my roots,  the materials with fancy physical properties. When you want to image them, the same challenges apply. So many of those things that are just. Now at Berkeley, as you can probably tell that that I really love Berkeley, is that I have the ability to do all of that, such a great breadth of sciences that are happening, which can also, you know, pull me into a different directions and that that that I wouldn't otherwise have gone into.  

Vivien Na Ji works on the brain, applies physics and chemistry. Tying all of this together for her, she says it's all about curiosity.

Na Ji [25:55] 
I benefit by finding new questions. Then given my personality curiosity is, if I define myself I would just say I am very curious, this is a great place to be.

 I'm just very curious about things , like anything.  What I do for fun: read I read widely about everything, pop culture, politics, history. So I think my broad interests make this interdisciplinary work feel very natural to me. It's not something I have to strive for, it's just something I really enjoy doing. 

Vivien That was conversations with scientists. Today's episode was with Dr. Na Ji , a researcher at the University of California, Berkeley. This music is Equinox by the band Split Phase, I'm Vivien Marx, thanks for listening.

Na Ji (Photo: E. Betzig) 

Vivien Marx

Journalist , Nature Research

No comments yet.