Interrogation of spinal networks during movement preparation using transcutaneous spinal stimulation

That’s not the name of the paper, but that’s the general idea. We tend to take movement for granted, I mean most of the time we do it without thinking. I don’t look down as I’m typing these words, yet my fingers know what to do to make it happen. Similarly, we don’t really think about balancing ourselves when we walk, we just do it. So if I’m not consciously thinking about every step I take when I walk down the street, who, or what is keeping me going?

Semi-controversial opinion, but I believe the spinal cord is just a really long brain. It’s made up of the same stuff as the brain and it’s been shown to act independently of the brain. So if it’s made of the the same stuff and thinks for itself, why then do we treat it like a nerve? The answer is, for a long time we didn’t realize the spinal cord did anything besides relay information back and forth. It was the freeway of the body, but that’s no longer the case. More and more we’re realizing just how much the spinal cord does for us, so now we get to poke it with an electrical stick (so to speak) and see what it does.

The idea of the paper was simple, how do we figure out what is going on in the spinal cord when a person wants to move? Well the average human reaction time is anywhere from a (for a human) fast 200 ms up to a slower 300 ms, but most of us will fall in the middle of that distribution despite our best efforts to move faster. And we’ll get into why that’s important in just a minute.

When hospital-PI approached me about this experiment I was amazed that we could indirectly figure out what the spinal networks were doing without invasively poking around in a person’s spinal cord to see first hand (here’s where I said yes to do this paper, spoiler it took far longer than he proposed it would).

When you move your brain sends signals to networks in the spinal cord. But what does that mean for those networks and how does different movement intentions change what the networks are doing? To answer this question we used transcutaneous spinal cord stimulation (TSS) to see how responses changed PRIOR to movement onset (hence the reaction time comments). We gave our participants either an auditory or tactile cue and they were asked to perform one of four tasks knee flexion, knee extension, plantarflexion (think the motion your ankle makes when you use to step on a gas peddle), or dorsiflexion (pulling the foot back using the ankle, or letting off the gas so to speak).

Now here’s where the magic happens. The brain hears (or the person feels) the cue and when the person tries to move the brain sends signals to the spinal networks responsible for doing those movements. We can figure out what is going on in these networks by applying a electrical stimulus via TSS. Normally, this causes the person’s muscles to flex or causes what we call evoked responses. We’re not 100% sure how TSS works or at least we’re not in agreement, but the idea is that (probably) the electrical current reduces the threshold for neurons to fire. If we lower the threshold far enough the neurons fire and cause the muscle to flex (evoke a response). Measuring that response tells us something about the state of the network because these are dynamic networks so they don’t always require the same amount of stimulus (via TSS or through signals in the body) to fire. A change in the amount required to make a network fire tells us a lot about what’s going on inside the spinal cord without having to peak under the hood.

In fact, it goes both ways. Some networks in the spinal cord will require less stimulus to fire and others will require more. This translates in changes to amplitude of the evoked responses we see when we apply TSS! So the idea in the paper is to ask the person to move as quickly as possible after the cue (remember tactile or auditory) and then we applied TSS at one of several intervals 50, 100, 150, 200, or 250 ms after the cue was given. Since most people don’t actually start moving until close to the 250 ms mark we can determine how the spinal networks are changing based on the responses we record. Oh and in the paper the average response time was 225 ms (ish), which is what we expected to find.

Now that I (hopefully) summarized well enough that you have the needed background we can discuss what we found. But, first thing to note, everything in the paper is scaled to the responses when the person was not moving. Meaning a response of 100% or 1 is what happens when no movement is occurring and we did find some muscles had no real change when movement occurred so we got a ~100% response. However, we would expect to find some muscles facilitated (higher response) and some inhibited (lower response) when a person is moving. More importantly we would expect different results depending on the movement.

What we found was some very cool stuff! Now I’m trying to avoid the technical language of the paper, but that can be hard so apologies if I fail. We found facilitation in the agonist muscles (think agony, or the muscle doing the work) as early as 50 ms after the cue was given! There were some very interesting changes and it was a LOT of data, so to make it easier (or at least more compact) to read I condensed it all into cord diagrams. Figures 3, 5, and 7 in the paper show our mean response across all the participants (13 in total). We repeated the experiment so we had 10 data points for each movement and “CTI” which is the delay between the cue and stimulation, so a lot of data points!

To read the cord diagrams (the auditory cue is shown above), the bottom shows the name of the muscle and the top shows the CTI used (again, delay) the space between two tick marks at the top is scaled to 100% and the thickness of the connection tells you how much of a response we got, so if the “LVL” muscle at CTI 150 was 1 and a half tick marks thick, that means there was a 150% response! So just by looking for the thickest lines you can see the largest changes and trace them back to the muscle and CTI used. Of course we get way more complicated than that, but for the average reader I think the biggest (and easiest) things to look at are my probability maps!

After we had all those data points we could determine the probability of seeing facilitation (a higher response) or inhibition (lower response) for a given movement, CTI and cue (auditory vs. tactile). So if all 1300 or so datapoints had a higher response than our baseline (no movement) then we would have 100% chance of facilitation, if all our data points were below the baseline then we would say we had 100% chance of inhibition. We would expect on average for half to fall above and half to fall below if there were no differences between the condition and baseline, but we had 100% facilitation (and inhibition too, but only for isometric contraction) for several cases.

You can see the pattern of excitability change as the movement progresses, or in the case of isometric contraction (where we applied TSS while the movement was done and held) we can see how the “fixed” network looks. Now I’m 98% confident hospital-PI would’ve noticed this (no idea why I’m referring to him as hopital-PI since I’m outing myself for this paper, habit I guess), but I saw an interesting pattern here for all conditions and it happened a bit by luck.

When we perform dorsiflexion (as an example) we’re using our tibialis anterior muscle (shown above in green, LTA) so that is our “agonist” muscle (remember agony, doing work!) the opposite muscle (since muscles come in pairs) is called the antagonist and in this case would the soleus (shown above in pink, LSOL), which should in theory start relax to let the other muscle do its work, if you notice above for the left dorsiflexion (E) both auditory and tactile we the probability tend toward inhibition in the soleus (so we would expect to find a smaller response). This is also seen in the isometric contraction with 100% inhibition (dark pink LSOL)!

We found this type of behavior in both dorsiflexion (E) and knee extension (B). For knee extension the agonist is the LVL (left vastus lateralis, blue) and the antagonist is the LMH (left medial hamstring, orange). Notice the isometric contraction for (B) also has a similar extreme probability of facilitation for the agonist and inhibition for the antagonist. My realization, we don’t see this for plantarflexion and knee flexion! In fact we see there is a probability of co-facilitation! In fact, in the isometric contraction conditions we have almost a 100% chance of co-facilitation.

Since this is getting a little long I’ll wrap it up by saying that our findings suggest that the brain doesn’t excite the motorpools of the spinal cord, insead it downregulates them (makes it easier for them to fire). This means the spinal cord can take control of the movement using information about what the legs are doing in space via for example proprioception, or the ability for you to close your eyes and know where your limbs are without seeing them. Which means your brain doesn’t have to do all the work while you’re walking, it just lets the spinal cord do the driving. Now obviously more research is needed to confirm our findings and this is just a hypothesis based on what we found, but it’s all very exciting, no pun intended!

There was a lot of other implications to the stuff we found, but this was a much longer post than I intended already. Hopefully it makes sense to everyone, particularly those who are not in my field.

Since we believe science should be for the people this paper is open access so you can read the entire thing if you want! This is a little scary for me, but to hell with it: Characterization of Spinal Sensorimotor Network Using Transcutaneous Spinal Stimulation during Voluntary Movement Preparation and Performance

Hello world.