Author Response to Letter Regarding IOVS Publication “Investigation of Selective Innervation of Extraocular Muscle”

We appreciate the opportunity to respond to the letter to the editor submitted by Drs. Demer and Clark regarding our recent publication – “Investigation of Selective Innervation of Extraocular Muscle Compartments.”¹ We thank Demer and Clark for their interest in our work and for complimenting us on conducting these difficult experiments. We also acknowledge and thank them for inspiring us to attempt these experiments as this recording project was indeed motivated by the series of magnetic resonance imaging (MRI) based studies in humans by the Demer laboratory. 

In their letter, Demer and Clark make several points to which we respond below. Although their points are indeed interesting and provoke thought, in the end, we maintain our original interpretation that our data do not support a role for selective innervation of extraocular muscle compartments. At the outset, we point out that our study does not refute existence of compartments or any potential biomechanical effects due to compartments (we cannot speak to that) – rather we do not believe the selective compartmental behavior observed in an MRI is due to selective activity of subpopulations of classical motoneurons that behave differently from other subpopulations during specific types of eye movements. Below we discuss the issues raised in the letter by Demer and Clark. 

• 1. Demer and Clark Point 1: In their letter, Demer and Clark make an important point about compartmental differences being small and evident only after they averaged MRI data from many subjects during steady fixation in eccentric gaze. Although some of the difficulty may be attributed to the spatial and temporal resolution of the MRI methods, the underlying question is whether small differences would be identifiable in motoneuronal responses. We would note that averaging MRI data from many subjects does not amplify the amount of compartmental activity; rather it serves to average out uncorrelated noise and pull out the signal, which in this case is the compartmental contraction behavior. We point out that averaging over multiple trials of the eye movement tasks and analyzing population data from many motoneurons serves the same purpose of unearthing small, correlated activity that could have supported selective innervation.
• More importantly, the choice of the vertical smooth pursuit and vertical vergence tasks were precisely to avoid having to work with small differences in firing rates overlaid over already modulating activity of motoneurons. These are eye movements tasks where the MRI studies from the Demer laboratory showed surprising behavior of certain compartments – (i) unexpected contraction in a horizontal (medial rectus and lateral rectus) compartment during vertical pursuit or vertical vergence, and (ii) no contraction in a vertical muscle (superior oblique and inferior rectus) compartment during vertical vergence (see the second paragraph in our Discussion section and table 1 in our original paper). In these specific conditions, in order to support a role for selective innervation, activity of motoneurons would run counter to what we would normally expect (we would not expect horizontal motoneurons to be modulated during vertical eye movements and we would expect all vertical motoneurons to be modulated during vertical eye movements) and therefore should have been relatively easy to identify. These are the specific predictions that we did not find support for. In short, Demer and Clark's argument about difficulty in investigating motoneuronal behavior that is correlated with small changes would indeed have validity for many of the predictions of compartmental behavior that they have published over the years but, in our opinion, do not have validity for the specific tasks and specific predictions that we chose to test. We posit that the neurophysiological method, even with relatively small numbers of motoneurons, is sensitive enough to identify “small change” where none is expected or identify “no change” when a lot is expected.
• Our methodology indeed involved first identifying a motoneuron by its classical burst-tonic firing pattern and then testing during a particular compartmental behavior. We accept that if compartmental differences are due to a completely different population of cells that do not resemble classical motoneurons in their burst-tonic characteristics, we would not have identified them. However, it seems difficult to imagine half of the lateral rectus (LR) or medial rectus (MR) muscle being able to actively contract even a small amount without motoneuron modulation; hence our suggestion of passive mechanisms in the Discussion section. To our knowledge, there have been no publications of neurons recorded within the motor nuclei (easily identified by the beehive of background neuronal activity related to on-direction of eye movements) whose activity does not resemble classical burst-tonic response characteristics.
• 2. Demer and Clark Point 2: In their letter, Demer and Clark suggest that the activity of 2/30 LR motoneurons (approximately 6%) is sufficient to explain the approximately 6% change in LR posterior partial volume (PPV) from MRI observed during vertical vergence. They generalize to suggest that only small percentages of motoneurons would be needed to account for compartmental changes in all muscles and take issue with our predictions (in the table) that, during vertical vergence, approximately 50% of the superior oblique (SO) motoneurons and approximately 50% of inferior rectus (IR) motoneurons might be expected to behave differently from the other 50% if there were selective innervation of compartments. The source of our predictions is the anatomic data published by the Demer laboratory, which shows bifurcation of the motor nerves into two almost equal parts, each sub-nerve separately innervating each compartment. In this scenario of each nerve branch innervating approximately 50% of the muscle, for selective innervation to work, approximately50% of the muscle would be getting a different signal from the other approximately 50% which would imply that approximately 50% of the motoneurons would be behaving differently from the other approximately 50%. To reiterate, small differences might be difficult to unearth if the muscle compartments were already contracting or relaxing (see point 1 above). However, in the particular choice of tasks in our study, half the muscle is behaving unexpectedly and therefore half of the motoneurons would be predicted to follow suit.
• We do not believe compartmentalization can work with only 6% of the motoneurons behaving selectively. Even if this were a consistent finding (it is not for several other predictions in the table), it would result in only a sliver (approximately 6%) of the muscle contracting differently from the other approximately 94% during a particular movement. Simply put, the observed 6% delta change in volume that occurs over 50% of the muscle is not equivalent to 6% of the muscle experiencing an unknown delta change in volume. Our motoneuron studies were predicated on the former assertion (i.e. change in volume occurred over 50% of the muscle).
• A general observation to be made regarding the assertions within the letter of Demer and Clark about percentages of motoneurons required to explain compartmentalization is that the relationship among motoneuron firing, muscle force generation, and eye movement is quite complex, especially if dynamics were also to be considered. The PPV obtained from static MRI measures is a proxy measure of muscle contraction and does not measure force directly and that relationship may be similarly complex and nonlinear. Therefore, the percentages being discussed above are approximations. What does appear to be true is that for a large section of muscle to change shape via an active process, even by a small amount, we should expect a significant subpopulation of motoneurons to modulate appropriately. We did not find support for such modulation.
• 3. Demer and Clark Point 3: In this point, Demer and Clark suggest that our data support a role for SO compartments because of the large range of sensitivities during vertical vergence as compared to vertical pursuit. One should note at the outset that the range of position sensitivities for motoneurons is known to be large (see classic papers by the Cullen laboratory, Fuchs Laboratory, Gamlin laboratory and others).^2–6 The estimated position sensitivity values are also known to be different across different types of eye movements and they are also different during conjugate versus vergence eye movements. These differences are among the evidence that has been used to argue against a so-called final common pathway. So, we are not surprised that SO motoneuron sensitivity is different during vertical pursuit and the asymmetrical vertical vergence that we tested or that there is a wide range of values for vertical vergence sensitivity. Demer and Clark wish to argue that neurons with larger vergence sensitivity might be selectively innervating one compartment. We do not know if this is true or not, and our data cannot speak to this point. Our interpretation of predictions for selective SO innervation proposed thus far is more limited: specifically, that roughly half the SO muscle would not be receiving any modulation over background innervation during vertical vergence if that compartment did not contract during vertical vergence and if the muscle has strict independent control of compartments. However, we found no SO neurons with zero vertical sensitivity during vertical vergence in our sample, the exact opposite of the prediction.

Other points raised in the Demer and Clark letter: 

• 4. We accept that there are differences in stimulus conditions between the MRI studies and our study and of course also acknowledge potential difference between humans and monkeys. We will leave it to the readers to determine if they believe these are critical differences that explain the differences between our finding and the predictions of Demer and colleagues. However, we respectfully do not believe they are considerations because it appears unlikely that compartmental changes over an entire half of the muscle appear during 30 degrees eccentric fixation but not 15 degrees smooth pursuit. Perhaps, a biomechanical modeling approach can be used to provide some insight that is difficult to achieve experimentally.
• 5. Demer and Clark raise an important issue surrounding Listing's law violation during the VOR and the failure of cyclovertical neurons to encode such violations. Our data do not speak to this paradox as we did not perform the elegant experiments that Demer and Clark propose in their letter. A task involving a head movement and therefore the vestibular system may be different from other head-fixed tasks that only activate the saccadic, smooth pursuit, or vergence oculomotor subsystems. However, the current theories of compartmentalization appear to generalize beyond vestibularly driven eye movements, and our data do not support selective innervation under those circumstances.
• 6. We apologize for any inadvertent missing of referencing of papers from Demer and colleagues, although a couple of key studies that they suggest we ignored were indeed referenced in the introduction and used to motivate our work. Note that minimal lateral force transmission and minimal side to side junctions amongst muscle fibers noted by Demer and colleagues^7,8 (questioned, however, by some other studies)^9–11 might allow for parts of the muscle to contract independently but if the central signal to these fibers are all the same (as our study suggests), then the observation of minimal lateral connection is largely irrelevant and the two compartments will contract similarly. We readily admit that the sections in the Discussion about alternative explanations about pulleys, orbital fat etc., are speculation and need further experimental study. Certainly, additional investigation is needed to sort out key issues, such as whether compartmentalization manifests only under specific stimulus conditions, what happens when a vestibular component is introduced and whether compartmental differences and consequently the biomechanics can be accounted for by some other mechanism that does not involve selective innervation from classical motoneurons.