| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1Abteilung für Kognitive Neurologie, Hertie-Institut für Klinische Hirnforschung and 2Graduate School for Neural and Behavioural Sciences, Universität Tübingen, Tübingen, Germany
Submitted 28 April 2008; accepted in final form 31 July 2008
 |
ABSTRACT |
|---|
|
|
|
INTRODUCTION |
|---|
|
). A typical whisker stroke follows a smooth ellipsoid trajectory with its principal axis aligned with the rostrocaudal axis of the rat (Bermejo et al. 2002). This is one reason why virtually all studies conducted so far employed latitudinal (horizontal or vertical) whisker stimulation protocols. However, already Zucker and Welker (1969) reported the qualitative finding that 
50% of neurons in the TG respond to longitudinal stimulation (push or pull) of the whisker. Observations of whisker motion using high-speed videography have revealed that whisker-object contacts consist of repetitive "stick-slip" events and frequently involve the bent whisker tips rather than whisker shafts (Ritt et al. 2008) with even large whisker tip deformations having comparatively small effects on whisker base angle (Voigts et al. 2008). In addition, a preliminary report found that rats "dab" their whisker tips against surfaces of interest during exploration (Prescott et al. 2005). Most importantly, Gopal and Hartmann (2007) demonstrated the presence of significant longitudinal forces by rotating a plucked whisker against a point object with force-sensing capabilities. Notably, these longitudinal forces were observed even though the direction of the contact force was perpendicular to the long axis of the whisker—a situation typical of object contact during whisking (Mitchinson et al. 2007). Moreover, depending on whisker rotational angle and radial object distance from the base, longitudinal forces were up to eight times larger than normal forces. Taken together, these findings suggest that longitudinal forces occurring during natural whisking could generate responses at least in a subset of primary afferent neurons. These responses may be behaviorally relevant to the animal and, due to mechanical processing in the follicle, could be both quantitatively and qualitatively different from those evoked by rostrocaudal stimulation. Here we investigated the quantitative relationship between kinematics of longitudinal whisker stimulation and neuronal responses in the trigeminal ganglion.
|
|
METHODS |
|---|
|
). Briefly, rats were under intraperitoneal ketamine and xylazine anesthesia (100 and 10 mg/kg, respectively). To expose the trigeminal ganglion at the base of the skull, the overlying bone was removed, the left cerebral hemisphere was aspirated and the dura covering the trigeminal ganglion was carefully teased away. The ganglion was kept moist with physiological saline (0.9% sodium chloride) throughout the experiment.
Recordings were made with laboratory-built, pulled and ground glass-coated platinum tungsten electrodes (4–8 M
). Band-pass filtered (200–5,000 Hz) voltage traces were recorded at 20-kHz sampling rate. An automated spike-sorting algorithm was applied to sort out stimulation artifacts and multiunit activity (Hermle et al. 2004). Only single-unit spike trains entered the data set. Once a unit was isolated, its receptive field was determined by deflecting individual whiskers using a cotton-tipped probe. The corresponding whisker was trimmed to 5 mm. The whisker stump was attached to a glass capillary glued to a piezoelectric actuator. Stimuli were constructed such that the stimulator pushed the whisker along its longitudinal axis into the follicle. The shortened whisker stumps were sufficiently stiff such that longitudinal push did not result in bending; for every cell recorded, we checked that the stimulator exerted its force at an angle identical to the long axis of each whisker using a dissecting microscope. We sampled exclusively caudal whiskers (whiskers 1–4 in rows A–E, and straddlers). Each stimulus was composed of a fast half cosine wave, pushing the whisker into the follicle, followed by a 500-ms plateau and another half cosine wave at 0.5 Hz for a very slow return (stimulus waveforms are depicted in Fig. 1). The steepness of the half cosine wave was varied to achieve different peak velocities. For longitudinal stimulation, we used a total of 15 stimuli (3 different amplitudes, 95, 155, 285 µm; 5 different peak velocities, 5, 22, 43, 87, 130 mm/s, where "peak velocity" refers to the maximum velocity achieved during the onset ramp). Ten presentations per stimulus were presented in pseudorandom order at an interstimulus interval of 2.5 s.
|
20° at a contact point 3 mm from the skin); eight different peak velocities were employed (5, 22, 43, 87, 130, 174, 260, 350 mm/s corresponding to 105, 415, 825, 1,655, 2,480, 3,325, 4,965, and 6,685°/s).
|
To tease apart the contribution of variations in peak velocity and amplitude to the variation in the units' spike responses, we employed two analyses. First 
2 was used as measure of effect size, representing the proportion of the total variance that is attributable to an independent variable (IV), (sum of squaresIV)/(sum of squarestotal). 2 for an IV (here, amplitude or peak velocity) varies between 0 and 1, with 1 implying that the entire variance in the dependent variable (DV; here, spike counts) is due to variation of the IV, and 0 implying that variance in the DV is unaffected by variations of the IV. Second, for each neuron, we entered amplitude and peak velocity as predictors into a multiple regression equation with number of spikes fired in response as criterion variable. The magnitude of the beta weights for the predictors, obtained from the multiple regression equation, can be interpreted as the standardized contribution of the given predictor's variance to variation in the criterion variable. Positive values of beta indicate a positive correlation between predictor and criterion, negative values indicate a negative correlation. The relative magnitudes of beta for the two predictors can be used to infer the relative importance of the predictors in explaining a unit's spike counts. See Kline (2004) for detailed descriptions of the preceding measures. All analyses were performed in Matlab 7.0 (Mathworks, Natick, MA).
It is well known that the velocity- and amplitude-sensitivities of trigeminal units depends on the time windows over which spikes are counted (Shoyket et al. 2000; Stüttgen et al. 2006). However, as a first approach, we decided to analyze spike counts during the whole stimulus period because we sought to conduct our analyses as free from prior assumptions as possible.
|
|
RESULTS AND DISCUSSION |
|---|
|
Tuning properties of the responsive neurons are presented in Fig. 2. Panel A depicts the neurons' spike counts as a function of peak deflection velocity, averaged over amplitudes, whereas B plots the neurons' spike counts as a function of amplitude averaged over peak velocity. Response strengths covered a wide range, from 0 to >150 spikes/stimulus. Comparing the neurons that generated low versus high spike counts (cf. plots on the right in Fig. 2, A and B), revealed two well-separated groups of neurons. The first group yielded high spikes counts and increased the response with high stimulus amplitudes but did not show any dependency on peak velocity (orange, cf. Fig. 1A). The second group was composed of neurons that responded sparsely and showed the opposite preference (green, cf. Fig. 1B): they increased their response with increasing peak velocity but were relatively independent from amplitude. To work out this possible classification more quantitatively, we asked how much of the response variance of each individual neuron is explained by either peak velocity or amplitude. To this end, we calculated 2 ('proportion of variance explained', see METHODS), entering amplitude and peak velocity as independent variables and number of spikes per stimulus as dependent variable. Inspection of the scatter plot in Fig. 2C revealed that TG neurons are either sensitive to amplitude (orange; amplitude2 > 0.2) or peak velocity (green; amplitude2 > 0.2). No TG neuron was sensitive to both. The impression of a response dichotomy was confirmed by plotting the standardized regression coefficients for amplitude and peak velocity for each neuron based on multiple regression analysis (Fig. 2D). From these results, we conclude that response profiles of TG neurons can be well classified into amplitude-responsive (AR) versus velocity responsive (VR) neurons, based on amplitude2 or velocity2 >0.2 (broken lines in Fig. 2C), respectively.
|
Next, we provide three more characteristics that in addition help to differentiate the two classes of cells that we have based so far on different responses to stimulus kinematics. First and most importantly, as suggested by the raw data shown in Fig. 1, the two classes display different adaptation profiles that resembled closely the classical scheme of slowly and RA cells found with latitudinal stimulation (SA vs. RA). To investigate this quantitatively, we adapted the classification scheme developed by Lichtenstein et al. (1990) designed for latitudinal stimulation for the present purpose to separate responses to longitudinal stimuli: According to this strategy, SA from RA responses are separated by assessing the spike count during the hold phase of a fast, high-amplitude stimulus. SA responses are characterized by spike counts significantly larger than zero (1-sample t-test at P < 0.025). For the longitudinal stimulation used here we chose a test stimulus of 285 µm amplitude and 130 mm/s peak velocity and found a perfect congruence of SAlongitudinal with AR as well as of RAlongitudinal with VR classes. The same holds true for all other peak velocities at this amplitude. The second characteristic that differentiated well between the classes was mentioned earlier: spike counts of AR cells were substantially higher than that of VR cells [grand averages across all stimuli are 53 (n = 22) and 2.7 (n = 11), respectively]. A third property was the velocity threshold, which is highlighted by the application of high-amplitude, low-velocity stimuli (e.g., 285 µm, 5 mm/s, Fig. 2E). To this kind of stimulus, neurons of the VR kind (asterisks in Fig. 2E) exhibit very low responses (in fact it was 0 for 10 of 11 VR cells and one for the remaining cell). In contrast, the AR cells all generated more than one spike in response to this stimulus. In summary, we found that small-amplitude longitudinal stimuli <100 µm routinely evoke responses in trigeminal ganglion units. The TG units were found to be superbly classifiable based on responses to amplitude and peak velocity, patterns of adaptation, and presence of a response to very low velocities (5 mm/s).
Next we posed the question if membership in one of the response classes derived from longitudinal stimulation (AR/SA vs. VR/RA) predicts membership in the corresponding response class derived from classic latitudinal stimulation. To test this, we recorded a subset of neurons under both stimulation conditions. To securely surpass the amplitude threshold of SA units in latitudinal direction, we presented high-amplitude stimuli (20°) (Stüttgen et al. 2006). Applying the criteria of Lichtenstein et al. (1990), as in the preceding text, showed that seven of eight cells that classified as RA using longitudinal stimulation were correctly classified as RA using latitudinal stimulation. Nevertheless, the overall match of the two classification methods was quite poor (52% correct) due to the fact that 2/3 of SAs as assessed with longitudinal stimuli were classified as RA using rostrocaudal stimuli (Table 1, 1st row). We believe that this result does not imply a genuine difference between response properties obtained with different stimulation axes; rather we suspect that our latitudinal classification method was deficient. As reported by Lichtenstein et al. (1990), we found that some SA neurons were directionally sensitive, characterized by a RA response in one direction and a SA response in the reverse direction (stimulus return; see Fig. 3A for an example). To capture these cases, we extended our SA classification criterion to include units with significant activity during a 100-ms period starting 50 ms after stimulus end. As can be seen in Table 1 (2nd row), this procedure increased the match of the two classification methods but still left many nonmatching cases (68% correct). In a previous paper (Stüttgen et al. 2006), we reported that SAs have markedly lower velocity activation thresholds compared with RAs (>250°/s for SAs vs. >750°/s for RAs). We therefore used this feature to further improve our ability to capture SA responses. Indeed classifying units as SA that responded to the slowest stimulus (105°/s) with more than one spike on average, improved the match considerably (88% correct) (Table 1, 3rd row).
|
, their Fig. 1) and to an another complicating feature of SA units, the positional sensitivity (also mentioned in the previous study). Latitudinal stimulation, identical in its kinematic parameters but starting from different start points were observed to yield qualitatively different responses. The responses in Fig. 3, A and B, were obtained by ramp-and-hold stimuli that differed only in the starting position (difference, 1 mm). As can be seen, the SA responses to the offset ramp visible in A are markedly reduced in B, leaving the impression of a RA neuron.
In summary, our results suggest that ganglion cells and their associated mechanoreceptors display matching response profiles under latitudinal and longitudinal whisker stimulation. However, this could be demonstrated only when three different classification approaches for latitudinal stimulation—rate of adaptation to stimulus onset or offset, and velocity threshold—were taken into account simultaneously. This fact highlights a noteworthy spin-off of the present results for the practical implementation of TG cell classification: Presentation of a single longitudinal ramp-and-hold stimulus completely suffices for perfect classification simply using "rate of adaptation" as a criterion. To reach the same classification result with latitudinal stimulation, a whole battery of stimulus directions has to be applied to secure stimulation of each vibrissa in a unit's preferred direction (Lichtenstein et al. 1990). In addition, even multi-angle stimulation is likely to suffer from offset position effects (cf. Fig. 3). These multiple measurements consume valuable experimental time and require special equipment (e.g., a multi-angle stimulator) (Simons 1983). Taken together, we conclude that classification based on longitudinal stimulation is convenient, time saving and much less prone to error.
As outlined in the INTRODUCTION, the richness of tactile sensory input generated by the rat during exploration cannot be reduced to latitudinal deflections. This is the first quantitative study providing evidence that the whisker system is highly sensitive to longitudinal movement. For instance, if one imagines a hypothetical object engaging the whisker at a distance 3 cm from the face, the known SA amplitude threshold of 3° with latitudinal stimulation (corresponding to 1.6 mm at the point 3 cm from the face) (Stüttgen et al. 2006) is far outrun by the longitudinal amplitude of 95 µm, sufficient in most cases to evoke vigorous responses from SA neurons (on average 35, maximally 113 spikes). While such considerations are helpful in shaping an intuition for the different sensitivities along longitudinal and latitudinal axes, it is clear that a realistic tactile situation is highly unlikely to engage one of the whisker axes in isolation as is the case in all studies employing precise whisker stimuli to date. Because all neurons we recorded were responsive to either force vector when presented in isolation, and in addition displayed remarkably similar response profiles, it seems that trigeminal neurons cannot tell the difference between the axes of force vectors. This is surprising given the abundance of different nerve endings at several locations inside the whisker follicle (Ebara et al. 2002) and suggests that the rat must base three-dimensional discrimination on a code in just one set of neurons.
An important hint to the functional significance of longitudinal forces comes from the insight gained from biomechanical measurements that this force component acting on the whisker while being swept (in rostrocaudal direction) over surfaces is high and potentially carries significant information about object distance (Gopal and Hartmann 2007), a parameter shown to be represented neuronally and of relevance behaviorally (Krupa et al. 2001; Szwed et al. 2006). Our present results suggest that a lot can be learned about tactile signal acquisition using whiskers by studying the interplay and dependencies of information captured by longitudinal and latitudinal forces acting on follicle receptors in a concerted way.
|
|
GRANTS |
|---|
|
|
|
ACKNOWLEDGMENTS |
|---|
|
|
|
FOOTNOTES |
|---|
Address for reprint requests and other correspondence: C. Schwarz, Hertie Institut für Klinische Hirnforschung, Universität Tübingen, Abteilung für Kognitive Neurologie, Otfried Müller Str. 27, 72076 Tübingen, Germany (E-mail: cornelius.schwarz{at}uni-tuebingen.de)
|
|
REFERENCES |
|---|
|
Ebara S, Kumamoto K, Matsuura T, Mazurkiewicz JE, Rice FL. Similarities and differences in the innervation of mystacial vibrissal follicle-sinus complexes in the rat and cat: a confocal microscopic study. J Comp Neurol 449: 103–119, 2002.[CrossRef][Web of Science][Medline]
Gopal V, Hartmann MJ. Using hardware models to quantify sensory data acquisition across the rat vibrissal array. Bioinspir Biomim 2: S135–S145, 2007.[CrossRef]
Hermle T, Schwarz C, Bogdan M. Employing ICA and SOM for spike sorting of multielectrode recordings from CNS. J Physiol 98: 349–356, 2004.
Kline RB. Beyond Significance Testing. Washington, DC: American Psychological Association, 2004.
Krupa DJ, Matell MS, Brisben AJ, Oliveira LM, Nicolelis MA. Behavioral properties of the trigeminal somatosensory system in rats performing whisker-dependent tactile discriminations. J Neurosci 21: 5752–5763, 2001.
Lichtenstein SH, Carvell GE, Simons DJ. Responses of rat trigeminal ganglion neurons to movements of vibrissae in different directions. Somatosens Mot Res 7: 47–65, 1990.[Web of Science][Medline]
Mitchinson B, Martin CJ, Grant RA, Prescott TJ. Feedback control in active sensing: rat exploratory whisking is modulated by environmental contact. Proc Biol Sci 274: 1035–1041, 2007.
Prescott TJ, Mitchinson B, Melhuish C, Dean P. Three-dimensional reconstruction of whisking patterns in freely moving rats. Soc Neurosci Abstr 625.3, 2005.
Ritt JT, Andermann ML, Moore CI. Embodied information processing: vibrissa mechanics and texture features shape micromotions in actively sensing rats. Neuron 57: 599–613, 2008.[Web of Science][Medline]
Shoykhet M, Doherty D, Simons DJ. Coding of deflection velocity and amplitude by whisker primary afferent neurons: implications for higher level processing. Somatosens Mot Res 17: 171–180, 2000.[CrossRef][Web of Science][Medline]
Simons DJ. Multi-whisker stimulation and its effects on vibrissa units in rat SmI barrel cortex. Brain Res 276: 178–182, 1983.[CrossRef][Web of Science][Medline]
Stüttgen MC, Rüter J, Schwarz C. Two psychophysical channels of whisker deflection in rats align with two neuronal classes of primary afferents. J Neurosci 26: 7933–7941, 2006.
Szwed M, Bagdasarian K, Blumenfeld B, Barak O, Derdikman D, Ahissar E. Responses of trigeminal ganglion neurons to the radial distance of contact during active vibrissal touch. J Neurophysiol 95: 791–802, 2006.
Voigts J, Sakmann B, Celikel T. Unsupervised whisker tracking in unrestrained behaving animals. J Neurophysiol 100: 504–515, 2008.
Welker WI. Analysis of sniffing of the albino rat. Behaviour 22: 223–244, 1964.[CrossRef]
Zucker E, Welker WI. Coding of somatic sensory input by vibrissae neurons in the rat's trigeminal ganglion. Brain Res 12: 138–156, 1969.[CrossRef][Web of Science][Medline]
This article has been cited by other articles:
|
|
 |
|
  H. J. Chiel, L. H. Ting, O. Ekeberg, and M. J. Z. Hartmann The Brain in Its Body: Motor Control and Sensing in a Biomechanical Context J. Neurosci., October 14, 2009; 29(41): 12807 - 12814. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
TOP
ABSTRACT
INTRODUCTION
