THE MORPHOLOGY, DISTRIBUTION AND BEHAVIORAL SIGNIFICANCE OF MECHANORECEPTORS IN THE GLABROUS PAN SKIN OF SQUIRRELS By Gene Louis Brenowitz A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Neuroscience Program and Department of Zoology T978 ABSTRACT THE MORPHOLOGY, DISTRIBUTION AND BEHAVIORAL SIGNIFICANCE OF MECHANORECEPTORS IN THE GLABROUS PAN SKIN OF SQUIRRELS By Gene Louis Brenowitz The relationship between sensory specializations and behavioral specializations in two ecologically distinct species of squirrels was examined. On the basis of the behavior and natural history of these species, it was predicted that the relative density of receptors in the glabrous forepaw skin of tree squirrels (Sciurus niger) would be higher than that in ground squirrels (Spermophilus tridecemlineatus). In addition to testing this prediction, several other aspects of the distribution of receptors were quantitatively examined in silver stained material. As predicted, the relative density of receptors in the glabrous fbrepaw skin of tree squirrels was significantly higher than that in ground squirrels. Receptors were randomly dis- persed and different types of receptors (corpuscular vs. non-corpus- CUlar) were intermingled in the palmar skin of both species. The proportions of the different receptor types did not differ among species. Another series of experiments examined the role that sensory input plays in the control of food handling behavior in the two Gene Louis Brenowitz species of squirrels. First, using different size food items, it was shown that fOOd handling (rate of manipulation) was subject to sensory control in both species. Second, comparison of sham operated groups with groups receiving median nerve lesions indicated that tactile input from the volar surface of the forepaw contributes to the sensory control in the two squirrels. Third, changes in behavior over the time taken to eat large food items indicated continuous sensory feedback rather than only an initial evaluation of the food item. Fourth, as predicted from the results of the anatomical studies described above, tree squirrels depended upon tactile input from the volar surface of the fOrepaw to a greater extent than ground squirrels in the handling of food. The first series of experiments shows that the relative densi- ties of receptors in the glabrous paw skin of ecologically distinct species of squirrels can be predicted from infOrmation about their behavior and natural history. The second series of experiments indi- cates that differences in the extent to which the behavior of those species is influenced by somatic sensory input can be predicted from infbrmation about differences in the organization of their somatic sensory systems. Together, these studies indicate that a species sensory specializations and behavioral specializations are closely related and that both reflect its ecology and natural history. This dissertation is dedicated to Nathan and Ruth Flaum Brenowitz ii ACKNOWLEDGMENTS I would like to thank my advisor Dr. John A. King for his guidance throughout this study. I would also like to acknowledge Drs. Martin Balaban and John I. Johnson for their continued interest in my research and well being; it was above and beyond the call of duty. I would also like to thank Drs. Rudy A. Bernard and Glenn I. Hatton who served on my committee. Pat Cromwell, an undergraduate student in the Department of Zoology, helped me by doing histology. This research was supported, in part, by funds from the Neuroscience Program. My personal support was provided by teaching assistantships from the Department of Zoology. TABLE OF CONTENTS Page LIST OF TABLES ......................... vii LIST OF FIGURES ........................ viii INTRODUCTION .......................... l LITERATURE REVIEW ....................... 4 Introduction ....................... 4 Relationships Between Central Somatic Sensory Projections and Behavior ...................... 4 Presentation of Mapping Data ............. 7 Relating Cortical Representations to Behavior . . . . 9 The Relationship Between Central Projections and Peripheral Receptor Density ............ ll Cutaneous Mechanoreceptors ................ l3 Merkel Cell - Neurite Complexes ........... 15 Free Endings ..................... 22 Corpuscular Endings ................. 24 Dendritic Bulboid Endings ............ 24 Encapsulated Corpuscles with Inner Cores ..... 28 Physiological Properties ............. 35 The Phylogenetic Distribution of Corpuscles . . . 37 Receptor Arrays ................... 38 The Sensory Control of Behavior ............. 4l Sexual Behavior ............ A ....... 43 Aggression ...................... 45 Habitat Exploration ................. 47 Feeding Behaviors .................. 48 Behaviors not Controlled by Somatic Sensory Input . . 49 Conclusions ..................... 50 Summary ......................... Sl CHAPTER I NEUROANATOMICAL EXPERIMENTS ............ 52 . Purpose ......................... 52 Page Methods and Materials ................... 52 Subjects ....................... 52 Histological Procedures ................ 52 Skin Shrinkage .................... 53 Receptor Density ................... 54 Receptor Dispersion .................. 59 Results .......................... 6l Receptor Density ................... Bl Receptor Dispersion .................. 66 Receptor Morphology .................. 69 Intraepidermal Endings .............. 72 Dermal Free Endings ................ 82 Meissner Corpuscles ................ 82 Simple Corpuscles ................. 87 Pacinian Corpuscles ................ 95 Discussion ........................ 95 Testing a Neuroethological Hypothesis ......... 95 Receptor Arrays .................... lOl Receptor Morphology .................. l03 CHAPTER II BEHAVIOR EXPERIMENTS ............... 105 Purpose .......................... l05 Methods and Materials ................... 105 Subjects ....................... l05 Surgical Procedures .................. l06 Preparation of Food Items ............... lO6 Testing Procedures .................. lO7 Data Analysis ..................... lO8 Autopsies ....................... 109 Inter-Observer Reliability .............. 109 Results .......................... 110 Description of Food Handling ............. llO Interbout Interval .................. ll] Pre-Eating Time .................... ll9 Total Time ...................... ll9 Discussion ........................ 126 Food Size ....................... l27 Median Nerve Lesion .................. 127 Species Comparison .................. 128 ‘Ll’ A"' l l" vgvb Page The Role of Somatic Sensory Input in Controlling Behavior ...................... 130 Dynamic Weighting of Sensory Inputs ......... 132 Conclusions ...................... 134 APPENDICES A RECORDING EXPERIMENTS .................. 135 Methods and Materials .................. 135 Results ......................... 136 B COEFFICIENT OF SEGREGATION ................ 138 BIBLIOGRAPHY .......................... 141 vi Table 0501-th LIST OF TABLES Page Analysis of Receptor Densities .............. 65 Receptors Found in Squirrel Glabrous Paw Skin ...... 73 Analysis of Interbout Intervals ............. 114 Analysis of Interbout Interval by Quarters ........ 118 Analysis of Pre-Eating Time ............... 122 Analysis of Total Time .................. 125 vii Figure 1. A diagram of a Merkel cell- neurite complex ....... 2. A diagram of a Meissner corpuscle ............ 3. A diagram of a simple corpuscle with inner core ..... 4. The procedure for determining skin surface area ..... Receptor densities in squirrel glabrous paw skin . . . . 6. Receptor dispersion ................... 7. A cross section of squirrel glabrous skin ........ 8a. Intraepidermal endings in tree squirrel tubercle skin . . . 8b. Intraepidermal endingsin ground squirrel tubercle skin 9a. A Merkel cell in tree squirrel digit skin ........ 9b. A Merkel cell in ground squirrel tubercle skin ..... 10a. A dermal free ending in tree squirrel tubercle skin . . . . 10b. A dermal free ending in ground squirrel tubercle skin . . . 11. A Meissner corpuscle in a dermal papilla ........ 12a. A simple corpuscle with inner core ........... 12b. A simple corpuscle in ground squirrel tubercle skin . . . . 13. A Pacinian corpuscle in tree squirrel tubercle skin . . . . 14. Interbout Intervals in tree and ground squirrels . . . . 15. Interbout Intervals by quarters of Eating Time ..... 16. Pre-eating Time in tree and ground squirrels ...... 17. Total Time in tree and ground squirrels ......... LIST OF FIGURES viii Page . 124 INTRODUCTION Interspecific variability in central somatic sensory projections reflects variability in behavioral and ecological specializations (Nel- ker and Campos 1963, Pubols and Pubols 1972, Nelker 1973, 1976, Johnson et a1. 1974, Johnson 1978). It is also thought to reflect the differen- tial distribution of receptors peripherally (Woolsey and Fairman 1946, Mountcastle and Henneman 1952, Nelker and Seidenstein 1959, Pubols and Pubols 1971, Welker 1973, 1976). This study was designed to test the hypothesis that the distribution of cutaneous receptors differs among two behaviorally and ecologically distinct species of squirrels (Sciur- idae): one a tree squirrel, the other a ground squirrel. Tree squirrels (Sciuru§_njger) occur in open forest and forest edge habitats and use their forepaws in a broad range of skilled motor patterns that require tactile discrimination. This behavioral reper- toire includes climbing and balancing on small diameter branches and procuring and processing food items that require extensive manipula- tion (Svihla 1931, Allen 1943, Moore 1957, Reichard 1976). The ground squirrels studied were thirteen-lined ground squirrels (Sperm: ophilus tridecemlineatus) which occur in prairie and other open habitats and use the digits and palms of their forepaws in excavating extensive underground burrow systems (Evans 1951, Rongstad 1965, Desha 1966, Hildebrand 1974, Nistrand 1974). Their diet consists largely of seeds and other items that do not require extensive 2 manipulation (Whitaker 1972). These behavioral comparisons were used to predict that the relative density of receptors in the glab- rous forepaw skin (defined here as fbrepaw receptor density/hindpaw receptor density ratio) of tree squirrels would be greater than that of ground squirrels. In addition to testing this specific neuroethological hypothesis, quantitative techniques were used to examine spatial relationships within receptor arrays. Array properties examined include: propor- tional representation of receptor classes, patterns of dispersion (random, clumped, uniform) and degree of segregation between recep- tor classes. Receptor arrays in the glabrous palm skin of tree squirrels and ground squirrels were compared. Descriptions of the specific receptors found in the glabrous paw skin of these two species are presented. Preliminary results from the above study indicated that the rel- ative density of receptors in glabrous forepaw skin was greater in tree squirrels than in ground squirrels. It was predicted that elim- inating somatic sensory (tactile) input from the volar surface of the forepaw by bilaterally lesioning the median nerve would affect the behavior of tree squirrels more than that of ground squirrels. Test- ing this prediction was the main purpose of this study. In addition to testing the above prediction, more general ques- tions about the role of sensory input in controlling food handling in squirrels were examined. First, varying food size, shown to affect food handling in chipmunks (Lockner 1970), was used to examine whether food handling is subject to control by sensory input, in general. It was predicted that large fbod items would be manipulated more per unit time than small ones. Second, the lesion described above was used to determine whether somatic sensory input contributes to the sensory control of food handling. Third, data obtaind in answering the first two questions were re-analyzed to determine whether sensory input acts via an initial evaluation of the food item or continued feedback from it. It was hypothesized that if feedback occurs, fbodhandling should change as a food item is eaten and gets smaller. LITERATURE REVIEW Introduction The primary function of this literature review is to provide back- ground for the experiments presented in this dissertation. The first of three sections concerns the relationship between an animal's behavior and the organization of its somatic sensory system, especially its cere- bral neocortex. The second section consists largely of a detailed dis- cussion of the morphology, physiology and phylogenetic distribution of cutaneous mechanoreceptors. The concept of a receptor array also is con- sidered. The third section contains a general introduction to the sen- sory control of behavior and a detailed discussion of the role of tactile input in controlling behavior. I have approched this literature with an interest in the interface between behavior and sensory neurobi- ology. A secondary function of this review is to criticize methods of looking at this particular “brain-behavior" relationship. Relationships Between Central Somatic Sensory Projections and Behavior Our ability to examine the proposed relationship between an anim- al's behavior and the relative development of its central somatic sen- sory pathways (Ariéns Kappers, et a1. 1936; pp. 261-262) increased greatly when Woolsey's group began a long series of cortical mapping studies in several species of mammals (Woolsey et a1. 1942, Haynes and Woolsey 1944, Woolsey and Wang 1945, Woolsey and Fairman 1946, Chang et a1. 1947, Woolsey and Le Messurier 1948, Wool sey et a1. 1952, Pinto Hamuy et a1. 5 1956, Woolsey 1958) . These early mapping studies frequently revealed dis- parities between the relative surface areas of different regions of an animal's body and the amount of cortical surface receiving input from them. For instance, in the primary somatic sensory area (SmI) of rats (Woolsey and Le Messurier 1948), rabbits (Woolsey and Wang 1945), pigs, sheep (Woolsey and Fairman 1946), and dogs (Pinto Hammy et al. 1956) parts of the head, face and intra-oral regions are represented to a much greater extent (the representations are disproportionately large) than would be expected based on their surface area relative to that of other regions of the body. Welker and Seidenstein (1959) hypothesized that such disproportionalities in cortical representations are related to a species' behavior. On this basis, they predicted that in the raccoon's (Procyon lotor) cortical map the forepaw, which is used extensively to manipulate food items and explore other objects, would be highly represented. Their results indicate that the forepaw representation constitutes fully 68% of SmI (Welker and Seidenstein 1959, Welker and Campos 1963), supporting their original hypothesis. Present evidence indicates that in mammals the relative sizes of afferent projections from different regions of the body surface onto the cortical surface correlate with behavioral-specia1izations rather than with the relative surface areas of those regions. How- ever, this correlation is straightfbrward only when behavioral spec- ializations and disproportionalities in representations are pronounced (Welker and Campos 1963). One of the clearest examples is the en- largement of facial, peri-oral and intra-oral representations in browsing and grazing species which have hoofed limbs (Woolsey and 6 Fairman 1946, JOhnson et al. 1974, Welker et a1. 1976). In sheep, Johnson et a1. (1974) found these representations to be so enlarged that they were unable to find any body or limb representations in SmI. Enlarged representations of the furry buccal pads in capyberas (Hydrochoerus hydrochaeris) and guinea pigs (Zeigler 1964, Campos and Welker 1976), peri-oral and intra-oral tissue in rabbits (Woolsey and Wang 1945, Woolsey 1958) and the bill of the platypus (Ornith- orhynchus anatinus) (Bohringer and Rowe 1977) are also considered to be related to foraging behavior. In rats and Three-toed sloths (Bradypus tridactylus) vibrissae and forelimbs, respectively.are used in more general exploratory behavior and have disproportionately large cortical representations (Vincent 1912, 1913, Welker 1964, Welker 1971, Saraiva and Magalhaes-Castro 1975). Welker (1976) suggested that the proportions in a species' cortical map are determined by a variety of selective pressures associated with specific behavioral-ecological parameters. For species with striking disproportionalities in their maps, it can be assumed that either these selective pressures are additive or that one is dominant. In other species in which selective pressures con- flict and none is clearly dominant one might expect to find a map in which disproportionalities are less striking. Different species of monkeys (WoolSey et a1. 1942, Pubols and Pubols 1971, 1972), cats (Haynes and Woolsey 1944, Woolsey 1958, Rubel 1971), beavers (Castor canadensis) (Carlson and Welker 1976), two species in the genus Didelphis (Lende 1953, Pubols et a1. 1976, Magalhges-Castro and Saraiva 1971), gray squirrels (Siurus carolinensis) (Nelson and Sur 1977, Sur et al. 1978), porcupines (Erethizon dorsatum) (Lende 7 and Woolsey 1956) and slow lorises (Nycticebus coucang) (Krishnamurti et a1. 1976) all fall in this category. In these cases it is more difficult to correlate cortical organization with behavior. The nature of the mapping and behavioral data also contribute to this difficulty and will be considered next. Presentation of Mapping Data Mapping data are presented in four different formats, three are graphic and one is numerical. The animalcule-like figure positioned on one cortical hemisphere is the best known graphical technique (Woolsey 1958, Johnson et a1. 1974, Krishnamurti et a1. 1976). It provides a concise summary of the relative sizes of representations of the different regions of the body surface. A second approach is to plot the representations as an areal map of the cortex (Welker and Campos 1963, Pubols and Pubols 1971, 1972, Sur et a1. 1978). What these maps lack in esthetics they gain back with improved accuracy. The last graphic format is the use of figurines to depict response properties for each recording site (Woolsey 1958, Johnson et a1. 1974, Pubols et a1. 1976). Use of figurines in lieu of other for- .mats (Lende 1964, Rubel 1971) makes it difficult to visualize sizes of representations. The fourth fbrmat is the numerical statement of the amounts of cortical surface to which different regions of the body surface project. Frequently tables containing a detailed breakdown of the actual or relative sizes of cortical projections from different body regions are presented (Welker and Campos 1963, Pubols and Pubols 1971, Sur et a1. 1978). A detailed table containing relative or actual sizes of 8 cortical representations is the single most useful data format for individuals interested in correlating behavior and cortical organiza- tion. Its great advantage is that it lends itself to the designing of quantitative behavioral experiments to test predictions that one might make from mapping studies. In addition some graphic presenta- tion of data is helpful in trying to visualize both sizes and loca- tions of various body surface representations. As indicated earlier, disparities between sizes of cortical representations and the relative surface areas of those regions of the body form the basis for correlations of behavior and somatotOpic maps (Welker 1973, Johnson 1978). Surprisingly, Pubols and Pubols (1971, 1972) appear to have published the only study in which surface areas for both cortical representations and the corresponding regions of the body surface were measured. Whereas cortical areas receiving input from glabrous hand skin and tail pad of spider monkeys (Atglg§_ §E,) were similar, cortical area/skin area ratios were very different (0.39 fbr hand and 0.20 for tail pad). Thus, the hand skin is more highly represented than tail skin, despite the fact that this was not at all evident from examining the cortical map alone. These findings formed the basis fbr an interesting behavior experiment discussed in the next section of this review. In summary, our under- standing of the relationship between behavior and a species' cortical organization would benefit from a more quantitative approach to the study of cortical representation and more rigorous analysis of the relationship between central and peripheral portions of the system. RelatingCortical Representations to Behavior Relating a species' behavior to its cortical map has been approached in a number of ways. Welker and Seidenstein (1959) used information about the raccoon's extraordinary use of its forepaws to develop a testable prediction about the structure of its cortical map. Unfbrtunately, the powerful approach of testing progressively refined g_prjgri_predictions based on prior experimental work has been fairly limited in this area (Herron 1978). In other studies behavioral documentation is presented to support a proposed correla- tion, however these correlations remain largely post-hoc (Welker and Campos 1963, c. Welker 1971, Saraiva and Magalhfies-Castro 1975, Welker and Carlson 1976). Occasionally, proposed correlations rest on post-hoe reasoning and lack behavioral documentation (Lende and Woolsey 1956, Johnson et a1. 1974, Welker et a1. 1976). There are also studies demonstrating the importance of somatic sensory input in controlling several types of behavior (see discussion of this literature in section three, below), but they are not addressed to different roles for input from differentially represented regions of the body. Studies designed specifically to relate a species behavior to the differential representation of the body surface in its cortex are rare. L. Pubols' preliminary experiments comparing the spider monkey's ability to perfbrm tactile discriminations with its tail vs. its hand are an exception (Pubols 1966, Pubols and Pubols 1972). She demonstrated that whereas monkeys could perform roughness dis- criminations with both parts of the body, learning ability and per- formance with the forepaw was superior. Recall that in spider monkeys 10 the forepaw representation is almost twice that of the tail pad (Pubols and Pubols 1971). The importance of these experiments is that they show that even when cortical maps do not show striking disproportionalities one can design critical experiments to test the relationship between an animal's behavior and the organization of its cortical representations. The comparative approach, which has proven extremely useful in examining cortical maps of different species (Welker and Campos 1963, Welker 1976, Johnson 1978), has been underexploited in looking at their behavior. Two closely related yet behaviorally and ecolog- ical distinct species that use a given part of their body in differ- ent ways (e.g. the use of the forepaw by burrow digging vs. tree climbing squirrels) could be compared in controlled behavioral tests. The results from the tests could then be used to develop testable predictions about the representation of the forepaw in the cortices of the two species. One might also begin with the mapping study and make predictions about behavior. Work on species in which a disproportionate cortical representation appears to correlate with more than a single behavior (e.g. foraging and general habitat ex- ploration in rats) is needed as well. One might block somatic sen- sory input from the body part in question and compare the effects of this deficit on the behaviors in question. In summary, there is little doubt that disproportionalities in cortical representations reflect a species behavioral repertoire, however, in most cases the correlations between the two are rela- tively general. Critical experiments, relating the differential representation of the body surface (within and between animals) to IL“ \ I Lib 11 specific behaviors are needed. Also, a shift towards the use of more quantitative procedures to collect and present data would be useful. The remainder of this section will serve as a bridge between the pre- sent discussion and a review of the cutaneous mechanoreceptor liter- ature. It will help to explain the significance of examining cutaneous mechanoreceptors pgr_§g, rather than central projections from them. The Relationship Between Central Projections and Peripheral Receptor Density A cortical map is a description of the end point of an orderly projection of afferent input from peripheral receptor tissue (Celesia 1963, Werner and Whitsel 1968). Both this orderliness and the rela- tive sizes of projections from different regions of the body surface can be fbund at other levels in the medial lemniscal pathway. For several species it is known that the relative sizes of projections onto cells in the ventrobasal complex of the thalamus match the rel- ative sizes of cortical projections (Rose and Mountcastle 1952, Emmers 1965, Welker and Johnson 1965, Pubols and Pubols 1966, Pubols 1968, Cabral and Johnson 1971). Studies on the organization of the dorsal column nuclei show that the relative sizes of central projec- tions are established by this level as well (Chang and Ruch 1947, Nord 1967, Johnson et a1. 1968, Woudenberg 1970, Hamilton and John- son 1973). Generally, recording sites in the regions of enlarged repre- sentations have smaller peripheral receptive fields than sites else- where (Nord 1967, Woudenberg 1970, Rubel 1971, Pubols and Pubols 1971, 1972, Johnson et a1. 1974, Campos and Welker 1976, Krishnamurti et a1. 1976). This differential distribution of receptive field 12 sizes is considered to reflect the differential distribution of receptors peripherally (Welker 1973). The central representation of the body surface is considered more directly related to peripheral receptor density than to body surface proportions (Mountcastle and Darian-Smith 1968). A logical extension of this argument is that the distribution of receptors in the skin of an animal is correlated with its behavioral specializations. As mentioned at the outset, Ariéns Kappers et a1. (1936; pp. 261-262) had suggested these rela- tionships by 1936. Since then they have played a central role in explanations of the relationship between peripheral and central por- tions of the somatic sensory system (Woolsey and Fairman 1946, Mount- castle and Henneman 1952, Mountcastle and Darian-Smith 1968, Welker and Seidenstein 1959, Welker 1973, 1976, Pubols and Pubols 1971, 1972, Rubel 1971, Johnson 1978) and in understanding relationships between the organization of that system and behavioral specializa- tions (Welker 1973, 1976, Johnson 1978). There are, however, few experimental data to support (or negate) the proposed relationships. The most frequently cited study is that of Zollman and Winkel- mann (1962). Staining for acetylcholinesterase activity in the glabrous digital skin of raccoons, they found more positive staining loci (which they assumed were receptors) in samples from the forepaw digits than in samples from corresponding hindpaw locations. These results match what is known about the relative sizes of central representations of these regions (Welker and Seidenstein 1959. Welker and Johnson 1965, Johnson et a1. 1968). The authors were not able to morphologically identify these structures from their prepar- ations nor was the study a rigorous quantitative analysis of the 13 distribution of acetylcholinesterase activity. If their positively staining loci were simple corpuscles, as Munger and Pubols (1972 suggest, then Zollman and Winkelmann were looking at one subpopula- .tion of receptors. In a more quantitative study, Lee and Woolsey (1975) fbund that the number of neurons in the cortical barrels of mice is highly correlated with the number of fibers innervating the vibrissa fbllicle providing input to a given barrel. They did not look at morphologically identified receptors per se, Although there is some suggestive evidence, the actual relationship between periph- eral receptor density and the relative sizes of central projections remains more a hypothesis than a firmly established fact. By deduc- tion, the relationship between behavioral specializations and the differential distribution of receptors also remains a largely un- tested hypothesis. Cutaneous Mechanoreceptors The nature of cutaneous innervation and sensation are topics of long standing interest evidenced, in part, by frequent discussion and review of that literature (Adrian 1928, Weddell et al. 1955, Winkelmann 1960a, b, Melzack and Wall 1962, Weddell and Miller 1962, Sinclair 1967, Catton 1970, Munger 1971, Andres and v. DUring 1973, Winkelmann and Breathnach 1973,Ha1ata1975, Bohringer 1977, Horch et a1. 1977, Montagna 1977, Munger 1977, Smith 1977). The primary objective of this section is to provide morphological and physiological descrip- tions of mechanoreceptors fbund in squirrel glabrous skin. The in- nervation of hairy skin will be considered only indirectly, in the context of understanding receptors found in glabrous skin. .Munger (1971), Andres and v. DUring (1973), Halata (1975) and Cauna (1976) 14 have recently discussed that literature in more detail. This section will close with a consideration of the receptor array concept. My use of the term mechanoreceptor does not preclude a recep- tor's response to other forms of stimulation, only that a receptor primarily responds to mechanical forces. This usage is based on a synthetic theory of cutaneous sensation (Melzack and Wall 1962) rather than on Von Frey's theory (each sensation is subserved by a different receptor type) (as discussed in Melzack and Wall 1962) or the pattern theory (patterns of input rather than receptor morphology are responsible for sensation) advanced by Weddell (1955) and Sin- clair (1955). As Munger (1965, 1971) and Halata (1975) point out, the identification, naming and classification of cutaneous receptors have been problematic for quite some time. Frequently names have implied unsubstantiated functional properties, variants of a single receptor type are given their own names (which often correspond to the investigator's) and successive reviews change classification schemes back and fOrth. Halata (1975) recently provided a reasonable class- ification of receptors based on ultrastructural characteristics. For instance, he is able to take Botezat's (1912) 38 classes of glabrous skin receptors and with very little information loss integrate them into three types. In the discussion that follows I will adopt Halata's (1975) general principles of classification and will avoid unjustified func- tional names. To be consistent with my experimental work I will divide receptors into two classes based on light microscopic char- acteristics: non-corpuscular endings and corpuscular endings. This scheme is also similar to Cauna's (1966). The first category, 15 including Merkel cell- neurite complexes and free endings (Halata's dermal simple bulboid endings and epidermal free endings) will be discussed first. Corpuscular endings (Halata's dendritic bulboid endings, simple encapsulated corpuscles with inner cores and Pacinian corpuscles) will be discussed last. Merkel Cell - Neurite Complexes Originally, Merkel (as discussed in Munger 1965) found special- ized, large vesicular cells with large pale nuclei in the bases of rete pegs (extensions of the epidermis down into the dermis) in the glabrous snout skin of moles (Ialpa_§pp,). He reasoned that these cells acted as transducers of mechanical stimuli to the disc-like terminal expansions of neurites that he found adjacent to many of them and named them Tastzellen (touch cells). Similar endings were found in both hairy skin (Brown and 1990 1963, Cauna 1954, Mann 1965, Smith 1967) and glabrous skin (Cauna 1954, Miller et a1. 1958, Miller and Kasahara 1959a). A consistent feature of Merkel cells in mater- ial embedded in paraffin is a vacuolated cytoplasm (Munger 1965). In 1965 Munger (1965) described the ultrastructure of these cells and their associated neurites and, removing functional implications, named the unit the Merkel cell- neurite complex. The following description, based largely on his study of the snout skin of opossums (Didelphis marsupialis Virginiana) fits virtually all of the Merkel cell- neurite complexes found to date (see Munger 1971, Winkelmann and Breathnach 1973, Halata 1975 and Smith 1977 for reviews) (see Figure 1 for a schematic diagram of a Merkel cell). Large (5-10um) fibers course up through the dermis, branch in 16 Figure l. A diagram of a Merkel cell- neurite complex. Granu- lar vesicles (G) are the same as dense-cored granules. (From Iggo and Muir 1969) 17 .=3 250.3 05 89a 33ch omamdaofmo .m 3:325 633533338 .2 “363 0:0: 05 warm—.895 8:25: 4 “momenta—w CCU “Buchanan 336 .06 mg 6.3% Ergo: 05 £33 masons." a .30: :8 330.3 05 E mo~ommo> 335.8 .0 3536:: :8 3:05.30 .fl 3:888va .Q m 235808 aaoaomdn .Sm u :83 Edam—ohm“ .4 .339 3.8: 633083 mom and :8 3303 a .«o 8325...» on» 9:32? 8% < .m .maA—NQB Figure 1. O .n. nun - - Au.» v I- , I.‘ \II I i 1 g..- .I .- '1 u 1 l I ‘e ~' 18 the superficial dermis and then lose their myelin sheaths shortly before reaching the epidermis. These unmyelinated fibers then reach the dermal- epidermal (D-E) junction where the basement membranes of their Schwann cells and epidermal cells interdigitate and become contiguous with each other. The neurite is then enveloped in a unique fashion by processes of epidermal cells (Munger 1965, Halata 1975). The neurite expands into a plate or disc-like terminal and contains microtubules, neurofilaments, mitochondria and lipid mater- ial (Munger 1965), as well as lysosome—like inclusions and sometimes small vesicles (Munger et a1. 1971, English 1977b). The Merkel cell itself lies in the basal layer of the epidermis, is less electron-opaque than surrounding cells and has a highly lobulated nucleus. A prominent Golgi apparatus is present in the cytoplasm on the side of the nucleus away from the neurite. The cell's most striking characteristic is the collection of osmiOphillic, dense-cored granules (measuring roughly 10002 in diameter) adjacent to the nerve terminal. The dark core is surrounded by a membrane- limited pale halo (Mustakillio and Kiistala 1967). Munger (1965) described them as secretory granules and McGavran (1964) likened them to granules in adrenergic cells in the adrenal medulla. Other char- acteristics of Merkel cells include desmosomal attachments to adja- cent cells, finger-like cytoplasmic extensions and tonofilaments (Munger 1965, Hashimoto 1972a, English 1977b). Some authors report synapse-like structures between Merkel cells and adjacent neurites (Andres 1966, Halata 1970, 1972a, 1975, Chen et a1. 1973) but others fail to find any (Munger 1965, Munger et a1. 1971, Smith 1970). Several aspects of Merkel cell biology have attracted considerable 19 attention. The granules have been examined as possible sites for neurotransmitter storage, however, there is no consensus concerning their chemical composition (Winkelmann and Breathnach 1973, Halata 1975). Pharmacological agents known to affect standard neurotrans- mitter activity usually fail to alter activity in Merkel cell-neurite complex afferents when applied to groups of Merkel cells (Smith and Creech 1967). The suggestion that the granules indicate a trophic relationship between the Merkel cell and neurite (Smith and Creech 1967, Kasprzak et a1. 1970) is presently being examined (Cooper et a1. 1975, 1977, Diamond 1976). The role of the neurite in the devel- opment and maintenance of Merkel cells is also under investigation (Brown and Iggo 1963, Smith 1966, 1967, English 1974, 1977a, b, Tweedle 1978, Brenowitz 1978). The embryological origin of Merkel cells (do they migrate into the skin or differentiate j__situ) is a matter of debate with no conclusion established yet (Breathnach and Robins 1970, Breathnach 1971a, Lyne and Hollis 1971, Hashimoto 1972a, b, Winkelmann and Breathnach 1973, English 1977a, b, Tweedle 1978, Brenowitz 1978). Merkel cell-neurite complexes occur in several configurations in both glabrous and hairy skin. In glabrous skin they are found singly or in clusters (up to 36/cluster) at the bases of rete pegs and ridges or in the epidermis above dermal papillae (Cauna 1954, Miller and Kasahara 1959a, Halata 1970, 1975, Munger et a1. 1971, Munger and Pubols 1972, Hashimoto 1972a, Chouchkov 1974, Loo and Kanagasuntheram 1972, 1973). They are also associated with special- ized receptor structures such as Eimer's organ in the mole (Halata 1972a) and the similar rod organ in platypus (Bohringer 1977). on“ o\. f 14.. or“ 9,- ‘4 To 1. 20 Clusters of up to 50 Merkel cell-neurite complexes occupy small (200- 400um in diameter) domed shaped elevations known as Haarscheiben, touch domes, touch spots or Iggo corpuscles in the hairy skin of mammals (Brown ‘and Iggo 1963, Mann 1965, Iggo and Muir 1969, English 1974, 1977a, b). These receptor complexes also are located within the root sheath of tylotrich or sinus hair follicles (Mann and Straile 1965, Andres 1966, Patrizi and Munger 1966, Gottschaldt et a1. 1972). The Merkel cell-neurite complex is phylogenetically the most widely distributed cutaneous receptor, yet in most species they are strikingly similar. Teleost fish have cells that fulfill ultra- structural criteria for Merkel cells (Lane and Whitear 1977). Urodele (Cooper and Diamond 1977, Cooper et a1. 1975, 1976, 1977, Diamond 1976, Tweedle 1978) and Anuran amphibians (Nafstad and Baker 1973), as well as reptiles (Order: Squamata) (v. DOring 1973) have them. Nafstad (1971) found cells that resemble Merkel cells in the hard palate of birds. And, in addition to Monotremes (Quilliam and Armstrong 1963b, Bohringer 1977) and Marsupials (Munger 1965, Bren- owitz 1978) the following placental mammals have Merkel cell- neurite complexes: Insectivores (Quilliam and Armstrong 1963b, Halata 1972a, 1975), Primates (Breathnach and Robins 1970, Breath- nach 1971a, b, 1977, Loo and Kanagasuntheram 1972, 1973, Chouchkov 1974, Halata 1975), Lagomorphs (Smith 1967), Rodents (Patrizi and Munger 1966, Smith 1966, 1967, Chen et a1. 1973), Carnivores (Brown and Iggo 1963, Iggo and Muir 1969, Gottschaldt et a1. 1972, Munger et a1. 1971, Munger and Pubols 1972, Stephens et al. 1973) and Artiodactyls (Mann 1965, Lyne and Hollis 1971, Nafstad 1971). 21 Intraepidermal fibers, probably associated with Merkel cells, are found in elephants (Order: Proboscoidea) (Montagna et a1. 1975). The physiological properties of the Merkel cell-neurite complex were first established as a result of studies of the Haarschieben in cats and primates. Iggo and Muir (1969) demonstrated that these receptors are the morphological correlate of the Type I slowly adapting (SA) mechanoreceptor unit (Chambers and Iggo 1968). Its characteristics include spot-like receptive fields, irregular firing rate (ISI varies greatly) and the lack of a resting discharge. After an initial decrease in firing rate following the dynamic phase of stimulation (while a mechanical stimulus is actively displacing the skin) these units continued firing for over 30 minutes (Iggo and Muir 1969). A decrease in temperature also evokes a discharge from Merkel cell afferents (Iggo and Muir 1969). These results agree with those of Tapper (1964, 1965), Lindblom and Tapper (1966) and Smith (1967). Additional studies in glabrous skin (Janig 1971, Munger et a1. 1971, Munger and Pubols 1972) clearly establish the Merkel cell-neurite complex as the Type I SA mechanoreceptor in mammals. Recently, however, studies in amphibians indicate that their Merkel cell-neurite complexes may be rapidly adapting (RA) (Cooper et a1. 1976, Parducz et al. 1977), although these results are not yet confirmed in other laboratories. In terms of behavioral correlates, Tapper (1970) has shown that minute displacement of one Haarscheibe is sufficient to elicit a behavioral response (conditioned avoidance) in cats. ’ A ”I... i ”‘0 It. ‘Q- 22 Freg'Endings By definition free endings lack the non-neural elaborations (Merkel cells or corpuscles) present in other receptor types. They were originally described in many light microscopic studies (Miller and Kasahara 1959a, b, Winkelmann 1960a, Palmer and Weddell 1964, Giacometti and Machida 1965, Cauna 1966), however, the precise def- inition of a free terminal is difficult (Munger 1975, Halata 1975). Failure to locate non—neural elaborations can be because one is not looking at a fiber's terminal or because they really do not exist. In the epidermis free endings seen in light microscopy are almost always associated with Merkel cells (Munger 1965, Breathnach 1977). In the glabrous snout skin of opossums fibers in the epider- 'mis appear to be continuations of neurites innervating Merkel cells (Munger 1965). In other studies free endings were always found relatively close to Merkel cells (Chouchkov 1974), suggesting that they were probably part of Merkel cell-neurite complexes situated in nearby sections. Despite these reservations, ultrastructural studies of Eimer's organs in the glabrous snout skin of moles (Halata, 1972a, 1975), the platypus' rod organ (Bohringer 1977) and human hairy skin (Cauna 1973, 1977) indicate that in some instances epidermal free endings may exist. In view of the light microscopic nature of the experimental part of this dissertation, all fibers found within the epidermis will be called intraepidermal endings and no attempt will be made to subdivide this category into free endings and Merkel cell-neurite‘complexes. Dermal free endings (Halata's simple bulboid endings) are found 23 in the corium (superficial dermis) of several mammals (Munger and Pubols 1972, Hensel et a1. 1974, Halata 1975). In such endings a large (5-7um) myelinated fiber courses up into a dermal papilla (an extension of the dermis up into the epidermal region) and ramifies into a number of terminal branches. These branches can form a tangled skein of fibers (Munger and Pubols 1972) and their terminals are frequently unmyelinated (Munger and Pubols 1972, Hensel et a1. 1974). Sometimes a Schwann sheath is absent and the terminals contact the surrounding connective tissue (Munger and Pubols 1972). Halata (1975) considers the presence of clusters of mitochondria in parts of the unmyelinated branches evidence that they are terminals. In raccoons these endings are intimately associated with vascular chan- nels (Munger and Pubols 1972). ' Because much of the information about free endings is based on light microscopy and is therefore subject to the criticisms detailed above, a discussion of the phylogenetic distribution of these endings seems unwarranted and will not be undertaken. Physiologically, free endings are poorly understood. In Horch et a1. (1977) free endings are suggested as the possible morphological correlates of different types of mechanoreceptor units. On the basis of fiber diameter and position Munger and Pubols (1972) consider dermal free endings part of the somatic afferent system. They were unable to associate these endings with a specific set of physiological parameters. There is at least one study that indicates that free endings may be thermoreceptors (Hensel et a1. 1974). Because of their probable role in the somatic afferent system and their potential mechanoreceptor function dermal free endings will be treated as non-corpuscular receptors along with Lb.- 6v fill ‘1 5v (s ‘11 1 Too 24 intraepidermal fibers. Corpuscular Endings, Corpuscular mechanoreceptors consist of one or more terminal neurites associated with a complex of epithelial cells and extracel- lular connective tissue, yielding a bulb-like appearance at the end of the neurite(s). Light microscopic studies have concentrated on the variability of such endings (Miller et a1. 1958, Miller and Kasahara 1959a, b, Winkelmann 1960a, PolaEek 1961, Kadanoff and Spassova 1962, Palmer and Weddell 1964, Malinovsky 1966a, b, c). Different classical names such as Krause end-buld, genital end-bulb, Meissner corpuscle, Golgi-Mazzoni corpuscle, and mammalian end organ are given to relatively similar endings. More recently, ul— trastructural studies have made it possible to identify similarities between many of these receptor types (Munger 1971, Andres and v. Dfir- ing 1973). Halata(1975) has proposed that there are essentially two types of corpuscular endings: dendritic bulboid endings and encapsul- ated corpuscles with inner cores. While his scheme underplays actual differences it provides a useful framework for this discussion. I will describe the main characteristics of the different types of corpuscles using information about the specific variations found in squirrels to fill in details. Dendritic bulboid endings are found in the papillary dermis and consist of one or more terminal fibers which, in close association with lamellar cells, form complex structures (Halata 1975). In the glabrous skin of mammals the Meissner corpuscle is the most widespread ending in this category (see Halata 1975 for a review of the early ‘Af‘ I. V out. an. .I.‘. '1'. ..,, VV ... 1.7 . . l- 3!. a L- - 25 literature). Large myelinated fibers originating in the corial plexus course up into dermal papillae and enter the Meissner corpuscle proper (see Figure 2). These fibers frequently ramify into terminal branches which fbllow a tortuous course, winding around within the corpuscle (Cauna 1954, 1956, Cauna and Ross 1960, Miller and Kasahara 1959a, b, Winkelmann 1960a, Idé 1976, 1977). In light microscopic material it was not clear whether an outer capsule was present (Miller and Kasahara, 1959a, b, Weddell and Miller 1962). The following description of the ultrastructure of Meissner corpuscles is based largely on Cauna and Ross' (1960) original de- scription. After entering the base or side of the corpuscle, a fiber(s) loses its myelin and Schwann cell sheaths and enters into a close, often appositional relationship with laminar (lamellar) .cell processes. Synaptic structures sometimes occur (thickenings and vesicles) (Cauna and Ross 1960, Hashimoto 1973). The bulk of the corpuscle in humans and other species is composed of flattened stacks of lamellar cells which stretch across the corpuscle parallel to the skin surface and have their nuclei at the edges of the cor- puscle (Cauna and Ross 1960, Munger 1971, Andres and v. Dfiring 1973, Halata 1975, Idé 1976). Based on ultrastructural and developmental data, lamellar cells are considered modified Schwann cells (Cauna and Ross 1960, Saxod 1970, 1973, Hashimoto 1973, Idé 1976, 1977). The fiber (or its terminal branches) meanders between lamellar cell processes (extracellularly). The nerve endings are described as non-ramified, ramified with discoid expansions or ramified with varicosities (Cauna and Ross 1960, Halata 1975, Idé 1976) and are filled with mitochondria (Cauna and Ross 1960, Idé 1976). Fibers 26 Figure 2. A diagram of a Meissner corpuscle. (from Andres and v. DUring 1973) 27 Fig. 10. Schematic representation of a Meissner corpuscle showing the tonofibrilis of the epithelial cells in continuity with collagen fibres of the corium, some of which enter the upper part of the corpuscle. Others are continuous with the endoneural sheath at the basal half of the corpuscle. The tonofihril-collagen system may act directly on the receptor axon (black arrow). The white arrow indicates a possible consecutive movement of the lower part of the corpuscle- which could eliminate the mechanical stimulus. Such a mechanism could explain the rapid adaptation of this receptor. Coiled receptor axon (ra); Schwann cells (sc): cup shaped peri- neural sheath (pn); myelinated axons (ax); capillary (cp) Figure 2. 28 sometimes exit the corpuscles and terminate in the epidermis above it (Hashimoto 1973, Idé 1976). Cauna and Ross (1960) found that Meissner corpuscles in humans are surrounded by collagen fibers and fine fibrils similar to those found between lamellar cell processes. Halata (1975) confirms this finding but adds that fibrocytes may also form part of the capsule. He emphasizes that the corpuscle is not surrounded by a true perineural capsule. Tonofibrils often radiate out from part of the corpuscle to the surrounding connective tissue and epidermis (Andres and v. DUring 1973). In mice it appears that there is a cup of perineural tissue around the bottom part of the capsule (Idé 1976). Occasionally, small unmyelinated fibers are associated with these corpuscles (Idé 1976). These endings reach 100um in length by 50um in width in primates (Halata 1975) but are iconsiderably smaller in other species (Idé 1976). Meissner corpus- cles are largely confined to glabrous skin. Encapsulated endings in the rat's penis (Patrizi and Munger 1965), genital end-bulbs (Polacek and Malinovsky 1971) and Ruffini endings (Chambers and Iggo 1967, Chamber et a1. 1972, Halata 1976) are other dendritic bulboid endings found in mammals. The Grandry corpuscle, an avian mechanoreceptor consisting of specialized cells similar to Merkel cells, dendritic nerve terminals and a capsular structure (Quilliam and Armstrong 1963a, b, Munger 1966, Saxod 1970, 1975, Gottschaldt 1974,Gottschaldt and Lausmann 1974, Idé and Munger 1978) is another similar corpuscle. Encapsulated corpuscles with inner cores are the second type of corpuscular ending and can be divided into simple and Pacinifbrm 29 corpuscles(Ha1atal975). This sensory ending consists of a neurite, an inner core, a subcapsular space (also called capsular space) and a capsule (Halata 1975). Both simple and Pacinian corpuscles will be discussed. Simple corpuscles are extremely variable, even within an individual: the axon may or may not be branched, the number of axons varies, the inner core and the outer capsule may branch separ- ately or together and the overall size may vary (Malinovsky 1966a, b, c). The size and form of simple corpuscles are thought to be a func- tion of where in the body they are found and how deep in the dermis they lie (PolaEek 1961). The ultrastructural description presented below is based on Halata's (1972b, 1975) studies of moles, Munger and Pubols' (1972) study of raccoons and Mac Intosh's (1975) study of rats. I A large myelinated fiber from the corial plexus proceeds up into the papillary dermis and enters a corpuscular structure (see Figure 3) either in a dermal papilla or below an epidermal extension (Munger and Pubols 1972, Halata 1972b,l975, Bohringer 1977). As the fiber enters the corpuscle it loses its myelin and then its Schwann sheath. The fiber becomes surrounded with a variable number of lamellar cells which appear to be modified Schwann cells (Halata 1972b, 1975, Saxod 1973). Successive lamellae are separated by connective tissue (Munger and Pubols 1972, Halata 1975, Mac Intosh 1975) and there are mixed opinions as to whether these lamellae are attached to each other by desmosomal structures (Munger and Pubols 1972, Halata 1975). These lamellae constitute the inner core. There are additional, concentric lamellae which are noticably less tightly packed than those of the inner core (Munger and Pubols Figure 3. 30 A diagram of a simple corpuscle with inner core. (1) Bead- like expansion of the axon with digitate processes. The middle portion of the axon (2) runs inside the inner core. The afferent nerve fibre (3) is myelinated. The inner core is formed of a lamellar system of Schwann cells (4). The lamellae are linked by desmosome-like structure (*). The sub- capsular space (5) contains fibrocytes and collagen fibres. The capsule (6) is a continuation of the perineurium and is lined and covered with a basal lamina (1). 31 . I! 11.1. Al 11. (I l 1 M or 1. I111: iii/a, ,/l///(llhI/l/ 11 .3. .. {Mir/fl .Pfliifl/i/yl’iIJ/lli hill“ T J. Airy/(01" 10/11 ”'1qu n : 1 1 I '0 . 3.. s. 1: .t r 1 .17 l . v .n.: -. we 1 ,armeexmsav, .- o u. . 31.. , r «I 14; .. ; rl w: I I 9 :r‘ 1. .. .1 1 .. 1 Figure 3. l'n, 99- I ‘4' not, q d d od.‘ 1- 0\1 32 1972, Mac Intosh 1975) and often a subcapsular space filled with connective tissue (Halata 1975). A capsule of cellular material thought to be of perineural origin encloses this structure (Munger and Pubols 1972, Halata 1975). Within the inner core the nerve fiber(s) forms a terminal expan- sion which is filled with mitochondria and sometimes clear vesicles (Halata 19726, Munger and Pubols 1972). Simple corpuscles in moles and raccoons generally measure 15-25um in diameter and 30-50um in length (Halata 1972b, Munger and Pubols 1972). The orientation of the longitudinal axis of simple corpuscles varies from parallel to the skin surface to perpendicular to it (Halata 1972b, Munger and Pubols 1972, Mac Intosh 1975). Endings in the naso-labial region in cats (Malinovsky 1966b, PoléEek and Halata 1970), mammalian end organs (Winkelmann 1957, 1960b, Loo and Kanagasuntheram 1972, 1973), innominate corpuscles (Quilliam 1966, Quilliam et a1. 1973, Loo and Kanagasuntheram 1972, 1973), Krause end-bulbs (Spassova 1973), Golgi-Mazzoni corpuscles (Polécek 1961) and mucocutaneous end organs (Winkelmann 1960a) are other endings that can be considered variants of simple corpuscles (Halata 1975). Pacinian corpuscles are larger and more complex than simple corpuscles. They were first discovered in the mid 1700's and de- scribed by Pacini in the 1840's (see Pease and Quilliam 1957 for a review of this early literature). The structure and function of Pacinian corpuscles have been discussed frequently (Weddell et a1. 1955, Pease and Quilliam 1957, Loewenstein 1966, 1971, Quilliam 1966, Munger 1971, Andres and v. Dfiring 1973, Halata 1975). While 33 they are found in the deep dermis of the skin and in subdermal tissue (Cauna 1958, Miller et al. 1958, Munger and Pubols 1972, Malinovsky and Sommerova 1972, Brenowitz 1978), they have been most thoroughly studied in mesenteric tissue (Pease and Quilliam 1957, Loewenstein 1971, Ilyinsky et al. 1976). Based on light microscopy the corpuscle is divided into an inner core containing the nerve terminal, an in- termediate zone and a capsule (Cauna 1958, Cauna and Mannan 1959, Miller et al. 1958a, Quilliam 1966). The description of Pacinian corpuscle ultrastructure presented below is based on Pease and Quil- liam's (1957) original description of corpuscles from the cat's mesentery. A large myelinated fiber (5-10um) enters the Pacinian corpuscle at its base, loses its myelin sheath and proceeds further into the corpuscle before losing its Schwann sheath and becoming surrounded by lamellae. The fiber terminates in an ellipsoid expansion twice the diameter of the fiber and contains numerous mitochondria. The lamellae, once again, are thought to be modified Schwann cells (Cauna 1958, Halata 1975). The inner core is composed of closely packed cytOplasmic lamellae which are bilaterally organized into two opposing groups on either side of the nerve. The nuclei of the lam- ellar cells are found at the outer edge of the core. Because of this arrangement of lamellae, there is a symmetrical bilateral cleft run- ning parallel to the ellipsoid nerve terminal in which no lamellae are found. Often blebs of the neural terminal extend into these clefts (Pease and Quilliam 1957, Nishi et a1. 1970). Vesicles are sometimes found in the terminal as well. Next, there is an intermediate zone which surrounds the inner .uv {OI 1. I... '2 . 1 '1 udh ‘II 1 ‘1 34 core (Munger and Pubols 1972) and appears to be the site from which the corpuscle grows during development (Pease and Quilliam 1957). In this zone, lamellae are much less tightly packed and are con- centrically arranged, with several cells contributing overlapping processes to each lamella so that no cleft is present. Inter-lamellar spaces are filled with connective tissue, including collagen fibrils (Pease and Quilliam 1957, Halata 1975). The outermost structure is the capsule, which was initially described as being composed of approx- imately six lamellae tightly packed together and underlying a conden- sation of connective tissue (Pease and Quilliam 1957). The outermost layer of the capsule is now thought to be composed of tightly packed layers of perineural cells (Halata 1975). Halata (1975) found that in general the deeper a Pacinian cor- puscle lies from the surface, the greater the number of layers in the capsule. Several studies show that Pacinian corpuscles can take on rather complex shapes in the skin and can reach sizes greater then 1mm long (Cauna and Mannan 1959, Brenowitz 1978). In addition to the large mechanoreceptor fiber, smaller, unmyelinated noradrenergic fibers have been seen entering the corpuscle in the cat's mesentery (Santini 1968). Corpuscles also contain vascular profiles, especially in the outer portion (Cauna 1958, Nishi et a1. 1970). Herbst corpuscles in the skin and other tissues in birds are strikingly similar to Pacinian corpuscles but are much smaller. They have an inner lamellar core surrounding a central unmyelinated fiber, an intermediate zone and ia capsule (Quilliam 1963, 1966, Quilliam and Armstrong 1963a, b, qunger 1966, Anderson and Nafstad 1968, Saxod 1970, 1973, 1975, Gottschaldt and Lausmann 1974). .a'w C l in F o '4‘: run Alp ’v- i. V an I!) n H. " V. N a ‘p 1 H ‘1 35 Physiologicalgproperties have been correlated with morpholog- ical characteristics for a relatively small number of corpuscular endings. Generally, the physiological response properties of a pri- mary afferent fiber are determined and the piece of skin containing its receptive field is examined histologically. By repeated sampling it is possible to identify the receptor type present in all or most pieces of skin associated with a particular set of response properties. These tehcniques are, however, subject to sampling error. Pacinian corpuscles have been studied singly jn_yit§9_and are an exception. The most thoroughly studied dendritic bulboid ending is the Ruffini corpuscle which has been identified as the morphological correlate of the Type II SA mechanoreceptor in hairy skin (Chambers and Iggo 1967, Chambers et a1. 1972, Biemesderfer et a1. 1978). It has a resting discharge, a larger receptive field than the Type I SA receptor (Merkel cell-neurite complex) and a regular firing rate (ISI is uniform): The discharge rate during the dynamic phase of mechanical stimulation is a function of velocity and final amplitude of displacement, contrary to Horch et a1.'s (1977) categorization. Ruffini corpuscles also respond to changes in temperature (a decrease in temperature increases discharge rate) (Chambers et a1. 1972). The Grandry corpuscle, which I include in the dendritic bulboid category, is a rapidly adapting (RA), receptor that is sensitive to the velocity of stimulation. It is not sensitive to displacement amplitude, lacks a resting discharge and acts as a vibration detec- tor (1 to 1 following of stimuli up to approximately ZOO/sec) (Gottschaldt 1974). In the absence of information about other re- ceptors in this category it is premature to draw any general Q lu‘fl'i oh i' ,. 11.: .t .‘.' \ velv- Op; A i‘uh .;,., \ I '5 0A”. VI '8‘ a... GI '0 "v 36 conclusions about their physiology. Among simple corpuscles with inner cores, the simple corpuscle of raccoons (Pubols et a1. 1971, Munger and Pubols 1972) and the Krause end bulb-like ending in cats (Iggo and Ogawa 1977) have been studied. Both are associated with RA afferent units responding to the dynamic phase of mechanical stimulation with a velocity sensitive discharge rate. The simple corpuscles of raccoons do not respond to temperature changes and these data are not given for Krause end bulbs. Tuning curves for sinusoidal vibratory stimulation of both endings show that threshold amplitudes are lowest at 20-100 Hz (Munger and Pubols 1972, Iggo and Ogawa 1977). ReSponse properties of Pacinian corpuscles are relatively well known (see Loewenstein 1966, 1971 and Ilyinsky et a1. 1976 for re- views). They are extremely rapidly adapting and follow stimulation rates of up to 1000 Hz with a 1 to 1 response. Tuning curves for Pacinian corpuscles have best frequencies of 100-300 Hz (Sato 1961, Iggo and Ogawa 1977) and they are described as vibration detectors (Sato 1961, Hunt 1961, Loewenstein 1966, Halata 1975). These tuning curves have been shown to be temperature sensitive (Sato 1961). Loewenstein (Loewenstein and Mendelson 1965, Loewenstein 1966) and Ilyinsky et a1. (1976) have demonstrated that the outer capsule with its lamellar structure is responsible for the rapid adaptation rate and the "off" response seen when a stimulus is withdrawn from a cor- puscle. The ellipsoid shape of the unmyelinated terminal (Pease and (fililliam 1957) appears to be responsible for directional sensitivity ir1 Pacinian corpuscle responses. This shape maximizes the efficiency 37 with which a mechanical stimulus is transduced by the terminal (Ilyinsky et al. 1976). The corpuscular structure also acts as a mechanical filter and is important in transferring mechanical dis- placements to the terminal (Loewenstein 1966, 1971). The unmyelin- ated terminal part of the nerve fiber is responsible for producing a generator potential as well as an action potential (Gray and Sato 1953, Sato and Ozeki 1966). Recently, Gottschaldt (1974) found that Herbst corpuscles which are morphologically similar to Pacinian corpuscles share many of the physiological properties with the lat- ter. For instance, they are RA vibration detectors capable of 1 to 1 following of stimulation rates up to 500 Hz. The phylogenetic distribution of corpuscles has been presented, in large part, in describing these endings. The discussion that follows will focus on establishing the presence of receptors in various groups and will not contain exhaustive lists of studies on a particular group (this applies to mammals especially). Lamellated receptors with inner cores occur in the skin of Anuran amphibians (Bolgarskij 1964, v. Dfiring and Seiler 1974) and appear similar to mammalian simple corpuscles. Reptiles (Order: Squamata) have a variety of encapsulated endings that resemble Ruffini and Pacinian corpuscles of mammals and Herbst corpuscles of birds (v. DUring 1973, Jackson 1977). As mentioned above, birds have dendritic bul- boid corpuscles known as Grandry corpuscles and Herbst corpuscles uniich resemble Pacinian corpuscles (Quilliam and Armstrong 1963a, 1). Gottschaldt 1974). They also have simple corpuscles with inner cores in their eyelid skin (Malinovsky 1968). 38 Dendritic bulboid endings occur in the following mammals: Insectivores (in hairy skin: Halata 1975), Primates (Cauna and Ross 1960, Munger 1975, Biemesderfer et a1. 1978), Rodents (Idé 1976, 1977), Carnivores (Malinovsky 1966a, Chambers et a1. 1972) and Artiodactyls (pig hairy skin: Halata 1975). Simple corpuscles with inner cores are more widely distributed and are found in: Monotremes (Quilliam and Armstrong 1963b, Bohringer 1977) and Marsupials (Brenowitz 1978) and in Insectivores (Giacometti and Machida 1965, Halata 1972b), Primates (Loo and Kanagasuntheram 1972, 1973), Rodents (Patrizi and Munger 1965, Mac Intosh 1975), Carnivores (Winkelmann 1960b, Palmer and Weddell 1964, Malinovsky 1966a, Munger and Pubols 1972) and Artiodactyls (Malinovsky 1968, Quillaim et a1. 1973). Pacinian corpuscles are the most restricted corpuscles and have been identi- fied in the skin of Marsupials (Brenowitz 1978) and in Primates (Cauna and Mannan 1959, Kanagasuntheram et a1. 1971) and Carnivores (Nilsson and Skoglund 1963, Munger and Pubols 1972). Receptor Arrays This review of the mechanoreceptor literature has concentrated on descriptions of individual types of receptors. In this regard it is an accurate reflection of the major emphases of investigators working in this area. As Munger (1971) suggests, understanding these receptors is a prerequisite for understanding cutaneous sensation in general. However, a natural tactile stimulus such as a food item, 13 potential mate, or a burrow entrance, will impinge upon large num- bers of receptors as it moves across a patch of skin (or the animal "lives the patch of skin across it). The individual receptors that 39 a stimulus excites will depend on the mechanical pr0perties of the skin (Cauna 1958, Quilliam 1966, 1975, Halata 1975) and upon the dis- tribution of receptors in the piece of skin being stimulated (Quil- liam 1966, 1975). Quilliam (1966, 1975) has approached the question of distribu— tion of receptors by considering complex, highly specialized aggrega- tions of receptors such as the Eimer's organ in the snout skin of moles which consists of free endings, Merkel cell-neurite complexes and simple corpuscles (Quilliam and Armstrong 1963b, Halata 1975). He has applied the term "array" to these orderly arrangements of re- ceptors. Other eXamples of complex arrays are the rod organ or the platypus (Quilliam and Armstrong 1963b, Bohringer 1977), the bill organs of geese (Gottschaldt 1974) and sinus hairs in mammals (see Halata1975 for a recent review). Quilliam has also applied the term to groups of receptors found in the ridged digital skin in primates (Quilliam 1975). In a larger sense, any patch of skin can be consid- ered to have an array of receptors, although defining units in the array and the pattern of the array will be more difficult where re- ceptors are not organized into striking, highly specialized and lo- calized aggregations. While the significance of specialized arrays is widely recog— nized (Quilliam 1966, 1975, Quilliam and Armstrong 1963b, Cauna 1958, Munger 1971, Pubols et a1. 1971, Andres and v. DUring 1973, Halata 1975, Montagna 1977), little research into the nature of these more diffuse arrays exists. In addition to understanding the relation- ships of individual receptors to other skin structures (Cauna 1954, (Mlilliam 1966, 1975, Halata 1975), the only array characteristic 40 examined in detail is the density of receptors (number per unit area) (Miller et a1. 1958, Miller and Kasahara 1959a, Fitzgerald 1961, Janig 1971, Gottschaldt and Lausmann 1974). There is also some data concerning the relative proportions of different receptor types in a variety of locations (Malinovsky 1966a, b, c, Malinovsky and Zemanek 1969, Gottschaldt and Lausmann 1974). Other aspects of distribution such as dispersion patterns (random, clumped or uniformly distribu- ted) and the degree to which different receptor types are segregated from each other are unknown. The advantage of having several different receptor types in a piece of skin is that they can act in combination and thereby re- spond to broad ranges of environmental stimuli (Quilliam and Arm- strong 1963a). While understanding individual receptors is important (Munger 1971), Freeman (1976) argues that understanding spatial rela- tionships between elements in neural systems and their interactions are crucial as well. He suggests that there are properties of neural systems that are not predictable on the basis of known properties of individual elements within those systems, but are predictable when complex interactions of elements are considered. Loewenstein (1966) suggested that our emphasis on individual elements within the per- ipheral somatic sensory system has led to an increase in entropy in that area. While only a starting point, attempts to define proper- ties of arrays rather than single elements within arrays seem like steps in the right direction. I‘." r- 5” s... I. ebb h: a b . '1 ii 1‘11 all ('81- ‘U‘. AN 5 r. l“ 41 The Sensory_Control of Behavior The role of sensory input in controlling behavior played a central role in the historical development of the field of animal behavior. Ethologists were particularly impressed by "instinctive" "reflex-like" Fixed Action Patterns (FAP's) which they considered immune to the effects of sensory stimuli (Lorenz 1950, Tinbergen 1951, Eibl-Eibesfelt 1970). They also concentrated on finding specific features in patterns of sensory input (such as a mother gull's beak) that elicit behavioral responses (Sign Stimuli) (Lorenz 1950, Tin— bergen 1951, Tinbergen and Perdeck 1951, Hailman 1967). It is ironic that Lorenz (1950) described FAP's as reflex-like for physiologists had cited reflexes as evidence that behavior (locomotion) is controlled primarily by peripheral sense organs (Sherrington 1906, 1910, Gray 1950). A question that developed as a result of early behavioral studies, as well as physiological work is to what extent is behavior controlled by the central nervous system and to what extent is it controlled peripherally (by sensoryinput)? Bullock (1961) considered behavior to be primarily centrally controlled with sensory input serving to trigger or modulate centrally generated activity. Exper- iments such as those by Hamburger and Balaban (1963) showed that rhythmic motor patterns occur prior to the time sensory-motor hookup occurs. Studies on vertebrates and invertebrates showed that patterns of activity could also be maintained in preparations deprived of ex- isting sensory input (Ikeda and Wiersma 1964, Kennedy et a1. 1966, Fentress 1973, Edwards 1977). On the other hand, there is consider- able evidence that sensory input is important in the development and —-——-— H—b._h _———h '- 1 fl 1. .1. J" \u ud'J‘ o'er we r . -::r: U h .FAF- I ‘qu "'13! - a re 1 U .y . ‘ 9"'1 “VI. a 1"! 1' l .11: P5,} 'w'. 1/$ 42 maintenance of behavior (Konishi 1965, King 1968, Gottleib 1971, 1976, Duysens 1977). Gray (1950) questioned whether central control had any role in controlling behavior (locomotion in amphibians). In practice, early and recent models of the organization and control of behavior involve roles for peripheral and central control mechanisms (Weiss 1950, Lorenz 1950, Tinbergen 1951, Andrew 1976, Baerends 1976, Dawkins 1976, Fentress 1976). Fentress (1976) has proposed that the much debated boundaries between central and per- ipheral control are dynamic rather than static. The relative con- tributions made by these two sources of control and, therefore, the precise location of the boundary lines depend upon a host of inter- acting factors including an animal's specific motivational state. One might carry his argument further and suggest that the relative contributions of the different sensory modalities to the sensory control of behavior also are variable depending on specific sets of circumstances. For instance, when a pigeon is able to see the sun it will use it to navigate with, however, when the sun becomes clouded over the pigeon is able to switch to other sensory cues in order to continue its flight (Keeton 1974). To study the role of sensory input in controlling a behavior one can change the stimuli reaching an animal or one can alter an animal's ability to detect that input (by altering the animal) . (Beach and Jaynes 1956, Welker 1964, Konishi 1965, Marler 1970, Webster and Webster 1971, Kow and Pfaff 1976). The most commonly employed experimental manipulation fer studying the role of tactile (somatic sensory) input in controlling behavior is cutting that input out all together. This can be done-for short periods of time with 43 local anesthetics or for longer periods by lesioning peripheral nerves or central structures. While the primary effects of these procedures appear to be somatic sensory (largely tactile) deficits, other sen- sory deficits generally are not looked for. Stimulation studies to examine motor changes are uncommon. In studies using local anes- thetics independent tests for the effect of the drug (e.g., record- ings) are not routinely performed. While these may be minor consid- erations, they suggest that more thorough definition of lesion effects would add weight to conclusions about the role of tactile input in controlling behavior. Bearing these considerations in mind, somatic sensory input has been shown to play a role in controlling sex be- havior, aggression, "predatory behavior," feeding behavior and habi- tat exploration. Each of these will be discussed in turn. Sexual Behavior The first category to be discussed is sexual behavior which has been more thoroughly investigated than any other behavior. In male cats sectioning the dorsal penile nerve (Cooper and Aronson 1962) and removing lumbosacral spinal cord segments (Root and Bard 1937) eliminate sensory input from the penis. They do not interfere with normal penile erection. Lesioned animals were capable of ejac- ulation and showed normal sexual excitement, however, their ability to guide their penises into place for intromission was reduced. Similar results were obtained with rats when they were given local anesthetics in the penis (Carlsson and Larsson 1964, Sachs and Bar- field 1970). In rats an important difference is that anesthetiza- tion interferes with normal penile erection (Carlsson and Larsson 44 1964). Sachs and Barfield (1970) found that while tetracaine pre- vented intromission it markedly increased mounting behavior. Input from the penis is necessary for modulating or orienting the motor patterns concerned with normal intromission. Lesioned animals com- pensate for the deficit by increasing motor output. Tactile stimulation of the lower back, rump, flanks and perineum initiates the reflex-chain leading to lordosis in female rates (Gerall and McCrady 1970, Diakow et a1. 1973, Pfaff et a1. 1974). Deaf, blind, anosmic females show strong lordosis reflexes in response to either a male's mounting or manual stimulation. Extensive cutaneous denervation or local anesthetization of the areas listed above re- duces lordosis reflexes under most hormonal regimes (Kow and Pfaff 1976). The length of time a female remains in lordosis is also de- pendent upon tactile input. Desensitization of the cervix via pel- vic nerve section shortens a female's time in lordosis, indicating that stimulation to the cervix maintains the reflex (Diakow 1970). Normally, the further through a mating sequence a female rat is allowed to proceed, the longer the interval before it will seek out additional sexual contact (Bermant and Westbrook 1966). Swabbing the genital region with lidocaine significantly reduces this interval and indi- cates that adequate stimulation temporarily inhibits further sexual behavior. The above results indicate that somatic sensory input (and cer- vical stimulation) initiates the normal mating sequence (from point of contact on). A female's failure to produce a normal lordosis re- flex results in a decrease in its mate's intromission performance (Kow and Pfaff 1976) and therefore seriously compromises the entire 45 mating sequence. Continued sensory input also plays a role in mod- ulating the timing of the motor patterns concerned with lordosis as well as in motivational components of sex behavior. In light of the importance of somatic sensory input in the con- trol of the mating sequence, it is particularly interesting to note results of studies on the effects of estrogen treatment of ovariectem- ized females. Estrogen replacement has been found to increase the size of the receptive field of the pudendal nerve (the entire nerve) to include regions on the hind legs which are stimulated (actually palpated) by males during mounting sequences (Komisaruk et a1. 1972, Kow and Pfaff 1973a, b). Receptive field size in untreated ovar- iectemized females are significantly smaller and generally do not include these hind-leg sites. These results might be taken as an indication that through hormonal influence females are maximizing their chances of receiving adequate stimulation to enable them to proceed through the mating sequence successfully. Aggression Tactile input plays an important role in the control of inter- male aggression in rats and mice (Flory et a1. 1965, Bugbee and Eichelman 1972, Thor and Ghiseli 1973a, b, 1974, Katz 1976). While studying the effects of visual impairment on inter-male aggression in rats, Flory et a1. (1965) noted that removal of vibrissae decreased levels of aggression beyond that of blinding alone. This finding is important because the deficit produced by cutting vibrissae off should be strictly tactile and moreover is reversible, as the vibrissae grow back out. In fact, Bugbee and Eichelman (1972) showed that 46 removing vibrissae in male rats significantly reduced the number of attacks in a shock-induced aggression test, but as the vibrissae grew back out aggression increased again. Essentially one can titrate levels of aggression by adjusting vibrissae length. Bilateral local anesthetization of the vibrissal regions produces deficits similar to those of vibrissae removal (Thor and Ghiselli 1973a, b). Reduction in aggression has also been demonstrated in tests where male rats receiving local anesthetization of the vibrissal pad and male mice receiving local anesthetization plus vibrissae removal were allowed to interact with male conspecifics without artificial induction of aggression (shock or drug) (Thor and Ghiselli 1973a, Katz 1976). In mice these procedures reduced the number of aggres- sive encounters and increased the latency to aggression. They did not alter more general social contact (Katz 1976). In summary, the effect of altering tactile input via the vibrissae is to inhibit specific motor patterns associated with aggression. Thor and Giselli (1973a) suggest that a rat uses vibrissal input to orient towards its opponent. In this case tactile input would play a modulating role. However, evidence on attack behavior presented in the next part of this discussion as well as Katz's (1976) study indicate that vibrissal input may also serve to trigger aggressive behavior. The role of somatic sensory input in controlling "aggression" has also been examined in experiments on hypothalamically induced pre- datory attack behavior in cats (MacDonnelland Flynn 1966, Flynn 1967, Flynn et a1. 1971, Bandler and Flynn 1972). While these studies suf- fer from use of extremely biased subpopulations of animals and bear a very tenuous relationship to adaptive (naturally occurring) behavior, 47 they contain some information of potential use in understanding the tactile (somatic sensory) control of behavior. Sectioning sensory branches of the trigeminal nerve reduces biting attacks against rats in hypothalamically stimulated cats. There are also data to suggest that peripheral somatic sensory receptive field prOperties may be altered by changing the status of the CNS. Stimulation of sites on the forepaws that do not normally elicit a striking reflex, produces this reflex when hypothalamic sites are stimulated (Bandler and Flynn 1972). The role of somatic sensory input in attack behavior is considered one of triggering a reflexive motor pattern (jaw move- ments) rather than orienting the animal towards its prey (modulat- ing motor patterns) (MacDonnell and Flynn 1966). Habitat Exploration Somatic sensory input has been implicated in the control of non-social behaviors, as well. Vincent, who studied tactile hairs (vibrissae) in rats (Vincent 1913) examined the effect of vibrissae removal on open maze running and tactile discrimination of surfaces (Vincent 1912). These behaviors should be relevant to general hab- itat exploration in this species. Animals without vibrissae took slightly longer to learn mazes, had a higher number of errors in their performance, moved through the maze more slowly and slipped and fell from it more frequently than animals with their vibrissae intact. Animals with their vibrissae were also able to learn tac- tile discriminations more rapidly and Spent less time in a given trial than animals without vibrissae. In an analysis of sniffing behavior in rats, Welker (1964) found that deprivation of normal snout somatic 48 afferent input decreased an animal's efficiency in finding food pellets but did not reduce vibrissae movements. In these instances tactile input apparently acts in orienting and modulating motor out- put rather than initiating it. In both cases preventing normal input increases the time taken to perform a given task, indicating a gen- eral decrease in efficiency. FeedinggBehavior Zeigler and co-workers have been examining the role that tac- tile (and proprioceptive) input plays in controlling feeding behavior in pigeons. They have looked at the organization and response pro- perties of trigeminal structures (Zeigler and Witkovsky 1968, Silver and Witkovsky 1973, Witkovsky et a1. 1973, Zeigler et a1. 1975) and have analyzed the effects of trigeminal lesions on both motiva- tional and sensorimotor components of feeding behavior (Zeigler 1973, 1974, 1975a, b, Zeigler and Karten 1974, 1975, Zeigler et a1. 1975: see Zeigler 1974 and 1976 for reviews). Their findings concerning sensorimotor components of feeding behavior are of particular inter- est. Normal feeding behavior in pigeons consists of three sets of motor patterns: pecking, mandibulating and swallowing (described in Zeigler 1974). Pecking is the downward movement of the head and terminates when the beak is opened and a food item is contacted. Mandibulating involves moving the food item from the front of the beak to the rear of the buccal cavity where it is then swallowed. Response properties and receptive field orientations of neurons in the nucleus basalis (a second order forebrain structure in the 49 trigeminal afferent pathway) indicate that tactile and proprioceptive input should be important in the mandibulating process (Witkovsky et a1. 1973). Peripheral trigeminal deafferentation does not affect general pecking responses in operant tasks (Zeigler 1975a), however, cine- matographic analysis indicates that pecking accuracy with grain feed is somewhat impaired (Zeigler et a1. 1975). Grasping of the food item and subsequent mandibulating are seriously impaired, resulting in an increase in the number of pecks needed to successfully consume a single grain of food (Zeigler 1974, 1975b). Swallowing appears to be unaffected if the food item can be moved into position at the back of the buccal cavity (Zeigler et a1. 1975). These changes occur after an initial post-lesion period characterized by depression of both moti- vational and sensorimotor components of feeding behavior (Zeigler 1975b). In conclusion, lesioning trigeminal structures leads to a short term decrease in motor output followed by a longer period dur- ing which there is an actual increase in motor output. This increase is due to a decrease in the efficiency of individual movements (e.g., the number of pecks needed to successfully eat one grain increases). The role of tactile (and proprioceptive) input in feeding behavior in pigeons is one of orienting and guiding motor output rather than triggering it. Behaviors Not Controlled by Somatic Sensory Input Somatic sensory input apparently has little or no effect on some behaviors. Beach and Jaynes (1956) examined the effects of enucleation, trigeminal deafferentation and anosmication on maternal r31 ‘1 3.1 50 retrieval of young in rats. Trigeminal deafferentation and its com- bination with enucleation both had only minor effects in comparison to anosmication. Fentress (1973) has shown that removal of the entire forearm of mice does not alter the execution of motor patterns used by mice in grooming their head and face. In normal animals, as the forelimb moves over the eye region, the eye is closed for protection. In forelimb amputees as the stump is moved in a fashion that would have brought a paw into position over the eye, the eye is still closed despite the absence of any physical contact with that area. Fentress uses these observations as an example of the central control of behavior (Fentress 1976). Conclusions In conclusion, somatic sensory (tactile) input plays an impor- tant role in the control of a variety of different behaviors, but its role is varied. In the cases of lordosis in female rats, "predatory attack" in cats, and possibly inter-male aggression in rats and mice, this input serves to initiate or trigger behavioral sequences. Eth- ologists would consider this a "releaser" role (Lorenz 1950, Tinbergen 1951). In these cases, failure to receive proper sensory input re- sults in an overall decrease in motor output such as a decrease in attacking behavior, a shortening of the time spent in lordosis, or a decrease in the lordosis quotient. On the other hand, in mounting and intromission by male rats and cats and feeding behavior in pigeons somatic sensory input serves to orient or modulate ongoing behavior. Failure to receive normal input appears to increase motor output in these situations: pecks/food grain increase and intromission attempts “qr-AM .CW I a red ansr vy“' rr‘ 115v c b2‘1'r a: 1 ‘1 9A.. — 'N-vv 51 become more vigorous. This increase can be taken as an indication of a reduced efficiency of individual movements. In the case of rats in open mazes and in tactile discrimination tests (Vincent 1912) it is not clear what is happening with respect to actual motor output. Summary In Stmnary, this literature review has concentrated on three basic areas and their inter-relationship. The first section estab- lished a relationship between an animal's behavior and the organiza- tion of its somatic sensory system, and set up the hypothesis that re- ceptor distribution should be predictable based on a species' behavior. The second section provided detailed descriptions of those receptors and considered the importance of receptor arrays to behavior. The third section examined the varied role that somatic sensory (tactile) input plays in controlling behavior. CHAPTER I NEUROANATOMICAL EXPERIMENTS Purpose The primary purpose of this study was to test the prediction that the relative density of receptors in the glabrous forepaw skin of tree squirrels would be greater than that of ground squirrels. Additionally, this study used quantitative techniques to examine spatial relationships within receptor arrays. Methods and Materials Subjects Subjects were 13 tree squirrels and 12 ground squirrels cap- tured in the vicinity of East Lansing, Michigan. Seven individuals per species were included in the quantitative analyses described be- low. Histologjcal Procedures Squirrels were anesthetized with sodium pentobarbital and per- fused intracardially with 0.9% saline solution followed by 10% form- alin or 10% neutral buffered formalin, in 0.9% saline. Neutral buf- fered formalin left skin less brittle than unbuffered formalin and was used on all animals included in quantitative analyses. Fore- paws and hindpaws were removed and immersed in fixative for at least 24 hours. In this study only glabrous skin from the ventral surfaces 52 53 of the paws was examined. Blocks of tissue were taken from four locations: 1) forepaw digit 3 or 4 (digit 1 is greatly reduced in size and tree squirrels often damaged digits 2 and 5 while attempting to escape from live traps), 2) ferepaw palmar tubercle 3, 3) hindpaw digit 3 or 4, and ' 4) hindpaw plantar tubercle 3. The side of the animal (left vs. right) and, where appropriate, the digit number for each block were completely randomized. Blocks of tissue were dehydrated through alcohols to xylene and embedded in paraffin (Paraplast). Serial sec- tions were cut perpendicular to the skin surface at a thickness of 15 um and sections totalling 1 mm (67 sections) were affixed to slides with a gelatin-albumin solution (Harleco). Sections were then stained with a modified Bielschowsky silver stain (Sevier and Munger 1965) using 2 drops of 37-40% formalin instead of 10 drops of 4% formalin as originally indicated (Munger, personal communication). Skin Shrinkage Because of the quantitative and comparative nature of this study, shrinkage in tree squirrel and ground squirrel skin due to paraffin processing was compared. One randomly chosen tissue block was taken from each of 7 squirrels per species. The length of each block was measured with an ocular micrometer in a dissecting micro- scope and then with vernier calipers. Tissue blocks were dehydrated, cleared and infiltrated with paraffin according to a schedule used in the quantitative portions of this study. Each block was remeasured and percent shrinkage for ocular micrometer and vernier caliper measurements were calculated individually and then averaged together. 54 Percent shrinkage fbr the two species were then compared using a Mann- Whitney U test. Skin shrinkage in tree squirrels (8 i 1% for ocular micrometer and 6 i 1% for vernier calipers; mean shrinkage = 7 i 1%) and ground squirrels (8 t 1% for ocular micrometer and 7 i 1% for vernier cal- ipers; mean shrinkage = 8 i 1%) did not differ (Mann-Whitney U = 26.0, p = NS). To avoid inaccurate portrayal of variance in shrinkage by uniformly applying correction factors based on mean shrinkage es- timates (based on an overlapping but different group of squirrels than used in the body of the experiment) and because shrinkage did not differ between species, the original uncorrected density estimates were used for all analyses. Receptor Density An estimate of the density of the following types of sensory endings (receptors) was made for each of the f0ur pieces of skin taken from a squirrel: l) corpuscular endings including Meissner, simple and Pacinian corpuscles and 2) non-corpuscular endings includ- ing dermal free nerve endings and intraepidermal endings (largely associated with Merkel cell-neurite complexes). Descriptions of these receptors are presented in the results section below. Sections were viewed under the light microscope at 125x magni- fication and 0.7mm skin surface lengths were measured off using an ocular micrometer grid measuring 0.7mm x 0.7mm. The grid was then placed over one 0.7mm length that was randomly chosen using a "coin toss" procedure. All receptors falling at least partially within the grid and meeting the criteria described below were counted. For 55 corpuscular endings to be counted, the neurite within the corpuscle had to be visible and the section being examined had to contain at least as much of that neurite as the surrounding sections. Dermal free endings were included if the endings per se were visible or if the terminal neurite could be seen within the upper 1/2 of a dermal papilla. Intraepidermal fibers had to be seen approaching and crossing the dermal-epidermal (D-E) junction or coursing through the epidermis and had to have at least part of their length in the same fbcal plane as epidermal cells under 300x magnification to be counted. The latter criterion helped to exclude neurites running along the D-E junction without actually crossing into the epidermis in that section. These criteria should lead to relatively conservative estimates of receptor density and were applied uniformly to squirrels of both species. The area of the actual skin surface under the grid was also de- termined (see Figure 4). The curvature of the skin surface required that it, along with the borders of the grid be traced onto a data sheet with the aid of a drawing tube. A map measuring wheel was then used to measure the length of skin surface under the grid and the grid length from the drawing. The area of the skin surface under the grid was calculated according to the following equation: Skin Surface length Grid length (0.7mm) X drawingi . X Gr1d length drawing known Skin surface width (15 um) = skin surface area. The above procedures and calculations were repeated for six Randomly chosen sections (section numbers were drawn from a random Figure 4. 56 The procedure for determining skin surface area. The skin surface length is determined by measuring the grid length from the drawing made on a data sheet. By multi- plying the ratio of the skin surface length/grid length by the known length of the actual grid (0.7mm) the actual skin surface length is calculated. The section thickness (15um) is used as the skin surface width. Skin surface length X skin surface witdth = skin surface area. 57 15111118 Skin Surface Width/Scctlon S K I N SURFACE ‘1 \0 ‘ $0 ’9 1. [Data Sheet SuRFACE \omo LENGTH— X'SECTION Figure 4. 58 numbers table) from each tissue block. The number of sections to be used was determined by plotting the standard error of the mean re- ceptor density for a given piece of skin (averaged across all members of a species) as a function of the number of sections from which that density estimate was derived (S.E. decreases as the number of sections used increases). The number of sections at which the slope of this curve approaches or fluctuates around 0 (4-6 sections for pieces of skin in this experiment) was then used for all tissue blocks. The number of receptors was summed over the six sections and then divided by the total skin surface area of the six sections to yield a single receptor density (receptors/mmz) for each tissue block. This approach for estimating density was chosen over calculating a mean density es- timate for the six sections because it preserves information about the actual amount of skin sampled until the final calculation of density. The total density of receptors was analyzed by analysis of variance (ANOVA) with the following design: a three factor mixed design with repeated measures on two factors (location and paw) Factor 1: species (SpermOphilus vs. Sciurus) Factor 2: paw (fOre vs. hind) Factor 3: location (digit vs. tubercle) The prediction that tree squirrels would have a greater density of receptors in its forepaws than ground squirrels was tested with a planned comparison as well. Duncan multiple range tests were used for appropriate pggtehgg_comparisons. The ratio of forepaw receptor density to hindpaw receptor density was compared among species with a Mann-Whitney U test. The directional prediction is that tree 59 squirrels will have a greater forepaw receptor density/hindpaw re- ceptor density ratio than ground squirrels. The remaining analyses were restricted to the forepaw tubercles of the two species of squirrels where receptor density is sufficient. First the proportions of the two classes of receptors were determined by calculating corpuscular receptor density/non-corpuscular receptor density ratios. Ratios for the two species were compared using a Mann-Whitney U test. Receptor Dispersion The pattern of dispersion of receptors over the skin surface (randomly distributed, clumped or uniformly distributed) was examined by calculating Coefficients of Dispersion (CD'S) for each animal. The sections used to estimate receptor densities also were used for this analysis. On the data sheet for each section, skin lengths of 0.70mm and 0.35mm, yielding skin surface areas of 11 x 10'3mm2 and 3mmz, were randomly chosen. The number of receptors in these 6 x 10- large and small "quadrats" were counted. In all, 6 large and 6 small quadrats per animal were sampled. The mean number of receptors within a quadrat and the variance between quadrats were calculated for each quadrat size, for each animal. A CD was then calculated for each quadrat size, for each animal according to the following equa- tion: CD = Variance/Mean (Pielou 1969, Sokol and Rohlf 1969). The relationship between CD and patterns of dispersion is discussed in the results section. TWo quadrat sizes were examined, rather than one, because it is reported that the dispersion pattern observed is at least partially a function of the quadrat size used (Pielou 1969, 6O Smith-Gill 1975). The CD's for the two species of squirrels were then compared using Mann-Whitney U tests. To study the pattern of dispersion (see Figure 6 for examples) of different types of receptors (corpuscular and non-corpuscular) in relationship to each other (rather than in relationship to the skin surface, as in CD), a Coefficient of Segregation (S) (Pielou 1969) was calculated for each animal (see Appendix B for additional infor- mation). The sections used for this analysis overlapped partially with sections used in previous analyses but additional sections were also used. Sections were scanned under 125x magnification until a “base receptor" was located. To be included the base receptor had to be surrounded by two other receptors or it had to have a receptor on one side and a length of skin at least as great as the distance between the base and second receptor on its other side. A receptor meeting these criteria had its type (corpuscular or non-corpuscular) and the type of the "Nearest Neighbor" (the receptor closest to it) recorded. The scanning then continued until a new receptor was lo- cated. It was designated a base receptor and its type, along with the type of its nearest neighbor, was recorded. In this case, the nearest neighbor could be a receptor examined earlier in the same section. This procedure was repeated until 10 base receptors of each type and their nearest neighbors were recorded. The data were then arranged in a 2 X 2 table with the following format: 10 CC” 61 Nearest Neighbor Corpuscular Non-corpuscular Corpuscular a c m Base Receptor: Non-corpuscular b d n r s N (a, b, c and d = cell frequencies; m and n = row totals; r and s = column totals; N = table total). S was then calculated for each animal according to the following equation: observed number of mixed pairs of receptors _ S = 1 ' expected number of mixed pairs or receptors _ N(b + c . l - ms + nr (Pielou 1969). These procedures were adopted from Pielou (1969). The relationship be- tween S and the relative distribution of different receptor types is discussed in the results section. The S calculated in this study is an estimate of the true population S and is, therefore, subject to sampling error. S's for tree squirrels and ground squirrels were compared using a Mann-Whitney U test. Results Receptor Density On the basis of differences in the behavior and natural history of the squirrels, it was predicted that the forepaw receptor density/ hindpaw receptor density ratio would be higher for tree squirrels than for ground squirrels. The mean forepaw receptor density/ hind- paw receptor density ratio for tree squirrels is 3.3 t 0.5 and only '1.3 i 0.3 fer ground squirels. As predicted, the ratio for the first sPecies is significantly greater than that of the second species 62 (Mann-Whitney U = 45.5, p < 0.005). Mean receptor densities (and their standard errors) are presented in Figure 5 and results of the 3-way analysis of variance and ppgtyhpp comparisons are summarized in Table 1. Because Figure 5 shows actual densities and the ANOVA is based on transformed data, direct compari- son of the two may be misleading. The analysis of variance shows that paw and position are significant main effects, and that Species X paw and paw X position two way interactions are significant. There is no significant species effect, nor are the species X position or the three way interactions significant. Ppgteppg comparisons indi- cate that the tree squirrel's forepaw tubercle has a significantly higher receptor density than its hindpaw tubercle. No other compar- isons between paws are statistically significant. In all cases, the tubercle of a paw has a higher receptor density than its correspond- ing digit, but for both species, this difference is significant for the forepaw only. Significant interactions between species and paw variables and between paw and position variables appear to be based largely on comparison of the tree squirrel forepaw tubercle, with its exceptionally high density of receptors (95.4 i 16.8 receptors/mmz), and other locations. In a planned comparison of average forepaw re- ceptor densities (digit + tubercle/2) the two species did not differ (t = 0.8, df = 12, p = NS). The proportions of the two classes of receptors in the fore- paw tubercles of the two species, expressed as corpuscular receptor density/non-corpuscular receptor density ratios, were compared. For tree squirrels this ratio is 0.5 i 0.1 and for ground squirrels it is 0.9 i 0.2. Whereas tree squirrels have proportionately more 63 Figure 5. Receptor densities in squirrel glabrous paw skin. Means (bars)plus one standard error (flags) are presented. 64 "0‘ CI scuunus I SPERMOPHILUS 100- Q0 0 1 m 0 l q 0 05113171! (RECEPTORS/M1102) «b 0’ CH O 1 9 9 (N O n 00 O 1 5 l °L DIGIT TUBERCLE DIGIT TUBERCLE FOREPAW H I ND PAW Figure 5. Table 1. Analysis of receptor densities Anova Table SS df MS F P Total 430.7 55 -- -- -- Between subjects 78.5 13 -- -- -- Species 1.4 l 1.4 0.2 NS Errorbetween 77.2 12 6.4 -- -- Within subjects 352.2 42 -- -- -- Paw 28.7 1 28.7 9.6 ** Position 100.7 1 100.7 20.1 *** Speices x Paw 18.2 1 18.2 6.1 * Species x Position 17.5 1 17.5 3.5 NS Paw x Position 42.0 1 42.0 6.0 * Sp. x Paw x Pos. 8.3 1 8.3 1.2 NS Error] 36.4 12 3.0 -- -- Errorz 60.4 12 5.0 -- -- Error3 83.9 12 7.0 -- —- Post-hpg_Comparisons r Spermophilus/forepaw/digit Spermophilus/forepaw/tubercle Sciurus/fbrepaw/digit Sciurus/forepaw/tubercle SpermOphilus/forepaw/digit SpermOphilus/hindpaw/digit Sciurus/forepaw/digit Sciurus/hindpaw/digit Spermophilus/hindpaw/digit : NS Spermophilus/hindpaw/tubercle: NS Sciurus/hindpaw/digit : NS Sciurus/hindpaw/tubercle : * Spermophilus/forepaw/tubercle: * Spermophilus/hindpaw/tubercle: NS Sciurus/forepaw/tubercle : * Sciurus/hindpaw/tubercle : NS «x p < 0.05 ** p < 0.01 *** p < 0.001 66 non-corpuscular receptors than ground squirrels, this difference is not statistically significant (Mann-Whitney U = 25.5, p = NS). Receptor Dispersion In a Poisson (random) distribution the mean number of receptors/ plot equals the variance between plots and, by definition, CD = 1. When CD > 1 (variance > mean) receptors are clumped and when CD < 1 receptors are overdispersed or uniformly distributed (Pielou 1969, Sokol and Rohlf 1969). These three possible dispersion patterns are illustrated in Figure 6. The mean CD's for tree squirrels are 0.8 i 0.2 for small plots and 1.1 i 0.3 for large plots. For ground squir- rels the mean CD's are 0.7 i 0.1 and 0.9 i 0.2, respectively. The CD's for the two speices did not differ regardless of the plot size considered (for small plots U = 27.0, p = NS; fbr large plots U = 27.5, p = NS). Receptors are randomly distributed across the skin surface in the glabrous palm of both species, for both of the plot sizes examined. These results indicate that of the potential sites fer a receptor (e.g. a dermal papilla for a Meissner corpuscle), those sites actually occupied are randomly distributed. Alternatively, if every potential site is occupied, the sites themselves would have to be randomly distributed. In both species the first situation ex- ists. These results do not imply that a receptor will be found outside its normal site (e.g. a Meissner corpuscle in a rete ridge). It is also possible to study the distribution of one receptor type relative to another receptor type, rather than in relationship to the skin surface, as in the analysis above. When corpuscular and non-corpuscular receptors are random1y intermingled, each receptor 67 Figure 6. Receptor dispersion. a) Potential dispersion patterns and their relationships to the Coefficient of Dispersion (C0). b) Examples of segregated and unsegregated patterns and their relationships to the Coefficient of Segregation (S). 68 00 OO o o o o o o CD