Charakterisierung von lokalen Interneuronen im Antennallobus der Honigbiene

The antennal lobe (AL) is the primary olfacory center of the honey bee. It is the site ofinteraction between olfactory receptor neurons, and two types of AL neurons: local interneurons (LNs) that are restrained to the AL, and projection neurons (PNs) that relay output to higher processing areas. The present work investigates physiological and morphological properties of honey bee AL neurons, LNs in particular. The individual studies, summarized here, are united by the underlying attempt to infer potentially functional LN sub-populations from the described characteristics.My first objective was to investigate how individual AL neurons encode a type of complexinformation content, the olfactory system is challenged with every day: an odour mixture(Chapter 2). I stimulated with mono-molecular odorants, their temporally perfect-,and imperfect binary mixture to reproduce a natural dynamic odour environment. Single cell's responses were recorded intracellularly.Response patterns between di erent neurons varied in many details but were generally speaking either rate changes to excitation or inhibition, respectively, or membrane depolarisation accompanied by one or a few single spikes (cf.: 2.4.2).Irrespective of its individual response pattern, each neuron, challenged with the binary mixtures, responded in one of two possible ways: elemental, or con gural. About half of the neurons responded elementally, i.e. responses evoked by mixtures re ected the underlying feature information from one of the components. The other half exhibited con gural responses, i.e. responses evoked by mixtures represented these as clearly di erent from their single components (cf.: 2.4.3).A question immediately arising is, whether these two types of encoding could be associatedwith di erent sub-populations of neurons. Referring to the neuron's latencies as an indicator of position within the circuitry, I found that elemental neurons divided in earl responders and late responders whereas latencies of con gural coding neurons concentrate in between these divisions (cf.: 2.4.4). From this nding one may infer that elemental odour coding is not con ned to only one sub-population of neurons. In fact, it is more likely that LNs and PNs, which have previously been shown to di er signi cantly in latency, can both exhibit elemental coding (cf.: 2.5.2). Latencies of neurons with con gural responses expressed a tendency torespond faster to single components than to imperfect mixtures. This nding made me thinkthat these neurons may participate in multiple processing circuits. Both of the above assumptions were con rmed by exemplary morphological data (cf.: 2.4.5).For each of the two groups of 'elemental neurons', early and late responders, I could obtain an exemplary staining. The early responding neuron was con rmed as an LN, while the late responding neuron was a PN. More surprisingly, however, I found that one of the 'con gural neurons' was a hetero LN, just like the short latency elemental neuron. By comparing the inter-glomerular innervation patterns of the two hetero LNs with odour-speci c glomerularactivation maps, derived from calcium imaging, I hoped to nd an explanation for the difference in odour coding. Indeed, the elemental hetero LN innervated one of the responsive glomeruli densely. The con gural hetero LN, in contrast, innervated glomeruli that were responsive to the chosen stimuli only sparsely. Based on the combined morphological and physiological evidence, I propose to consider hetero LNs as multi-function neurons. Multifunctionneuron here means, the possibility to be recruited by di erent circuits such thatelemental as well as con gural odour-processing are performed by the individual neuron in a stimulus-context dependent manner (cf.: 2.5.3).The proposed multi-function hetero LN would require the ability to receive sensory inputand give output in both, sparsely and densely branching arbours. Are these requirementsmet on the local scale of individual intra-glomerular arborisation? And judging by the global inter-glomerular innervation, in how far are LNs generally tailored to receive and redistribute direct sensory input? Or are innervation patterns rather oriented towards PNs? These are questions I hoped to answer by analysis of LN morphologies under functional aspects (Chapter 3).I reconstructed morphologies of single neurons from stainings of different LNs. Neuron reconstructions were transformed to fit into a common reference frame. By this means, it gotpossible to evaluate which of the four sensory tracts (afferent fi elds) and which of the twoAL hemilobes (efferent fi eld) received innervations. To analyse intra-glomerular arborisation, I reconstructed cap, and where possible core of single glomeruli as well as sparse or dense arborisations of the LNs innervating them.The inter-glomerular innervation pattern of LNs differed widely on an individual scale, butappeared to correlate with the division of afferent fields rather than efferent fields (cf.: 3.3.1).This observation suggests that at least a considerable sub-population of LNs is tailored to collect and integrate meaningfully related ORN input (cf.: 3.4.1). Investigating intra-glomerular arborisation I found markedly different branching patterns for sparse as well as dense arbours. All of these had the potential to establish contact with sensory neurons in the glomerular cap as well as PNs and other LNs in the core (cf.: 3.3.2). On these grounds, morphological requirements for a multi-functional hetero LN are ful lled.Having come so far it seems that LNs express a certain functional communality. Still, differences in local and global branching patterns, amongst other characteristics, are obvious. Continuing from a perspective of functional morphology, I assembled a toolbox of morphological descriptors based on which I di erentiate six LN-phenotypes (cf.: 3.3.4). Clearly, under the objective of finding functional LN sub-populations the convenient division in only two broad groups of homo- and hetero LNs has to be reconsidered (cf.: 3.4.2).Systematic differences in morphology, like those based on which I had distinguished sixdifferent LN Phenotypes, are indicators for functional differences between neurons. But, asthe function of a neuron is inevitably the result of all its properties, the same is true for measures of differences in physiology.Classifi cation based on spiking properties of single neurons has decidedly facilitated the investigation of inter-neurons in the mammalian neocortex. I wondered whether the activity patterns observed in the honey bee, by me and others, could likewise be separated in conclusive groups (Chapter 4).To approach this question I analysed single cell recordings from a set of AL neurons whichincluded different LNs, PNs and morphologically unidenti ed neurons. Collected descriptive values of spiking and sub-threshold activity that could be extracted from odour-evoked responses were decorrelated and reduced by means of PCA and subsequently clustered by means of an automated hierarchical algorithm.Referring back to the original descriptive values, some of the suggested groups were immediately conclusive whereas others appeared more heterogeneous (cf.: 4.3.1). Repeating theclustering on PCs derived from spiking activity only, the allocation of neurons within clusters was largely preserved while the inter-relationship between groups changed (cf.: 4.3.2).On this ground I conclude that electro-physiological classifi cation of honey bee AL-neurons is feasible and that the information contained in spiking activity alone is su cient for this purpose (cf.: 4.4.1).If it is possible to classify groups of AL neurons according to their discharge patterns, could the classi cation criteria be used to predict neuron morphology? Examining the distribution of morphological identi ed LNs, mPNs and lPNs between the suggested groups, I had to conclude that, there is no simple one-to-one relationship between electro-physiological pro les and neuron morphology (cf.: 4.3.3/4.3.1). Still, trends are present in that discharge patterns seem to be typically associated with mPNs and others are strongly suggested to be correlatedwith LNs (cf.: 4.4.2). As a whole, LNs and PNs are better distinguished based of spikingactivity alone (cf.: 4.3.3)