All animals need to know what is going on in the world around them; thus, brain mechanisms have evolved to gather and organize sensory information so as to build transient and sometimes enduring internal representations of the animal’s surroundings. Using relatively simple animals and focusing primarily on olfaction, we combine electrophysiological, anatomical, behavioral, and other techniques to examine the ways that intact neural circuits, driven by sensory stimuli, process information. In the past year, our research program has begun to address several questions, such as what mechanisms, including the transient oscillatory synchronization and slow temporal firing patterns of ensembles of neurons, underlie information coding and decoding; how multimodal stimuli are integrated into unified perceptions; and how innate sensory preferences are determined. Our work has revealed basic mechanisms by which sensory information is transformed, stabilized, and compared as it makes its way through the nervous system.
Encoding a temporally structured olfactory stimulus with a temporally structured neural code
Brown, Joseph, Stopfer
Projection neurons (PNs) in the locust antennal lobe (AL) respond to odor puffs with odor identity- and concentration-specific sequences of spiking, inhibition, and quiescence. We are interested in how this spatiotemporal mechanism encodes odors in rapid trains of nearly overlapping brief pulses, as could occur in a natural odor plume, and whether the temporal structure of the stimulus interferes with the temporal structure of the neural representation.
In adult locusts, we made intracellular recordings from PNs and extracellular “tetrode” recordings from PNs and Kenyon cells (KCs, followers of PNs). We delivered 100 ms (milisecond) odor puffs in trains of three or 10 pulses, with interpulse intervals ranging from 0.5 to 2 seconds. For each pulse pattern, we delivered blocks of 10 trials (15- or 20-second intertrial interval) in random order. We used four odorants at two concentrations each. For most PN-odor combinations, numbers of odor pulse–elicited spikes changed reliably and often significantly with pulse position, as responses to one puff interfered with responses to subsequent ones. Despite the interference observed in individual PNs and regardless of tested interpulse interval, we found that the across-ensemble response of 117 PNs (pooled from 14 experiments) was sufficiently informative to achieve greater than 80 percent classification success for any given 50 ms time window; trajectory representations of the ensemble response repeatedly circled nearly identical regions of space for each odor pulse in the train. For trains of 10 pulses, we could identify three 50 ms “templates” corresponding to the train’s onset, middle, and offset, each of which could reliably classify odors and concentrations. Consistent with this observation, odor-specific KCs most often responded to all pulses in a train, but some KCs responded mainly at train onset or offset. In our recordings, no single PN reliably coded for the pulse timing of all odors. Yet, a template-based classification scheme using the ensemble response could reliably detect peaks in classification performance for each pulse in the train, thus encoding odor pulse arrival time and duration as well as odor identity and concentration.
Brown SL, Joseph J, Stopfer M. Encoding a temporally structured stimulus with a temporally structured neural representation. Nat Neurosci 2005;8:1568-1576.
Stopfer M, Jayaraman V, Laurent G. Intensity versus identity coding in an olfactory system. Neuron 2003;39:991-1004.
Central and peripheral plasticity in early olfactory processing
Brown, Stopfer
In nature, odorants are often encountered repeatedly owing to turbulence-generated discontinuous patches of odor and iterative sampling behaviors (such as sniffing or antennal flicking). In the locust, a model system for exploring the neural coding of sensory experiences, we found that repeated odor presentations caused several forms of neural plasticity acting on different time scales.
We delivered odor pulses to the antennae of intact locusts while recording from several brain locations. We delivered 100 ms odor pulses with short (500 to 2000 ms) interpulse intervals (IPIs), similar to natural odor plumes in turbulent environments. Pulse trains of three or 10 were presented randomly in blocks of 10 trials (15- to 30-second intertrial intervals). To track sensory neuron input, we made intracellular recordings from PNs and local neurons (LNs) together with simultaneous electroanennogram (EAG) recordings; to track odor-elicited neural oscillatory synchronization, we made local field potential (LFP) recordings from the mushroom body.
EAG responses showed a decrease of about 25, 15, 10, and 5 percent for 500, 750, 1,000, and 1,250 ms IPIs, respectively, when the antennae were presented with trains of 10 pulses. The decrease reflected olfactory receptor adaptation occurring over the short intervals. Oscillatory power measured by the LFP and LN recordings (10 to 40 Hz) decreased significantly with IPIs of less than 750 ms but increased with the longer IPIs. For all IPIs, 75 percent of the odor-PN combinations that elicited spiking showed average decreases in spiking of 40 percent while 15 percent increased their spiking by an average of 85 percent. Interestingly, despite the IPI-dependent changes in afferent strength during a stimulus train, the overall numbers of spikes across the PN population remained relatively constant, varying by about 5 percent. Similarly, the amplitude of odor-induced slow hyperpolarization in the PNs remained, on average, constant across the population. The changes reveal interactions between central and peripheral plasticity in the olfactory system. We are currently studying the underlying mechanisms of such interactions. It appears that, for shorter IPIs, peripheral adaptation dominates and that, for longer IPIs, central facilitation dominates, ensuring a relatively constant amount of PN spiking regardless of odor pulse train characteristics.
Spike train integration: contributions of synchronous neural activity to sensory representation
Ito
The olfactory system encodes information about odor identity and concentration in the spatiotemporal firing patterns of ensembles of principal neurons. The downstream neurons integrate action potentials from principal neurons across time and space. To test directly whether some spikes in the spatiotemporal firing patterns are more important than others for olfactory coding, it is crucial to understand how downstream neurons integrate their inputs. Coincidence detector neurons detect highly synchronized firings (narrow time window). Integrator neurons sum up inputs over time (wide time window). Using electrophysiological techniques, we are examining the integration properties of the downstream olfactory neurons in the tobacco moth Manduca sexta. By identifying and characterizing the properties of third- and fourth-order olfactory interneurons, we will improve our understanding of the contribution of synchronous neural activity to sensory representation.
Information reformatting through several interneuronal relays in the Drosophila olfactory system
Tanaka
The responses of primary sensory neurons to sensory stimuli are usually simpler than those of interneurons in the brain. Neural representations of the sensory stimuli are formed through the complex activity of interneurons. Such activity is determined not only by simple sensory input but also by parallel inputs from interacting networks of neurons in the brain. To explore the reformatting mechanisms and functions of these networks, we are employing a combination of genetic and electrophysiological tools in Drosophila. Drosophila is useful for this task for several reasons. First, several existing transgenic lines allow us to visualize specific types of neurons and modulate their activities. Second, a great deal of information about olfactory sensory neurons already exists. Third, the anatomy of the olfactory neural network has undergone detailed study. However, given Drosophila’s small size, electrophysiological study of function has only recently been possible.
We established techniques to record intracellularly from olfactory interneurons in the Drosophila brain. By means of genetically encoded fluorescence, we can visualize specific types of olfactory interneurons in transgenic flies and have successfully recorded intracellularly from them. We found a variety of response patterns in different cell types as elicited by different odorants. Using genetic techniques, we are now examining the mechanisms underlying these olfactory responses by conditionally inhibiting synaptic transmission of specific types of neurons.
Cellular and behavioral analysis of multimodal integration
Joseph
Strong behavioral evidence suggests that insects, like other animals, integrate information from several modalities, such as olfaction and vision. Using anatomical and electrophysiological techniques, we are exploring neural mechanisms for multimodal integration by identifying neurons in the locust brain that respond to several stimulation modes. In addition, we are working to identify behaviors that correlate with and are perhaps caused by the activities of these neurons.
Temperature and olfactory multimodal encoding and decoding in the insect brain
Ong
The antennae of insects not only serve as primary olfactory organs but also sense temperature and humidity in the insect’s environment. The antennae then transmit sensory information on olfaction, temperature, and humidity to higher brain centers. Using the tobacco moth Manduca sexta as a model system, we will initially examine the effect of temperature change in odor-evoked activities in principal neurons within the antennal lobe, which is the moth’s primary olfactory center. By revealing the anatomical projection patterns and physiological firing patterns of single-modality and multimodal neurons, we are exploring unimodal and multimodal coding and decoding in the brain.
Innate preferences for food odorants: I
Ong, Cowansage
Newborn, naive animals often display sensory preferences, perhaps reflecting an innate template for important stimuli. As a first step in a mechanistic analysis of innate preferences, we studied odor learning in newly eclosed moths, which we raised in our laboratory under carefully controlled conditions. We tested the ability of newly eclosed moths to learn associations between flower odorants and odorants that are innately repellent to adult moths. We adjusted the concentrations of all odorants to elicit approximately equivalent electroantennogram responses, indicating equivalently intense olfactory sensory neuron activation. Using a proboscis extension reflex conditioning paradigm that temporarily paired odor puffs with sugar water rewards, we found that moths could learn to associate both odor types with the reward. However, we found that moths learned the different odors to different extents. Our work demonstrates that moths are born with knowledge of the olfactory world. Given that moths are relatively simple and suitable for behavioral and electrophysiological study, we will be able to investigate mechanisms underlying innate coding of sensory information.
Innate preferences for food odorants: II
Sun, Feldman,1 Chiriboga2
We used electrophysiological and behavioral procedures to examine the odor preferences of newly hatched locusts. Open field tests showed that emerging hatchlings chose to move toward natural grass rather than toward a plastic replica of grass; further, the hatchlings preferred filter paper rubbed with fresh grass to nonodorized, green paper. A new electroantennogram (EAG) technique revealed that the hatchling antenna could detect and respond vigorously to a wide variety of stimuli, including food and nonfood odors. We matched food and nonfood odors by EAG response intensity and then allowed the hatchlings to choose between the odors in a new open field test procedure. Hatchlings preferred the food odors. We are now beginning an electrophysiological study of neural coding for innate odor preferences.
Improving extracellular, multiunit recording techniques
Carlton, Joseph, Zhu
We have been working to improve techniques for multiunit “tetrode” recordings, a recording technique that uses small clusters of electrodes and statistical analyses of the resulting recordings to permit the simultaneous recording of many neurons. The more neurons that are available for recording at one time, the better will be our understanding of how the brain interprets odor information. First, we have been working to increase the efficiency and versatility of the electrode design commonly used for such recordings. By exploring several parameters in the electrode probes, we are optimizing our design for specific recording configurations. Our new probes are more compact, durable, and easier and much less expensive to produce. Second, we are exploring new computational strategies to analyze the complex and informative recordings obtained from using the probes.
Publication Related to Other Work
Bazhenov M, Stopfer M, Sejnowski TJ, Laurent G. Fast odor learning improves reliability of odor responses in the locust antennal lobe. Neuron 2005;46:483-492.
1Jacklyn Feldman, former High School Student
2April Chiriboga, BS, former Technician
Collaborator
Maxim Bazhenov, PhD, Salk Institute for Biological Studies, La Jolla, CA
For further information, contact stopferm@mail.nih.gov.