Two Parallel Olfactory Pathways for Processing General Odors in a Cockroach

Abstract

In animals, sensory processing via parallel pathways, including the olfactory system, is a common design. However, the mechanisms that parallel pathways use to encode highly complex and dynamic odor signals remain unclear. In the current study, we examined the anatomical and physiological features of parallel olfactory pathways in an evolutionally basal insect, the cockroach Periplaneta americana. In this insect, the entire system for processing general odors, from olfactory sensory neurons to higher brain centers, is anatomically segregated into two parallel pathways. Two separate populations of secondary olfactory neurons, type1 and type2 projection neurons (PNs), with dendrites in distinct glomerular groups relay olfactory signals to segregated areas of higher brain centers. We conducted intracellular recordings, revealing olfactory properties and temporal patterns of both types of PNs. Generally, type1 PNs exhibit higher odor-specificities to nine tested odorants than type2 PNs. Cluster analyses revealed that odor-evoked responses were temporally complex and varied in type1 PNs, while type2 PNs exhibited phasic on-responses with either early or late latencies to an effective odor. The late responses are 30-40 ms later than the early responses. Simultaneous intracellular recordings from two different PNs revealed that a given odor activated both types of PNs with different temporal patterns, and latencies of early and late responses in type2 PNs might be precisely controlled. Our results suggest that the cockroach is equipped with two anatomically and physiologically segregated parallel olfactory pathways, which might employ different neural strategies to encode odor information.

Keywords: antennal lobe; insect; intracellular recording; olfaction; parallel processing; projection neurons; simultaneous intracellular recording; temporal pattern.

FIGURE 1

Two major types of uniglomerular projection neurons (PNs) in the cockroach brain. (A,B) Mass staining of PNs in the antennal lobe (AL). PNs (green) were stained by dye injection into the medial ALT (m-ALT) and olfactory sensory neurons (OSNs) (magenta) were labeled by anterograde staining from the antennal nerves. Axons of PNs form two different bundles in the AL (arrowheads and arrows in A,B), and their cell bodies separately cluster in the dorsal region of the AL. Loci of clusters of cell bodies correspond to the two major types of PNs; type1 and type2 PNs. Depth of serial optical image from anterior surface is indicated in each panel. (C,D) 3D images of two major types of uniglomerular PNs. Axons of both type2 PN (C) and type1 PN (D) run through the m-ALT, anteriorly to the mushroom bodies (MB) peduncle (PED) and terminate in both MB calyces (CA) and lateral horn (LH). (E,F) PN type-specific glomerular organization, viewed anteriorly (E) and medially (F). Glomeruli innervated by intracellularly stained PNs are colored according to PN types. Type2 PNs specifically arborize in the antero-dorsal group glomeruli (magenta in E,F), whereas type1 PNs arborize in the postero-ventral group glomeruli (green in E,F). Glomeruli innervated by ALT2 and ALT3 uniglomerular PNs, which have been identified by Malun et al. (1993), are colored blue. Two macroglomeruli innervated by sex pheromone-sensitive PNs are colored red. Bars in (A,E) = 100 μm, bar in (C) = 200 μm.

FIGURE 2

The segregated parallel pathways from the periphery to higher brain centers in the cockroach brain. (A–D) A type1 PN (green) and a type2 PN (magenta) differentially labeled in the protocerebrum. Serial optical sections (A–C) and a stacked image (D) reveal that terminal regions of type1 PN are segregated from those of the type2 PN in the MB calyces (CA) and the LH. Depths of the serial optical images from anterior to posterior are indicated in each of panels (A–C). Twenty-five serial optical images obtained every 4-μm are stacked in (D). Bar in (A) = 100 μm. (E) Schematic drawing of parallel pathways from the periphery to higher brain center in the cockroach brain. OSNs in perforated basiconic sensilla (single-walled A [sw-A] and sw-B) selectively terminate in antero-dorsal group glomeruli (AD glomeruli), whereas those in trichoid sensilla (sw-C) and grooved basiconic sensilla (double-walled A [dw-A]) projects to the postero-ventral group glomeruli (PV glomeruli: Watanabe et al., 2012b). Therefore, type1 PNs with dendrites in PV glomeruli and type2 PNs with dendrites in AD glomeruli form two segregated parallel pathways from the periphery to higher brain centers. Nomenclatures of MB calyces were described in previous articles (Mizunami et al., 1998; Strausfeld and Li, 1999a,b). MGs, macroglomeruli; PED, pedunculus.

FIGURE 3

Classification of PNs according to odor-specificities. (A) Response intensities to nine tested odorants in recorded PNs. We summarized response intensities to nine odorants, PN types, innervating glomeruli and odor spectra groups of 178 identified PNs. Response intensities are colored according to the values of “R-R₀” values (see “Materials and Methods”). PNs are classified into five odor spectra groups by cluster analyses using Ward’s method (Supplementary Figure S1). Post-recording visualization revealed innervating glomeruli and PN types. Innervating glomeruli are identified based on the 3D-map of the cockroach AL (Watanabe et al., 2010). (B) Odor-specificities of PNs. A zero effective odor number means that PNs did not show any excitatory responses to all tested nine odors. The percentages of responding PNs were calculated in each of pathways (type2 PNs, green bars, n = 60; type1 PNs, red bars, n = 107). Within type1 PNs, PNs arborizing in the T8–T10 group glomeruli (dark red bars, n = 50) show higher odor-specificity than those arborizing in the T5–T7 group glomeruli (light red bars, n = 57). (C,D) Glomerular maps of odor-specificities, viewed anteriorly (C) and medially (D). Glomeruli innervated by recorded PNs are colored according to the number of effective odors. The colder color represents higher odor-specificity. When PNs innervated the same glomerulus and exhibited different effective odor numbers, we colored the glomerulus based on the largest effective odor number. (E) Recruitment rates of PNs. The percentages of responding type2 PNs (green bars) and type1 PNs (red bars) per odor are shown as recruitment rates. Numbers denoted in bars are sample numbers used in the analysis.

FIGURE 4

Typical olfactory responses of two different types of PNs. (A–F) Olfactory responses of three type2 PNs (A–C) and three type1 PNs (D–F). Each of the recorded PNs (green) has its dendrites in a single glomerulus (magenta) belonging to different glomerular groups [laser scanning microscope (LSM) images in A–F]. During intracellular recording, nine odors were presented to the antenna. The 1-s olfactory stimuli are indicated by gray boxes. White bars in LSM images = 100 μm; vertical bars = 20 mV.

FIGURE 5

Temporal activity patterns of type1 and type2 PNs. (A–D) Temporal activity patterns of type1 and type2 PNs elicited by a given odor. Responses of type2 PNs and type1 PNs to hexanol (A, 28 responses from 14 type2 PNs; B, 40 responses from 20 type1 PNs) and cineol (C, 38 responses from 19 type2 PNs; D, 46 responses from 23 type1 PNs) are shown as raster plots (top) and cumulative histograms with a bin of 10 ms (bottom). All PNs were stimulated with a 1-s pulse of odor stimulus (gray box). Raster plots show that both hexanol and cineol evoke on-phasic responses in type2 PNs and various response patterns in type1 PNs. (E,F) Differences in temporal activity patterns between type1 and type2 PNs. Based on the peri-stimulus time histograms (PSTHs) with a bin of 20 ms, 68 responses to hexanol (E) and 84 responses to cineol (F) are, respectively, clustered into four clusters (Supplementary Figures S2, S3). Responses of type1 and type2 PNs are, respectively, plotted red and green using the first two PCs (PC 1 and PC 2). Distributions of PC 1 and PC 2 in type1 and type2 PNs are shown as box plots. Each marker represents the response clusters (Supplementary Figures S2, S3). PC 1 and PC 2 explain 36.67% of the point variability in (E) and 30.79% in (F). (G,H) Differences of temporal activity patterns across odors. Based on PSTHs of 284 responses to four different odors, we performed PCA. The distributions of the first two PCs (PC 1, G; PC 2, H) are displayed as box plots. The PC 1 and PC 2 explain 22.92 and 6.75% of the variability of responses, respectively. In (E–H), the line in the box and the box represents the median and the quartiles, respectively. Outliers are shown as dots.

FIGURE 6

Clustering of temporal activity patterns of PNs. (A) Results of cluster analysis. We classified temporal activity patterns of 284 PN responses into five response clusters based on the cluster dendrogram (left panel) formed by Ward’s method. The heat map shows PSTHs with a bin of 20 ms, and the heater color represents the higher spike activities within the bin. Red and green circles represent responses from type1 and type2 PNs, respectively. Responses to hexanol, octanol, phenyl acetate, and cineol are, respectively, denoted as blue, pink, orange, and yellow triangles. (B) Averages of PSTHs in each of five response clusters.

FIGURE 7

Early and late responses in type2 PNs. (A) Temporal dynamics in odor-induced action potentials of a type2 PN. On-phasic responses to six different odors (12 responses) are shown as raster plots (top) and time-courses of instantaneous spike frequencies (bottom). In each response, we identified a “peak spike” (dots; see “Materials and Methods”). Magenta and green responses indicate typical early and late responses, respectively. (B,C) Time distribution of peak spikes obtained from 160 responses. We identified peak spikes from 160 olfactory responses recorded from 16 different type2 PNs. The timing of each peak spike is corrected as follows; t_peak – t_first, where t_peak is a time of the peak spike from odor onset and t_first is a time of earliest odor-induced spike in the specimen (exampled in A). A total of 160 peak spikes are plotted in a scatter diagram (upper in B), and the instantaneous spike frequencies ranged from >300 Hz, 200–300 Hz, and <200 Hz, which are colored as magenta, green, and black dots, respectively. In the histogram (bottom in B), the number of peak spikes are counted every 5 ms after t_first, and there are two prominent peaks of the histogram (arrows in B). Peak spikes with high instantaneous spike frequencies (>300Hz; magenta boxes in B,C) are distributed significantly earlier than those with low instantaneous spike frequencies (200–300 Hz; green boxes in B,C). In (C), the line in the box and the box represents the median and the quartiles, respectively. Outliers are shown as dots. Results of statistical comparison using ANOVA and post hoc Tukey test are shown as asterisks (t-test, ^∗∗P < 0.01). (D,E) Olfactory responses of two different type2 PNs. To show rising points of early (white arrowheads) and late responses (gray arrowheads), excitatory responses elicited by different odor stimuli (top) are processed by a low-pass filter set at 50 Hz (bottom). Olfactory responses are arrayed based on odor onsets (black arrowheads). (F) Response type-specific glomerular organization. Glomeruli innervated by recorded type2 PNs are colored according to response types. When type2 PNs exhibited early or late responses to a given odor (octanol, cineol, or terpineol), glomeruli innervated by the PNs are colored orange or blue, respectively. Glomeruli innervated by PNs that did not show any excitatory responses to the odor are colored gray.

FIGURE 8

Simultaneous intracellular recordings from a type1 PN and a type2 PN. (A) Current injection to a PN. The artificial excitation of one PN by an inward current injection (5.0 nA) did not affect the membrane potential of the other PN. (B–D) Three pairs of response traces of simultaneous recordings from a type2 PN (upper traces) and a type1 PN (lower traces). The glomeruli innervated by PNs are denoted in each panel. We selected typical combinational responses elicited by three different odorants (black lines). Olfactory responses are arrayed based on timings of odor onsets (black arrowheads). Based on type2 PN response patterns, we identified latencies of early (white arrowheads and broken lines) and late responses (gray arrowheads and dotted lines) in each simultaneous recording. In each trace, there was considerable cross-talk (1–5 mV) between the two simultaneously recorded signals. Vertical bars = 20 mV.

FIGURE 9

Putative neural models to elicit olfactory on-responses of type2 and type1 PNs. (A–C) Three typical combinations of temporal activity patterns revealed by simultaneous recordings from a type2 PN (top) and a type1 PN (bottom). When a given odor stimulus activated both type1 and type2 PNs, combinational response patterns are categorized into one of three typical patterns: “type2 PN-early and type1 PN-early responses” (A), “type2 PN-late and type1 PN-early responses” (B) and “reciprocal responses” (C). White and gray arrowheads, respectively, indicate latencies of early and late responses of type2 PNs. Type1 PNs often fired after a break phase (hatched arrowhead in C). (D–F) Putative models to elicit three typical combinational response patterns of type2 and type1 PNs. A given odorant activated specific neurons (magenta lines) and activated type2 and type1 PNs with different temporal patterns. Based on results of transmission electron microscopy (Distler and Boeckh, 1997a,b) and electrophysiologies from OSNs (Fujimura et al., 1991) and LNs (Husch et al., 2008, 2009; Watanabe et al., 2012a), we assumed excitatory (triangles) and inhibitory (t-shaped bar) synapses in the AL. AD-glo, anterior-dorsal group glomerulus; PV-glo, posterior-ventral group glomerulus.