Supersensitive odor discrimination is controlled in part by initial transient interactions between the most sensitive dorsal olfactory receptors and G-proteins

Biomedical Research Institute, National Institute of Advanced Industrial Science and Technology, Amagasaki, Hyogo 661-0974, Japan Institute of Biomedical Science, Kansai Medical University, Hirakata, Osaka 573-1010, Japan Takasago International Corporation, Hiratsuka, Kanagawa 254-0073, Japan Laboratory for Behavioral Genetics, Brain Science Institute, RIKEN, Wako, Saitama 351-0198, Japan Research Institute of Frontier Medicine, Nara Medical University, Kashihara, Nara 634-8521, Japan Department of Applied Chemistry, Graduate School of Engineering, Osaka University, Suita, Osaka 565-8571, Japan

transform, and build upon the material for any purpose, even commercially, as long as the author and original source are properly cited or credited.
The olfactory system discriminates enantiomers sensitively and rapidly within seconds.The olfactory information processing is initiated by stereospecific ligand binding to olfactory receptors (ORs) and allows the perception of the odor identity of a given ligand.ORs are members of the G-protein-coupled receptor (GPCR) superfamily, which includes visual receptors and neurotransmitter receptors.The number of OR family members is more than 100-fold greater than that of visual receptors.In mice and humans, each olfactory sensory neuron (OSN) expresses one of approximately 1,000 and 350 different types of ORs [1][2][3][4] , respectively, that operate as independent coding channels.The sensory profile of an odor stimulus may include several distinct elemental odors if multidimensional input is segmented through parallel pathways [5] .It is likely that, analogous to the process by which the color orange is hierarchically decoded as yellowish red or reddish yellow by signals from three types of the visual receptors, elemental odors are decoded by multiple receptors hierarchically through a temporal coding scheme that prioritizes the most sensitive, best-tuned receptors [6][7][8] .These receptors would dominate the perceived odor qualities by relaying the earliest ascending signals to cortical targets, evoking specific elemental odors, and recruiting feedforward inhibition to suppress competing odors evoked by longer latency OR inputs that are initially weaker and become stronger at later times.We previously estimated that enantiomers of carvone (at 100 μM) activated 70 types of murine ORs with >80% overlap [6] .The high overlap at higher odorant concentrations suggested that a minority of the most sensitive receptors plays a key role in determining odor quality differences between enantiomers and enabling their discrimination.Our hierarchical odor coding model predicts selective shifts in perceived odors by mutual inhibition when different stimuli are mixed [8] .For example, 2,5-dihydro-2,4,5-trimethylthiazoline (TMT) induces stress responses in mice when recognized as a predator odor [9] .The stress responses are reduced in different ways through feedforward inhibition when TMT is mixed with rose or hinokitiol odors, but not with (S)-(+)-carvone [10,11] .These led us to formulate a model of odor quality coding in which signals transduced by cognate receptors and relayed as inputs through segregated channels in the olfactory bulb [12][13][14] are processed in the olfactory cortex to evoke "elemental" perceived odor qualities by feedforward, feedback, and associative connections [6][7][8][15][16][17][18] .
Recently, we discovered a "supersensitivity" of wild-type (WT) mice to sub-ppq (<10 -15 ) level enantiomeric pairs of wine lactones and carvones in both detection and discrimination tasks in a Y maze (Figure 1A) [8] .This supersensitivity may be surprising given that these compounds are usually regarded as general odorants.Extreme sensitivity is often associated with innate responses to semiochemicals such as allomones and pheromones [19] .We previously showed that ΔD mice, which lack all dorsal olfactory receptors, are able to detect an allomone of rodent predators, TMT, but cannot recognize it as a predator odor [9] .The reported thresholds of WT and ΔD mice for TMT were equally low at 1.3×10 -8 w/w, and much lower than for 2-methylbutyric acid (5.7×10 -5 and 5.7×10 -4 w/w, respectively) in the habituation-dishabituation test [8,9] .Instead of TMT, we used non-dihydrogenated TMT (nTMT) to avoid potential problems arising from exposing WT mice to >2 weeks of repeated assays with the innate stressor TMT.We observed supersensitive detection of nTMT in both WT mice (10 -21 w/w) and ΔD mice (10 -19 w/w) (Figure 1A) [8] .These findings were archived by employing negative pressure-guided odor plume-like flows in the Y-maze with modified experimental procedures.Moths and other insects supersensitively navigate their way to scent sources along odor plumes [20,21] .The Y-maze design was optimized to direct plumes of odorized air along the central axis of each maze arm, maintaining radial concentration gradients between the central axis and arm walls.Dorsal receptors may play special roles in odor coding and recognition.As expected, our approach revealed an "enantiomer odor discrimination paradox" in ΔD mice that showed >10 10 -fold reductions in enantiomer discrimination sensitivity compared to wild-type mice [8] .ΔD mice were able to detect, but not discriminate one or both of the (-)-and (+)-enantiomers over a wide concentration range.This result strongly supports our hypothesis of hierarchical odor coding at least for enantiomer identification [8] and likely for fine odor discrimination between genetically determined body odors and their disease-induced alterations [22][23][24] .Notably, ΔD mice showed selective major loss of sensitivity to the (+)-enantiomers.The resulting 10 8 -fold differential sensitivity of ΔD mice to (-)-vs.(+)-wine lactone matched that observed in humans (Figures 1A, 1B) [8,25] .This suggests that humans lack highly sensitive orthologous dorsal receptors for the (+)-enantiomer, similarly to ΔD mice.Elemental odors encoded by subsets of orthologous ORs may be broadly conserved across species.Patterns of similarity among murine OR codes for 12 odorants resembled groupings of human percepts of the same odorant set [26] .Notably, three distinct subsets of murine ORs completely matched human odor percepts of vanilla, creamy, and cinnamon.The human olfactory system detects (-)-and (+)-enantiomers of carvones with nearly equal sensitivity [27] .Moreover, humans are roughly 10 5 -fold less sensitive to (R)-(-)-carvone than to (-)-wine lactone.The observed 10 4 -fold differences in sensitivities of ΔD mice were much greater than the corresponding sensitivity difference in humans (Figure 1) [8] .These results indicate that humans Odorant threshold concentrations (w/w); 10 -17 -10 -21 w/w (10 -2 -10 -6 ppq) for WT mice, 10 -11 -10 -19 w/w (10 1 ppt-10 -4 ppq) for ΔD mice.For wine lactones, the sensitivity difference between enantiomers was similar in humans and ΔD mice (10 8 -fold, green and red arrows between broken lines, respectively).However, for carvones, the differences in WT and ΔD mouse strains, and humans, were inconsistent (10 2 -, 10 4 -, and 8.6-fold, black, red, and green arrows, respectively).Despite retention of high detection sensitivity to (-)-enantiomers, ΔD mice showed a >10 10 -fold reduction in discrimination sensitivity.This leads to an enantiomer odor discrimination paradox, in which ΔD mice detected one or both of the (-)-and (+)-enantiomers but could not discriminate them over a wide concentration range (blue arrows).B, Detection thresholds in humans (black-red asterisks) as reported previously [25,27] .The differences in odorant detection thresholds are about 10 8 -and 8.6-fold for wine lactone and carvone enantiomers, respectively.Reprint with permission for authors [8] .
express neither the orthologous murine ORs most sensitive to (S)-(+)-carvone, nor the dorsal and ventral ORs most sensitive to (R)-(-)-carvone.Humans express fewer than half the number of ORs found in mice.Absence of the most sensitive orthologous ORs likely accounts for the much poorer detection and discrimination performance of humans compared to mice.
The basic concept of hierarchical odor coding is that the earliest arriving signals from the most sensitive, short latency, cognate receptors are the first to activate inhibitory feedforward pathways in the olfactory cortex through short-latency olfactory bulb tufted cells [8] .These early inhibitory signals from the ventro-rostral part of the anterior piriform cortex [15] , receiving signals from a minority of the most sensitive receptors, will trigger synchrony of cognate receptor signal inputs to pyramidal cells that selectively evoke "elemental" odor percepts by engaging associative neural pathways.The processing cascade may also suppress other odors corresponding to less sensitive, long latency, non-cognate ORs.The model hypothesizes that unique elemental odors correspond to a relatively small set of narrowly tuned ORs with highest sensitivities to a target ligand, whereas common elemental odors correspond to more broadly tuned ORs with overlapped sensitivity to multiple ligands.Primary qualities of odor percepts are determined by unique elemental odors and are modulated by secondary qualities from common odors.
To understand the origins of striking sensitivity differences between enantiomers of wine lactones and carvones, we compared dose-dependent changes in receptor codes.Using Ca 2+ imaging to profile odorant responses of OSNs of WT mice, we found that OR coding was sparser for wine lactones than carvones, i.e. wine lactone-sensitive ORs were about three-fold less numerous than carvone-sensitive ORs [8] .Sparse coding is consistent with the greater impact of dorsal OR ablation on behavioral detection thresholds of wine lactones.The largest subpopulations of the most sensitive ORs were those best-tuned to (-)-wine lactone or those with overlapping sensitivity to (-)-/(+)-wine lactones.These proportions held at all tested concentrations, except for (+)-wine lactone at 1 μM [8] .Moreover, we did not observe (+)-wine lactone-sensitive ORs, except for one OR at concentrations <100 μM [8] .These data are consistent with sparse coding of (+)-wine lactone by a small set of the most sensitive and specific dorsal ORs.
The large reduction in the detection sensitivity to (S)-(+)-carvone in ΔD mice may be explained by a loss of most of the most sensitive (S)-(+)-carvone-best ORs.In the absence of highly sensitive (S)-(+)-carvone-best ORs, weak subthreshold signals from common ORs may fall below the detection threshold for pyramidal cells of the olfactory cortex, and prevent ΔD mice from perceiving any difference between (R)-(-)-and (S)-(+)-carvones.In our model, the simplest interpretation is that signals of the most sensitive (R)-(-)-/(S)-(+)-carvone-best ORs dominate the principal elemental odors for both (R)-(-)-and (S)-(+)-carvones with no emphasis on a weak but unique elemental odor, so that ΔD mice only perceive a common (R)-(-)-/(S)-(+)-carvone odor and fail to discriminate the enantiomers [8] .In insects, fine discrimination of similar odorants is impaired by desynchronization of antennal lobe output neurons by picrotoxin, which blocks GABAA receptor-mediated synchrony [31] .In mammals, such inhibitory signal-mediated synchronization of olfactory bulb mitral/tufted cells could serve to bind signals from selected subsets of cognate (or other) ORs for downstream readout by coincidence detection in cortical pyramidal cells of the anterior piriform cortex.At present, the precise synaptic and network mechanisms in the olfactory bulb and cortex that could underlie hierarchical odor coding remain to be elucidated, besides OR-dependent differences in response onset latencies.
In the heterologous functional expression system, response onset latencies of ORs were shortened by using a chimeric Gα15_olf that possessed the C-terminal tail of the Gαolf [28] .This means that mutations of the interactive sites of the most sensitive dorsal ORs to Gα15_olf may change temporal ordering of OR input to the brain as well as hierarchical odor coding.To identify residues responsible for rapid and robust response of murine olfactory receptor S6 (mOR-S6) via the chimeric Gα15_olf, mutations of the C-terminal helix 8 were analyzed in the functional expression system.The N-terminal hydrophobic core between helix 8 and TM1−2 of mOR-S6 is important for Gα activation (Figure 2) [32] .A point mutation of a helix 8 N-terminal acidic residue eliminated the improved response dynamics via the chimeric Gα15_olf [32] .Atomic resolution structures of GPCR complexes with G proteins were reported for rhodopsin and β2 adrenergic receptor [33][34][35][36] .GPCRs stably interact with G proteins through their intracellular domains including the DRY motif in the transmembrane domain 3 (TM3).Significant residues controlling receptor−G-protein coupling are believed to be generally located at the intracellular end of TM5, the N-terminal region of intracellular loop 3 (IL3), the junction of TM3 and IL2, the C-terminal TM6, and the junction of TM7 and helix 8 [37] .In many GPCRs, an amphipathic helix 8 in the C-terminal domain plays several key roles in protein/lipid interaction [38,39] , receptor internalization [40] , dimerization of receptors [41] , and coupling with G proteins [42, 43]   .Understanding intra-and intermolecular interactions in GPCRs and conformational changes between intramembrane and cytoplasmic domains are crucial for elucidation of GPCR activation mechanisms.It was reported that helix 8 interacted with the NPxxY motif in TM7, highly conserved residues in GPCRs [34,44,45] .Mutation within this motif caused a significant reduction in signal activity [45,46] .It was also shown that a proximal dibasic motif in helix 8 was important for GPCR signaling [47] .These results suggest that an N-terminal acidic residue of helix 8 is responsible for rapid Gα activation via initial transient interaction.The supersensitive odor discrimination is largely governed by signals from the most sensitive dorsal olfactory receptors with the shortest onset latencies, which are controlled in part by initial transient interactions between the C-terminal helix 8 of receptors and the C-terminal region of Gα.

Figure 2 .
Figure 2. Model of mOR-S6 generated by homology modeling.A, A whole model of mOR-S6 (left).The right figure represents enlarged views around the helix 8, which are crucial for the experimental design.The residues in hydrophobic interactions surrounding helix 8 are shown as transparent CPK spheres and labeled.The residues of helix 8 are magenta, while TM1 and TM2 residues are green.B, The detailed interfaces of helix 8 and TM1−2, rotated 45° from the top panels.The residues in hydrophobic interactions surrounding helix 8 are shown as stick models.Reprint with permission for authors[32] .