Odorant receptor alterations in β3GnT2−/− mice
Five separate samples of OE were dissected from adult β3GnT2−/− mice and littermate controls and mRNA was isolated from each sample. Following preparation of cDNA and labeling, the samples were each hybridized to an Affymetrix Mouse Gene 1.0 ST Array (Afftymetrix, Santa Clara, CA USA). A median value was calculated for each probe set for each OR gene in the 10 samples. Median values of the five null samples were compared pairwise with median values from the five WT samples. A null/WT ratio was generated that represents the mean of the five pairwise comparisons, Of 1,144 probe sets for OR transcripts, results indicated that mRNAs for 867 of the ORs were expressed at similar levels in WT and β3GnT2−/− OEs, including OR37 and M72. Results also indicated that expression of 256 OR genes was decreased by more than 30% in all five samples obtained from β3GnT2−/− OEs compared to the five WT OE samples. In addition, 21 OR transcripts, including mOR 256–17 and mOR 125–1 were increased by greater than 25% in all β3GnT2−/− OEs. The null/WT ratios of several of the 21 ORs that were increased the most, and a few of the many other ORs that decreased or were unchanged are presented in Table . In order to examine these differences in greater detail, we selected several ORs with mean values that were significantly decreased, increased or about the same in β3GnT2−/− OEs compared to littermate controls and performed real time RT-qPCR or in situ hybridization analyses.
Comparison of odorant receptor gene expression in β3GnT2−/− and WT OEs
Using the gene array data as a guide, we chose primer sets for six ORs with null/WT ratios that were either greater than 1.9 or were less than 0.6. After isolating RNA from OE samples, we performed RT-qPCR to find the relative differences in the expression of these ORs (Figure ). Three ORs, mOR267-17, mOR125-1, and mOR256-7 were analyzed that increased significantly in β3GnT2−/− OEs compared to controls and three ORs, mOR13-4, mOR18, and mOR28 were significantly decreased. qPCR analysis showed that mOR256-17 expression was increased by 71% in β3GnT2−/− OEs compared to controls, and mOR28 was decreased by 89%, validating the gene array generated values of a 90% increase for mOR256-17 and a 79% decrease for mOR28.
Figure 1 Quantitative PCR analysis for relative expression of ORs. In β3GnT2−/− mice the expression of mOR256-17 (Olfr15), mOR125-1 (Olfr43), and mOR256-7 (Olfr136) increased while that of mOR13-4 (Olfr640), mOR16 (Olfr1264), and mOR28 (more ...)
As the RT-qPCR results confirmed the gene array analyses that many ORs were either under- or over-expressed in β3GnT2−/−
OEs, we made riboprobes to these and other ORs in order to further examine OR gene expression in these mice, and to quantify the changes in the number of OSNs expressing specific receptors. As a control, to confirm a previous observation [8
], the number of neurons expressing the P2 receptor was reduced substantially in β3GnT2−/−
mice compared to WT. In contrast, the number of cells expressing M72 and OR37 showed no evidence of change (Figure , Tables and ). In no case was the distribution of positive neurons within zones and turbinates changed in β3GnT2−/−
mice (arrows in Figure ). RT-qPCR analysis indicated that several receptors were up-regulated in β3GnT2−/−
OEs. In situ
hybridization studies of these same receptors confirmed that the number of OSNs expressing mOR256-14 (olfr1370), mOR256-17 (olfr15) and mOR125-1 (olfr43) were significantly increased in β3GnT2−/−
OEs compared to WT OEs (Figure and Table ).
Figure 2 In situhybridization of ORs in the olfactory epithelium of β3GnT2−/−mice. The expression of some individual odorant receptor mRNAs is decreased in the OE of β3GnT2−/− mice compared to controls. A, B. The (more ...)
Odorant receptor expression in olfactory sensory neurons
Figure 3 In situhybridization of ORs in the olfactory epithelium of β3GnT2−/−mice. Several mRNAs for OR genes are increased in expression in the olfactory epithelium of β3GnT2−/− mice compared to controls. These (more ...)
Interestingly, as is the case in rats [15
], the number of cells expressing different ORs varies significantly, often by several fold in WT mice. For example, there are more than 10 times more mOR256-17+
cells than M72+
neurons in the adult OE. However, although our sample is small, changes in expression in β3GnT2−/−
mice do not appear to be related to the absolute number of cells expressing a given OR. For example, although their expression levels are about the same in WT mice, the number of mOR125-1+
cells increased by about one third in β3GnT2−/−
mice, whereas the number of P2+
neurons decreased by more than 75% (Table ).
Analysis of mOR256-17 axon guidance in β3GnT2−/− mice
Antibodies to the mOR256-17 odorant receptor were used to map the position of glomeruli in adult OBs. In WT mice, the antibody reacted with axons in the nerve layer and glomeruli within a narrowly circumscribed area of the OB. The antibody consistently labeled two glomeruli on the lateral surface of the OB, midway between the dorsal and ventral extent of the OB. The two immunoreactive glomeruli were always positioned side by side (arrows in Figure ). Antibodies to mOR256-17 also reacted with axons in the nerve layer and two glomeruli on the medial surface of the OB. These two medial mOR256-17+
glomeruli were positioned midway between the dorsal and ventral extent of the OB and were always side by side, identically to the lateral immunoreactive glomeruli (arrows in Figure ). In previous studies, immunocytochemical analysis using this same antibody normally revealed just one lateral and one medial glomerulus per OB [16
], although other studies have also found two medial and two lateral mOR256-17+
]. It is not clear whether this reveals a second mOR256-17+
glomerulus or if the antibody cross-reacts with closely related ORs that normally project to adjacent glomeruli. The total number of mOR256-17+
neurons is a extremely high compared to some other ORs (Table ), thus it would not be surprising if there is normally more than one mOR256-17+
glomerulus in each half OB.
Figure 4 mOR256-17 Immunocytochemical analysis of the adult olfactory bulb. In WT mice antibodies to the mOR256-17 OR identify two lateral glomeruli (arrows in A) and two medial glomeruli in the extreme caudal OB (arrows in B). In β3GnT2−/− (more ...)
OBs, mOR256-17 immunoreactive glomeruli are present in the OB but in more scattered positions on the lateral and medial surfaces. The mutant OBs are shaped somewhat differently from WT OBs in that they are shorter along the rostrocaudal axis and along the dorsoventral axis but not mediolaterally [8
]. In this example, which is typical of four OBs examined, a mOR256-17 immunoreactive glomerulus is present on the lateral surface of the OB (arrow in Figure ) in a position similar to those seen in the wild type OB. Three additional mOR256-17 positive glomeruli are also detected in more caudal sections of the ventral OB (arrows in Figure , E, F). Even further caudally, three additional mOR256-17 glomeruli are visible in the ventral OB that appear to represent medially projecting axons (arrows in Figure ). One of these mutant glomeruli is in the approximate correct position compared to wildtype OB, but the others are located either more ventrally or more dorsally from their normal position. The mOR256-17 positive glomerulus in Figure is the same very large glomerulus as the more dorsal glomerulus in Figure . The aberrant positions of glomeruli in null mice is very similar from mouse to mouse, just as OR-defined glomeruli are found within a very restricted area of the OB in WT mice. We previously showed that this was also the case for M72 glomeruli [9
]. These results suggest that other guidance information is maintained in β3GnT2−/−
mice that continues to steer axons to specific but aberrant targets.
A map of the positions of mOR256-17 glomeruli in the adult OB clearly shows the difference between the WT and β3GnT2 mutant mouse. The map is a two-dimensional representation of the glomerular layer, looking down at the dorsal surface of the OB after it has been flattened by unrolling it from the ventral midline, rostral is at the top. The map of the mOR256-17+ glomeruli in WT OBs (green circles Figure ) shows two lateral glomeruli rostral to the AOB, midway between the dorsal and ventral surfaces of the OB and two medial glomeruli close to the caudal end of the OB, midway between the dorsal and ventral surfaces of the main OB.
Figure 5 Composite map of mOR256-17 glomerular positions in adult β3GnT2 mice. The position of glomeruli in one OB of WT (A) and β3GnT2−/−(B) mice was plotted as a two dimensional map. The OB was flattened and unrolled from the (more ...)
The map of the β3GnT2−/− OB (Figure ) is smaller along the rostrocaudal axis but has similar dimensions to the WT OB along the mediolateral axis. As shown in this map, there are seven glomeruli in total (red circles in Figure ), compared with four glomeruli in the WT OB. Although glomeruli are not located in an absolutely fixed position from mouse to mouse, most glomeruli map to within 100 to two hundred microns of the position shown on this map. In addition, three of the mOR256-17 positive glomeruli on the lateral surface of the β3GnT2−/− OBs are near the ventral surface of the OB. On the medial surface of the mutant OB, one of the glomeruli is positioned identically to the WT medial glomeruli, but the other two glomeruli are scattered ventrally and dorsally.
Axons from under-represented ORs extend into the OB
Although the level of expression of many ORs is significantly decreased in the OE of β3GnT2 mice, independent analysis using transgenic mice, in situ hybridization or immunocytochemistry reveals that expression of these ORs is not completely lost; rather the number of OR expressing cells may be greatly reduced (Figure , Table ). For example, gene array data showed that the I7 odorant receptor was decreased by more than 50% and in situ hybridization and RT-qPCR results indicated that the number of P2 neurons and mOR28+ neurons were decreased by 78% and 80% respectively in β3GnT2−/− mice compared to controls. In order to examine the fate of axons expressing under-represented ORs in β3GnT2−/− mice, we studied two OR subsets independently, I7-GFP+ axons using GFP fluorescence combined with synaptophysin labeling of presynaptic terminals and mOR28+ axons by immunocytochemistry. In WT mice, the medial I7+ glomerulus is normally visible near the ventral midline (arrow in Figure ). Although we have never seen an I7-GFP+ glomerulus in postnatal β3GnT2−/− mice, I7-GFP+ axons are detectable in the nerve layer and glomerular layer of adult null mice. In Figure , an I7+ axon can be seen entering a synaptophysin+ glomerulus (arrow) in approximately the correct location of the β3GnT2−/− OB. However, this glomerulus is for the most part I7 negative but synaptophysin positive, indicating that it is principally populated by axons expressing other ORs.
Figure 6 Reduction in OR expression does not prevent axons from correctly targeting the OB. I7+ axons normally form a medial glomerulus in the OB near the ventral midline (arrow in a). In β3GnT2−/− mice, there are insufficient numbers of (more ...)
Gene array analysis showed that mOR28 mRNA was reduced more than any other OR and RT-qPCR results indicated that mOR28 mRNA was significantly decreased in β3GnT2−/− OEs compared to wild type OEs (Figure ). Immunocytochemical analysis (Figure , D) confirms that the number of mOR28+ neurons is decreased by about 75% in β3GnT2−/− OEs. In spite of the severe loss of mOR28 expression in β3GnT2−/− mice, immunoreactive axon projections are visible in the nerve layer of the OB and can be seen targeting glomeruli in the same location that mOR28+ axons normally form glomeruli in WT mice (arrows in Figure ).
Neuronal activity is unchanged in β3GnT2 null mice
Previous studies have identified two distinct mechanisms for the regulation of OSN gene expression by olfactory signaling reviewed in [7
]. One class of genes, including axon guidance cues such as Nrp1 and Sema3A, are directly modulated by cAMP levels established by AC3 [4
]. For a second class of genes, transcriptional levels are determined by neuronal activity downstream of AC3. This molecular subset exhibits OE expression changes in either naris occluded mice or anosmic cyclic nucleotide gated channel subunit A2 (CNGA2) null mice that are not directly cAMP-dependent [5
]. For example, microarray analysis studies revealed that RNA levels of the Leucine Rich Repeat Containing 3b (Lrrc3b) gene are decreased more than ten-fold in CNGA2 null OEs relative to WT mice, while those of the calcium binding protein calretinin are elevated nearly four-fold by the loss of activity.
We reasoned that β3GnT2−/−
mice should show similar gene alterations if the perturbations in olfactory connectivity and cAMP signaling compromised odor-evoked activity. Surprisingly, this was not the case (Figure ), as the activity-dependent genes we analyzed by in situ hybridization were unchanged in β3GnT2−/−
mice except for the loss of mature OSNs described previously [6
]. Lrrc3b expression is unchanged in OSNs of the thinner null OE, despite the dramatic loss of Lrrc3b evident in CNGA2 null OEs. S100A5, which is down-regulated eight-fold in CNGA2 null mice, remains robustly expressed in β3GnT2 knockouts. Likewise, calretinin expression is unchanged in the basal layer of the β3GnT2−/−
OE, despite the fact that its expression is significantly increased and redistributed throughout the OE of CNGA2 KOs [18
Figure 7 In situ hybridization of odor-evoked activity dependent genes in the OE.A. The expression of genes that are regulated by activity-dependent processes were analyzed in adult WT and β3GnT2−/− OEs. Calretinin, which is expressed in (more ...)
To confirm these results (Figure ), we used qPCR to determine expression levels of activity-dependent genes relative to known markers for the neuron populations that undergo dynamic changes in β3GnT2−/−
mice. Expression levels of Lrrc3b, S100A5 and Kirrel2 decrease in parallel with those of OMP, a marker for the mature neurons that are lost in β3GnT2−/−
]. Because of the loss of mature OSNs, immature neurons expressing GAP43 increase proportionally in the β3GnT2−/−
]. Despite this, calretinin levels are not significantly altered by the loss of β3GnT2, and there is no redistribution to more apical layers, as reported for CNGA2 knockout mice [18
]. These results suggest that neuronal activity in the localized environment of olfactory cilia is not severely compromised in β3GnT2−/−
mice compared with CNGA2 null mice.
β3GnT2 null mice have minor olfactory discrimination deficits
The significant loss of OR gene expression and the axon guidance errors, coupled with the dramatic decrease in AC3 activity reported earlier [6
], we expected β3GnT2−/−
mice to have a significant loss in olfactory discrimination ability. We, therefore, examined the performance of 9– to 13-week-old β3GnT2+/+
mice in an olfactory discrimination task using structurally unrelated odors. In this assay, mice are trained to associate a food reward with one of two odors. Training took place over a two-day period and testing followed on Days 3 and 4. This test was designed to provide intermittent reinforcement during the first test sessions of Days 3 and 4 so that both wild type and null mice improved their discriminatory ability on Day 4 compared to Day 3 (Figure ). An initial t-test comparison of the animals with or without a food reward, within a given day, showed no differences on either of the testing days (P
0.24 to 0.47). Therefore, the data within each day were combined and compared using a two-way repeated measures ANOVA using genotype and day as factors (there is no significant statistical interaction between genotype and day, P
Figure 8 β3GnT2−/−mice retain the majority of their olfactory discriminatory ability. Mice were tested against pairs of odors for their ability to associate odor with a food reward. β3GnT2−/− mice can discriminate (more ...)
The analysis shows that β3GnT2−/− mice were able to discriminate all six odor pairings but that they had a significant deficit in their ability to discriminate three of the six odor pairings either on Day 3 or Day 4 compared to littermate controls (asterisks in Figure ). On Day 4, null mice showed a deficit to two of the six odors, citronellal and decanal, both 10 carbon aldehydes but with several structural differences. Decanal was the only odor to which β3GnT2−/− mice consistently displayed a deficit on both Days 3 and 4. Had the null mice been tested for their ability to discriminate between enantiomers or mixtures of enantiomers, it is likely that further deficits would have been revealed.