|Home | About | Journals | Submit | Contact Us | Français|
The trpc2 gene codes for an ion channel found in the vomeronasal organ (VNO). Studies using the trpc2−/− (KO) mouse have exploited the gene's role in signal transduction to explore the VNO's role in pheromonally-mediated behaviors. To date, no study has evaluated the impact of the trpc2 gene on activity within the brain. Here, we examine the gene's effect on brain regions governing maternal aggression. We intruder-tested lactating dams and then quantified Fos immunoreactivity (Fos-IR) in the vomeronasal amygdala, hypothalamus, olfactory regions, and the accessory olfactory bulb (AOB). Our data confirms previous reports that loss of the trpc2 gene severely diminishes maternal aggression. We also show that deletion of the gene results in differential hypotrophy of the glomerular layer of the AOB (GlA), with the anterior portion the GlA resembling that of wild type mice, and the posterior portion reduced or absent. This anatomy is suggestive of residual functioning in the apical VNO of these animals. Our Fos study describes an impact of the deletion on a network of 21 brain regions involved in emotion, aggression and olfaction, suggesting that signals from the VNO mediate activity throughout the brain. Home-cage observations of KO dams reveal specific deficits in nest-building, suggesting a role for pup pheromones in inducing and maintaining pup-directed maternal behaviors, as well as maternal aggression.
In many species, a vomeronasal organ (VNO) mediates critical communication between conspecifics, binding chemical cues emitted by another animal – pheromones – and sending information about these cues to the brain. These cues influence both the behavior and physiology of the receiving animal. In the context of maternal behavior, the VNO is largely known for its role in promoting maternal defense (offspring protection) in mice; lesioning it eliminates this behavior entirely (Bean & Wysocki, 1989). More current studies have also shown a role for the VNO in mediating the time lactating mice spend on nest (Kimchi et al., 2007). However, less is known about how signals from the VNO shape activity within the brain to promote these maternal behaviors.
The trpc2−/− mouse has been a popular tool for exploring questions related to VNO signaling, since deleting this gene for a highly localized VNO ion channel reduces signaling from the VNO significantly without damaging adjacent structures or fibers of passage (Kelliher et al., 2006, Kimchi et al., 2007, Leypold et al., 2002, Liman et al., 1999, Stowers et al., 2002). Like VNO-lesioned females, trpc2−/− dams display a striking absence of maternal aggression behavior (Leypold et al., 2002), making them especially useful for exploring the effects of VNO signaling on the neural circuitry producing maternal defense behavior.
One possibly confounding variable in the behavior of trpc2−/− mice is the low levels of maternal aggression typical in the C57BL6/J strain (Parmigiani et al., 1999), which has been used in all published studies on the trpc2−/− mouse. For example, in the first study demonstrating loss of maternal aggression in trpc2−/− mice, the mean maternal attack duration for control animals in a 15 minute test was only 12 seconds and the mean number of attacks was only 6 (Leypold et al., 2002). These levels are considerably lower than those displayed by animals of other strains (Parmigiani et al., 1999), leading us to wonder if the lack of maternal aggression displayed by trpc2−/− mice in these studies might reflect a specific interaction of overall low aggression and the mutation.
In this study, we capitalize on the advantages of the trpc2−/− mouse while mitigating possible effects of low aggression, by shifting the mutation into a line of mice selected for robust maternal defense (Gammie et al., 2006). After testing both trpc2+/+ (WT) and trpc2−/− (KO) dams for maternal aggression, we quantify Fos immunoreactivity (Fos-IR) within the accessory olfactory bulb (AOB) and specific subsets of brain regions. Each subset mediates an aspect of maternal defense behavior or pheromonal cue processing. In each case, we begin with a hypothesis about how the gene deletion will impact Fos-IR within the brain region, based upon earlier studies of trpc2−/− behavior. We also measure changes in brain morphology associated with the deletion. To our knowledge, these studies are the first to directly explore the impact of the trpc2 gene on brain activity in the context of maternal behavior.
The focal animals in this study were derived from crosses of male KO mice bred in a C57BL6/J background (Stowers et al., 2002)(generously donated by Dr. Catherine Dulac, Harvard University) and females from our colony of hsd:icr mice (Harlan, USA) selected for high levels of maternal defense (S) (Gammie et al., 2006). The heterozygous offspring of these initial pairings were then bred to each other; sibling matings were avoided. Offspring of the heterozygous pairings were genotyped using PCR analysis of DNA obtained from ear snips. The resulting KO males and females were bred back into our S mice to create heterozygous offspring with a greater proportion of the S genetic background. These back crosses were performed in three consecutive generations, with the offspring of each crossing genotyped as before. After the final cross, the S background represented approximately 87% of the genetic background of the offspring. All focal female animals, including WT controls, came from this third crossing. WT controls were littermates of KOs.
Males used to evaluate AOB morphology were bred from homozygous pairings and intended for another study; we decided to use the AOBs of these males to analyze the size of the glomerular layer (GlA) after discovering an effect of genotype on this structure in the females.
Males used for intruder testing were hsd:icr mice from our colony; they were either purchased directly from Harlan (Harlan, Madison, WI, USA) or bred from non-transgenically altered mice in our colony.
All animals were housed in polypropylene plastic cages with ad lib access to tap water and food (Mouse Breeder Chow, Harlan) on a 14:10 light/dark cycle with lights on at 0600 CST. Cages were changed weekly, except during the post-partum period, when they were left undisturbed until the conclusion of the experiment for that animal. Virgin males and females were group housed until mating, except for males used for examination of AOB anatomy. These animals were singly housed for 4 weeks. Breeding pairs remained together for 10 days, at which point males were removed.
Starting 18 days post-pairing, cages were checked for litters with any litters born before 1600 h considered born that day. Day of birth was considered postpartum day (PPD) 0. Dams and litters were weighed on PPD 1, and litters were culled to 11 pups. No dams with fewer than 8 pups in a litter were included in the study. Dams and litters were weighed again on PPD 6, prior to sacrifice.
Mice were genotyped by PCR. To amplify the wild-type gene, we designed two primers to identify the pore region of the Trpc2 ion channel: sense, 5′ – ACA GAG GAC CCC CAG TTT CT – 3′, and antisense 5′- ACA GAG GAA GGC AGT CAG GA – 3′. The deleted gene was replaced with a PGK- neo cassette in the knock-out mice (Stowers et al., 2002) which we amplified using the following primers: 5′ – AAT ATC ACG GGT AGC CAA CG – 3′ and 5′ – TGC TCC TGC CGA GAA AGT AT – 3′.
The behavior of each dam was recorded at 1 min intervals for 1 hour of the light cycle between 1400 and 1600 hours on PPD 2 and 3. Behavior was divided into on-nest and off-nest categories; on-nest behaviors were: arched back nursing, supine nursing, licking of pups, self-grooming, eating/drinking, nest-building (in this case, nest-building consisted of repositioning material already in the nest). Off-nest behaviors were: pup retrieval, self-grooming, eating/drinking, nest-building (in this case, nest-building consisted of bringing material from other sites in the cage to the nest). These behaviors could be combined; for example, a dam whose nest was close to the food hopper might eat while on nest, and so could be considered “on nest” and “eating/drinking” in one measurement.
On PPD 4 and 5, dams were brought to the testing room in their home cages between 0800 and 1200 hours. Pups were removed from the cage (removing pups from the cage within an hour of an intruder test does not diminish maternal defense on the part of the dam (Broida et al., 1981)) and an unrelated, group-housed, sexually mature and naïve male was introduced. The resulting interaction was videotaped for 7 min and the male was then removed. No male was used more than once for testing on any given day, and no male was used more than three times across all days of testing. After the male was removed, the female's pups were distributed around the cage to stimulate retrieval behavior in the dam and resulting interactions between dam and pups were videotaped for a further 2 min, after which cages were returned to the home room.
On PPD 6, each dam was once again brought into the testing room and pups removed from each cage. However, a male intruder was introduced into only half the KO and half the WT dam cages forming the Test (T) groups for the subsequent Fos immunoreactivity (Fos-IR) study. The remaining dams formed the Untested (U) group. Interactions between T dams and the intruder male were recorded for 10 min, at which time males were removed. Pups were returned to T cages at this point, and to U cages 10 min after being removed. Mother-pup interactions were recorded for both T and U dams for 2 min, and the cages returned to the home room.
Videotaped behavior was scored by individuals blind to the genotype of each dam. Defense behaviors scored were: latency to first attack, number of attacks (any interaction including a bite was considered an attack), total time attacking, number of sniffs (an interaction in which the female's snout came into physical contact with the male but no bite occurred was considered a sniff), total time sniffing. Pup retrieval was quantified by measuring the time elapsed to retrieval of first and fourth pups after the pups were returned to the cage. In any case where a behavior never occurred (e.g. no attacks or no retrievals), the latency recorded was the full time of the test.
Dams were sacrificed 90 min following the start of their PPD 6 intruder test (in the case of T dams) or 90 min following the removal of pups from the cage (in the case of U dams). After brief isoflurane anesthesia, the mice were deeply anesthetized with 0.15 mls of sodium pentobarbital and perfused pericardially first with 0.9% saline and then with 4% paraformaldehyde in PBS. Brains and olfactory bulbs were dissected, fixed overnight in 4% paraformaldehyde, and then cryoprotected in 30% sucrose in PBS.
Olfactory bulbs and the most rostral 2 mm of tissue from the brains were removed and cut sagittally into 41 μm sections using a sliding microtome (Leica, Microsystems, Heidelberg, Germany), and sorted into 2 alternate sets of sections in a 24-well plate containing cryoprotectant solution (PBS containing 30% sucrose, 30% ethylene glycol, and 10% polyvinylpryrrolidone) and stored at −20 C° until processing for immunoreactivity. The remaining caudal portions were frozen and sectioned frontally into 41 μm sections, sorted into 2 alternate sets, and stored identically to the olfactory sections.
To validate whether anatomical differences in AOB between genotypes also occurred in males, we examined male olfactory bulbs (other data on these animals will be reported in a separate publication). Adult male brains were collected and processed exactly as described for females. Males had been singly housed for one month. Within 90 min of tissue collection, males were exposed to either nothing, clean bedding or dirty bedding before being sacrificed.
Procedures to visualize Fos-IR replicated those reported previously (Hasen & Gammie, 2005). Briefly, sections were incubated with anti-rabbit Fos antibodies (1:20,000; Calbiochem, San Diego, CA), incubated in biotinylated goat anti-rabbit secondary antibodies (1:500; Vector Laboratories, Burglingame, CA), exposed to an avidin-biotin complex (Vector), and stained with diaminobenzidine (0.7 mg/ml; Sigma, St. Louis, MO) as a chromagen, enhanced with 0.008% nickel chloride. Stained sections were then mounted, dehydrated, and coverslipped.
Fos-IR was determined by quantifying the staining in a given region of the brain. We projected sections in bright-field from an Axioskop Zeiss microscope (Zeiss, Gottingen, Germany), via an Axiocam Zeiss digital camera, to a computer running KS300 software (Zeiss). The software performed thresholding and counting with a paradigm similar to one previously employed (Gammie et al., 2004, Hasen & Gammie, 2005, Rhodes et al., 2003). The KS300 software identifies each spot on the digital photo that meets specific criteria as a cell, and counts the number of these spots. This number represents the amount of Fos staining in that region of the brain for that animal. Each brain section was at least 41um distant from its adjacent sections. Given that the nuclei of neurons and glia are rarely larger than 30um, we could be confident that we were not counting the same cell nucleus twice in any region. Each region of the brain was identified and located using standard brain landmarks (Paxinos & Franklin, 2001) and counted in a frame of uniform size, with the exception of the mitral (MiA) and granular (GrA) layers of the accessory olfactory bulb (AOB). Because these cell layers are small and irregularly shaped, Fos-IR was quantified in these regions using a macro built in KS300 to manually outline the layer on a digital image. The software then calculated amount of Fos staining within the outlined area, using the same criteria used in the rest of the brain. The outlining macro also calculated the area, in square pixels, of the area selected and was used to measure the size of the GlA of the AOB. Measures taken to ensure Fos-IR was measured consistently between samples are identical to those previously reported (Hasen & Gammie, 2005, Hasen & Gammie, 2006, Rhodes et al., 2003).
Dam weights were compared by genotype using one-way ANOVAs; average individual pup weights were calculated for each litter by dividing total litter weight by number of pups and these average pup weights were compared by genotype using one-way ANOVAs. Pup counts were compared by genotype using one-way ANOVAs or Kruskal-Wallis ANOVAs on ranks when data were non-normally distributed.
Maternal behavior observations were grouped into categories and quantified as a percentage of total observations. For example, time on nest would be quantified by dividing the number of observations where the dam was on nest with the total observations in that testing period. These percentages were compared for each group using one-way ANOVA. Non-normally distributed data was transformed where possible. Maternal aggression behavior was compared using one-way ANOVAs, or Kruskal-Wallis one way ANOVAs on Rank where data was of non-equal variance.
When analyzing Fos-IR, we performed planned one- and two-way ANOVAs between the control and tested groups of each genotype, as well as between genotypes in each test condition. The one-way comparisons address a common problem of unequal variance in two-way ANOVA comparisons of Fos-IR in transgenic and wild-type animals. Non-normal data and data of unequal variance was analyzed using Kruskal-Wallis ANOVAs on ranks or Mann-Whitney rank sum tests. Results for all two-way ANOVA tests are included as supplemental tables. The cutoff value for significance was set at p < 0.05. In order to account for false positives in groups with many tests, we determined the global, experiment-wise, false discovery rate that occurred for our data when applying the standard 0.05 p-value cut off to reject the null hypothesis (Storey, 2002). This was done using the open source software, Qvalue (see URL: http://faculty.washington.edu/~jstorey/qvalue/). This technique typically yields a much more stringent cutoff for significance. These more stringent cutoff values are included in data tables, and the reader is encouraged to take the possibility of false positives into account when considering the data.
No difference was found in mean weight of dams or pups on PPD1 (dams: WT 35.9 g, KO 35.9 g, F(1, 43) = 0.08, p = 0.781; pups: WT 1.8 g, KO 1.8 g, F(1, 43) = 0.1, p = 0.724) or PPD6 (dams: WT 38.8 g, KO 39.7 g, F(1, 38) = 0.8, p = 0.371; pups: WT 4.1 g, KO 4.0 g, F(1, 39) = 0.3, p = 0.592), nor was there a difference in litter size between the WT and KO animals (day 1 pup count, WT 12.0, KO 12.1, H = 1.5, p = 0.218).
When averaged across all days and both off and on-nest categories, KO dams nest-built less frequently than WT dams (frequency of nest-building, H = 5.2, p = 0.022) (Fig. 1D). Mean percentage of nest-building bouts for WT dams was 14.5%, for KO dams, 6.0%. No other significant differences between KO and WT maternal behaviors emerged in these daytime observations. There was no effect of the tprc2 gene on pup retrieval in any test on any day (e.g. latency to 1st pup on PPD4 on ranks, H = 1.0, p = 0.310).
Of 20 KO dams tested for maternal aggression only 2 were aggressive towards a male intruder. Mouse 34 was less aggressive (time aggressive PPD5 4s) and was not tested on PPD6. Mouse 40 had levels closer to that of the WT average (time aggressive PPD5 32s) and was tested on PPD6. Twenty-two of 24 WT animals were aggressive, with a mean total time aggressive on PPD 4 of 41.5 ± 7 s. This difference in how many animals showed aggression within each group was statistically significant (χ-square of 26.1 with 1 degree of freedom, p < 0.0001), although comparisons of total time aggressive between just the aggressive animals in each group on PPD 4 (F(1, 23) = 1.89, p = 0.183) and PPD 5 (F(1, 23) = 0.69, p = 0.413 ) were not (Fig. 1C). Comparisons between all animals in each group show that WT dams are more aggressive towards intruders, showing shorter latencies to bite, attacking more frequently and for more time than KO animals (Fig. 1A,B). Of the two aggressive KO dams, the more aggressive one showed an average nest-building frequency of 10%. The KO dam that showed a low level of aggression nest-built with a average frequency of 5%.
In the process of quantifying AOB Fos-IR, it became apparent that the glomerular layer of the AOB (GlA) in many animals was both smaller and less uniformly shaped than the norm (Fig. 2). Analysis of the size of the GlA showed a main effect of genotype (entire GlA, F(1, 33) = 5.3, p = 0.028), but no effect of test (F(1, 33) = 0.004, p = 0.952) or any interaction of effects (F(1, 33) = 2.2, p = 0.146). These differences appeared in both the left (L) and right (R) GlA, (L, F(1, 37) = 29.9, p < 0.0001; R, F(1, 35) = 12.0, p < 0.001, non-equal variance) but were concentrated in the posterior portion (GlAp) (GlAa F(1, 34) = 0.7, p = 0.421; GlAp F(1, 34) = 50.1, p < 0.001) (Fig. 2G). In order to determine if this defect was restricted to females, we examined the AOBs of 24 WT and 21 KO male mice (Fig. 2A) and found similar effects, except that in the males the GlAa is also significantly smaller in KO-Than in WT AOBs (entire GlA F(1,44) = 83.8, p < 0.001; GlAa F(1,44) = 19.2, p < 0.001; GlAp F(1,44) = 126.5, p < 0.001). These males had been exposed to either nothing, clean bedding or dirty bedding prior to sacrifice, but there was no effect of this treatment on GlA size (data not shown).
Fos-IR was used to indirectly evaluate neuronal activity in the AOBs of KO and WT dams. We hypothesized that testing would increase Fos-IR in the AOBs of WT but not KO dams, since deleting the trpc2 gene would sharply reduce inputs from the VNO to the AOB. In comparisons of the AOB between T and U dams of both genotypes, we found a consistent pattern of differences, with WT dams showing significantly higher levels of Fos staining than KO dams, across both the mitral (MiA) and granular (GrA) layers, regardless of test condition (see Supplemental Table 1 for two-way ANOVA F and p values and Table 1 for one-way ANOVA means and p values). In most but not all regions, WT dams showed higher levels of Fos-IR in the tested condition versus the untested. For example, WT-T dams showed significantly higher levels of Fos than WT-U dams in the right MiA, (p = 0.037), but not the left MiA (p = 0.896).
The vomeronasal amygdala is a subset of brain regions that process signals from the vomeronasal organ, via the mitral layer of the AOB (Dulac & Wagner, 2006). It includes the anterior medial amygdala (MeA), the posterodorsal medial amygdala (MePD), the posteroventral medial amygdala (MePV), the posteromedial cortex of the amygdala (PMCo), the bed nucleus of the accessory olfactory tract (BAOT), and the dorsal and ventral parts of the bed nucleus of the stria terminalis (BNSTD and BNSTV). We examined Fos-IR in the vomeronasal amygdalas of U and T mice, hypothesizing that basal levels of Fos-IR would be identical in the two U groups, but that testing would increase Fos-IR only in the WT-T group. This hypothesis held in the BNST-D, the PMCo, the MeA, and the MePD (F and p values for two-way ANOVAs for all brain regions are reported in Supplemental Table 2, one-way ANOVA mean and p values are reported in Table 2). Testing increased Fos-IR in the BAOT and BNST-D of WT mice and not KO mice, but not enough to create a significant difference between the two T groups in either region. No significant effect of testing on Fos-IR appeared in the BNST-V, though there are trends in the predicted directions.
The central (CeM) and basolateral (BLA) regions of the amygdala are associated with maternal defense behavior (Gammie & Nelson, 2001), as well as with fear- and aggression-motivated behaviors, more generally (Hans-Christian Pape, 2003, Veenema & Neumann, 2007). We hypothesized that these regions would show an increase in Fos-IR with testing in both WT and KO animals, since the intruder male would present olfactory, visual and auditory cues that would stimulate these emotional responses regardless of the status of vomeronasal signaling. However, while testing increased Fos-IR as predicted in the WT dams, it had no impact on Fos-IR in the KO dams (CeM WT-U vs WT-T, p = 0.006, KO-U vs KO-T p = 0.861; BLA WT-U vs WT-T, p = 0.001, KO-U vs KO-T p = 0.647).
We measured Fos-IR in 10 sub-regions of the hypothalamus: the lateral (LPO) and medial (MPA) pre-optic area – including just the dorsal part of the MPA, MPAd – the medial pre-optic nucleus (MPOM), the paraventricular nucleus (PVN), the ventromedial hypothalamus (VMH), the posterior part of the anterior hypothalamus (AHP), the lateral hypothalamus (LH), an anterior point in the hypothalamic attack area (HAAa), and a posterior point in the hypothalamic attack area (HAAp). We hypothesized that regions previously associated with maternal aggression – LPO, PVN, VMH, HAAa, HAAp, LH, MPA, and MPOM (Hasen & Gammie, 2005, Hasen & Gammie, 2006) – would show increased Fos-IR with testing in all aggressive animals, although increases would be more modest in KO animals, since the hypothalamus is a main target of signals from the AOB (Keverne, 1999). Only a subset of these regions showed increased Fos-IR with testing in the WT animals (Table 3), and there was no effect of testing on Fos-IR in any part of the hypothalamus in KO dams. We also hypothesized that Fos-IR in some regions associated with nursing or other pup-directed behaviors might be different between the two genotypes in the untested state, since previous work has shown decreased maternal care in trpc2−/− dams and our study had found lower levels of nest-building in KO dams. However, none of these regions – the PVN, MPOM, MPA, LH – showed any difference without testing.
The lateral septum is a key regulator of emotion-driven behavior (Sheehan et al., 2004), and has been implicated in the down-regulation of maternal aggression by corticotropin releasing factor (Bronikowski et al., 2004). We have previously found increased Fos-IR in the lateral septum with testing in some (Gammie et al., 2004), but not other studies (Hasen & Gammie, 2005). We have also found increased activity in the septum associated with the injection of aggression-inhibiting compounds (Gammie et al., 2004). Therefore, we hypothesized that both WT and KO mice might show increased Fos-IR in the septum with testing, though not necessarily in the same cell populations. Instead, we found that in the lateral ventral septum (LSV), testing increased Fos-IR of WT animals (WT-U vs WT-T, p = 0.049), but that basal levels of Fos-IR were higher in KO animals (WT-U vs KO-U, p = 0.019) (Fig. 3). No effect of testing or genotype was found in the lateral dorsal septum (LSD, data not shown).
We measured Fos-IR in two regions of the brain associated with the processing of olfactory cues: the piriform cortex (Pir) and the olfactory tubercle (Tu). Because the deletion of the trpc2 gene has no direct impact on olfaction, we expected to find increased Fos-IR with testing in mice of both genotypes. However, there was no significant difference between the U and T groups in either genotype. There were significant differences between the two T groups in the lateral portion of Tu (TuL) that drove a significant difference in the whole region (Table 4).
Several regions in the midbrain/brainstem have been closely correlated with the production of maternal defense behavior, including the dorsal and lateral periaqueductal grey (DPAG, LPAG), the dorsal raphe (DR), and the locus coeruleus (LC) (Gammie & Nelson, 2001, Hasen & Gammie, 2005, Hasen & Gammie, 2006). We hypothesized that testing would increase Fos-IR in all these regions in both genotypes. This hypothesis held true for WT animals everywhere except the DR. There was no increase in Fos-IR with testing in any brainstem region for KO mice (Table 5).
We measured Fos-IR in the paraventricular nucleus of the thalamus, a region where we have previoulsy found increased activity in association with maternal aggression. We found no effect of testing or genotype in either a two-way ANOVA (Supplemental Table 2) or in one-way ANOVAS (data not shown). We also examined the anterior cingulate cortex, another region associated with maternal aggression in our previous studies, and again found no effect of testing or genotype in a two-way ANOVA, though we did find a significant effect of genotype within the tested animals in a one-way ANOVA (Mann Whitney Rank Sum, H = 3.89, p = 0.049). Finally, we examined the caudate putamen (CPu) as a control and found no effect of testing or genotype in the two-way ANOVA (genotype F (1, 38) = 1.2, p = 0.285, testing F (1, 38) = 0.1, p = 0.706, interaction F (1, 38) = 1.8, p = 0.185) or in any one-way ANOVA (data not shown).
There were too few aggressive KO mice to perform statistical analysis. Supplemental Table 3 presents Fos-IR values for both of these mice next to the means for the 4 groups. Mouse 34 had Fos-IR staining in her AOB lower than the means for the KO-U group, but higher than mean Fos-IR in LSV, PVN, HAAa, BAOT, CeM, MePD, and VMH. Her GlAa and GlAp were close to the average for KO mice. Mouse 40 had levels of Fos-IR resembling WT-T means in the AOB, BNST, and PMCo; Fos-IR in other regions of her brain were low or normal in comparison to other KO-T dams. Both her GlAa and GlAp were larger than that of the average WT animal (this animal was genotyped twice to confirm her KO status).
In this study, we probe the role of the VNO in both maternal behaviors and the brain regions that govern them. We employ a novel transgenic mouse; a trpc2 mutant in a background strain that shows robust levels of maternal aggression. We report that, in these mice, deletion of the trpc2 gene leads to decreases in nest-building, as well as maternal aggression. We show that the accessory glomerular layer is deformed in both sexes of mice carrying the deletion. Our data also show an impact of VNO signaling on regions of the brain that mediate maternal aggression, but not on regions that mediate other aspects of maternal behavior or olfaction.
The choice to move the tprc2 deletion onto another strain was driven by concerns about a possible confound between strain and transgene. Existing trpc2 lines were all on C57/B6 background, and this strain shows very low levels of maternal aggression compared to wild or outbred lab mice (Parmigiani et al., 1999). It seemed likely that the elimination of maternal aggression seen in other studies of trpc2−/− dams (Kimchi et al., 2007, Leypold et al., 2002) might be partly a function of low aggression, and that in mice with more normal levels of maternal aggression, the deletion of the gene might have subtler effects.
In fact, only two of 20 KO mice showed any aggression, one at levels comparable to WT dams, another much lower. The aggression in these two mice might represent a rescue of function by genetic background. Auditory signals can elicit maternal aggression (Bean et al., 1986), and these mice may have been more sensitive to auditory cues from males. Variables such as the intruder's weight, his previous experience of intruder-testing, or his social status may also have enhanced the response of these females. However, in our experience, the female's previous encounters with males and her genetic background are the most important variables in determining her response.
Signaling from the relatively intact GlAa of these mice (see below for more discussion) might be enough to trigger aggression in a minority of mice. Deletion of genes coding for V1R receptors – which are found only in the apical VNO (VNOa ) – severely reduces maternal defense behavior (Del Punta et al., 2002). Fos-IR in the AOB of the more aggressive KO mouse was very high compared to the mean of the KO-T group (Supplemental Table 3), suggesting that her VNOa might have signaling via her GlA to induce aggression. However, it should be noted that this animal's GlAp more closely resembled the mean size of GlAp in the WT animals than that of the other KO dams. While Fos-IR levels in her brain were often lower than the average for the KO-T group, her Fos staining in PMCo and BNST-D, both part of the vomeronasal amygdala, were high. Taken together, this data suggests that there was some rescue of function in the VNO of this mouse, which drove maternal aggression via signaling from the AOB, although no conclusions can be drawn from this single sample.
The mildly aggressive KO dam was in the U group, thus her levels of Fos-IR do not reflect her brain's response to male intruders. Fos-IR staining in her AOB is lower than the means for the KO-U group (See Supplemental Table 3), making it unlikely that residual VNOa function was responsible for her aggression. However, several regions in her brain show much higher Fos-IR than typical for KO-U, including LSV and HAAa. In previous work, we show that lactating mice have significantly higher Fos-IR in LSV and HAAa than do virgin mice (Hasen & Gammie, 2005). Perhaps the elevated Fos-IR in this aggressive KO mouse reflects higher basal levels of activity – a more “primed” state – which allowed her to muster a modestly aggressive response to an intruder.
The reductions in nest-building in KO mice support recent findings that KO dams spend less time on the nest (Kimchi et al., 2007). This result indicates that nest-building behavior in dams is sensitive to input from the VNO; the most logical source of this input would be pheromonal cues from pups. In contrast with other studies, we found no other differences in maternal behavior, but evaluated these behaviors only during the daylight hours, when mice are typically quiescent (a subsequent study will examine the impact of the mutation on maternal behavior exhibited during the dark cycle).
Our finding of biased deformity in the GlA of both female and male KO mice (Fig. 2) supports a hypothesis presented elsewhere (Kelliher et al., 2006) that deletion of the trpc2 gene does not eliminate signaling from the VNO. An earlier study showed biased reductions in axonal projections from the VNO to the AOB in trpc2−/− mice, with more neurons from the VNOa surviving (Stowers et al., 2002). This study also found that trpc2−/− mice were unable to sense pheromonal cues from urine. However, trpc2−/− females maintain the capacity for pregnancy block, a form of pheromonally-mediated memory reliant upon cues in male urine (Kelliher et al., 2006). In this study, electrophysiological recordings showed activity in the basal VNO (VNOb) of trpc2−/− females, suggesting that a trpc2-independent mechanism allows these mice to sense pheromones during mating. The relatively normal size of the GlAa in our KO mice suggests that the VNOa of these animals may also be functional. Volatile, low molecule weight fractions of male urine have been shown trigger Egr-1 expression in the GlAa (Peele et al., 2003), and the anterior AOB of female mice is especially sensitive to cues from male mice (Brennan et al., 1999, Halem et al., 1999). Mating trpc2−/− females may be creating pheromonal memories of males via V1R receptor-mediated signaling in the VNOa. A recent study shows that small peptide ligands of the major histocompatibility complex class I molecules are sufficient to induce pregnancy block and that these peptides bind V2R receptors (Leinders-Zufall et al., 2004), confirming that the VNOb and posterior AOB are typically involved in these processes. It remains to be determined whether these peptides are always necessary for pregnancy block, or if signals activating the VNOa can also perform this function.
This study uses Fos-IR as an indirect marker of neuronal activity to identify a subset of the brain regions involved in maternal aggression that are specifically triggered by pheromonal cues. Previous studies show altered Fos expression in neurons in response to neurotransmitters (Hetzenauer et al., 2007), that Fos expression can be linked to changes in cAMP, a common neuronal second messenger (Cho et al., 2007), and that Fos activity in neurons correlates with electrical excitability changes in some neurons (Batterham et al., 2002). However, Fos expression does not discriminate between increased or decreased excitability (Dragunow & Faull, 1989) and in some cases glial cells can show altered Fos in response to a neurotransmitter (Edling et al., 2007). Here, we present profile densities for each region. It is possible that these provide an estimate of the number of activated neurons, but because our methods use a defined threshold, the numbers counted could either be higher or lower depending upon where the threshold is set. Thus, we see our approach as useful in identifying regions with significant changes in activity, but not necessarily as providing information on the total number of neurons showing this change.
We began with the AOB, confirming that signaling in both layers was lower in KO mice than in controls, in both the tested and untested states (Table 1). Surprisingly, only the R MiA showed increased Fos-IR in comparisons of WT-U to WT-U mice. Previously, we had shown increases in both MiA and GrA Fos-IR with testing in a similar paradigm (Hasen & Gammie, 2005). This difference may result from lighter Fos staining in the current experiment, or from a difference in the animals used in the two studies.
In the vomeronasal amygdala, a group of brain regions within the striatum and amygdala receiving inputs from the VNO via the AOB, we expected to see increased Fos-IR with testing in the WT but not the KO mice. This hypothesis held true for all regions except the BNST-V, where testing did not lead to significant increases in the WT dams (Table 2). In other regions of the amygdala, we expected to see similar increases in Fos-IR with testing in both genotypes, but this proved not to be the case. In both the CeM and BLA, testing increased Fos-IR only in WT mice – a pattern mimicked throughout much of the brain, including regions in the midbrain/brainstem (Table 5), that likely reflects the much lower levels of aggression in the KO mice (Fig. 1).
In the hypothalamus, our hypothesis that KO animals would show increased Fos-IR with testing was also not supported (Table 3). More surprising was the similarity in Fos-IR in these regions in comparisons between the WT-U and KO-U dams. This might have been where differences in nest-building between these two groups would be reflected in brain activity, especially in the dorsal MPA, a region associated with nest-building (Jacobson et al., 1980).
Our results in lateral septum (Fig. 3) demonstrate an effect of genotype consistent with our previous work. We have shown that Fos-IR in this area increases both with testing and with exposure to substances that inhibit maternal aggression (Gammie et al., 2004) and decreases with exposure to substances that enhance aggression (Lee & Gammie, 2007), suggesting that different cell populations in LS may play restrictive and permissive roles with regard to the behavior. Thus, higher levels of Fos-IR in our WT-T might reflect increased activity in the permissive population, while higher levels of Fos-IR in the KO-U mice (relative to the WT-U) may indicate more activity in a restrictive population. Future studies might test the hypothesis that pheromonal cues from pups facilitate maternal aggression by inhibiting activity in this restrictive population of LS cells.
We hypothesized that both KO-T and WT-T dams would show higher levels of Fos-IR than KO-U and WT-U dams in regions associated with olfaction, since the gene deletion should have no effect on olfactory processing. Surprisingly, this was not the case. Instead, we found no significant differences in Fos-IR with testing in any olfactory processing region, regardless of genotype (Table 4). However, this may be a function of factors specific to this experiment, such as slightly lighter than normal overall Fos staining, since previous studies comparing Fos-IR between U and T dams have found responses in Pir (Hasen & Gammie, 2005). We were also surprised to see an effect of genotype on Fos-IR in the Tu of tested animals. This finding may be an artifact, but warrants further study; it seems plausible that cues from the VNO might upregulate activity in the olfactory system as a way to enhance the animal's response to a potent stimulus.
One confound in the cFos study was the behavioral response of the aggressive animals; the increased Fos-IR we see in WT-T animals reflects brain activity in the production of aggressive behavior, as well as the processing of pheromonal cues. One way to tease apart these variables might be to examine the Fos-IR of aggressive KO animals, where presumably there was no pheromonal input. The aggressive KO-T dam had levels of Fos-IR resembling WT-T means in the AOB, BNST, and PMCo – all regions highly associated with signaling from the VNO; Fos-IR in other regions of her brain were low or normal in comparison to other KO-T dams (Supplemental Table 3). Both her GlAa and GlAp were actually larger than that of the average WT animal (this animal was genotyped twice to confirm her KO status), suggesting her aggression may have arisen from rescued activity in the VNO. These data seem to reveal more about the importance of the VNO in triggering maternal aggression than they do about the brain production of the behavior. Future research might address this question by offering WT and KO females a pheromonal cue in the absence of an intruder, and measuring Fos-IR in this context.
This study of the VNO demonstrates its importance to the production and maintenance of maternal behaviors. While recent studies show that some pheromonal cues are processed via the main olfactory bulb (Kelliher, 2007, Wang et al., 2006), this study demonstrates the critical function of the VNO in eliciting and supporting nest-building and offspring defense. It seems likely that lactating dams integrate pheromonal data from both their pups and from other mice via the VNO, and that this integration is necessary for maternal aggression.
This work was supported by National Institutes of Health Grant R01 MH066086 to S.C.G and an AAUW American Fellowship to N.S.H. The authors wish to thank Dr. Catherine Dulac for generous donation of animals; Kate Skogen and Jeff Alexander for animal care; and Sharon Stevenson for technical assistance.