The sudden infant death syndrome is associated with a reduction in brainstem serotonin 5-hydroxytryptamine (5-HT) and 5-HT1A receptor binding, yet it is unknown if and how these findings are linked. In this study, we used quantitative tissue autoradiography to determine if post-natal development of brainstem 5-HT1A receptors is altered in two mouse models where the development of 5-HT neurons is defective, the Lmx1bf/f/p, and the Pet-1−/− mouse. 5-HT1A receptor agonist-binding sites were examined in both 5-HT-source nuclei (autoreceptors) and in sites that receive 5-HT innervation (heteroreceptors). In control mice between post-natal day (P) 3 and 10, 5-HT1A receptor binding increased in several brainstem sites; by P25, there were region-specific increases and decreases, refining the overall binding pattern. In the Lmx1bf/f/p and Pet-1−/− mice, 5-HT1A-autoreceptor binding was significantly lower than in control mice at P3, and remained low at P10 and P25. In contrast, 5-HT1A heteroreceptor levels were comparable between control and 5-HT-deficient mice. These data define the post-natal development of 5-HT1A-receptor binding in the mouse brainstem. Furthermore, the data suggest that 5-HT1A-heteroreceptor deficits detected in sudden infant death syndrome are not a direct consequence of a 5-HT neuron dysfunction nor reduced brain 5-HT levels.
To elucidate the developmental relationship between serotonin (5-HT) levels and 5-HT1A receptors in the brainstem, we examined 5-HT1A binding in two 5-HT-deficient mouse models. In nuclei containing 5-HT neurons, 5-HT1A binding was decreased (autoreceptors), while binding was maintained in projection sites (heteroreceptors). Thus, brainstem 5-HT1A-heteroreceptor-binding sites do not appear developmentally sensitive to reduced brain 5-HT levels.
Serotonin 5-hydroxytryptamine (5-HT) is a major monoamine neurotransmitter that communicates with multiple receptor subtypes throughout the neuroaxis to regulate key physiological and behavioral functions. Importantly, it helps regulate brainstem respiratory and autonomic responses to homeostatic challenges, in part by virtue of actions at the 5-HT1A receptor (Darnall et al. 2005; Curran and Leiter 2007; Audero et al. 2008; Pham-Le et al. 2011). 5-HT1A receptors are located somatodendritically on 5-HT neurons, where they function as autoreceptors. In addition, 5-HT1A receptors are located on non-5-HT neurons and act as heteroreceptors in regions of the brainstem and forebrain that receive 5-HT innervation. Autoreceptor and heteroreceptor populations are known to exhibit region-specific adaptations to the over- or under-availability of ligand [reviewed by (Hensler 2003)].
Alterations in brainstem 5-HT1A receptors are associated with several human disorders (Kinney et al. 2011) including the sudden infant death syndrome (SIDS) (Saito et al. 1999; Panigrahy et al. 2000; Ozawa and Takashima 2002; Paterson et al. 2006; Duncan et al. 2010; Waters 2010). The leading cause of post-neonatal mortality in the United States today, SIDS is defined by sleep-related death in the first post-natal year of life that is unexplained by a complete autopsy and death scene investigation (Kinney and Thach 2009). The majority of SIDS deaths are associated with asphyxia-generating circumstances that appear to trigger death, for example, rebreathing exhaled gases in the face-down (prone) sleep position (Pasquale-Styles et al. 2007; Kinney and Thach 2009). Accordingly, a leading hypothesis in SIDS research today is that SIDS is caused by a brainstem abnormality that impairs the ability to generate protective responses to life-threatening challenges (Kinney and Thach 2009; Kinney et al. 2009). Indeed, 5-HT1A receptors reductions have been observed in the brainstem of SIDS infants that involve both auto- and heteroreceptor populations (Kinney et al. 2003; Paterson et al. 2006; Machaalani et al. 2009; Duncan et al. 2010). Lower 5-HT1A-receptor binding in SIDS cases is associated with decreased medullary tissue content of 5-HT and tryptophan hydroxylase 2 (TPH2), the rate-limiting biosynthetic enzyme for 5-HT, suggesting a 5-HT-deficient disorder (Duncan et al. 2010).
Two murine models have been instrumental in examining the developmental role of 5-HT on cardiorespiratory function and thermoregulation, that is, homeostatic processes potentially relevant to the pathogenesis of SIDS (Erickson et al. 2007; Hodges et al. 2009; Kinney and Thach 2009; Buchanan and Richerson 2010; Cummings et al. 2011a; Richerson and Buchanan 2011). These 5-HT-deficient models include a conditional removal of the transcription factor Lmx1b selectively in Pet1-expressing cells (Lmx1bflox/flox;ePet-cre) resulting in the ‘Lmx1bf/f/p’ mouse in which there is nearly a complete loss of 5-HT neurons by midgestation and consequently 5-HT in the CNS (Zhao et al. 2006). The other model is the Pet-1−/− mouse in which 5-HT neurons persist in the brain but the majority fail to differentiate appropriately and do not produce 5-HT (Hendricks et al. 2003; Erickson et al. 2007; Cummings et al. 2011a). Detectable 5-HT synthesis appears to occur in only about 30% of the normal number of 5-HT neurons, and these residual neurons innervate a selective subset of brain regions, predominantly regions associated with stress responses (Kiyasova et al. 2011).
In this study, we examined how 5-HT1A-receptors-binding patterns develop in the brainstem in these two mouse models with 5-HT deficiency. We applied the methodology of quantitative tissue autoradiography, a technique used to study SIDS tissue, in part to allow for comparisons to 5-HT1A-receptor binding in human brainstem disorders (Paterson et al. 2006). We tested the hypothesis that decreases in 5-HT1A-heteroreceptor binding in the brainstem during development may be caused by 5-HT neuron dysfunction or associated 5-HT deficiency. Given the absence of 5-HT neurons in the Lmx1bf/f/p mouse, and the demonstrated loss of 5-HT1A receptor gene expression in the dorsal and median raphe nuclei of Pet-1−/− mice (Liu et al. 2010; Jacobsen et al. 2011), we also anticipated finding a large decrease in medullary 5-HT1A autoreceptors. As tissue autoradiography does not reveal the cellular location of receptors, we operationally defined 5-HT1A autoreceptors as those within the 5-HT source nuclei and 5-HT1A heteroreceptors as those localized to nuclei receiving 5-HT projections.
Materials and methods
Two different mouse strains were used in this study: Lmx1b conditional knockout in Pet1-expressing cells [Lmx1bflox/flox;ePet-Cre/+ (Lmx1bf/f/p)] and Pet-1−/− mice. Lmx1bf/f/p mice were bred at Yale University in New Haven, Connecticut, and Pet-1−/− mice at the Geisel School of Medicine at Dartmouth, Lebanon, New Hampshire. Animal protocols were approved by the Institutional Animal Care and Use Committees at these institutions, and were consistent with guidelines of the National Institutes of Health. For Lmx1bf/f/p mice, controls were siblings lacking Cre recombinase on a C57BL/6 background. For Pet-1−/− mice, controls were heterozygote and control siblings from a mixed C57BL/6 and 129 background; heterozyotes have been shown to have normal number of 5-HT immunolabeled neurons (Hendricks et al. 2003; Cummings et al. 2011a). Both male and female mice were included and were analyzed at P3, P8-10, and P25. These ages were selected based on the severe respiratory deficits in Lmx1bf/f/p mice at neonatal (< P4) ages (Hodges et al. 2008) and vulnerability of Pet-1−/− mice to repeated hypoxic challenge at P8/10 (Cummings et al. 2011a). Group size for the P3 and P8-10 groups averaged 7.5 and 7 individuals, respectively, with a minimum of five per group; for the P25 age groups, the average group size was 4.75 individuals, with a minimum of four individuals per group.
Under deep anesthesia, mouse brains were removed, frozen, and stored at −80°C until sectioned 20 μm thick with a cryostat. Sections were collected serially onto sets of four slides such that each slide contained every fourth section. Slides were stored at −80°C until processed for autoradiography.
Slides were gradually warmed to 18–24°C before tissue incubations for autoradiography, using a protocol adapted from human studies (Paterson et al. 2006; Duncan et al. 2010). One quarter of the slides collected for each mouse was used for total radioligand binding, while another quarter containing the adjacent sections was processed to determine non-specific-binding levels. Buffer solution consisted of 0.17 M Tris with 0.01% ascorbic acid and 4 mM CaCl2 (pH 7.6). At 18–24°C, slides were pre-incubated in buffer solution for 30 min. Then, sections were incubated in 4 nM 3H-8-hydroxy-2-(di-n-propylamino)tetralin and 10 μM pargyline diluted in fresh buffer solution for 1 h for total specific binding. For non-specific binding, 10 μM 5-HT was added to the 2-(di-n-propylamino)tetralin-containing solution, and slides were incubated for 1 h. Incubated slides were washed in ice-cold incubation buffer solution twice for 5 min. All sections were then air dried for 15 min using a cold air stream, and then placed on a slide warmer for 30 min to dry completely.
Quantitative data analysis
Slides for total and non-specific binding were exposed on 3H-sensitive film with one 3H-standard slide (Amersham, Pittsburgh, PA, USA) for 3 weeks. For quantification of receptor-binding density, NIH Image J software (National Institutes of Health, Bethesda, MD, USA) was used. To convert optical density into femtomoles/milligram tissue (fmol/mg), a calibration curve was generated using the 3H-standards. Seven sites were selected for analysis (Fig. 1) that were representative of auto- and heteroreceptor sites: three nuclei that contain 5-HT source neurons [raphe obscurus, dorsal raphe, and lateral paragigantocellular nucleus (LPGi)], and 4 sites that receive 5-HT projections (spinal trigeminal nucleus, inferior colliculus, interpeduncular nucleus, and cerebellum). At least in adults, the majority of 5-HT1A receptors co-localize with markers of serotonin neurons in the raphe nuclei; thus, we infer these areas largely represent autoreceptor populations (Czesak et al. 2012). For each region, 5-HT1A-receptor-binding density was measured in three sections/mouse and the values averaged to determine total and non-specific binding. Non-specific-binding values were subtracted from those of total binding to obtain specific binding. Subsequently, individual values were averaged to yield group means and standard error of the mean.
For each mutant group, data were first analyzed using gender as a factor. When there was no significant effect of gender, males and females were pooled and analyzed together. The data were compared using a three-factor anova for age (at three levels), genotype (at two levels), and region, with region as a repeated measure. Following a significant main effect of region, each region was analyzed independently using slice anovas for age and genotype with Tukey or Student's-t post hoc tests, depending on the number of comparison groups. Mutant groups were not directly comparable to each other because of differences in genetic background, rearing location, and ages at harvesting, which were associated with some modest differences between the two control groups in 5-HT1A-receptor binding. Therefore, to evaluate the relative loss of receptor binding in each mutant line (Fig. 5), the binding values were normalized to the appropriate control values yielding a ‘percent retained binding’ value (i.e., mutant binding/mean control binding expressed as a percent). These values were averaged, and a standard error was determined for each genotype and age. 5-HT1A-receptor-binding deficiencies in each mutant in the dorsal raphe nucleus and the raphe obscurus were then compared using anova followed by Student's t-test.
In control mice at P3, the dorsal raphe, inferior colliculus, and spinal trigeminal nucleus had the highest levels of 5-HT1A binding (greater than 30 fmol/mg) (Figs 1-3). Intermediate binding, in the range from 20 to 30 fmol/mg, was found in the raphe obscurus and LPGi, and low binding (< 20 fmol/mg) in the remaining sites analyzed, including the cerebellum. Between p3 and P10, in every site measured except the cerebellum, there were significant increases in 5-HT1A-receptor binding (Figs 1, 2). Between P10 and P25, control mice showed region-dependent changes in 5-HT1A–receptor-binding levels (Figs 1, 3). Two sites exhibited further increases in binding, including the interpeduncular nucleus and the dorsal raphe nucleus. In four sites, the inferior colliculus, LPGi, spinal trigeminal nucleus and raphe obscurus, there were decreases in 5-HT1A-receptor binding. These patterns of developmental changes in the control mice for Lmx1bf/f/p were the same in the Pet-1−/− control group (Fig. 4).
Genotype Effects, Lmx1bf/f/p mice
At P3, in comparison to control mice, Lmx1bf/f/p mice showed significantly reduced binding only in two sites, the dorsal raphe nucleus and raphe obscurus (Figs 1-3), that is, major sites of 5-HT source neurons in the brainstem. At P10, 5-HT1A binding was reduced in the dorsal raphe, inferior colliculus and raphe obscurus in mutants compared to their controls (Fig. 3). At P25, the magnitude of the reduction of 5-HT1A binding in the dorsal raphe nucleus and raphe obscurus was severe, with binding in the Lmx1bf/f/p more than two-thirds reduced from control values (Fig. 3). The maintenance of heteroreceptor binding in the quantified areas appeared representative of additional heteroreceptor populations, such as in the nucleus of the solitary tract (Fig. 2).
Genotype effects, Pet-1−/−
In Pet-1−/− mice at P3, the only site with a significant reduction compared to control was in the dorsal raphe nucleus (Fig. 4). At P8, both the dorsal raphe and raphe obscurus, however, demonstrated significant reductions in binding compared to controls. At P25, the additional site of the LPGi in Pet-1−/− mice had significantly less binding then control values (Fig. 4). In comparison to Lmx1bf/f/p mice (Fig. 5), the loss of 5-HT1A binding in the dorsal raphe nucleus and raphe obscurus was comparable at P3 and P8/P10, however, at P25 5-HT1A-binding levels in Pet-1−/− mice were half that seen in their controls, a less severe loss than that seen in Lmx1bf/f/p mice.
Previous studies have reported an association between altered brainstem 5-HT neurochemistry and SIDS. Specifically, there is evidence for aberrant 5-HT neuron function including decreases in tryptophan hydroxylase 2 and tissue 5-HT content (Duncan et al. 2010). In addition, there are consistent reports of decreases in 5-HT1A-receptor binding (Paterson et al. 2006; Machaalani et al. 2009; Duncan et al. 2010). This not only involves 5-HT1A autoreceptors but also 5-HT1A heteroreceptors in multiple medullary sites. The question remains as to whether 5-HT1A-heteroreceptor deficits can be explained by a primary dysfunction in 5-HT neurons themselves. In this study, we examined the post-natal development of 5-HT1A receptors in two mouse models with genetically induced deficits in 5-HT neuron function. Our data indicate a decrease in 5-HT1A binding in both mouse strains in areas that are normally enriched with 5-HT neurons (autoreceptors), confirming and extending previous observations to younger ages and medullary sites (Liu et al. 2010; Jacobsen et al. 2011). However, heteroreceptors, or 5-HT1A receptors located in areas containing non-5HT neurons, were largely comparable to control values in the brainstem at all post-natal ages examined.
5-HT1A-receptor autoradiography is a classic approach used to quantify binding sites in the brain. Strengths of this approach include the ability to compare results to studies in human pathology cases, such as those employed in the study of SIDS. In addition, many studies using animal models have examined changes of 5-HT1A receptor levels in response to manipulations of the 5-HT system or to drug exposure, particularly in the case of serotonin-selective reuptake inhibitors (Welner et al. 1989; Hensler et al. 1991; Hensler 2002; Rossi et al. 2008). However, it should be noted that there are potential adaptations in 5-HT1A-receptor function, including changes in the ability to activate G-proteins and subsequent intracellular signaling cascades that are not detected using receptor autoradiography, and require alternative methods to evaluate (Fairchild et al. 2003; Hensler 2003). Indeed, receptor autoradiography provides view of 5-HT1A receptors that is distinct and complimentary to methods that assay protein, mRNA levels, or intracellular signaling.
In both 5-HT-deficient mutant mouse lines, there were decreases in 5-HT1A-autoreceptor levels. These reductions are likely a direct consequence of the relevant genetic mutations and associated cell loss or transfating rather than a compensatory change in response to the loss of the endogenous ligand, 5-HT. In Lmx1bf/f/p mice, the loss of Pet1 expression and indeed, 5-HT neurons themselves would be expected to produce a parallel loss of 5-HT1A autoreceptors (Ding et al. 2003; Zhao et al. 2006). Any residual 5-HT1A receptor binding in the raphe nuclei of Lmx1bf/f/p mice could be associated with non-5-HT neurons or glia. With respect to Pet-1−/− mice, loss of autoreceptor binding in the Pet-1−/− mice is consistent with previously detected decreases in mRNA expression in the dorsal raphe nucleus and loss of agonist-evoked effects on currents (Liu et al. 2010; Jacobsen et al. 2011). Indeed, in 5-HT neurons Pet1 is a direct transcriptional enhancer of the 5-HT1Areceptor gene, while 5-HT1Areceptor transcription in non-5-HT neurons is Pet1-independent (Jacobsen et al. 2011). Therefore, lack of transcriptional activation of the 5-HT1Areceptor gene would be expected to result in a lower level of 5-HT1A receptor expression in Pet1-dependent 5-HT neurons. However, in the Pet-1−/− mice, there is a residual population of 5-HT neurons that appear Pet1-independent, and these neurons could potentially express 5-HT1A receptors independent of Pet1 function (Hendricks et al. 2003; Kiyasova et al. 2011). Alternatively, a very low level of expression could persist in residual neurons within the area. These explanations could account for the greater magnitude loss of 5-HT1A-receptor binding in the Lmx1bf/f/p mouse compared to the Pet-1−/− mouse detected at P25.
Loss of the endogenous ligand, 5-HT, in both of these mutants does not appear to consistently impact 5-HT1A–heteroreceptor-binding levels in brainstem projection sites during post-natal development. Adaptations in 5-HT1A-receptor binding in response to manipulations that increase or decrease available 5-HT have primarily been studied in the adult forebrain and in the primary sources of forebrain 5-HT, the dorsal and median raphe nuclei [reviewed by (Frazer and Hensler 1990; Hensler 2003)]. These studies show that changes in 5-HT1A receptors can be complex and may not only involve changes in receptor number or distribution but also in signaling capacity, and furthermore, the specific changes that occur appear to be region dependent (Hensler 2003). In the most relevant example, the adult Pet-1−/− mouse using Western blot, compensatory increases of 5-HT1A receptors were detected in the hippocampus while decreases were found in the striatum (Jacobsen et al. 2011). Increases in hippocampal 5-HT1A receptors are also found in cases where 5-HT projections were chemically lesioned, or 5-HT1A-receptor signaling was blocked by administration of an antagonist (Patel et al. 1996; Abbas et al. 2007). Our results would extend these observations using genetic models with constitutive developmental losses of 5-HT and show that at several brainstem sites no changes in 5-HT1A-receptor binding are detected. Likewise, the results confirm that transcriptional control of 5-HT1A receptors is region dependent in that heteroreceptors throughout the brainstem is Pet-1-independent, as indeed, Pet1 expression is restricted to serotonin neurons (Hendricks et al. 1999; Pfaar et al. 2002; Albert et al. 2011; Jacobsen et al. 2011).
The maintenance of 5-HT1A heteroreceptors in the mutant mouse lines also suggests that 5-HT1A-heteroreceptor levels are unaffected by the severe physiological defects displayed in these mouse lines. Specifically, both mouse lines have lower weight gain during the first few weeks of life. In addition, respiratory defects in both lines are likely to impact brain oxygenation and pH: as neonates, Lmx1bf/f/p have dramatic and prolonged (1–2 min) spontaneous apneas (Hodges et al. 2009), and as adults, both Lmx1bf/f/p and Pet-1−/− have abnormally low respiratory responses to CO2 (Hodges and Richerson 2008; Hodges et al. 2011). Finally, both lines are severely compromised in their ability to thermoregulate during environmental cooling (Hodges and Richerson 2010; Cummings et al. 2011b; Hodges et al. 2011). While these effects are interpreted as consequences of the altered state of 5-HT neurons, the current results provide evidence that these physiological deficits do not independently impact 5-HT1A heteroreceptor binding in these mice.
This study further describes the developmental time-course of 5-HT1A-receptor binding in the brainstem during the early post-natal period in mice. Analysis of the 5-HT1A-binding sites in several brainstem areas revealed two phases of development in the mouse. Before P8-10, there are broad increases in 5-HT1A-receptor binding. After P8-10, there are region-dependent increases or decreases that serve to refine the overall pattern of 5-HT1A-receptor binding. Consistent with the time-course of changes observed, the first three post-natal weeks are known to constitute an important period of synaptic development and refinement in mouse (Bhatt et al. 2009). Indeed, P8, an age of apparent rapid changes in 5-HT1A-receptor binding, corresponds to an age of peek vulnerability to repeated hypoxic stress in Pet-1−/− mice (Cummings et al. 2011a). Likewise, an extended post-natal development of 5-HT1A-receptor-binding sites has been described in the forebrain (Miquel et al. 1994). In the rat hypoglossal nucleus, a similar biphasic development of 5-HT1A receptors is observed where mRNA and binding increase until day 7 and decrease thereafter (Talley et al. 1997). However, it should be noted that our results differ from those of (Liu and Wong-Riley 2010) who showed with immunohistochemistry in rat that 5-HT1A receptor levels in several brainstem nuclei are high at birth, only decreasing after 2 weeks of age. These differences may be related to species differences or the different methodology employed.
Our data suggest that the decreased 5-HT1A-heteroreceptor binding in medullary regions of SIDS cases may not be caused by dysfunction of 5-HT neurons nor the associated reduced brain 5-HT levels. 5-HT1A receptors are known to be regulated by multiple transcriptional enhancer and repressors, including NF-kappa-B, Freud, REST, Deaf1, Hes proteins, and glucocorticoid receptors (Albert et al. 2011). Future studies may hone in on the mechanism of dysregulation of 5-HT receptors relevant for SIDS. Importantly it should be noted that in these mutants there is a profound loss of ligand, thus although present, 5-HT1A heteroreceptors may not endogenously function as normal. Therefore, these mouse lines continue to be useful to understand the importance of 5-HT to homeostatic functions relevant to SIDS. The results also reveal that the post-natal period in mouse development is associated with rapid and dramatic changes of 5-HT1A receptor binding that may be relevant to age-dependent disorders in humans including SIDS.
Supported by NIH grant PO1 HD-036379. The authors declare no conflicts of interest.
Study concept and design: KGC, HCK, EEN, GBR, SMD, KJC. Acquisition of data: CAM, GK, AEC, RLH, DSP. Analysis and interpretation of data: CAM, GK, KGC. Drafting of the manuscript: CAM, GK, KGC. Critical Revision of the manuscript for important intellectual content: HCK, KJC, EEN, AEC, RLH, DSP, GBR, SMD. Statistical analysis: CAM, GK. Obtained funding: HCK, EEN, GBR, SMD, KGC. Administrative, technical and material support: CAM, GK, AEC, RLH, DSP. Study supervision: KGC.