Gene Targeting Approaches to Serotonin Receptor Function

Laurence Tecott, M.D., Ph.D.

Serotonin (5-hydroxytryptamine; 5-HT) systems widely innervate the central nervous system (CNS) and are implicated in the regulation of a large number of behavioral processes such as sleep, feeding, aggression, depression, anxiety, and sexual responses. Serotonergic mechanisms may be implicated in the etiology and treatment of numerous neuropsychiatric disorders, including depressive disorders, anxiety disorders, schizophrenia, eating disorders, substance abuse, epilepsy and migraine. Uncovering the mechanisms through which serotonin systems modulate brain function is clearly an important challenge in neuropsychopharmacology.

Serotonin systems are complex, as evidenced by the diversity of receptor subtypes through which this transmitter acts (25). Molecular characterization of 5-HT receptors has clarified their classification and provided evidence for at least 14 distinct subtypes. 5-HT receptors are diverse with regard to their regional and synaptic localization, developmental regulation, affinity for serotonin, and effects on intracellular signaling pathways. To understand the mechanisms through which serotonin alters behavior, it is necessary to determine the functional significance of individual serotonin receptor subtypes. This task is complicated by a paucity of available subtype-selective pharmacological probes of receptor function. Rapid advances in the manipulation of the mouse genome have recently provided a complementary approach to the study of receptor function: the use of gene targeting technology to generate mouse strains that completely lack individual receptor subtypes. The analysis of phenotypic abnormalities in such animals provides clues to the biological roles of the targeted receptors and to their contributions to the actions of psychiatric drugs.

A mutational approach has proved to be invaluable to investigators examining the roles of gene products in complex biological processes within prokaryotic and cultured eukaryotic cells. Recently, it has become possible to apply this approach to a mammalian system. Gene targeting procedures enable the precise (site-specific) introduction of a mutation to one of the estimated 100,000 murine genes. Typically, mutations have been designed to eliminate gene function, resulting in the generation of "knockout" or "null mutant" mice. Mutations that produce more subtle alterations in gene function can also be achieved. Two major developments have made gene targeting experiments feasible: 1) the generation of totipotent embryonic stem (ES) cells, and 2) the elucidation of techniques to achieve homologous recombination in mammalian cells.

Embryonic stem cells are derived from 3.5-day-old mouse embryos at the blastocyst stage of development (). Blastocysts are cultured individually under conditions that permit the proliferation of the inner cell mass cells (those cells that would normally become the fetus). These cells are then disaggregated, and individual ES cell clones are grown. Under optimal conditions, ES cells retain the ability to contribute to all of the tissues of the developing fetus. The derivation of ES cells was pioneered using embryos derived from the 129/Sv strain of mice, a strain that is commonly used in studies of early embryonic development. Although this mouse strain is not ideal for the study of behavior (see below), most ES cell lines in current use are 129/Sv-derived.

This is the process by which a mutation is targeted to a precise location in the genome. A targeting construct is generated, which typically consists of a long target gene sequence into which a loss-of-function mutation has been engineered (). Most targeting constructs are designed to achieve homologous recombination events in which recombination at the target locus results in replacement of native target sequences with construct sequences. In mammalian cells, fragments of DNA preferentially integrate into the genome at random locations at rates that greatly exceed homologous recombination. Therefore, targeting constructs are designed for use in selection strategies that enrich for ES clones in which homologous recombination has occurred. In the commonly-used positive-negative selection strategy (14), a portion of a protein-coding exon is replaced by sequences that confer resistance to the drug neomycin (). This mutation serves two purposes: 1) to inactivate the gene product and 2) to provide a marker that enables the selection of cells that have integrated the construct. This exogenous DNA fragment is flanked by regions of DNA that are homologous to the native gene. Adjacent to one of these homologous regions is a gene encoding thymidine kinase. Treatment with the drug ganciclovir will kill cells that express this gene.

The targeting construct is typically introduced into ES cells by electroporation (). In this step, cells are subjected to an electrical current that facilitates the internalization of the DNA construct. Those cells that failed to incorporate the targeting construct are killed by the addition of neomycin to the culture medium (positive selection). The majority of the remaining cells have incorporated the entire DNA construct (including the thymidine kinase gene) at random sites throughout the genome. By contrast, during homologous recombination, nonhomologous regions of the construct that are not flanked by homologous sequences are excluded from the integration event. Therefore, homologous recombinant clones will not contain the thymidine kinase gene. Thus, the addition of a second drug, ganciclovir, will selectively kill cells that have randomly incorporated the construct (negative selection), thereby enriching for targeted clones. ES cell clones that survive this double drug selection are then screened for homologous recombination by polymerase chain reaction (PCR) or Southern blot analysis. The homologous recombinant clones, which are heterozygous for the introduced mutation (except for X-linked genes), are then used to generate chimeric mice.

Following the isolation of homologous recombinant ES cell clones, these cells are microinjected into the fluid-filled blastocoele cavity of 3.5-day-old embryos at the blastocyst stage (). The injected embryos are then surgically transferred into the uterus of a pseudopregnant female. Pseudopregnant females are generated by matings with vasectomized males. The act of copulation initiates the endocrine changes of pregnancy, providing a suitable uterine environment for the survival and implantation of the transferred embyros. These animals will then give birth to chimeric mice, which are partly derived from the injected ES cells, and partly derived from the host embryo. For example, ES cells derived from a brown strain of mice are often injected into embryos derived from black C57BL/6 mice, resulting in chimeras with coats containing black and brown patches. The extent to which the ES cells have colonized the animal may be roughly approximated by the extent of the brown contribution to the coat. It is most important that ES cell derivatives colonize the germ cells of the chimera, so that the targeted mutation can be propagated to subsequent generations. Typically, chimeras are mated with C57Bl/6 mice. The germ line transmission of ES cell-derived genetic material is indicated by the generation of brown offspring. Half of these brown mice will be heterozygous for the targeted mutation. These heterozygous mice are then bred to produce homozygous mutant mice that completely lack the normal gene product.

To date, two serotonin receptor subtypes have been studied using a gene targeting approach. Null mutant lines lacking the serotonin 5-HT1B and 5-HT2C receptor subtypes have been generated. Mutant mice lacking these receptor subtypes survive and reproduce, providing models to investigate the functional roles of these receptor subtypes.

The 5-HT1B receptor is found predominantly on presynaptic terminals and acts to inhibit neurotransmitter release (4,8,15). It is expressed in numerous brain regions, including the hippocampal formation, basal ganglia, amygdala, and raphe nuclei (2,20). Investigations of its functional roles have been hindered by a lack of highly selective agonists and antagonists. The generation of a 5-HT1B receptor mutant mouse line has provided an additional tool to determine the functional significance of this receptor (23).

The absence of functional 5-HT1B receptors in the homozygous mutants was confirmed by the reduction of 125I-cyanopindolol binding to brain sections. The animals exhibited no overt abnormalities in brain morphology or in the expression of related serotonin receptor subtypes. The absence of 5-HT1B receptors did not appear to produce marked alterations in appearance or baseline behaviors of the mutant mice. However, they were insensitive to the hyperlocomotor effects of the 5-HT1A/1B receptor agonist RU24969, confirming that 5-HT1B receptors contribute to this particular effect. By contrast, baseline activity levels of mutant and wild type animals did not differ, indicating that either these receptors are not routinely involved in the regulation of activity, or that activity levels had been normalized by compensatory mechanisms.

Due to the proposed role of serotonin systems in aggression, aggressive behaviors were also examined in 5-HT1B receptor knockout mice. As for locomotion, baseline levels of aggression appeared normal in the mutants. However, marked differences were observed in the provocative "resident-intruder" aggression paradigm. Animals were housed individually for a 4-week period. This isolation procedure enhances aggressive responses to "intruder" mice placed in the "resident's" home cage. Following isolation, resident mutant mice displayed hyperaggressive behavior toward intruders, as evidenced by reduced attack latencies and increased frequencies of attack, relative to wild type animals. This result led to the suggestion that 5-HT1B receptors contribute to the serotonergic regulation of aggression and to the actions of "serenics", a class of nonspecific 5-HT1 receptor agonists with antiaggressive properties.

A potential role for 5-HT1B receptors in the serotonergic modulation of alcohol intake was also supported in a recent study of 5-HT1B receptor mutants (3). Mutant mice displayed elevated ethanol consumption in a two-bottle choice situation. In addition, they exhibited reductions in ethanol-induced ataxia and in the development of tolerance to ethanol, compared with wild-type mice. This study demonstrated that the phenomena of ethanol sensitivity, tolerance, and drinking could be genetically dissected in this knockout model.

The 5-HT2C receptor is abundantly expressed in the CNS, and has been implicated in numerous behavioral and physiological actions of serotonin (17,28). As with the 5-HT1B receptor, highly selective 5-HT2C receptor agonists and antagonists have been unavailable. Through gene targeting procedures, a null mutation was introduced into the X-linked 5-HT2C receptor gene (27). The absence of intact receptor protein in hemizygous mutant male mice was verified by the loss of 5-HT2C receptor immunoreactivity and by the absence of functional 5-HT2C receptors encoded by brain mRNA, as determined in a Xenopus oocyte expression assay. The mutant mice exhibited no overt abnormalities in appearance or in brain morphology.

Two major phenotypic abnormalities were reported in 5-HT2C receptor mutants: epilepsy and obesity. Mutant mice were prone to occasional spontaneous episodes of tonic-clonic seizure activity. Furthermore, these animals displayed a markedly elevated sensitivity to the convulsant actions of the GABAA receptor antagonist metrazol. These unexpected results indicated that inherited perturbations of serotonergic systems can influence seizure susceptibility, and that the 5-HT2C receptor may play an important role in the regulation of neuronal network excitability. Moreover, the mutant phenotype was mimicked in wild-type mice by the administration of the 5-HT2C receptor antagonist mesulergine, suggesting that the mutant phenotype did not arise from a developmental defect.

Mutant animals also exhibited an obesity syndrome, manifested by a 50% increase in the deposition of white adipose tissue in young adult animals. The obesity was associated with elevated food intake and did not appear to result from metabolic alterations. This phenotype supported the suggestion that 5-HT2C receptors are involved in the serotonergic inhibition of appetite, and that stimulation of this receptor accounts for the anorectic effects of nonspecific serotonergic agonists such as dexfenfluramine and m-chlorophenylpiperazine (mCPP) [1,10,11,21,22,24]. Moreover, mutant mice were found to be resistant to the anorectic effects of mCPP, indicating that this drug reduces food intake through its action at 5-HT2C receptors. More generally, this demonstrates that receptor knockout models may be useful for determining the extent to which particular subtypes mediate the effects of nonspecific drugs. The obesity syndrome in 5-HT2C receptor mutant mice provides support for a major role of 5-HT2C receptors in the serotonergic regulation of appetite, and demonstrates that an inherited perturbation of serotonin systems can predispose animals to obesity.

Interestingly, weight gain and lowered seizure thresholds are side effects associated with a number of psychiatric drugs that have antagonist actions at 5-HT2C receptors. These include both typical and atypical antipsychotic drugs (such as chlorpromazine and clozapine) and a number of antidepressants (such as amoxapine and amitryptiline) [9,16]. Further work will be required to test the hypothesis that these side effects result from 5-HT2C receptor antagonism.

Null mutant mice generated through standard gene targeting approaches lack the targeted gene product throughout embryonic and postnatal development. Therefore, the potential for developmental perturbations is a major caveat to the interpretation of mutant phenotypes in adult animals. It may be difficult to determine whether a mutant phenotype reflects a normal adult role for the targeted gene or an indirect effect of the mutation attributable to perturbed development. Such an effect may lead to an overestimate of the functional significance of the gene product in the adult animal. Although no structural abnormalities have been found in the brains of 5-HT1B receptor and 5-HT2C receptor mutant mice, potential developmental effects are difficult to exclude. Conversely, if significant compensation for the loss of a gene product occurs during development, then the severity of the mutant phenotype may underestimate the functional significance of the gene product. The nature of such compensatory mechanisms and the extent to which they exist are poorly understood. Interestingly, null mutations of neurotransmitter receptor genes have not been associated with the upregulation of related receptor subtypes (26). However, other mechanisms of compensation are possible. One approach to assess whether a mutant phenotype results from altered development is to determine whether it can be mimicked in wild type animals by the administration of antagonists. This strategy is limited to those receptor subtypes for which selective antagonists exist.

In interpreting behavioral phenotypes, attention must also be paid to the effects of genetic background. Analyses are often performed with hybrid animals that are the progeny of matings between mice derived from two strains with different behavioral profiles. The resulting genetic heterogeneity introduces variability that may mask the effects of the mutation. The placement of mutations on an inbred genetic background will therefore allow for the detection of more subtle behavioral abnormalities. Another potential problem arises from the common use of ES cells derived from 129/Sv mice. Some 129 substrains are susceptible to structural abnormalities of the CNS, such as agenesis of the corpus callosum, and are impaired in several behavioral assays (18,19). This potential problem may be addressed through breeding programs to place targeted mutations on different inbred backgrounds, and by the generation of ES cell lines derived from other inbred strains.

The power and general utility of gene targeting approaches will be greatly enhanced by technical refinements currently under development. For example, it is possible to generate mouse strains bearing mutations that are cell type-specific. This approach is important because the expression of many serotonin receptor subtypes is not highly restricted, and it may be difficult to link phenotypic abnormalities with the dysfunction of particular brain regions. Under some circumstances, it would therefore be desirable to restrict a genetic disruption to subsets of neurons. Cell type specific gene targeting has been successfully applied using the loxP-cre site-specific recombinase system (7), and efforts are underway to apply this approach to neurotransmitter receptors. In addition, inducible knockout strategies that permit the modification of genes in the adult animal are under development. The ability to induce genetic disruptions in animals that have developed normally will eliminate developmental perturbations that may complicate the interpretation of results from mutant phenotypes. The feasibility of this approach has been suggested in studies using tetracycline- and interferon-regulated promoters (5,6,12). Finally, gene targeting procedures will also be used to introduce subtle mutations that modify, but do not eliminate, receptor function. For example, the functional consequences of allelic variation found in human serotonin receptor subtypes (e.g., a Cys₂₃-Ser₂₃ polymorphism in the 5-HT_2C receptor coding sequence [13]) may be investigated in mouse models. Such strategies may also be used to investigate the consequences of altering receptor properties such as ligand binding affinity, second messenger coupling, internalization and desensitization.

The generation of mutant mice by gene targeting provides the means to examine the consequences of a precise molecular lesion in the context of the intact organism. When applied to serotonin receptor genes, it provides a powerful complement to pharmacological studies of receptor function. 5-HT receptor knockout mice may be used to determine the extent to which a particular action of a nonselective drug is mediated by the targeted receptor subtype. The phenotypic analysis of null mutant strains will also uncover novel roles for receptor proteins. The knockout approach will be particularly useful for obtaining clues to the function of serotonin receptor subtypes for which there are no selective pharmacological probes. Furthermore, these models may provide useful tools for the development of subtype-selective drugs. Such agents should be devoid of behavioral and physiological effects in mutant mice lacking the corresponding receptor subtypes.

The two strains of serotonin receptor mutant mice described above provide model systems of relevance to human neuropsychiatric disorders such as epilepsy, obesity, hyperaggressivity and substance abuse. It is clear that the generation of additional mouse strains with null, cell type-specific, inducible, and subtle mutations of serotonin receptor subtypes will provide valuable tools for evaluating receptor function and for modeling pathological states.

Work by L. Tecott described in this review was supported by fellowships from the Veterans Administration and the National Alliance for Research on Schizophrenia and Depression.

published 2000