4.1 Pet1/Cre transgene drives Cre-mediated recombination in serotonergic neurons within the raphe nuclei

4. DISCUSSION

There is growing interest in the study of serotonergic system development due to the multiple roles serotonin plays in the whole central nervous system and to the known link between dysfunction of 5-HT signalling and the onset of several neurological diseases (Sodhi and Sanders-Bush, 2004). Moreover, in addition to its physiological function in the mature brain, serotonin has been shown to play a role regulating brain development.

The possibility to manipulate the mouse genome, and the generation of transgenic animals, have made available several genetic tools suitable for the study of brain development and function. In this study, I took advantage of the Pet1/Cre transgenic mouse line and Tph2::eGFP knockin mouse line previously generated in our lab to shed new light on some aspects of serotonergic system development and to address the consequences of lack of 5-HT on brain development.

4.1 Pet1/Cre transgene drives Cre-mediated recombination in serotonergic neurons within the raphe nuclei

Transgenic mouse lines are commonly generated via pronuclear injection of DNA constructs, resulting in random integration of the transgene into the host genome. The resulting pattern of transgene expression depend both by the specificity of the promoter and its integration site within the genome. Transgene expression can be influenced by the regulatory elements flanking the integration site, which might drive transgene expression outside promoter-specific regions and cell types (Bronson et al., 1996; Wallace et al., 2000; Chandler et al., 2007).

The large insert size of BAC clones likely guarantees the presence of

long-range acting regulatory elements required for correct temporal and

tissue-specific expression (Chandler et al., 2007). These large constructs

allow the expression of the transgene mirroring the expression of the endogenous gene independent of the integration site (Schedl et al., 1993;

Casanova et al., 2001; Giraldo and Montoliu, 2001; Gong et al., 2003). The use of BAC clones is particularly important when the regulatory elements driving the expression of a certain gene are well characterized or a precise transgene expression is required, such as for promoters driving Cre recombinase (Lee et al., 2001).

Cre mouse lines are used in combination with a second mouse line harbouring a floxed sequence. When Cre recombinase is expressed in cells harbouring the floxed gene, the desired gene inactivation is restricted to certain cell types and precise developmental stages, depending on the tissue-time specificity of the promoter driving Cre activity. Cre/loxP system thus represents a powerful tool that solves major concerns of conventional knockout approaches. First of all, gene ablation may exert its effect in multiple tissue and cell types, creating a complex phenotype difficult to unravel. The possibility to conditionally knockout a gene in a tissue-specific manner ideally prevents secondary effects resulting from altered gene function in multiple tissues. Moreover, Cre/loxP enables to overcome the problem of the embryonic lethality caused by gene deletion expressed in the germ line, that otherwise would preclude analysis of gene function in adult tissues. Keeping in mind these considerations, it appears clear that an exhaustive characterization of the Cre expressing transgenic mouse line is crucial for any further application.

During the last 10 years, Pet1 promoter has been an ideal sequence

in transgenic application, as it represents a known specific marker of 5-HT

neuron differentiation from embryonic stages to adulthood (Hendricks et al.,

2003). Previously, a knockin of the Cre recombinase into the Sert gene has

been used to obtain recombination in serotonergic neurons (Zhuang et al.,

2005). However, Sert control of Cre activity has revealed many expression

domains in extraserotonergic neurons during development (Gaspar et al.,

2003), thus compromising the use of this line in conditional knockout

strategies aimed to serotonergic-specific gene inactivation. Furthermore, despite the knockin approach results in a more faithful expression of the Cre recombinase, the endogenous gene is simultaneously disrupted, such that both fate-mapping and potential future gene inactivation studies are performed on a heterozygous genetic background.

A transgene in which the regulatory elements of the gene Pet1 drive the expression of Cre recombinase represents therefore the ideal strategy to recombine sequences specifically in serotonergic neurons.

Scott and colleagues have shown that the regions at the 5’ end of Pet1 gene are necessary to drive a LacZ transgene expression to 5-HT neurons, while those at the 3’ end are not (Scott et al., 2005a). They demonstrated that a 1.8 kb fragment region 5’ the Pet1 mouse locus is sufficient for 5-HT neuron-specific expression, even though the transgenic mouse line generated displayed significant ectopic expression due to positional effect. Moreover, the same research group demonstrated that only a 40 kb upstream region was able to recapitulate the expression profile of the endogenous Pet1 within the brain, with still some ectopic expression (Scott et al., 2005a; Scott et al., 2005b). Such a 40 k region was therefore used to generate several different transgenic mouse lines, including a ePet1- Cre mouse line (Deneris, 2011).

In order to avoid possible positional effects, since cell-type specific

enhancers are often located several kilobases away either upstream or

downstream of the gene promoter, we decided to use a BAC clone

containing 210 kb genomic fragment of Pet1 gene locus to drive Cre

expression (BAC_ rp23_165_D11). In particular, it contains a 170 kb region

upstream Pet1 gene, and additional 50 kb downstream it. To date, it

represents the largest Pet1 genomic region used to drive Pet1-like transgene

expression. The expression pattern of Cre recombinase, highlighted by Cre-

mediate activation of reporter genes, matched accurately the spatio-temporal

expression of Pet1 gene within the hindbrain. Indeed, in E12.5 Pet1/Cre-

ROSA26R mouse embryos X-gal staining confirmed the presence of a

stronger transgene expression in the rostral domain than in the caudal raphe group, consistently with the timing of Pet1 expression, which starts earlier (10.5dpc-11dpc) in the rostral raphe, and only about one day later (11.5dpc- 12dpc) in the caudal group (Hendricks et al., 1999). In postnatal mice, the presence of X-gal staining in the nine serotonergic B nuclei further confirms that the 210 kb region driving Cre recombinase expression is able to recapitulate the developmental timing of endogenous Pet1 expression.

Moreover, lack of reporter expression in extraserotonergic domains, particularly in Pet1/Cre FF9 line, guarantees a Pet1 reliable spatial control of transgene expression.

4.2 Pet1/Cre transgenic line identifies a non-serotonergic Pet1- expressing cell population within the raphe nuclei

Cre/loxP system offers a more comprehensive view of all the structures that express Pet1 during development, and particularly those in which the expression is either at sub-threshold levels or transient.

Unexpectedly, the present study allowed assessing that in Pet1/Cre- ROSA26YFP despite in the raphe nuclei almost all 5-HT-positive neurons were also YFP-positive (Pet1-positive), a good percentage of YFP- expressing neurons did not show 5-HT immunoreactivity. This discrepancy, observed in P1 pups was particularly marked in the B8 median raphe nuclei, and persisted up to P30 mice.

The maintenance of a YFP-positive/5-HT-negative population raised

the question about its possible origin. At least two hypotheses that can be

made. Taken into account that Pet1 is known to be expressed in

serotonergic precursors and that its expression anticipates the acquisition of

marker of terminal differentiation, such as Tph2 gene, one possibility is that

lack of 5-HT in some YFP-positive neurons may indicate that serotonergic

system is not yet fully mature at the postnatal stages analyzed. This view is

in contrast with the available data present in literature, describing that

serotonergic system full maturation is acquired during the first stages of postnatal development (Lidov and Molliver, 1982). In addition, double ISH experiments shown in the section “Results” demonstrated that discrepancy between YFP and 5-HT immunolabelling matches with a discrepancy in the expression of Pet1 and Tph2 transcripts. Moreover, as neurons that express Pet1 but not Tph2 are still present at P60, with their number not diminished as compared to that observed at earlier stages (P14 and P30), it is likely to exclude the existence of serotonergic immature neurons.

An alternative possibility is that YFP is expressed in a non- serotonergic raphe cell population. Indeed, even though serotonergic neurons constitute the most abundant cell population within raphe nuclei, they are not the sole. Indeed, approximately 50–75% of dorsal raphe neurons cells are nonserotonergic (Moore, 1981). Non-serotonergic raphe projections are neurochemically heterogeneous, and may contain a number of transmitter substances, including glutamate (Kiss et al., 2002, reviewed in Soiza-Reilly and Commons, 2011), dopamine (Pohle et al., 1984; Descarries et al., 1986; Yoshida et al., 1989; Stratford and Wirtshafter, 1990; Hasue and Shammah-Lagnado, 2002), GABA (Nanopoulos et al., 1982; Stamp and Semba, 1995; Ford et al., 1995; Jankowski and Sesack, 2004), nitric oxide (Xu and Hokfelt, 1997; Simpson et al., 2003), and neuropeptides such as vasoactive intestinal polypeptide (VIP) (Petit et al., 1995; Kozicz et al., 1998) and cholecystokinin (van der Kooy et al., 1981; Seroogy and Fallon, 1989).

Importantly, distinct neuronal fates may share some transcription factors for

their identity acquisition. For example, Lmx1b is an important transcription

factor that collaborates with Pet1 during embryogenesis in the specification

of serotonergic neurons (Cheng et al., 2003; Ding et al., 2003). It has been

demonstrated that, within the raphe nuclei, a subset of Lmx1b-expressing

neurons is 5-HT-negative, suggesting that Lmx1b is expressed in some non-

serotonergic neurons (Cheng et al., 2003). Thus, Pet1 and Lmx1b may

share not only the common role as important transcription factors for

serotonergic neuron development, but also their extraserotonergic expression within the raphe nuclei.

The observation that YFP-positive/5-HT-negative neurons were more abundant in the most rostral raphe nuclei, led us to speculate that these neurons could be dopaminergic neurons. Ontogenetically, both dopaminergic and serotonergic neurons develop from the ventral precursors adjacent to the floor plate, and their development is regulated by some common signals from the floor plate, such as Shh (Hynes and Rosenthal, 1999; Goridis and Rohrer, 2002). Moreover, it has been demonstrated that Lmx1b is also expressed in midbrain dopaminergic cells (Smidt et al., 2000).

However, none of the YFP-expressing neurons showed a positive immunostaining for TH, which encodes the rate limiting enzyme for the synthesis of dopamine, demonstrating that, unlike Lmx1b expression, Pet1 is not expressed in both serotonergic and dopaminergic neurons. Therefore, Pet1, conversely to what previously described, is not restricted to serotonergic neurons but it is also expressed in another yet uncharacterized neuronal population.

4.3 Pet1/Cre transgenic line identifies new Pet1 expression domains in the developing pancreas and kidney

The use of Pet1/Cre transgenic mouse line allowed the identification

of novel unexpected domains outside the brain as highlighted by strong X-

gal staining in Pet1/Cre-ROSA26R embryos: the developing pancreas and

kidneys. Taken into account that transgenic mouse line may be subjected to

positional effect, the presence of Cre recombinase activity ouside the known

expression domains of Pet1 could be explained with an ectopic expression

of the transgene. However, the presence of comparable patterns of X-gal

staining in pancreas and kidneys in all the three founder lines gave the first

clue that the observed reporter expression was due to a specific Pet1

promoter activity. ISH experiments demonstrated that Pet1 is actually expressed in the pancreas and in the kidney.

While this thesis was being written, a paper by Ohta and collaborators demonstrated the existence of common transcriptional cascade that drives the differentiation toward serotonergic neurons and β-pancreatic cells (Ohta et al., 2011). In particular, the authors found that islet cells express some serotonergic transcription factors such as Nkx2.2 and Pet1 and have the complete machinery necessary for synthesizing, packaging, and secreting serotonin, including both Tph1 and Tph2. Likely, the absence of Tph enzymes in our ISH experiments reflects the timing of appearance of their transcripts within the pancreas. We performed the analysis on E15.5 embryos, a stage in which Pet1 is clearly detectable, consistently with what observed in Ohta’s paper, but Tph1 and Tph2 mRNA levels are probably sub-threshold, as their expression become substantial only in adult animals (Ohta et al., 2011).

Interestingly, to date Pet1 expression in the kidney has never been

reported. This may be due either to a sub-threshold expression level or to a

transient expression within this structure, both of these situations being

difficult to detect with canonical ISH experiments. Indeed, it has been

necessary to perform radioactive-ISH to detect Pet1 expression in the

developing kidney. Our results are in contrast with the previously generated

ePet1-Cre transgenic mouse line, in which presence of Pet1 in the kidney

was excluded (Scott et al., 2005b). It is likely that regulatory elements

necessary to drive Pet1 expression in the kidney are not present in the 40 kb

region used to generate the ePet1-Cre transgene, but are instead included

in the 210 kb Pet1 entire regulatory region of the Pet1/Cre transgene. In

good agreement with our observation that Pet1 is expressed in mammalian

kidney, microarray analysis performed on mouse and rat transcriptomes

indicated that Pet1 is one of the 20 different transcripts that were enriched in

the ureteric bud compared with metanephric mesenchyme (Schmidt-Ott et

al., 2005).

Thus, our Pet1/Cre transgenic mouse line in combination with Cre- mediated reporter lines has been a suitable tool to for further fate-mapping studies on the expression of Pet1 in the kidney. Analysis of the progression of Cre activity in Pet1/Cre-ROSA26R mouse embryos, together with co- labelling experiments with specific markers such as calbindin-D28k and cadherin16, showed that Pet1/Cre is most widely expressed in UB- derivatives during development and in the collecting duct system in postnatal animals. Pet1/Cre transgene can mediate epithelial-specific recombination in the developing kidney, thus representing a novel experimental tool for producing tissue-specific gene knockouts in the kidney.

Interestingly, as observed for pancreas development (Ohta et al., 2011), also in the kidney Pet1 is not the unique serotonergic transcription factors. It has been reported that LMX1B mutation causes nephropaties and, often, renal failure in human patients affected by Nail-Patella syndrome (Dreyer et al., 1998; McIntosh et al., 1998). Mouse Lmx1b is expressed in podocytes of the kidney, and its genetic ablation results in kidney defects resembling those observed in Nail-Patella syndrome patients (Chen et al., 1998; Rohr et al., 2002). On the other hand, the functional role of Pet1 in the kidney has not been fully characterized yet. It is known that chromosomal traslocations in which EWS gene is fused to a variety of transcription factors, including human FEV, can lead to the onset of different subsets of Ewing tumours (Peter et al., 1997; Ng et al. 2007), with some of them occasionally occurring in the kidney (Berg et al, 2009). Given the expression of Pet1 in the developing collecting duct system, it will be interesting to address the possible consequences on renal functionality in Pet1 knockout mice.

On the whole, the use of Pet1/Cre transgenic mouse line allowed the

identification of novel Pet1 expression domains both in the hindbrain and in

the developing pancreas and kidney. As the expression of the reporter is

permanently induced following recombination, this line will allow long-term

analysis of Pet1-expressing cell types. Moreover, this mouse line constitutes

a valuable model for studying molecular mechanisms of renal and pancreatic

development by induction or silencing of specific genes in epithelial UB- derivatives and islet cells, respectively. Finally, great care must be taken using a Pet1/Cre-mediated gene ablation directed to serotonergic neurons, as Pet1 is expressed also in non-serotonergic cells, so that the phenotype observed may result from a combined, rather than serotonergic-specific, effect.

4.4 Developmental role of serotonin on the serotonergic circuitry formation

Studies from both vertebrate and invertebrate organisms have supported the hypothesis that serotonin regulates neurite outgrowth and establishment of neuronal connectivity during development (Hydon et al., 1984; Budnik et al., 1989; Sikich et al., 1990; Diefenbach et al., 1995;

reviewed in Daubert and Condron, 2010). The availability of the Tph2::eGFP knockin mice allowed to highlight serotonergic fibers independently of serotonin immunoreactivity and made possible to visualize directly the development and the formation of the serotonergic neuronal circuitry.

Thanks to the presence of strong eGFP fluorescence and by virtue of the absence of regulatory feedback mechanisms on the reporter expression as a consequence of altered serotonin levels, our mouse model made possible for the first time to perform accurate analyses of axonal projections of serotonergic neurons in animals lacking brain serotonin.

Absence of brain serotonin resulted in overt alterations of

serotonergic innervation in several regions of the rostral brain. Intriguingly,

besides the apparently normal distribution observed in cortex and striatum,

development of serotonin axon terminal arborisation exhibited, as a

consequence of Tph2 inactivation, a diametrically opposite response in

specific target districts. Specifically, we reported a dramatic reduction of

serotonergic innervation in the suprachiasmatic nucleus and in the thalamic

paraventricular nucleus. Conversely, an extensive axon terminal arborisation

resulting in a striking serotonergic hyperinnervation was observed in the nucleus accumbens and in the hippocampus of Tph2 mutant mice. These results demonstrate that brain serotonin has an important role in regulating the wiring of the serotonergic system during brain development and that serotonin can act either promoting or inhibiting serotonergic axon terminal arborisation relying on specific target brain districts.

An intriguing question is whether serotonin acts throughout the development of the serotonergic innervation or in specific periods among the three proposed by Lidov and Molliver. In Tph2 mutant animals serotonergic projections still reach all brain districts weakening the possibility of a specific serotonergic pathway rerouting. This evidence, rather, strongly suggests that serotonin exerts its role during the period of terminal field development either promoting or inhibiting sprouting of serotonin axon terminals. Support to this hypothesis is corroborated by the appearance of the abnormal sprouting of serotonin axon terminals matching the temporal progression of terminal field development of selective target structures. In fact, suprachiasmatic nucleus and thalamic paraventricular nucleus, in which the establishment of serotonergic innervation is completed by P10 (Lidov and Molliver, 1982), showed at P14 a reduction of fiber density that was comparable in extent to what observed in adult Tph2 mutant mice. Whereas, in the hippocampus in which serotonergic terminal field mature over a more protracted time span (Lidov and Molliver, 1982), we observed a progressive increasing hyperinnervation that peaks in adult mice.

Although the precise mechanism through which serotonin acts to

modulate the development of axon terminal arborisation is unknown,

activation of specific serotonin receptors is likely to play a pivotal role. Both

pharmacological and genetic results provided evidences that specific

serotonin receptors mediate the trophic effects of serotonin (Sikich et al,

1990; Ase et al., 2001; Dudok et al., 2009). Therefore, the complex and vast

operational heterogeneity of the serotonergic system signalling (Hannon and

Hoyer, 2008) together with the dynamic spatiotemporal distribution of

serotonin receptor expression during brain development (Lauder et al., 2000;

Lein et al., 2007; Bonnin et al., 2006) might be responsible for originating the divergent phenotypes we observed. This view suggests that serotonin has an autoregulatory role on the morphology of serotonergic system but not serotonergic neurons, and acts with different brain region-dependent trophic effects on serotonergic axon terminal arborisation depending on the presence of distinct combinations of serotonin subtype receptors. Similarly, in several invertebrate systems, serotonin depletion altered serotonergic fiber development without precluding normal differentiation of serotonergic neurons (Vallés and White 1986; Budnik et al., 1989; Sze et al., 2000). Thus, these data disclose the presence of an autoregulatory control mediated by serotonin that is conserved throughout animal evolution.

It is interesting to ask which are the factors impinging on the development of serotonergic system in central 5-HT-depleted animals and how they are regulated by serotonin. We identified two factors whose expression is modulated in response of serotonin depletion, such as BDNF and S100b, known to be specifically involved in promoting the development of serotonergic fibers.

Indeed, cerebral infusion of BDNF promotes sprouting of serotonergic

axons at the site of injection (Mamounas et al., 1995, 2000). Conversely,

BDNF downregulation observed in BDNF +/- mice causes a decrement in

the serotonergic fiber density in the hippocampus (Lyons et al., 1999). The

integration of our results about serotonin depletion effects on BDNF

expression together with the well-validated BDNF function on serotonergic

system unmasks a possible regulatory feedback mechanism through which

serotonin itself might regulate the formation of the serotonergic neuronal

circuitry in the hippocampus. Reduction of serotonin levels may trigger an

increment of endogenous BDNF signalling that in turn exerts a neurotrophic

effect upon serotonergic axons. A higher density of serotonergic axon

terminals would result in additional serotonin release sites and, likely, in an

augmented availability of serotonin thus restoring the original BDNF levels

that, consequently, would turn down the serotonergic axon sprouting. A recent report by Hill and collaborators provides strong support to this hypothesis and suggests that serotonin would modulate BDNF levels via the activation of the 5-HT2C receptor (Hill et al., 2011).

Similarly, S100b has significant and selective effects on terminal growth of serotonin neurons (Azmitia et al., 1999; Lui and Lauder, 1992;

Whitaker-Azmitia, 2001).
 For this reason, the serotonergic hyperinnervation observed in Tph2 mutants may be a direct consequence of S100b up- regulation in serotonergic neurons. In this view it can be hypothesized that serotonin may function with an autocrine action via serotonergic autoreceptors to control the expression levels of S100b that, in turn, regulate the sprouting of serotonergic fibers.

Thus, the up-regulation of BDNF in target districts where hyperinnervation of serotonergic fibers has been detected and the increased expression level of S100b in serotonergic neurons may account for the presence of a non-cell-autonomous and a cell-autonomous mechanism, respectively, through which 5-HT controls the development of serotonergic system itself.

In conclusion, our results demonstrate that serotonin, besides its role

as a neurotransmitter, represents an important signal required during CNS

development involved in the formation of the highly precise architecture of

the nervous system and in particular in the fine tuning for patterning the

correct serotonergic branching at selected regions of the brain. The evidence

that serotonin levels are crucial for the proper wiring of the brain contribute to

confirm that exposure to genetic, pharmacological or environmental insults

altering the serotonergic signalling produce long-lasting changes that may

prompt to neurodevelopmental disorders thought to have developmental

origins such as schizophrenia, autism and mental retardation (Chugani et al.,

1999; Sodi and Sanders-Bush, 2004).