| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
4* nAChR in adult mice increases the loss of midbrain dopaminergic neurons
,1,1
* Department of Neurology, University of Leipzig, Leipzig, Germany;
Division of Biology 156–29, California Institute of Technology, Pasadena, California, USA; and
Molecular Biology Institute and Department of Microbiology, Immunology and Molecular Genetics, University of California, Los Angeles, California, USA
2Correspondence: 156–29 Caltech, Pasadena CA 91125, USA. E-mail: lester{at}caltech.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
4 nicotinic receptors (nAChR) at near-normal levels displayed dominant neonatal lethality and dopaminergic deficits in embryonic midbrain, because the hypersensitive nAChR is excitotoxic. However, heterozygous L9’S mice that retain the neomycin resistance cassette (neo) in a neighboring intron express low levels of the mutant allele (
25% of normal levels), and these neo-intact mice are therefore viable and fertile. The neo cassette is flanked by loxP sites. In adult animals, we locally injected helper-dependent adenovirus (HDA) expressing cre recombinase. Local excision of the neo cassette, via cre-mediated recombination, was verified by genomic analysis. In L9’S HDA-cre injected animals, locomotion was reduced both under baseline conditions and after amphetamine application. There was no effect in L9’S HDA-control treated animals or in wild-type (WT) littermates injected with either virus. Immunocytochemical analyses revealed marked losses (> 70%) of dopaminergic neurons in L9’S HDA-cre injected mice compared to controls. At 20–33 days postinjection in control animals, the coexpressed marker gene, yellow fluorescent protein (YFP), was expressed in many neurons and few glial cells near the injection, emphasizing the neurotropic utility of the HDA. Thus, HDA-mediated gene transfer into adult midbrain induced sufficient functional expression of cre in dopaminergic neurons to allow for postnatal deletion of neo. This produced increased L9’S mutant nAChR expression, which in turn led to nicotinic cholinergic excitotoxicity in dopaminergic neurons.—Schwarz, J., Schwarz, S. C., Dorigo, O., Stützer, A., Wegner, F., Labarca, C., Deshpande, P., Gil, J. S., Berk, A. J., Lester, H. A. Enhanced expression of hypersensitive 4* nAChR in adult mice increases the loss of midbrain dopaminergic neurons.
Key Words: nicotinic receptor • substantia nigra • cholinergic excitotoxicity • knock-in mouse • helper-dependent adenovirus • Parkinson’s disease
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
4 subunits are highly expressed on cell bodies of dopaminergic neurons in the midbrain (1)
. nAChRs on dopaminergic neurons regulate dopamine release in striatum and frontal cortex (2
, 3)
, and play a role in nicotine addiction and motor control. Chronic treatment with nicotine facilitates dopamine release and locomotor responses, which is consistent with an up-regulation of nAChRs in dopaminergic neurons (4)
. Chronic continuous treatment with nicotine does not alter locomotor behavior (5)
but induces up-regulation of dopamine D2 receptors, which was interpreted as a consequence of down-regulation of dopamine turnover (6)
. Thus, nicotinic receptors on midbrain dopaminergic neurons and the 4 subunit in particular exert various biological functions in addition to development of tolerance and addiction to nicotine.
Knock-in mice carrying a leucine to serine mutation at the 9' position of the pore-lining second transmembrane domain (M2) in the 4 subunit represent the only known instance of dopaminergic cell death caused by chronic activation of nicotinic receptors (7
,8)
. While heterozygous mice expressing nearly normal levels of the mutant receptor die shortly after birth, heterozygous animals of a strain that displays reduced expression of the hypersensitive allele, because of a deliberate insertion of neo in an intron adjacent to the mutated exon, are viable and fertile. These viable and fertile heterozygous mice are termed "L9’S". L9’S mice do not show a significant reduction of dopaminergic neurons during embryonic development (8)
. Adult L9’S mice have a moderate but specific reduction of substantia nigra dopaminergic neurons and a reduction of amphetamine-induced locomotion. Dopaminergic neurons in acute brain slices of L9’S mice show increased excitation after application of low doses of nicotine (7)
.
Gene transfer using adenoviruses has been introduced into clinical studies due to efficient and long-lasting ectopic gene expression (9
10
11)
. However, there have been complications secondary to inflammatory reactions directed against adenoviral proteins (12)
. Helper-dependent adenoviruses (HDA) may be safer but adequately efficient vectors mediating long term expression of foreign genes in many tissues, including brain (13)
.
This study uses HDA-mediated gene transfer to convert the hypomorphic neo-intact L9’S allele to the more normally expressed, excitotoxic neo-deleted allele in adult mice. This manipulation utilizes the fact that the neo allele is flanked by loxP sites ("floxed") (Fig. 1
). Here, we show that this cre recombination, inducing elimination of neo, augments the loss of dopaminergic neurons in adult L9’S mice and also augments the behavioral deficit. HDA confer expression of cre and YFP predominantly in neurons when injected into the midbrain. These data render the L9’S knock-in mice a valuable model for studying dopaminergic neuron death after excitotoxicity. In addition, efficient cre-recombination in L9’S mice via virus-mediated gene transfer or mouse genetic engineering may result in a first genetic mouse model with selective death of dopaminergic neurons and a hypokinetic behavioral phenotype.
|
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
4 subunit of the nicotinic acetylcholine (Ach) receptor (nAChR) with a mutated form, was previously described (8
,14)
. Two lines of knock-in mice were generated by injection of mutated ES cells into C57BL/6 blastocysts. The strain with the neomycin-resistance cassette (neo) still present (neo-intact) was studied here and termed L9’S. (In the other, nonviable strain, neo was deleted via cre-mediated recombination in ES cells. The neo-deleted strain is not studied in this paper). Young adult (3–8 months) male and female mice heterozygous for the L9’S mutation and their wild-type littermates were used. Experimental groups comprised equal numbers of male and female animals. There were no differences in age. All experiments were performed following NIH and local institutional guidelines for the humane treatment of laboratory animals.
Detection of cre-deleted allele
The polymerase chain reaction (PCR) primers, 5' CAC ACA TGA TCC AGA AAC CAA G 3' and 5' CAG GTG TTA GAA TCA TAG GGC TAG 3', were complementary to genomic sequences surrounding both the floxed neo sequences as well as a SwaI site used to insert the neo cassette. This yielded a PCR product of 2.6 kb for the neo-intact allele (not present on the gel of Fig. 1B
), 218 bp for the WT allele, and 260 bp (a SwaI site, plus a single loxP site) for the neo-deleted allele.
Propagation of HDA
HDA vectors were propagated using the Parks and Graham method (15
, 16)
(Fig. 1A
). Briefly, pHDA-Cre was linearized with PmeI and transfected into 293Cre4 cells in tissue culture plates. 24 h posttransfection, cells were infected with helper virus, AdLC8cLuc, at sufficient titers to cause cytopathic effects after 72 h, at which point cells were concentrated and lysed. Two-thirds of the lysate and additional helper virus were then serially passaged on 293Cre4 cells until sufficient titers were produced to purify through a CsCl step and subsequent linear gradient. Vector was titered by infection of 293 cells, then counted YFP transduced cells on a hemacytometer 72 h later.
Virus injection
All mice were prepared and anesthetized according to procedures approved by local institutional committees and in accordance with NIH guidelines. Mice were anesthetized using ketamine/xylazine. After anesthesia, animals were placed in a David Kopf stereotactic frame. The head was shaved and a skin cut made above the skull between bregma and lambda. A 10 µl Hamilton syringe was positioned according to bregma and lambda and a bore-hole drilled above the expected injection track. Viral suspension was loaded into the syringe and the needle was advanced to substantia nigra according to appropriate coordinates derived from a mouse brain atlas (TB+0.7 mm; A –3.1 mm; L ± 4.0 mm) (Fig. 1)
. After injection, the needle remained in place for at least 2 min before removal. The skin was closed using tissue glue. After surgery, animals were placed on a heating blanket until full recovery.
Two separate experiments were performed. Experiment 1 comprised 7 WT (4 HDA-cre and 3 HDA-ctr injected) and 13 L9’S mice (8 HDA-cre and 5 HDA-ctr injected). These animals received 2 x 107 YFP transducing units suspended in 2 µl. Behavior was assessed 3 days and again 20 days after surgery. Experiment 2 comprised 15 WT (8 HDA-ctr and 7 HDA-cre injected) and 15 L9’S mice (8 HDA-cre and 7 HDA-ctr injected). These animals received 107 YFP transducing units suspended in 2 µl. Motor behavior was assessed 1 day prior to, 3 days after, and 33 days after surgery. For statistical analyses, both experiments were combined and only differences that developed between the first and the second postoperative assessment were calculated. During experiment 1, two animals (1 WT HDA-cre and 1 WT HDA-ctr) and during experiment 2, four animals (1 L9’S HDA-cre, 1 L9’S HDA-ctr and 2 WT HDA-ctr) were lost perioperatively.
Drugs
All drugs were purchased from Sigma-Aldrich (Sigma-Aldrich, St. Louis, MO, USA) and dissolved as suggested by the supplier.
Histology and immunocytochemistry
Each mouse was fixed by cardiac perfusion (0.1 M PBS and 4% formaldehyde), postfixed for 2 h, and dehydrated in 15% sucrose and 30% sucrose. 20 series of cryosections of 50 µm thickness were prepared in coronal orientation. All stainings were performed on free floating sections. Brain sections were qualitatively examined at low magnification (50x and 100x) to assess the gross neuroanatomical structure of L9’S and wild-type brains. To assess the number of surviving neurons in substantia nigra, sections were stained in a heated 0.1% aqueous solution of toluidine blue (Merck, Darmstadt, Germany) until water started to evaporate. Sections were then washed in dilute HCl and later transferred to 96% ethanol. Finally, sections were mounted onto slides, dehydrated with propanol, bathed in xylene, and coverslipped with Canada balsam. Simultaneous staining was performed to visualize dopaminergic neurons (TH), YFP-immunoreactive transfected cells (GFP), cAMP response element (CRE) -immunoreactive transfected cells (CRE), neurons (NeuN), astrocytes (GFAP), and macrophages (ED1) with antisera for TH 1:500 (Pelfreez, Rogers, AR, USA), GFP 1:500 (Abcam, Cambridge, MA, USA), CRE 1:1000 (Chemicon, USA), NeuN, GFAP, and ED1 1:200 (Chemicon, El Segundo, CA, USA) and fluorescent secondary antibodies (1:500, Alexa 594, Alexa 488, Molecular Probes, Eugene, OR, USA), respectively. Procedures to verify absence of "bleed through" have been described (17)
.
Free floating sections were washed in 0.5% BSA diluted in 0.1 M PBS and blocked in 3% serum. The sections of 3 series were incubated overnight at 4°C in primary antibodies (GFP-TH, CRE-TH, GFP-NeuN, GFP-GFAP, GFP-ED1). After additional washing the sections were incubated for 1 h in secondary antibody conjugated with Alexa dyes (green 488 or red 594), washed, and mounted in Prolong antifade reagent (Molecular Probes). Images were obtained from a total of 16 L9’S and 16 wild-type mice (8 receiving HDA-ctr or HDA-cre). Cell counts were performed in selected midbrain sections containing substantia nigra and ventral tegmental area with fibers of the third cranial nerve separating the two regions. We counted TH-positive SNc neurons in two coronal sections from two representative levels of the SNc of 6–11 animals per experimental group, as described previously (18)
. Neurons were imaged and counted with a Zeiss wide-field DMLS microscope with a 20 or 40x objective. An experimenter (AS) blind to the animal genotype and treatment selected the counting areas on scanning the entire structure of interest on the right and on the left of each section. TH-labeled neurons were scored as positive only if their cell body image included a defined nuclear counterstaining. Nissl positive neurons were counted in alternate sections. Analysis used the Neurolucida package (MicroBrightField, Inc., Williston, VT, USA). Cell counts according to this method reveal 200–250 TH-positive neurons per SNc section. For illustration (Fig. 2
, Fig. 3
, Fig. 4
), areas of interest were imaged with a confocal laser scanning microscope SP2 AOBS (Leica, Bensheim, Germany) using 488 nm and 594 nm wavelengths. Subsequently images were processed with Photoshop 7.0 (Adobe).
|
|
|
Intensity of nerve terminal TH staining in nucleus accumbens and striatum was determined by tracing circular areas (400 µm diameter) and measuring light transmission using NIH Image software. To correct for nonspecific staining, for each section, a circular area in the adjacent cortex was also traced and measured. The final intensity value was obtained as (TH positive value –cortex value)/TH positive value. TH intensity measurements were performed on both hemispheres for each brain section, and three sections each from nucleus accumbens and striatum were analyzed from each mouse.
Locomotion measurements
Baseline locomotor behavior was assessed for 30 min after the mouse was placed in the novel environment of the cage equipped with photobeam monitors (San Diego Instruments, USA) followed by an intraperitoneal (i.p.) injection of amphetamine (5 mg/kg D-amphetamine sulfate dissolved in 0.9% saline) in a vol of 100–200 µl (100 µl/20 g bodyweight) and another measurement for 45 min. Recorded events represent disruption of two adjacent light beams 10 cm apart in the cage. Animals were monitored simultaneously in 6 activity cages. Experiments were always done at the same time of the day and in the same order of animals. Control animals were WT littermates of the heterozygous mice. Data were analyzed within and between subjects by ANOVA using Origin (OriginLab, Northhampton, MA, USA).
Data are displayed as mean ± SE.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
). The genotyping strategy provided a band (260 bp) specific to the mutant neo-deleted gene (Fig. 1B
). This cre recombined mutant allele was detected only in substantia nigra of L9’S mice injected with HDA-cre, while there was no recombination in striatum or cortical tissue of these mice or any brain region in L9’S HDA-ctr injected mice (Fig. 1B
).
Neuronal tropism of HDA
We investigated the neuronal tropism of HDA-cre and HDA-ctr in our knock-in mice after virus injection into substantia nigra bilaterally. Here, L9’S or WT mice were stereotactically injected with either HDA-cre or HDA-ctr and sacrificed 20 or 33 days postsurgery. We found an appreciable concentration of HDA-mediated YFP expression in all injected animals (Fig. 2A, D
). Surprisingly, YFP expression was predominantly(>90%) detected in neurons in all experimental groups (summarized in Fig. 2I
). YFP-expressing neurons were detected as far as 5 mm from the injection site or needle track. However, we observed no YFP-expression in cell bodies of neurons that project to substantia nigra, indicating that there was little retrograde axonal transport. The overall infection efficiency without regard to cell type (neurons, glia) was 5% in substantia nigra (Fig. 2)
; however as noted below, dopaminergic neurons were apparently infected with much greater efficiency. There was almost no YFP expression in GFAP-positive cells (Fig. 3)
.
Mild inflammatory response after HDA injection
Gene transfer using viral vectors may be complicated by immune reactions directed against viral proteins that are not expressed by HDA (19)
. However, there is always a concern about helper virus contamination, which could induce such an immune response.
Therefore, we investigated the activation of astrocytes and macrophages at the site of virus injection. In all injected mice there was only mild activation of astrocytes as well as minimal infiltration of macrophages except for < 0.5 mm around the needle track (Figs. 3
, 4)
. We observed no differences in these mild inflammatory responses between HDA-cre and HDA-ctr injected animals.
Only L9’S animals injected with HDA-cre showed a few infiltrating macrophages within the substantia nigra (Fig. 4)
. Overall, immune responses within the midbrain (and >0.5 mM away from the needle track) were not observably different from immune reactions in other areas of the brain.
Death of dopaminergic neurons after cre recombination
Immunoreactivity for TH, the rate-limiting enzyme for the production of dopamine or noradrenaline, is the appropriate method to quantify numbers of catecholaminergic neurons (20
21
22)
. Stereological investigation showed that L9’S HDA-ctr mice have a moderate reduction, 35%, of TH-positive cells compared to their WT littermates, corresponding to our previous semiquantitative data suggesting a similar loss of dopaminergic neurons in young adult L9’S mice (7)
(Fig. 5
). Injection of HDA-cre and HDA-ctr had no effect on TH immunoreactivity in WT mice.
|
L9’S mice injected with HDA-cre, however, showed a dramatic loss of TH-positive neurons in substantia nigra, significantly exceeding the loss in L9’S HDA-ctr injected animals (Fig. 5)
. Injection of HDA-cre into substantia nigra led to a > 70% reduction of TH-positive neurons compared to WT littermates injected with the same virus (Fig. 5F
). In addition, YFP and TH expressing neurons were almost absent in L9’S HDA-cre injected animals (Fig. 6
). There were a few remaining dopaminergic neurons in L9’S HDA-cre injected animals, and these showed swollen perikaria and vacuoles resembling characteristics of excitotoxic cell death (Fig. 5E
).
|
Midbrain dopaminergic neurons predominantly project to the striatum. Thus, striatal density of dopaminergic nerve terminals also provides a measure of function of these cells. We therefore examined TH-stained striatal slices and compared relative anti-TH staining between experimental groups. Again, there were no relevant differences between WT mice injected with either HDA-ctr or HDA-cre or L9’S mice injected with HDA-ctr. However, L9’S mice injected with HDA-cre showed a significant reduction of striatal dopaminergic innervation (Fig. 5G-J
, summarized in Fig. 5K
).
Nissl staining showed no gross abnormalities in the injected area or in any other brain area. Quantification of Nissl positive cells in SNc showed a modest decrement in only one group, L9’S HDA-cre (WT HDA-ctr: 834.6±45.4, n=8; WT HDA-cre: 808.7±56.1, n=6; L9’S HDA-ctr: 832.9±32.5, n=6; L9’S HDA-cre: 734.6±44.2, n=8; mean±SE). This observation is consistent with a selective loss of dopaminergic cells.
Since recombination in the knock-in mice depends on expression of cre, we also performed double immunocytochemistry using antibodies directed against cre and TH. However, anti-cre staining showed rather low sensitivity; there were only a few cre-positive cells in HDA-cre injected animals. Nevertheless, we only detected cells immunoreactive for both, TH and cre in the substantia nigra of wt but not L9’S HDA-cre injected animals; and as expected, there were no cre-positive cells in HDA-ctr injected mice (data not shown). To assess infection of dopaminergic neurons in all experimental groups, we also performed YFP and TH double labeling, showing many cells that coexpressed YFP and TH in WT (HDA-ctr and HDA-cre injected) as well as L9’S mice injected with HDA-ctr. We were unable to find similar cells in L9’S mice injected with HDA-cre (Fig. 6)
. To assess the efficiency of our method to infect dopaminergic neurons with HDA, we also compared the number of cells expressing TH and YFP to those expressing TH alone. These data indicate that 40% of all dopaminergic neurons in substantia nigra expressed virus-mediated YFP: wt, HDA-ctr 39.8 ± 4.3%; wt, HDA-cre 43.5 ± 12.0%; L9’S, HDA-cre 46.3 ± 8.5%. We did not count sections of L9’S HDA-cre injected animals, since there were too few cells that expressed both proteins in substantia nigra (Fig. 6M
).
Locomotor effects of dopaminergic deficits
Locomotion was tested after transfer of the mice into a new environment within a cage equipped with photobeam monitors and 30 min later, after systemic administration of amphetamine, which induces release of most of dopamine stored in vesicles of dopaminergic nerve terminals. This dopamine release induces a robust increase in locomotor behavior, and quantification of amphetamine-induced locomotor behavior may represent an indirect measure of the amount of dopamine stored in the vesicles of nigrostriatal dopaminergic nerve terminals. The initial report on the L9’S 4 knock-in mice found a nonsignificant reduction of amphetamine-induced locomotion in L9’S mice compared to their WT littermates (8)
but subsequent investigations on larger groups showed that there is a significant reduction of amphetamine-stimulated locomotion 30 to 45 min after injection (7)
.
Unfortunately, the perioperative effect of HDA injection on locomotor activity was assessed in the second experiment only. In this limited number of animals, there was a perioperative reduction in all experimental groups. However, there were no differences between WT and L9’S mice injected with either HDA-ctr or HDA cre viruses (data not shown). Cre expression and recombination was expected to occur 3–7 days after injection. Thus, we compared baseline and amphetamine stimulated locomotion in both experiments at 3 days and at 
20 days after surgery. There was no significant change of baseline or amphetamine-stimulated locomotion in either of the three control groups, for locomotion measurements 3 days and 20–33 days after surgery. In contrast, spontaneous and amphetamine stimulated locomotion was significantly reduced in L9’S mice injected with HDA-cre 20–33 days after surgery compared to locomotion 3 days after surgery (Fig. 7
D).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Helper-dependent adenoviral vector
Long term efficient therapy of chronic neuropsychiatric disorders such as Alzheimer’s disease or PD remains a major challenge. Affecting the intricate pathophysiologies of such diseases may demand modification of central nervous system (CNS) cells (23
, 24)
. Gene therapy may be a promising strategy to achieve efficient modulation of CNS pathology. However, the highly compartmentalized CNS is difficult to access due to the blood-brain barrier, and direct delivery of genes can be unpredictable due to the complex and fragile neural circuitry. If one wants to modify neurons genetically for fundamental research or therapeutic reasons, the vector should preferentially modify neurons, which make up only one cell population of the primate brain. Vectors derived from neurotropic viruses were initially thought to be the most appropriate tools to target neuronal structures. Herpes simplex virus vectors efficiently transduce some neuronal subsets (25)
. However, rapid loss of expression, likely due either to the host’s immune response or to viral-induced cytotoxic functions, restricts their use (26)
. Many vectors derived from non-neurotropic viruses such as lentiviruses, alpha-viruses, oncoviruses, adeno-associated viruses, and adenoviruses are also capable of achieving gene transfer into the CNS, where they can transduce both neuronal and glial cells. However, these gene delivery systems suffer from their poor diffusion within the parenchyma, thus vitiating their use in extended neuroanatomical structures (24
, 27
28
29)
. So far, neuron-specific gains of functions have been achieved using neuron-specific promoter elements (30
, 31)
. Although these systems do confer neuron-restricted transgene expression, glial cells may still be infected. This may create adverse effects due to vector entry into cells involved in the inflammatory response (32
, 33)
.
This study uses adenovirus vectors that are helper-dependent (also known as fully deleted, high-capacity, or gutless) (15
, 19
, 34
, 35)
. These seem to be among the most efficient and safest viral vectors (13)
. Their neurotropism renders these viruses very interesting candidates for gene therapy of neuropsychiatric disorders. HDA vectors have eliminated the majority of the problems associated with the induced CD8+-mediated destruction of transduced cells in naive immunocompetent animals, and long-term, stable transgene expression is possible in many tissues (36
37
38)
, including the CNS (13)
. HDA vectors seem to have unique neurotropic properties and are retrogradely transported to neuronal cell bodies in rat brains when injected in various structures (39
, 40)
.
Using a human HDA, we were able to detect transgene neuronal expression without any relevant immune response in the host mouse brain after direct virus injection into the brain. For gene therapy in humans an immune reaction could nonetheless occur, because there may be preexisting immunity to human adenoviruses that also interact with v integrins (41)
. However, the lack of immune response in our animals agrees with previous reports (29)
and may encourage using HDA in clinical trials.
The purpose of these experiments was to produce a single irreversible event, cre-mediated recombination, in the genomic DNA of postnatal neurons near the injection site. One also asks whether HDA can confer sustained gene expression. It is likely that this was the case, because YFP fluorescence was still strongly detectable in dopaminergic neurons even 33 days after injection (Figs. 3
4
5
6
7
). This point is particularly evidently in the several control experiments, where excitotoxicity did not kill cells. However we acknowledge that this prolonged signal could arise from the stability of the YFP protein, rather from continued gene expression from the HDA.
Cellular and circuit basis of cholinergic toxicity
In the present study, we used such a vector to increase the expression of a hypersensitive 4 nAChR allele by excision of a neo cassette that had been introduced into an adjacent intron. This cassette interfered with the transcription of the mutant allele. This effect of the neo cassette enabled heterozygous mice (L9’S) to survive and reproduce, while L9’S mice died shortly after birth and showed dramatic deficits of dopaminergic neurons when this neomycin cassette was removed in ES cells (8)
.
Postnatal cre recombination and excision of the neo cassette was also expected to damage dopaminergic neurons in substantia nigra. Our initial report on these mice found that choline at CSF levels (10 µM) activates L9’S receptors in oocytes, and we suggested that such activation would ultimately lead to excitotoxic cell death of embryonic substantia nigra dopaminergic neurons in L9’S mice (8)
. In adult L9’S mice, there also is a moderate deficit of dopaminergic neurons, probably as a consequence of such increased activation by choline (ref 7
and the present data). It remains formally possible that the TH-positive cells decrease simply because they become nondopaminergic; but frank cell death appears more likely in view of (a) the small decrease in total cell number and (b) the obvious pathology occasionally observed in dopaminergic ells (Fig. 5E
). Cytisine-sensitive [125I]epibatidine binding data show that the neomycin cassette reduces expression by 75% (14)
. In another strain that differs from L9’S by only a single codon, removal of the neomycin cassette apparently increases expression of the mutant L9’S allele by > 2-fold (42)
; and it may be assumed that a similar factor operated in the present experiments before cell death. Unfortunately, all attempts to quantify 4 nAChR protein directly have been unsuccessful (14)
.
Dopaminergic neurons in substantia nigra seem to be quite vulnerable when hypersensitive 4 nAChR are expressed. Our present data agree with previous findings suggesting that death of dopaminergic neurons correlates with the concentration of expression of hypersensitive receptor multimers (8
, 14)
. In addition to the direct excitotoxic effect, there may also be circuit-based damage. Loss of dopaminergic neurons in substantia nigra results in striatal dopamine deficiency. Further downstream, striatal dopamine deficiency leads to increased activity of the subthalamic nucleus (STN) and the internal segment of the globus pallidus. In rodents, there is both a direct pathway from the STN to substantia nigra pars compacta (43)
, and also an indirect excitatory pathway from STN via the pedunculopontine nucleus (PPN) to substantia nigra pars compacta (44)
. Increased activity in STN due to striatal dopamine deficiency also increases activity in PPN (45)
. The latter uses both glutamate and ACh as neurotransmitters. In states of dopamine deficiency, one expects the increased activity in STN to produce an increased cholinergic drive of dopaminergic neurons in substantia nigra. Lesions to the pedunculopontine nucleus have substantial protective effects in MPTP-treated parkinsonian monkeys (46)
. Thus, dopamine deficiency may lead to an increased cholinergic drive and excitotoxicity of substantia nigra dopaminergic neurons via circuit based reinforcement.
In the present study, we detected >70% loss of dopaminergic neurons in L9’S mice after HDA-cre injection compared to their wt littermates. The infection efficiency of these cells using our technique was 40% in control groups (Fig. 6)
. Since L9’S mice show a reduction of dopaminergic cells by 35% without virus injection, we conclude that a cell-autonomous mechanism, direct excitotoxicity via enhanced expression of the mutant 4 receptor, accounted for much of the dopaminergic neuron death after virus injection. However, it is quite possible that circuit based mechanisms, described above, also contributed to the profound cell loss.
A model for inducible dopaminergic degeneration
Our data show that adult cre recombination in L9’S mice is a powerful model for excitotoxic damage to midbrain dopaminergic neurons. After recombination, these mice not only show a dramatic loss of dopaminergic neurons but also exhibit behavioral changes that are compatible with dopamine deficiency and simulate bradykinesia, a cardinal symptom of PD. Thus, L9’S mice represent a first genetic mouse model where loss of dopaminergic neurons is paralleled by an appropriate behavioral phenotype; and the model is enhanced by its inducible nature: degeneration is only slight until cre recombinase is expressed from an injected viral vector. Although the behavioral effects in the present experiment remain moderate, we believe that it will be possible to use more efficient mechanisms for adult cre recombination in L9’S mice. One would have a mouse model with rapidly progressive death of dopaminergic neurons and a severe phenotype with loss of spontaneous locomotion. Injection of lentiviral or adenoviral vectors expressing human -synuclein can also induce relatively selective death of dopaminergic neurons in rats and primates (47
48
49
50)
; however, in injected rats there is substantial recovery even when -synuclein levels remain high. Furthermore, we believe that mouse models have significant advantages with respect to genetic manipulation.
Current genetic models have not reproduced death of dopaminergic neurons or relevant behavioral effects. Transgenic mice expressing human wild-type -synuclein show deficits of striatal dopaminergic innervation, intracellular inclusions, and impairment in motor learning. The -synuclein transgenic mouse with highest copy number displayed a 50% reduction of striatal TH immunoreactivity and time spent on the accelerating rotarod at 12 mo of age, while there was no difference in the number of TH-positive neurons in substantia nigra (20)
. Although these findings help to understand the role of -synuclein in PD, these mice have limited value as an animal model for PD since those changes are rather subtle. In addition, another laboratory has reported a transgenic mouse expressing wild-type human -synuclein under control of the Thy1 promoter. This mouse also shows deficits in a related rotarod experiment although neurons in the substantia nigra pars compacta did not express the transgene (51)
indicating that accumulation of -synuclein causes motor deficits independent of substantia nigra pathology.
We emphasize that no known human PD mutations map to the 4 nAChR. However, it is quite likely that increased excitation contributes to dopaminergic neuron death via circuit based reinforcement (see above). Thus, even the L9’S mice may not model the pathogenesis of PD, they clearly provide a strong tool to study the consequences of cholinergic excitotoxicity in substantia nigra. Cholinesterase inhibitors have been suggested as adjunctive therapy in PD for the treatment of cognitive deficits, dyskinesias, and/or drug induced hallucinations. Some cholinesterase inhibitors that are effective within the CNS may also allosterically modulate nAChRs (52)
. Although no data are available yet, one may raise concerns regarding a long term increase of cholinergic signaling in substantia nigra in patients with PD.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Received for publication November 28, 2005. Accepted for publication December 19, 2005.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
4 subunit of the neuronal acetylcholine nicotinic receptor in the rat substantia nigra. J. Neurosci. 19,6475-64873ß4 nicotinic acetylcholine receptors and nicotine-evoked norepinephrine release. J. Neurosci. 18,8571-85794 nicotinic receptors show dopaminergic deficits and increased anxiety. Proc. Natl. Acad. Sci. USA 98,2786-27914* receptors: sufficient for reward, tolerance and sensitization. Science 306,1029-1032
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
