 |
The Journal of Neuroscience, September 20, 2006, 26(38):9780-9793; doi:10.1523/JNEUROSCI.0840-06.2006
 Previous Article | Next Article 
Cellular/Molecular
Identification of Sequence Motifs That Target Neuronal Nicotinic Receptors to Dendrites and Axons
Jian Xu,
Yongling Zhu, and
Stephen F. Heinemann
Molecular Neurobiology Laboratory, The Salk Institute, La Jolla, California 92037
 |
Abstract
|
|---|
Neuronal nicotinic acetylcholine receptors (nAChRs) belong to a family of ligand-gated ion channels that play important roles in central and peripheral nervous systems. The subcellular distribution of neuronal nAChRs has important implications for function and is not well understood. Here, we analyzed the targeting of two major types of neuronal nAChRs by expressing epitope-tagged subunits in cultured hippocampal neurons. Surprisingly, the  7 nAChR (7) and 4/ 2 nAChR (42) displayed distinct patterns of expression, with 7 targeted preferentially to the somatodendritic compartments, whereas 42 was localized to both axonal and dendritic domains. When fused to CD4 or IL2RA (interleukin 2 receptor subunit) proteins, which are normally distributed ubiquitously, the M3–M4 intracellular loop from the 7 subunit promoted dendritic expression, whereas the homologous M3–M4 loop from the 4 subunit led to surface axonal expression. Systemic screening and alanine substitution further identified a 25-residue leucine motif ([DE]XXXL[LI]) containing an axonal targeting sequence within the 4 loop and a 48-residue dileucine and tyrosine motif (YXXØ) containing a dendritic targeting sequence from the 7 loop. These results provide valuable information in understanding diverse roles of neuronal nAChRs in mediating and modulating synaptic transmission, synaptic plasticity, and nicotine addiction.
Key words: acetylcholine; receptor; dendrite; axon; synapse; nicotine
|
Introduction
|
|---|
Nicotinic acetylcholine receptors (nAChRs) are pentameric membrane proteins that form cation-selective ion channels. There are two broad classes of nAChRs based on their general location: muscle nAChRs and neuronal nAChRs. Five genes coding for muscle nAChR subunits (, ,  ,  , and  ) and 11 genes coding for neuronal nAChRs (2–7, 9–10, and 2–4) have been identified so far in mammalian organisms (Sargent, 1993 ; Lukas et al., 1999; Corringer et al., 2000).
Neuronal nAChRs are present on many cells in the central and peripheral nervous systems. These receptors play important physiological roles and have been implicated in disease states such as Alzheimer's disease and nicotine addiction (Dani and Heinemann, 1996; Role and Berg, 1996). Structurally, neuronal nAChRs are composed of various pentameric combinations of 11 subunits with distinct pharmacological properties and anatomical localizations (Sargent, 1993). At the subcellular level, neuronal nAChRs may be localized in many parts of the cells, including cell bodies, presynaptic terminals, and postsynaptic sites (Role and Berg, 1996; Wonnacott, 1997; Dajas-Bailador and Wonnacott, 2004). Although physiological functions of neuronal nAChRs depend critically on their locations in specific regions of the neuron, little is known about the subcellular distribution of neuronal nAChRs, and even less is known about the mechanisms responsible for their assembly and targeting process.
We chose to study neuronal nAChRs containing either the 7 subunit (7) or combinations of the 4/2 subunits (42), because they are the most abundant in the brain and because they participate in important processes, such as synaptic transmission, cognitive function, gene expression, and drug addiction (Dani and Heinemann, 1996). Although the subject has been pursued intensively in the past, the subcellular location of 7 remains elusive and controversial. Electrophysiological and immunochemical evidence supports a mainly presynaptic role and location for 7 at synapses (McGehee et al., 1995; Gray et al., 1996; Fabian-Fine et al., 2001; Jones and Wonnacott, 2004), although other evidence suggests that 7-containing nAChRs are also concentrated to somatodendritic and postsynaptic locations (Alkondon and Albuquerque, 1993; Frazier et al., 1998b; Hefft et al., 1999; Levy and Aoki, 2002; Khiroug et al., 2003). The subcellular distribution of 42 is also poorly understood. According to functional and immunochemical studies, 42 may be present at presynaptic locations (Jones et al., 2001; Zoli et al., 2002; Salminen et al., 2004) and postsynaptic/somatodendritic locations (Alkondon et al., 1999; Arroyo-Jiménez et al., 1999; McQuiston and Madison, 1999; Nashmi et al., 2003). Despite these findings, the question of how 42 receptors are expressed remains unclear. In fact, it is not known whether 42 may coexist in both presynaptic and postsynaptic locations in the same cell.
Studies of the distribution and function of the neuronal nAChR have been hindered by the lack of subunit-specific antibodies (Herber et al., 2004) and the low expression of endogenous receptors in the brain. The complexity of the brain structure further prevents precise mapping of neuronal nAChRs in a single neuron. To circumvent these problems, we expressed tagged receptors in primary hippocampal cultures. Because neurons in such cultures develop morphological polarity in vitro and because their neurites are arrayed essentially in two dimensions at relatively low densities, compared with the densely packed condition in the brain, hippocampal cultures were widely used for examining polarized expression of neuronal membrane proteins (Craig and Banker, 1994). The presence of at least two subtypes of neuronal nAChRs on hippocampal cultures, as demonstrated by electrophysiological studies (Alkondon and Albuquerque, 1993) and immunohistochemical studies (Zarei et al., 1999; Liu et al., 2001; Kawai et al., 2002), further suggests that these cultures can be applied to identify conditions that control the distribution of nAChRs within neurons. In this study, we demonstrate that neuronal nAChRs composed of different subunits are expressed with distinct patterns on the membrane of cultured neurons. We further identified an axonal targeting motif within the 4 loop and a dendritic targeting motif from the 7 loop. These findings may help to elucidate cellular mechanisms involved in the precise targeting of nicotinic receptors, a prerequisite for their proper physiological functions.
|
Materials and Methods
|
|---|
Molecular biology.
Rat 7, 4, and 2 nAChR subunits were each tagged by adding hemagglutinin (HA) or green fluorescent protein (GFP) at their C terminals. All new restriction sites were generated by PCR mutagenesis with the QuikChange protocol (Stratagene, La Jolla, CA). nAChR–HA and nAChR–GFP constructs were expressed in the mammalian expression vector pcDNA3 (Invitrogen, Carlsbad, CA) or pEGFP (Clontech, Mountain View, CA). The human interleukin 2 receptor subunit (IL2RA) coding sequence was amplified by PCR from the pMLSV-N1/N4-s expression vector (American Type Culture Collection, Manassas, VA) and was further subcloned into pcDNA3. Dr. Gary Banker (Oregon Health & Science University, Portland, OR) kindly provided transferrin receptor (TfR)–GFP. Postsynaptic density-95 (PSD-95)–GFP was provided by Dr. David Bret (University of California, San Francisco, CA). CD4 construct was provided by Dr. Lily Jan (University of California, San Francisco, CA). Tau–GFP was provided by Dr. John Thomas (The Salk Institute, La Jolla, CA), and GFP–NMDA receptor 1 (NR1) was provided by Dr. Stefano Vicini (Georgetown University, Washington, DC).
Cell culture preparation, DNA transfection, and protein staining.
Hippocampal neurons from postnatal day 0 rats were cultured using standard techniques. After  1 week in culture, cells were transfected using the calcium phosphate method. To detect proteins expressed on the cell surface, living cells were incubated with primary antibody diluted in culture medium for 10–15 min at room temperature, fixed (4% paraformaldehyde and 4% sucrose for 20 min), and processed for additional immunostaining. To detect total proteins, cells were fixed before permeabilization (1.5% Triton X-100 in PBS buffer) and immunostaining. To detect internalized proteins, live neurons were incubated with primary antibody for 30 min at 37°C, chilled in 4°C TBS to stop endocytosis, and then exposed to 0.5 M NaCl/0.2 M acetic acid, pH 3.5, for 4 min on ice to remove surface-bound antibody. Cultures were subsequently rinsed, fixed, and permeabilized, followed by secondary antibody staining of internalized proteins. Antibodies/probes were used as follows: mice anti-HA (Covance, Princeton, NJ); rabbit anti-HA (Santa Cruz Biotechnology, Santa Cruz, CA); rabbit anti-GABA (Sigma-Aldrich, St. Louis, MO); chicken anti-GFP antibody (Clontech); rabbit anti-synapsin (Millipore, Billerica, MA); anti-PSD-95 monoclonal (Affinity BioReagents, Golden, CO); anti-CD4 monoclonal (Caltag Laboratories, South San Francisco, CA); anti IL2RA monoclonal (Covance); anti-microtubule-associated protein 2 (MAP2) monoclonal (Sigma-Aldrich); rabbit anti-MAP2 (Millipore); rabbit anti-neurofilament (Millipore); Alexa 488 -bungarotoxin (Invitrogen, Eugene, OR); Alexa dye-labeled secondary antibodies (Invitrogen); and cyanine 3 (Cy3)- and Cy5-labeled secondary antibodies (Jackson ImmunoResearch, West Grove, PA).
Image analysis and processing.
Neurons were imaged on an LSM 510 microscope (Zeiss, Thornwood, NY). Each cell was imaged as a stack of optical sections, 1 µm in depth apart. The cell body was approximately in the center of each image. All calculations were performed on compressed images. All measurements were made with NIH ImageJ software. VolumeJ from the BIJ plugin packages was installed to ImageJ for three-dimensional (3D) reconstruction. The CorrelationJ 1e and JACoP plugins were used to calculate the overlap coefficient.
We used several methods to distinguish axons and dendrites. To facilitate morphological distinctions, GFP was coexpressed along with neuronal nAChRs to trace and visualize neurites. Previously defined morphological criteria were applied to distinguish neurites as axons or dendrites (Dotti et al., 1988). For example, dendrites have irregular contours, taper gradually, and branch at relatively acute angles. Axons, in contrast, display a smoother contour, have a relatively even diameter along their course, and extend much farther away from the cell body. Among these traits, particularly valuable in our study is the distinction that all of the long processes in the hippocampal culture represent axons (Dotti et al., 1988). This distinction was confirmed in our own experiments. We examined >30 GFP-transfected cells. In every case, neurites extending at least 250 µm from the cell body corresponded to axons, based on their negative immunostaining of MAP2. In this study, we identified a neurite as an axon only when it extended at least 250 µm from the cell body and also fitted other morphological distinctions described above (an example was provided in supplemental Fig. 5G–J, available at www.jneurosci.org as supplemental material). In addition to morphological criteria, axons and dendrites were further identified in some experiments based on a differential distribution of MAP2 (a dendritic marker) (supplemental Fig. 5A–F, available at www.jneurosci.org as supplemental material) and neurofilament (an axonal marker) (see Fig. 4). Dendritic protrusions were classified into two broad categories, spines and filopodia, based on criteria defined previously (Papa et al., 1995; Hering and Sheng, 2001). Briefly, a filopodia-like structure was identified if it was >2 µm in length with a base wider than the tip and did not terminate in enlarged heads, which is typical for spines.
Xenopus oocyte preparation, cDNA injection, and whole-cell current recording.
Standard protocol was used for oocyte preparation. Subunit cRNAs were synthesized in vitro (Message Machine; Ambion, Austin, TX) from linearized plasmid templates. Whole-cell currents were measured at room temperature, 1–2 d after RNA injection, using a GeneClamp 500 amplifier (Molecular Devices, Union City, CA) in a standard two-microelectrode voltage-clamp configuration. Oocytes were clamped between –40 and –65 mV and superfused in ND96 containing 1.8 mM CaCl2 and 1 µM atropine.
|
Results
|
|---|
C-terminally tagged receptors are expressed at the surface of transfected neurons and are capable of forming functional channels
To facilitate labeling, we first tagged the neuronal nAChRs by adding HA or GFP to their C terminals. Based on the topology of the receptor, tags at the C terminal are less likely to interfere with receptor trafficking than insertions at the N-terminal or intracellular loop domains. We confirmed that tagged neuronal nAChRs were capable of forming functional channels with biophysical properties similar to those of wild-type receptors when expressed in Xenopus oocytes (Fig. 1A–D). Surface expression of tagged receptors could not be detected in human embryonic kidney (HEK) cells (data not shown), although internal expression was visualized by immunofluorescence and detected by Western blot analysis (data not shown).
View larger version (61K):
[in this window]
[in a new window]
|
Figure 1. Tagging the receptors at C termini does not affect function and assembling of neuronal nAChRs. A–D, Representative examples of nicotine-induced currents from oocytes expressing wild-type 42 (A), 4–HA/2–HA (B), wild-type 7 (C), and 7–HA (D). Voltage clamping was performed 24 h after injecting RNA into Xenopus oocytes. E–J, Surface staining of recombinant 7 is consistent with -bungarotoxin labeling. Hippocampal neurons were transfected with 7–HA at 8 DIV and stained at 12 DIV. E, Surface expression of recombinant 7 stained by HA antibody. F, Surface expression of 7 stained by -bungarotoxin (a-BGT). G, Same view as E and F. Cell nuclei were labeled with 4',6'-diamidino-2-phenylindole dihydrochloride (DAPI) to show that one neuron of many from this view was transfected and that the surface expression level of endogenous 7 from nontransfected cells in this view was below detection using fluorescence-labeled -bungarotoxin in our experiments. H, Overlay of E and F shows the consistent labeling of surface 7 by HA antibody and -bungarotoxin, except for occasional nonspecific labeling of apparent cell debris by Alexa 488–-bungarotoxin. I, Enlarged boxed region from H. Ia, HA stain. Ib, a-BGT stain. Ic, Overlay of Ia and Ib. J, Immunofluorescence intensity profiles along the 100 µm line in H were measured from the original TIFF (tagged image file format) images with NIH ImageJ. Scale bars: (in G) E–H, 50 µm; I, 15 µm.
|
|
We transfected cultured primary neurons with tagged 7 using the CaPO4 method. In marked contrast to HEK cells, neuronal surface expression of recombinant 7 was readily detectable by immunofluorescence (Fig. 1E–J). Furthermore, 7 surface expression detected by antibody staining resembled staining by -bungarotoxin (Fig. 1E–J), which binds with high affinity to assembled 7 receptors (Couturier et al., 1990; Kawai et al., 2002). Therefore, adding an epitope tag does not appear to interfere with receptor assembly. However, HA antibody staining yielded a higher signal/background ratio and better spatial resolution and was more suitable for examining the subcellular distribution in our experiments (Fig. 1J).
Polarized expressions of 7 and 42 receptors in hippocampal neurons
We first examined the axonal/somatodendritic distribution of neuronal nAChRs in cultured hippocampal neurons. GFP was coexpressed along with the tagged receptors to visualize cell morphology. To avoid labeling intracellular receptors that might confound our results, we initially stained surface receptors only by applying antibody to live neurons for a short period of time at room temperature (see Materials and Methods). The distributions of expressed receptors were examined in transfected neurons that had clearly identifiable axons and dendrites (see Materials and Methods). Surprisingly, surface 7–HA immunostaining was detected almost exclusively in dendrites (Fig. 2B). Much lower surface expression was detected at proximal axonal regions (within 100 µm from the cell body), but the signal decreased sharply along the axonal process (Fig. 2B, arrows). Weak axonal immunoreactivity away from the cell body (beyond 200 µm) was detected in 5% of the transfected neurons (10 independent transfections) (Fig. 2E). In contrast, surface 42 was localized to both axons and dendrites (Fig. 2D) in all of the transfected neurons (Fig. 2E). As expected, surface expression was not detectable in cells transfected with 4 or 2 alone (data not shown).
View larger version (40K):
[in this window]
[in a new window]
|
Figure 2. Differential targeting of 7 and 42 expressed in cultured hippocampal neurons. A–D, Neurons were transfected at 8 DIV with 7–HA plus GFP. Surface receptor was labeled at 16 DIV with HA antibody. GFP fluorescence was used to outline neuronal morphology. Multiple images were taken from the LSM 510 confocal microscope and assembled to obtain a larger view. Unlike GFP (A), surface 7 (B) was targeted to dendrites and excluded from axons (arrowheads). Low surface expression of 7 was detected at initial regions of axons. This signal sharply decreased along the axonal process. Also note that surface expression of 7 in transfected glia cells (A, top left) was below detection in B. Surface expression of the 42 receptor (D) was localized to both soma-dendrites and axons in the transfected neuron also expressing GFP (C). Scale bar, 50 µm. Note that the surface 42 level was considerably lower than that of 7. Higher amplification in microscope scanning was necessary to acquire 42 imaging. E–G, Quantitative analysis of receptor polarization in neurons. E, Percentage of cells expressing recombinant proteins in axons. Unlike GFP and 42, surface expression of 7 at axons 200 µm away from the cell body was detected in only 5% of transfected cells (10 independent transfections). F, Normalized ratios of average fluorescence intensity for dendrites versus axons (see methods in supplemental Fig. 1, available at www.jneurosci.org as supplemental material). G, Averaged intensity profiles of surface 42 and 7 along axons. The method used is described in supplemental Figure 1 (available at www.jneurosci.org as supplemental material). The decay constant for 7 traveling along axons was estimated to be 45.6 µm. There is no reduction of 42 signal along axons. Error bars are SEM.
|
|
Quantitative examination of protein polarization is complicated by differences in the diameters of axons and dendrites, which result in a higher intensity in dendrites versus axons, even for proteins that are expressed ubiquitously, such as GFP. To more accurately assess receptor polarization, we used two methods for quantitative analysis (supplemental Fig. 1, available at www.jneurosci.org as supplemental material).
First, coexpressed GFP was used for normalization, as described previously (Rivera et al., 2003; Sampo et al., 2003). A value for normalized ratio of dendrite-to-axon expression (D/A) was derived by dividing the ratio of receptor density in dendrites versus axons by the corresponding ratio of GFP from the same cell (supplemental Fig. 1A–D, available at www.jneurosci.org as supplemental material). This normalized D/A ratio should reflect the extent of polarization expression more precisely. In our study, D/A value was calculated for at least 20 different cells for each construct. The average D/A value for 7 was 12.94 ± 0.68 (n = 20), indicating a preferential localization to the dendrite, significantly different from the D/A value for 42 (0.76 ± 0.069; n = 20; p < 0.001) (Fig. 2F).
Second, we used GFP to visualize the axonal processes, and we analyzed the surface expression of recombinant proteins along the axons (supplemental Fig. 1E–G, available at www.jneurosci.org as supplemental material). As seen in Figure 2B and supplemental Figure 1E (available at www.jneurosci.org as supplemental material), the intensity of surface 7 decreased drastically down the axons. In fact, such a decline fits well with single-exponential decay with a distance constant of 45.6 µm (n = 14) (Fig. 2G). Unlike 7, 42 surface expression level did not vary along axons (Fig. 2D,G).
It has been established that distinct mechanisms underlie specific targeting of neuronal membrane proteins (Silverman et al., 2001; Sampo et al., 2003). In the case of 7, the absence of axonal surface expression may arise from either the absence of axonal transport or the absence of subsequent membrane insertion and stabilization. To distinguish between these two possibilities, we used permeabilized cells to stain intracellular as well as surface proteins. If 7 was transported but not inserted into the membrane, then axonal staining would be expected in permeabilized cells. Instead, we observed a predominant dendritic localization of total 7 (supplemental Fig. 2D–F, available at www.jneurosci.org as supplemental material) (D/A, 12.4 ± 0.55; n = 20), similar to surface 7 (D/A, 12.94 ± 0.68; n = 20). Thus, selective transport is likely to account for polarized expression of 7, resembling the mechanism of dendritic targeting observed for Kv4.2 potassium channels (Rivera et al., 2003) and TfR–GFP (Silverman et al., 2001).
To more directly compare the expressions of 7 and 42 within one cell, we cotransfected GFP-tagged 7 (7–GFP) and HA-tagged 4 and 2 into primary neurons. We used double labeling with antibodies to GFP and HA to visualize surface 7 and 42 simultaneously. Consistent with previous results, subcellular expression profiles of 7 and 42 in the same neurons were quite distinctive, with surface 7 localized preferentially to somatodendritic compartments, in contrast to surface 42, which was detected at both axons and dendrites (Fig. 3A–D). Even in dendrites in which both 7 and 42 were expressed, a fraction of 7 clusters did not colocalize with those of 42 (Fig. 3E–F) (Pearson's coefficient, r = 0.56 ± 0.04; n = 6). This result once again indicates that distinct targeting mechanisms are used for different nAChR subunits.
View larger version (57K):
[in this window]
[in a new window]
|
Figure 3. Direct comparisons of surface 7 and 42 in the same transfected neuron. A, B, Hippocampal neurons were cotransfected with 7–GFP and HA-tagged 4 and 2 (4–HA/2–HA) at 8 DIV. Surface 7 was stained by chicken anti-GFP antibodies, and surface 42 was stained by mouse anti-HA antibody at 16 DIV. Neurons were fixed and blocked before incubating with secondary antibody (Alexa 568-labeled goat anti-mouse and Alexa 647-labeled goat anti-chicken antibodies). Images were scanned from the LSM 510 confocal microscope. Pseudocolors were initially applied to represent surface signals of 7 (green) and 42 (red) before converting to gray channel. Multiple images were taken from the LSM 510 confocal microscope and were assembled to obtain a larger view. Surface 7 was mostly expressed in soma-dendrites; a faint signal can be detected at initial axonal regions, but it decreased below detection along the axons (A). Surface 42 was expressed at both axons and soma-dendrites (B). Arrowheads indicate axon regions. C, D, Intensity line profiles of 7 and 42 along axons from the transfected neuron shown in A and B. Arrows mark the start of blank regions. E, F, Neurons were transfected and stained similarly to those in A and B, except that staining was performed at 22 DIV. A neuron expressing both 7–GFP and 4–HA/2–HA is shown in E. The boxed region from E is enlarged in F. Although 7 and 42 were both expressed in somatodendritic compartments, a large fraction of surface clusters from different receptors did not overlap (F). Scale bars: (in B) A, B, 100 µm; E, 30 µm; F, 10 µm.
|
|
Comparisons of surface nAChRs to neuronal subcellular markers
In addition to morphological criteria, we also used MAP2 staining and neurofilament staining to discriminate between soma-dendrites and axons. Consistent with previous conclusions, 7 was present within dendrites marked by MAP2 antibody (Fig. 4A,B, supplemental Fig. 5A–C, available at www.jneurosci.org as supplemental material) but was absent from axons labeled with neurofilament antibody (Fig. 4E–G). When coexpressed with other markers, 7 expression overlapped with TfR–GFP (supplemental Fig. 3A–C, available at www.jneurosci.org as supplemental material), a dendritic marker (Silverman et al., 2001), but not with Tau–GFP (Fig. 4A,C), a protein marker highly expressed in axons (Callahan et al., 1998). Altogether, these results confirmed that surface-expressed 7 was highly restricted to somatodendritic compartments. In contrast, 42 staining was found in axons labeled with neurofilament antibody (Fig. 4H–J) as well as dendrites labeled by MAP2 antibody (supplemental Fig. 5D–F, available at www.jneurosci.org as supplemental material). Using a different set of markers, 42 staining was found in both dendrites and axons, labeled by TfR–GFP (supplemental Fig. 3D–F, available at www.jneurosci.org as supplemental material) and Tau–GFP (supplemental Fig. 4D–F, available at www.jneurosci.org as supplemental material), respectively.
View larger version (90K):
[in this window]
[in a new window]
|
Figure 4. Comparison of 7 and 42 with dendritic and axonal markers. A–D, 7–HA was cotransfected with Tau–GFP into cultured neurons at 8 DIV and stained at 12 DIV. Surface 7 (A) was MAP2 positive (B) but did not overlap with Tau–GFP in axons (C). D, Merged images of A–C. The asterisk in B indicates the transfected neuron. E–G, HA-tagged receptor subunits were cotransfected with GFP into cultured neurons at 8 DIV. Neurons were triple labeled with surface 7 or 42 (red), GFP (green), and neurofilament (blue) at 12 DIV (F, I). Surface 7 (E) was mostly excluded from axons labeled by neurofilament antibody (F, G). In contrast, surface 42 (H) was found in neurofilament-positive axonal fibers (I, J). Boxed regions in F and I are shown with separated channels at higher magnification in G and J, respectively. Arrowheads mark axons. NF, Neurofilament. Scale bars: (in D) A–D, 50 µm; (in F, G, I, J) E–J, 25 µm.
|
|
Surface expression of nAChRs in hippocampal GABAergic neurons
In the hippocampus, neuronal nAChRs are distributed on both the excitatory pyramidal neurons (McGehee et al., 1995; Gray et al., 1996) and the inhibitory GABAergic interneurons (Jones and Yakel, 1997; Frazier et al., 1998a; Alkondon et al., 2000; Khiroug et al., 2003). A large body of evidence indicates that neuronal nAChRs in the GABAergic neurons can modulate synaptic transmission and influence inhibitory activity in the brain (Alkondon et al., 1997; Frazier et al., 1998b; Ji et al., 2001). We next examined the distribution of receptors in GABAergic neurons (4–8% of the neurons in our preparation). Hippocampal cultures were stained with anti-GABA antibody for the identification of GABAergic neurons (Fig. 5D,H). Analysis of transfected GABAergic cells (>20 cells from multiple transfections) demonstrated the predominant surface expression of 7 in the somatodendritic compartment (Fig. 5A–D), compared with ubiquitous expression of 42 in both axons and dendrites (Fig. 5E–H). Thus, we found that the distribution of neuronal nAChRs was similar in both inhibitory and excitatory neurons.
Synaptic locations of nAChRs in hippocampal neurons
Because neuronal nAChRs play important roles in synaptic transmission (McGehee et al., 1995; Gray et al., 1996; Mansvelder and McGehee, 2000; Ji et al., 2001), we next inspected the synaptic locations of 7 and 42 in cultured hippocampal neurons. We cotransfected HA-tagged 7 with postsynaptic marker PSD-95–GFP (Fig. 6). On neurons 12 d in vitro (DIV) in culture, the patches of 7 subunits on dendrites often colocalized with the postsynaptic marker PSD-95–GFP (Fig. 6A1–A3, arrows), in agreement with previous findings of the postsynaptic location of 7 (Liu et al., 2001; Kawai et al., 2002). Interestingly, 7 was also frequently observed at long necks and tips of the filopodia-like structures (Fig. 6A1, arrowheads), a phenomenon described previously from GABAergic neurons (Kawai et al., 2002). Because it is not always easy to distinguish filopodia from spines based on morphology standards, we also used glutamate receptor 1 (GluR1) staining, which was shown previously to form clusters almost exclusively at dendritic spines (Rao et al., 1998), as a marker. We found that the majority of PSD-95–GFP puncta were positive in 7 and GluR1 staining (arrows). However, at the tip or neck of thin structures that look like filopodia (arrowheads), only 7 staining was found, whereas both PSD-95 and GluR1 were absent or did not appear as punctate clusters (Fig. 6A1–A4). To determine whether the location of 7 in dendrites follows a developmental pattern, we examined its distribution from neurons with different days in culture. In confirmation of previous studies (Papa et al., 1995; Boyer et al., 1998), the fraction of filopodia-like structures decreased and the density of spine synapses increased as cultures matured (Fig. 6A2,B2,C2). Consistent with this time course, the distribution of 7 was also shifted to structures that showed concentrated clustering of PSD-95 and GluR1. The overall Pearson's coefficient of 7 to PSD-95–GFP increased from 0.71 ± 0.024 (n = 10; 12 DIV) to 0.78 ± 0.022 (n = 10; 17 DIV; p = 0.037; t test) and up to 0.82 ± 0.025 (n = 10; 22 DIV; p = 0.26; t test, 17 vs 22 DIV). Even at 22 DIV, when dendritic spines with distinct spiny heads were more evident (Fig. 6C2), 7 can still be seen in the few visible long filopodia. Together, these results confirm that 7 may be expressed at both mature synapses and filopodia-like structures in hippocampal neurons.
View larger version (47K):
[in this window]
[in a new window]
|
Figure 6. The locations of 7 at spines and filopodia-like structures. Shown are dendritic regions from hippocampal neurons cotransfected with 7–HA and PSD-95–GFP at 8 DIV. Cultures at 12 (A1–A4), 17 (B1–B4), and 22 (C1–C4) DIV were triple labeled with surface 7, PSD-95–GFP, and endogenous GluR1. Arrowheads label colocalized puncta of 7, PSD-95, and GluR1. Arrows point to locations at the tip or neck of filopodia-like structures at which 7 was present, but PSD-95 and GluR1 were not. Spine synapses were more evident in older neurons. The fraction of filopodia-like structures decreases progressively as the neurons mature. Consistent with this time course, 7 distribution was shifted to punctate clustering similar to that of PSD-95 and GluR1. The overall Pearson's coefficient of 7 to PSD-95 increased from 0.71 ± 0.024 (n = 10; 12 DIV) to 0.78 ± 0.022 (n = 10; 17 DIV; p = 0.037, t test) and farther to 0.82 ± 0.025 (n = 10; 22 DIV; p = 0.26, t test, 17 vs 22 DIV). Scale bars, 15 µm.
|
|
To further analyze the synaptic locations of 7, 3D reconstructions were performed from confocal image stacks. Surface-expressed 7 was often concentrated in discrete clusters within dendritic spines of mature neurons (Fig. 7C). Furthermore, 7 clusters usually were in close contact but not overlapped with presynaptic clusters labeled with anti-synapsin antibody (Fig. 7B,D) (Pearson's coefficient, r = 0.369; Manders' coefficients, M1 = 0.141, which represents the fraction of synapsin overlapping 7). (Note that calculation based on compressed images tends to overestimate the degree of overlapping. The actual overlapping coefficient is likely to be even smaller.) Finally, double labeling of tagged 7 (Fig. 7E,F, green) and endogenous PSD-95 (Fig. 7E,F, red) revealed a high degree of colocalization in mature neurons (22 DIV; note that PSD-95 puncta outside of 7 were from nontransfected cells). All of these results are consistent with the location of 7 at or close to postsynaptic sites.
View larger version (67K):
[in this window]
[in a new window]
|
Figure 7. Comparisons of recombinant 7 with endogenous synaptic markers. A–D, 3D reconstruction from image stacks scanned from a cell (22 DIV) transfected previously by 7 and GFP. 7 surface expression (C, red) was concentrated on spines (arrowheads) and often juxtaposed to presynaptic clusters labeled with anti-synapsin antibody (B, blue; D, merged). A small fraction of 7 puncta was also seen at filopodia-like structures (A, C, arrows) that usually do not have presynaptic counterparts (D). Matured mushroom-like spines are marked by arrowheads. E, F, Double labeling of neuron for surface-expressed 7 and endogenous PSD-95. 3D reconstruction (E) from boxed region in F showed colocalization of 7 (green) and endogenous PSD-95 (red). Image stacks were scanned from a cell (22 DIV) transfected previously by 7. Note that the PSD-95 antibody stained transfected and nontransfected cells. Scale bars: (in D) A–D, 10 µm; (in F) E, 20 µm; F, 40 µm.
|
|
We also examined possible synaptic locations of 42, in addition to 7 (Fig. 8). In the somatodendritic compartments, a number of 42-positive clusters were also apposed to endogenous presynaptic clusters labeled by anti-synapsin antibody (Fig. 8B,D), similarly to 7 (Fig. 8A,C). However, 42 clusters were also distributed with presynaptic clusters in axons (Fig. 8F), in contrast to the absence of 7 in these regions (Fig. 8E). Like 7 in dendrites (Fig. 8G), fractions of 42 clusters were also found to overlap with PSD-95–GFP clusters (Fig. 8H), with a degree of correlation not significantly different from that of 7 (Pearson's r = 0.67 ± 0.04 for 42, n = 8, 12–14 DIV; r = 0.71 ± 0.02 for 7, n = 10, 12–13 DIV; p > 0.05, t test; Pearson's coefficients were calculated by ImageJ with soma regions excluded for analysis).
View larger version (110K):
[in this window]
[in a new window]
|
Figure 8. Synaptic locations of recombinant 42. A–F, Neurons transfected with tagged 7 (A) or 42 (B) were surface labeled with mouse anti-HA antibodies (red), followed by labeling with rabbit anti-synapsin antibodies at 16 DIV (green). C, D, Enlarged images from the boxed regions in A and B, respectively. Note that the clusters of 7 from the dendritic region were often in close contact with presynaptic clusters (C, arrowheads). Fractions of 42 clusters were also in close contact with presynaptic marker (D, arrowheads). Small regions 300 µm from the cell body in which axons were located were shown from 7-transfected cells (E) or 42-transfected cells (F). G, H, Neurons transfected at 8 DIV with tagged 7 (G) or 42 (H) plus PSD-95–GFP (middle) were surface labeled with anti-HA antibodies (top) at 14 DIV. A number of overlapped clusters were indicated by arrowheads (bottom). Within the somal area, PSD-95–GFP was observed in a perinuclear distribution, typical for intracellular proteins. Scale bars: (in B) A, B, 50 µm; (in D) C, D, 20 µm; (in F) E, F, 20 µm; (in H) G, H, 20 µm.
|
|
Finally, we cotransfected tagged 7 or 42 with presynaptic marker synaptophysin–GFP (Fig. 9). There is very little overlap between 7 and synaptophysin (Fig. 9A,B), other than in the cell bodies (Fig. 9A,B, inset), consistent with the absence of 7 in axons (Fig. 2). Both 42 and synaptophysin–GFP were expressed along axon fibers, although 42 was distributed more uniformly compared with synaptophysin–GFP (Fig. 9C,D). More than 60% of the synaptophysin–GFP clusters, the expression of which often associates with functional presynaptic compartments (Y. Zhu and C. F. Stevens, unpublished data), were colocalized with 42-positive clusters (Fig. 9E). Colocalization analysis also revealed a high degree of correlation between synaptophysin–GFP and 42 (Pearson's coefficient, r = 0.65 ± 0.06; n = 6 cells, >2000 clusters) (Fig. 9F).
M3–M4 loops contain subcellular targeting sequences
Sequences from several groups of membrane proteins have been identified to mediate dendritic or axonal targeting (Stowell and Craig, 1999; Ruberti and Dotti, 2000; Garrido et al., 2001; Gu et al., 2003; Rivera et al., 2003). It is thus natural to speculate that specific sequences act as signals within neuronal nAChRs to mediate subcellular targeting. We examined the M3–M4 intracellular loop for several reasons. First, primary sequences in the M3–M4 loop are the most diverse among different receptors (Le Novere et al., 1999; Corringer et al., 2000) (see Fig. 11C for sequence alignment). This may be consistent with the diversity of polarized expression observed for different receptors. Second, M3–M4 loops compose the largest cytoplasmic domain within neuronal nAChRs that is likely to interact with the machinery involved in the protein sorting, trafficking, and membrane insertion. Indeed, it has been demonstrated that the M3–M4 loop from chick 3 subunit targets this receptor to subdomains within individual synapses on developing chick ciliary ganglion (CG) neurons in vivo (Williams et al., 1998; Temburni et al., 2000).
We constructed a series of chimeras by swapping M3–M4 loops among the 4, 7, and 2 subunits. However, surface expression was not observed from cells expressing chimeras alone or in various combinations (data not shown). To determine whether M3–M4 loops contain subcellular localization motifs, we attached loop sequences to the C terminal of the CD4 protein, a transmembrane protein that localizes nonspecifically in neurons (see Fig. 10J) (Gu et al., 2003). Consistent with previous reports, surface expression of CD4 was not polarized in our study (D/A, 0.78 ± 0.10; n = 20) (Fig. 10A–C). The addition of the M3–M4 loop from 7 caused the resulting fusion protein, CD4–7, to localize to the soma-dendrites (D/A, 6.52 ± 0.46; n = 20) (Fig. 10G–I). This was in marked contrast to the localization of CD4–4 (D/A, 0.09 ± 0.015; n = 20) (Fig. 10D–F). The difference in the pattern of expression was significant (p < 0.001). The CD4–2 was not expressed to the surface of transfected neurons (data not shown). In summary, these results demonstrate that the M3–M4 loop can function out of the context of neuronal nAChRs to mediate axonal/dendritic targeting and strongly suggest that M3–M4 loops are involved in the targeting of neuronal nAChRs.
Identifying sequence motifs that target neuronal nAChRs to dendrites and axons
To further identify the critical determinants in the 4 and 7 loops, a series of C-terminal or N-terminal truncations or internal fragments of M3–M4 loops from the 4 (Fig. 11A) and 7 (Fig. 11B) subunits were fused to another nontargeted protein, IL2RA (IL2RA was chosen instead of CD4 to facilitate screening because of its higher surface expression in our system). Positions of amino acids are based on the sequence alignment of the loop sequences shown in Figure 11C. All IL2RA fusion proteins were transfected into cultured hippocampal neurons. Ratios of polarized expression were determined by comparing with coexpressed GFP (Fig. 11A,B, right columns). Analysis of surface distributions from these constructs identified a 25-residue axonal targeting sequence within the 4 loop (residues 30–54) (Fig. 11A) and a dendritic targeting sequence from the 7 loop (residues 33–80) (Fig. 11B). The amino acid sequences for both motifs are illustrated in Figure 11D.
View larger version (27K):
[in this window]
[in a new window]
|
Figure 11. Identification of motifs in M3–M4 loops from the nAChR 4 and 7 subunits that mediate targeted polarization in neurons. A, B, Series of C-terminal or N-terminal truncations or internal fragments of M3–M4 loops from 4 (A) or 7 (B) were fused to IL2RA. Red, black, and blue bars indicate "polarized surface expression," "nonpolarized" (NP), and "not available" (NA), respectively. Relative ratios of polarization for each construct were also represented by + symbols, with +++ representing the most polarized and + representing slightly polarized. Low surface expression was observed from some constructs (A, B, right column); for most of these constructs, surface polarization ratios were labeled as NA because of the difficulties in image analysis. C, Sequence alignment of M3–M4 loops from 4, 7, and 2. The axonal sequence in 4 is highlighted in yellow, the dendritic signal of 7 is highlighted in pink, and the first 30 aa from 2, which abolishes surface expression, is highlighted in blue. D, Identified targeting sequences. Analysis of the surface distribution from fusion proteins defined a 25-residue sequence within 4 loops that governed axonal expression (residues 30–54) and a 48-residue segment from the 7 loop (residues 33–80) that was necessary for dendritic expression. Potential key targeting motifs within the identified segments are underlined, with the consensus motif marked above or below.
|
|
Multiple signals that consist of short sequences of amino acid residues have been identified previously to mediate membrane sorting and endocytosis (Bonifacino and Traub, 2003). Interestingly, some of these signals have also been implicated in basolateral targeting in polarized epithelial cell and axon/dendrite targeting in neurons (Rivera et al., 2003). Close examination of targeting sequences revealed several potential motifs, such as a leucine-based signal and a tyrosine-based signal (Fig. 11D). To test the hypothesis that these short motifs might be involved in the targeted expression of the receptor, mutations were introduced to either 7–HA or CD4–4, which we found to be the most polarized to either dendrites or axons, respectively (Figs. 2F, 10K). Three amino acid changes (LIE/AAA) in the leucine-based motif ([DE]XXXL[LI]) from the 4 axonal targeting sequence (Fig. 11D, top) almost abolished axonal targeting of CD4–4 (Fig. 12A,C). In contrast, mutation of the FW sequence (FWP/AAA) had little effect (Fig. 12C). The 7 dendritic targeting sequence (Fig. 11D, bottom) comprises a stretch of 6 aa (LLYIGF) that may include both dileucine and YXX motifs. Mutating three critical residues in these motifs (LLY to AAA) led to a fivefold decrease in D/A ratio (Fig. 12B,D), whereas mutations in the diarginine sequence (PRR/AAA) had little influence on D/A ratio (Fig. 12D). Furthermore, mutations within either the YXX motif (YIG/AAA and IGF/AAA) or the dileucine motif (LL/AA mutant) all resulted in reductions of D/A ratios (Fig. 12D), suggesting that both motifs participated in dendritic targeting of 7, although the YXX motif might play more important roles, because larger decreases in D/A ratios were observed from the YIG/AAA and IGF/AAA mutations than from the LL/AA mutation (Fig. 12D).
Axonal targeting signal in the 4 loop is inactive in the assembled 42 complex
Although the M3–M4 loop from 4 was able to confine nonpolarized proteins (CD4 and IL2RA in our experiments) exclusively to the axonal surface (Figs. 10D,E, 11A, 12A), assembled 42 was expressed more uniformly in axons and dendrites. There are two possibilities: either the axonal targeting signal in 4 was not active (for instance, it might be "masked" in the assembled 42 complex) or there might be competing signals in the 2 sequence or other regions in the 4 sequence that promote dendritic expression.
If 4-targeting signal is actively involved in targeting of 42, then mutations within this signal are predicted to affect 42 subcellular distribution. However, unlike the (LIE/AAA) mutations in CD4–4, which almost abolished axonal targeting (Fig. 12A), the same mutations introduced in the 4 subunit only resulted in a modest change in D/A ratio when coexpressed with the 2 subunit (0.83 ± 0.073, n = 20, compared with 0.76 ± 0.069 from wild-type 42, n = 20). This difference was not statistically significant (p = 0.49; t test), suggesting that axonal signal from 4 does not play active roles in the 42 complex.
Next, we sought to determine whether there might be targeting signals in the M3–M4 loop of 2 subunits. Because attaching a full 2 loop to CD4 and IL2RA precluded surface expression, we adhered IL2RA to different regions from the 2 loop. Fusion of IL2RA to the first 30 aa from the 2 loop abolished surface expression (data not shown). In contrast, the chimeric protein with a truncated 2 loop that lacks the first 30 aa (IL2RA-2 31–126) was expressed on neuronal surface in a nonpolarized pattern (D/A, 0.89 ± 0.13; n = 10) (supplemental Fig. 5G–J, available at www.jneurosci.org as supplemental material), suggesting that the 2 loop region (31–126) does not contain a subcellular targeting signal. Although we cannot positively exclude the possibility of the existence of a subcellular targeting signal in the first 30 aa of the 2 loop, sequence examination suggests that it is less likely, because this region is rather hydrophobic (Ren et al., 2005) and thus unlikely to be exposed in folded protein. In fact, residues in this region were implicated in early stages of receptor biogenesis, such as subunit assembly (Yu and Hall, 1994) and endoplasmic reticulum (ER) export (Ren et al., 2005), that take place before vesicle sorting, transporting, and endocytosis events, which regulate polarized expression (Horton and Ehlers, 2003).
With very few exceptions (Sampo et al., 2003), almost all known neuronal targeting signals were found in cytoplasmic regions of membrane proteins. Another cytoplasmic domain in neuronal nAChRs, the M1–M2 loop, might be too short (only 6 aa, with an almost identical sequence among all neuronal nAChRs) to hold functional targeting signals, as was confirmed by nonpolarized expression pattern when these short loops were attached to CD4 (data not shown). Whether or not targeting signals may exist in other regions in the 2 and 4 sequences remains to be determined experimentally. In summary, we found that mutation of key residues in the axonal targeting motif of 4 did not affect surface distribution of the 42 receptor. We did not find other targeting signals in 2 and 4 sequences from the regions we examined. Altogether, it is possible that the difference of axonal expression of fusion proteins containing the 4 loop (CD4–4 and IL2RA–4) versus nonpolarized expression of the 42 receptor may be caused by "inactivation" or masking of axonal signal in the 4 subunit in the assembled 42 receptor complex.
Dendritic targeting signal in the 7 loop mediates selective delivery, whereas axonal targeting signal in the 4 loop mediates fast dendritic endocytosis
One of two general mechanisms may be responsible for the selective localization of polarized proteins on the surface of neurons: selective transport or selective retention. In the first scenario, proteins destined for either axons or dendrites are segregated from one another into different carriers and only delivered to appropriate subcellular domains. Alternatively, proteins could be delivered equally to axonal and dendritic domains but retained on the surface only in the destined domain.
Quantitative analysis revealed that both surface CD4– |
Top
Abstract
Introduction
