rat retinal dopaminergic neurons: differential maturation of somatodendritic and axonal compartments

11
Rat Retinal Dopaminergic Neurons: Differential Maturation of Somatodendritic and Axonal Compartments PAUL WITKOVSKY, 1,2 * BLANCA ARANGO-GONZALEZ, 3 JOHN W. HAYCOCK, 4 AND KONRAD KOHLER 3 1 Department of Ophthalmology, New York University School of Medicine, New York, New York 10016 2 Department of Physiology and Neuroscience, New York University School of Medicine, New York, New York 10016 3 Experimental Ophthalmology, University Eye Hospital, 72076 Tuebingen, Germany 4 Department of Biochemistry and Molecular Biology, Louisiana State University Health Science Center, New Orleans, Louisiana 70119 ABSTRACT We examined developmental changes in dopaminergic (DA) neurons of rat pups between postnatal (P) days 3 and 21. DA cell bodies and dendrites grew progressively between P3–15. Voltage-sensitive sodium channels were present in axons at P11, but the ring-like DA axon terminals appeared only during the third postnatal week. The density of ring terminals increased markedly between P15 and P21. The vesicular monoamine transporter (VMAT2) was absent before P13 and became concentrated in DA ring terminals after P17. A steady increase in VMAT2-containing rings around AII amacrine cells occurred during the third postnatal week. The presynaptic membrane protein SNAP-25 colocalized with DA terminals, but several other presynaptic proteins tested, including synaptotagmin I, synapsin, bassoon, syntaxin, and synaptogyrin, appeared not to be associated with DA neurons. Our study shows that the somatodendritic compartment of DA neurons matures before the DA axon terminals do. Maturation of DA axons during the third postnatal week corresponds to the period of onset of visual function. J. Comp. Neurol. 481:352–362, 2005. © 2004 Wiley-Liss, Inc. Indexing terms: Na channel; SNAP25; monoamine vesicular transporter Dopaminergic (DA) neurons of the mammalian retina play an important role in the transition from rod- dominated to cone-dominated vision that occurs as light levels increase in the early part of the day (reviewed in Witkovsky, 2004). The anatomical organization of the ret- inal DA system reflects its functional role in effecting global changes in retinal performance. Thus, although the DA perikarya are found at a low density (10 –100 cells/ mm 2 in different vertebrate retinas; reviewed in Wit- kovsky and Schuette, 1991), their processes overlap to form a continuous network that spreads throughout the retina in the most distal sublamina of the inner plexiform layer (Kolb et al., 1990). From this centrally positioned stratum, released dopamine diffuses to neurons and non- neuronal cells distributed throughout the retina (Nguyen- Legros et al., 1999), reaching target receptors by volume conduction (Bjelke et al., 1996). The physiological behavior of the retinal DA system that underlies dopamine release is better understood as a result of recent investigations. Dacey (1990) showed that the mammalian retinal DA neurons are polyaxonal; a related physiological property is that retinal DA cells pro- Grant sponsor: National Institutes of Health/National Eye Institute; Grant number: EY 03570; Grant sponsor: Richard H. Chartrand Eye Research Foundation (to P.W.); Grant number: NS 25134 (to J.W.H.); Grant sponsor: Deutsche Forschungsgemeinschaft; Grant number: GU259/ 12-1 (to K.K.). *Correspondence to: Paul Witkovsky, Dept. Ophthalmology, New York University School of Medicine, 550 First Ave., NY, NY 10016. E-mail: [email protected] Received 7 July 2004; Revised 26 August 2004; Accepted 27 August 2004 DOI 10.1002/cne.20389 Published online in Wiley InterScience (www.interscience.wiley.com). THE JOURNAL OF COMPARATIVE NEUROLOGY 481:352–362 (2005) © 2004 WILEY-LISS, INC.

Upload: uni-tuebingen

Post on 10-Nov-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

Rat Retinal Dopaminergic Neurons:Differential Maturation of

Somatodendritic and AxonalCompartments

PAUL WITKOVSKY,1,2* BLANCA ARANGO-GONZALEZ,3 JOHN W. HAYCOCK,4

AND KONRAD KOHLER3

1Department of Ophthalmology, New York University School of Medicine,New York, New York 10016

2Department of Physiology and Neuroscience, New York University School of Medicine,New York, New York 10016

3Experimental Ophthalmology, University Eye Hospital, 72076 Tuebingen, Germany4Department of Biochemistry and Molecular Biology, Louisiana State University Health

Science Center, New Orleans, Louisiana 70119

ABSTRACTWe examined developmental changes in dopaminergic (DA) neurons of rat pups between

postnatal (P) days 3 and 21. DA cell bodies and dendrites grew progressively between P3–15.Voltage-sensitive sodium channels were present in axons at P11, but the ring-like DA axonterminals appeared only during the third postnatal week. The density of ring terminalsincreased markedly between P15 and P21. The vesicular monoamine transporter (VMAT2)was absent before P13 and became concentrated in DA ring terminals after P17. A steadyincrease in VMAT2-containing rings around AII amacrine cells occurred during the thirdpostnatal week. The presynaptic membrane protein SNAP-25 colocalized with DA terminals,but several other presynaptic proteins tested, including synaptotagmin I, synapsin, bassoon,syntaxin, and synaptogyrin, appeared not to be associated with DA neurons. Our study showsthat the somatodendritic compartment of DA neurons matures before the DA axon terminalsdo. Maturation of DA axons during the third postnatal week corresponds to the period ofonset of visual function. J. Comp. Neurol. 481:352–362, 2005. © 2004 Wiley-Liss, Inc.

Indexing terms: Na channel; SNAP25; monoamine vesicular transporter

Dopaminergic (DA) neurons of the mammalian retinaplay an important role in the transition from rod-dominated to cone-dominated vision that occurs as lightlevels increase in the early part of the day (reviewed inWitkovsky, 2004). The anatomical organization of the ret-inal DA system reflects its functional role in effectingglobal changes in retinal performance. Thus, although theDA perikarya are found at a low density (10–100 cells/mm2 in different vertebrate retinas; reviewed in Wit-kovsky and Schuette, 1991), their processes overlap toform a continuous network that spreads throughout theretina in the most distal sublamina of the inner plexiformlayer (Kolb et al., 1990). From this centrally positionedstratum, released dopamine diffuses to neurons and non-neuronal cells distributed throughout the retina (Nguyen-Legros et al., 1999), reaching target receptors by volumeconduction (Bjelke et al., 1996).

The physiological behavior of the retinal DA systemthat underlies dopamine release is better understood as aresult of recent investigations. Dacey (1990) showed thatthe mammalian retinal DA neurons are polyaxonal; arelated physiological property is that retinal DA cells pro-

Grant sponsor: National Institutes of Health/National Eye Institute;Grant number: EY 03570; Grant sponsor: Richard H. Chartrand EyeResearch Foundation (to P.W.); Grant number: NS 25134 (to J.W.H.);Grant sponsor: Deutsche Forschungsgemeinschaft; Grant number: GU259/12-1 (to K.K.).

*Correspondence to: Paul Witkovsky, Dept. Ophthalmology, New YorkUniversity School of Medicine, 550 First Ave., NY, NY 10016.E-mail: [email protected]

Received 7 July 2004; Revised 26 August 2004; Accepted 27 August 2004DOI 10.1002/cne.20389Published online in Wiley InterScience (www.interscience.wiley.com).

THE JOURNAL OF COMPARATIVE NEUROLOGY 481:352–362 (2005)

© 2004 WILEY-LISS, INC.

duce action potentials (Gustincich et al., 1997) like theircounterparts in the nigrostriatal system (Grace and Bun-ney, 1983). Witkovsky et al. (2004) showed that vesicularmonoamine transporter 2 (VMAT2), the protein responsi-ble for packaging dopamine into vesicles, is located pri-marily in the DA axons and their ringlike terminals. Ul-trastructural studies (Pourcho, 1982; Kolb et al., 1990)indicate that DA axon terminals are devoid of synapticinput, suggesting that it is the spike-firing frequencywhich controls dopamine release. Blocking spike produc-tion by TTX prevents dopamine release (Puopolo et al.,2001) and, by inhibiting phosphorylation of tyrosine hy-droxylase in DA axons and their terminals (Witkovsky etal., 2004), slows the rate of dopamine production (Iuvoneet al., 1982). These diverse data point to the importance ofthe axonal network of DA neurons in regulating the ex-tracellular concentration of dopamine.

In the present study we describe the development of theretinal DA system in rat pups during the first 3 weeks ofpostnatal life. Although many prior studies have exam-ined DA neuronal development in mammalian retinas(Lam et al., 1981; Nguyen-Legros et al., 1983; Parkinsonand Rando, 1984; Wulle and Schnitzer, 1989; Wang et al.,1990; Casini and Brecha, 1992a,b), we utilized a battery ofantibodies that permitted us to distinguish between soma-todendritic and axonal components of the retinal DA sys-tem. Our main conclusion is that DA perikarya and asso-ciated dendrites appear first and achieve adultdimensions within the first 2 postnatal weeks. In contrast,the DA axons and their terminals appear relatively late inpostnatal development and undergo rapid maturationduring the third postnatal week, a period during whichrats open their eyes and begin to function visually.

MATERIALS AND METHODS

Pigmented rats (Brown Norway strain) were obtainedfrom the breeding colony of the University Eye Hospital(Tuebingen, Germany). The animals were maintained un-der a 12-hour light/dark cycle (lights on at 6 AM) withaccess to food and water ad libitum. For developmentalstudies, the day of birth was taken as postnatal day 0 (P0)and retinas were studied between P3 and P21. Animalswere sacrificed between 10 AM and 1 PM by exposing the ratpup to CO2, followed by decapitation. All experimentswere performed in accordance with the ARVO Statementfor the Use of Animals in Ophthalmic and Vision Re-search.

Eyes were enucleated and their anterior portions cutaway. The posterior poles were fixed by a 1-hour immer-sion at room temperature in 4% freshly prepared parafor-maldehyde buffered in 0.1 M phosphate, pH 7.4. Thereaf-ter, the retina was dissected free and washed 3 � 20 minin phosphate-buffered saline (PBS). In most cases thetissue was “defatted” by dehydration in a series of gradedalcohols followed by 30 minutes in propylene oxide, thenin reverse order through the alcohol series to PBS. Thedefatting procedure diminished a haze of fluorescencesometimes observed in whole-mount preparations. There-after, the retina was placed in blocking solution (PBScontaining 0.1% Na azide, 0.3% Triton X-100, and 100mg/10 cc bovine serum albumin).

The primary antibodies used were mouse monoclonalanti-TH (1:500; Chemicon, Temecula CA); rabbit poly-clonal anti-TH (1:500–1:800 Chemicon,); mouse monoclo-

nal anti-sodium channel (pan), (1:500 Sigma St. Louis,MO); mouse monoclonal anti-parvalbumin (1:1,000,Sigma), and rabbit polyclonal VMAT2, 1:3,000 (Haycock etal., 2003). We also tested the following antibodies againstsynaptic proteins: mouse monoclonal anti-synaptophysin1 (1:500), rabbit polyclonal anti-synaptotagmin 1,2(1:1,000), mouse monoclonal anti-synaptogyrin (1:1,000),mouse monoclonal anti-SNAP-25 (1:1,000), rabbit poly-clonal anti-SV2A (1:800), all obtained from Synaptic Sys-tems (Goettingen, Germany). Mouse monoclonal anti-syntaxin (1:500) was obtained from Sigma and mousemonoclonal anti-bassoon (1:250) was purchased fromStressgen Biotechnologies (Victoria, BC, Canada).

For preparation of whole mounts, the retina was cutinto eight pieces; for sections, the retina was cryoprotectedin 30% sucrose, after which 16–18-�-thick frozen sectionswere cut. Invariably the retinal piece or section was placedin a mixture of the appropriate anti-TH and one of theother primary antibodies utilized. After 16–20 hours inprimary antibody solution at room temperature, the tis-sues were washed 3 � 20 minutes in PBS and left for 2hours in secondary antibody solution. The secondary an-tibodies used were Alexa 488 goat antimouse (MolecularProbes, Eugene, OR) and Cy3 goat antirabbit (JacksonImmunoResearch, West Grove, PA). After a final series of3 � 20 minute washes in PBS, the pieces were mountedflat, vitreous side up, in Vecta Shield (Vector Laboratories,Burlingame, CA) and viewed in a Nikon PM800 confocalmicroscope equipped with a digital camera controlled bythe Spot software program. Digital files were processed inAdobe PhotoShop 5.5 (San Jose, CA) and Adobe Illustrator9.0. Auto-Quant software providing 2D deconvolution(Auto-Quant Imaging, Watervliet, NY) was used to im-prove image quality in relation to colocalization of TH andthe presynaptic protein, SNAP-25.

To estimate DA perikaryal area, the retina was viewedin flat-mount with a 40� objective and the cell broughtinto best focus. The outline of the cell body (excludingprocesses) was traced onto a clear acetate sheet. Fiftyrandomly selected DA neurons from each developmentalstage were traced (one retina/timepoint). Subsequently,the cell profiles were cut out, weighed on a microbalance,and the perikaryal areas calculated by reference to theweight of an acetate standard calibrated by an ocularmicrometer.

Specificity of the antibodies was tested by omitting theprimary antibodies and noting that no specific staining ofcells or cell processes occurred. In addition, it was ob-served that VMAT2-immunostaining invariably colocal-ized with anti-TH immunostaining and was seen in noother place.

RESULTS

Growth of DA perikarya and dendritesprecedes that of DA axons

We used TH immunoreactivity (TH-IR) as a marker forDA neurons. In principle, TH-IR is found also in norad-renergic and adrenergic neurons (Cooper et al., 1986), butin the mammalian retina the TH-IR, large cell bodies(perikaryal diameter �15 �m) situated at the border ofinner nuclear and inner plexiform layers are dopaminer-gic (Wulle and Schnitzer, 1989). A second group of small(perikaryal diameter �10 �m) TH-IR cells whose pro-

353DEVELOPMENT OF RAT RETINAL DA NEURONS

cesses arborize in the middle of the inner plexiform layer(Mariani and Hokoc, 1988 ) were observed in our prepa-rations but not studied in detail. Whether or not thesesmaller TH-IR neurons are dopaminergic is still not fullyresolved (Dos Santos and Gardino, 1998). We focused ourattention on the large neurons showing robust TH-IR.

At P3, the first stage of development we examined,perikarya showing TH-IR were an inhomogeneous popu-lation of small neurons whose morphology differed sub-stantially from that associated with more mature DA neu-rons. The TH-IR cell bodies were small and round (4–6�m in diameter), and although some TH-IR neuronslacked processes altogether, others showed relativelylarge, multibranched arbors extending 50–100 �m fromthe cell body (Fig. 1). It is noteworthy that at P3 all TH-IRcells had about the same staining intensity, whereas wecould readily distinguish larger, bright, and smaller, dim,neurons with the TH antibody at P5 and later develop-mental stages. This suggests that at P3 the two types ofTH neuron characterized in adult retinas by Hokoc andMariani (1988) may be intermingled but lack the differ-ence in TH-IR which is apparent at later stages of devel-opment. A similar finding was reported in developing catretina by Wang et al. (1990).

By P5 (Fig. 2a), the brightly immunostained TH-IR neu-rons, which we presume to be DA neurons, typically had athick primary dendrite from which finer dendrites branch,although additional fine dendrites emerged directly from theperikaryon. A similar profile was observed at P7 (Fig. 2b),although by P9 a more adult pattern began to emerge inwhich most DA neurons have lost the thick primary den-drite, instead emitting 4–5 processes from the cell body (Fig.2c). In our whole-mount preparations in which the wholepopulation of DA neurons is immunostained it is hard todefine the degree of dendrite maturation, because not all theprocesses emitted by a given DA neuron could be visualizedover their full extent. In many individual cases, however,processes could be traced for more than 200 �m, which isequivalent to the dimensions noted for DA dendrites in P21rat pups (Nguyen-Legros et al., 1983; this report). The in-creasing extension of growing dendrites noted by PN9 beganto create the overlapping of processes that underlies, in part,

the DA plexus. The contribution to this plexus of growing DAaxons is described below.

Dopaminergic perikarya grow markedly between devel-opmental stages P5 and P15. Figure 3 illustrates theprogressive increase of DA perikaryal areas in a samplepopulation of 50 randomly selected cells from each devel-opmental stage studied. Between P5 and P15, there is aprogressive shift of mean area from 86 � 3 �m2 to 258 �10 �m2. The distribution of perikaryal areas is statisti-cally different (P � 0.01) between adjacent ages testeduntil P15. There is no statistically significant difference (P� 0.05), however, between P15/P17 or P15/P21. The graphillustrates that in addition to a rightwards shift of themain population of cells on the size axis, there is a pro-gressive addition between P15 and P21 of a few very largeDA neurons (area �350 �m2), which constitute 5–10% ofthe DA population.

Characterization of DA axons

When DA neurons are examined at high magnification(Fig. 4) at age P11, two types of DA process are apparent.One, the presumed axon, is characterized by a fine diam-eter of 0.2–0.4 �m and has a smooth appearance, i.e.,lacks varicosities or sudden changes in course. DA axonsare readily distinguished from dendrites, which are 1–3�m thick, have an irregular shape, and often changecourse abruptly, at least when observed close to theperikaryon. Axons maintain a fine diameter and a smoothshape, occasionally interrupted by varicosities throughouttheir course, whereas dendrites tend to narrow with in-creasing distance from the cell body. DA axons sometimesarise directly from the cell body (Fig. 4e) but in other casesbranch from a primary dendrite (Fig. 4d).

Further evidence that the fine processes are axons isprovided by their immunoreactivity to an anti-sodiumchannel (anti-Na) antibody, which has been shown tostain ganglion cell axons in adult retinas (Boiko et al.,2003). In Figure 4, pairs of photos illustrate the same DAneurons reacted with either anti-Na (Fig. 4a–c) or anti-TH(Fig. 4d–f) antibodies. It can be seen that the fine axonsare immunoreactive to the anti-Na antibody (Fig. 4a–c,arrows). Dendrites which give rise to an axon (compare

Fig. 1. TH-IR retinal neurons at postnatal day 3. Three views froma whole-mount preparation showing an inhomogeneous population ofTH-IR neurons. a: Two neurons with approximately round perikarya,prominent primary dendrites, and abundant radiating processes. b: Aneuron with a round perikaryon, but lacking a prominent primary

dendrite. c: TH-IR neurons viewed at the edge of the preparation inapproximately vertical view. Note the prominent primary dendritefrom which multiple processes radiate. Scale bar � 10 �m for allpanels.

354 P. WITKOVSKY ET AL.

Fig. 2. Maturation of TH-IR neurons between postnatal days 5and 15. Views of TH-IR neurons in whole-mount preparations. a: AtP5, most TH-IR neurons have a single stout dendrite from whichradiate finer dendrites. b: At P7, multiple slender dendrites emanatefrom the cell body. There is little overlap of processes from adjacentcells. c: At P9, the DA plexus created by overlapping processes begins

to form. d: At P11, TH-IR perikarya achieve adult dimensions anddendrites extend up to 200 �m from the cell body, creating a rich DAplexus. No axonal rings are present, however. e: At P13, many TH-IRprocesses have a beaded appearance. f: At P15, the first TH-IR ringsbegin to form in the DA plexus. Scale bar in f � 20 �m applies to allpanels.

355DEVELOPMENT OF RAT RETINAL DA NEURONS

Fig. 4a,d) also show anti-Na-IR, whereas other dendritesthat do not emit an axon either lack anti-Na-IR or areweakly immunoreactive. At P11, the DA perikarya showanti-panNa-IR (Fig. 4a–c), presumably because largenumbers of sodium channel proteins are being synthe-sized and transported to the axons. An axon taking alooping course around its perikaryon of origin is illus-trated in Figure 4g.

Both the axonal and dendritic processes of DA neuronsoccupy the same narrow, distalmost sublamina of the in-ner plexiform layer, where together they create a plexus ofincreasing complexity between P11 and P21, the oldestdevelopmental stage we examined. At an early stage (P11)axons manifest occasional varicosities, but these local

swellings in the axons become much more numerous andparticularly notable at P13 (Fig. 2e). By P15 (Fig. 2f), theoverlap of DA processes in the plexus is extensive, but thering-like structures characteristic of the adult eye (seebelow) are still quite sparse.

Developing axons form rings

Between P17 and P21 the axon terminals mature pro-gressively, as noted by the increased density of rings (Fig.5a–c). Each ring is 7–8 �m in diameter, and consists of4–8 puncta �0.5 �m in diameter separated by finer, tu-bular segments (Fig. 5c, inset). It is well established thatVMAT2, a protein responsible for packaging dopamineinto vesicles, is concentrated at dopamine release sites(Hoffman et al., 1998). In the retina, VMAT2 has beenshown to associate with axonal rings of the dopaminergicplexus as well as with individual varicosities along theaxons (Witkovsky et al., 2004). In the present study weutilized VMAT2-IR as a marker of maturation of the axonterminals. VMAT2-IR is barely discernible at P 13 (Fig.6a), consisting of a few, widely spaced particles, some ofwhich are relatively large (0.5 �m in diameter), whereasothers are about half that size. It is noteworthy that atthis same stage of development the DA somatodendriticcompartment has an adult configuration, as assessed byperikaryal dimensions and the number and spatial distri-bution of the dendrites. At P15, VMAT2-immunoreactivityis relatively unchanged (Fig. 6b), but the number ofVMAT2-IR particles increases sharply beginning at P17(Fig. 6c), and they begin to organize into rings. The num-ber of such rings increases progressively at P19 (Fig. 6d)and P21 (Fig. 6e), but the rings are still at a lower densitythan that in the adult eye (Fig. 6f), where they achieve aconcentration of about 3,000/mm2 (data not shown). If oneassumes that, on average, there are about six varicosities/ring, the number of putative dopamine release sites in theadult retina is on the order of 18,000/mm2, without takinginto account dopamine release from the perikarya (Puo-polo et al., 2001) or from the DA processes that ascendtowards the outer plexiform layer (see below).

Maturation of the dopamine to AII amacrinecell synapse

It is well documented that a primary target of the do-paminergic axonal network is the AII amacrine cell, anintegral component of the rod pathway (Bloomfield andDacheux, 2001). Figure 7 illustrates the maturation of theDA neuron to AII amacrine synapse between PN13 andPN21. In these whole-mount views the plane of focus is atthe distalmost sublamina of the inner plexiform layer.VMAT2 profiles are seen in red, and the AII amacrine cellsare immunostained (green) for parvalbumin (Waessle etal., 1993). At P13 (Fig. 7a), although VMAT2 particles arepresent, they are not organized into coherent rings aroundthe AII amacrines. By P15 (Fig. 7b), a clear ring structureis still not evident, although, by comparison with P13, agreater number of VMAT particles is arrayed around theperiphery of the AII amacrine cells. By P17, however (Fig.7c), rings of VMAT varicosities are arranged around theperiphery of the AII neurons, and this is seen even moreclearly at PN 21 (Fig. 7d). The ring of VMAT particles issituated just proximal to the AII cell body, at a pointwhere the AII neuron emits a vertically directed primarydendrite that passes into the inner plexiform layer.

Fig. 3. Growth of DA perikaryal areas between postnal days 5 and21. The perikaryal areas of 50 TH-IR neurons were measured for eachdevelopmental stage from tracings (see Materials and Methods) andtheir distribution plotted. Between P5 and P15, there is a steadyrightward shift along the scale of abscissae. At later developmentalstages, a broadening of the distribution is seen due to the appearanceof a few (�10%) very large TH-IR neurons.

356 P. WITKOVSKY ET AL.

Presynaptic proteins at dopamine releasesites

A large number of specialized proteins are found inrelation to presynaptic release sites and many of thesehave been identified in vertebrate retinas (Koontz and

Hendrickson, 1993; Morgans, 2000; Heidelberger et al.,2003). We first tested a few such proteins on adult ratretinas, looking for one or more that colocalized with theringlike axon terminals of DA neurons. We found thatsynaptotagmin I, syntaxin, synapsin, synaptogyrin, andbassoon did not appear to colocalize with DA neurons,

Fig. 5. Maturation of DA axonal rings between postnatal days17 and 21. a–c: Views of TH-IR neurons and the DA plexus inwholemount. Note the increasing density of DA rings betweenpostnatal days 17 and 21. The inset in (c) shows axonal rings at

higher magnification. Each ring consists of a few granules sepa-rated by smooth tubular segments. Scale bar in b � 20 �m appliesto a– c; 5 �m for the inset in c.

Fig. 4. Presence of Na channels in axons of TH-IR cells. Pairs ofphotos (a/d; b/e; c/f) illustrate the same neuron immunostained withanti-Na antibody (a–c) or with anti-TH (d–f). Anti-Na-IR is visible in theperikarya and is concentrated in slender axonal processes (arrows); thecorresponding processes in d–f are indicated by arrows. Note that den-

drites have much lower anti-Na-IR than axons, unless they give rise toan axon (large process from which axon emerges in (a) Two dendriticprocesses are indicated by asterisks in (e) and (f). g: A TH-IR neuronemits a fine axon (marked by arrowheads) which takes a looping coursein the DA plexus. Scale bar in g � 10 �m for all panels.

357DEVELOPMENT OF RAT RETINAL DA NEURONS

although all of the above showed robust staining through-out the inner plexiform layer (ipl) and, in some cases, theouter plexiform layer (opl) as well. Two proteins that didcolocalize with DA neurons were SNAP25 and synapticvesicle protein 2a. We selected SNAP25 and examined itsappearance in DA neurons during postnatal development.

Figure 8 illustrates the results obtained with SNAP25.At P11, SNAP25-IR was present in both ipl (Fig. 8a) andopl (not illustrated). SNAP25-IR was robust in the proxi-mal portion of the ipl, but was weak or absent in the distalipl where the DA processes arborize. No colocalization ofTH and SNAP25 was evident at P11. By P15 (Fig. 8b),SNAP25-IR was distributed evenly throughout the ipl, butnevertheless there was no apparent colocalization withTH-IR. Considering that the DA axon terminals are stillquite undeveloped at P15, and that VMAT2 is not yetabundant, the absence of colocalization is not surprising.At P19, however, a colocalization of SNAP25 and TH wasobserved (Fig. 8c), consistent with the maturation of theDA axonal terminals by this postnatal age (cf. Fig. 5b), theclear association of VMAT2 with those terminals (cf. Fig.6d), and the maturation of the DA to AII amacrine syn-apse between P17 and P21 (Fig. 7c,d). Nevertheless, atP19 the synapse evidently has not achieved its mature

configuration, because SNAP25 was not found in associa-tion with every ring, as it is in the adult retina (Fig. 8d).

Distally directed processes of DA neurons

In a survey of DA neurons in vertebrate retinas (Wit-kovsky and Schuette, 1991) it was noted that in someretinas the DA cells are interplexiform, whereas in othersthey are amacrine cells only, and in still other retinas theyare a mixture of both types. In the rat retina, DA neuronsare primarily of the interplexiform subtype (Nguyen-Legros et al., 1982). We examined the development of theascending processes, as illustrated in Figure 9. At P13(Fig. 9a,b), ascending processes have moved only a shortdistance from the cell body and are characterized as fineprocesses with a terminal swelling. By P15, the processeshave advanced towards the outer plexiform layer, but lacken passant varicosities. The distal terminal of the processis unbranched and resembles a growth cone (Fig. 9c). AtP17 (Fig. 9d) the distal processes have begun to branchand manifest multiple en passant varicosities. By P21(Fig. 9e), the terminal branching can be quite extensive,consisting of multiple branches, which each bear a fewvaricosities. In the adult eye at least a fraction of suchvaricosities contain VMAT2 (Witkovsky et al., 2004), in-

Fig. 6. Increased presence of VMAT2 between postnatal days 13and 21. Flat-mount views focused on the border between inner nu-clear and inner plexiform layers. a: At P13, VMAT2-IR consists of alow density of particles not organized into larger structures. b: At P15,the density of VMAT2-IR granules has increased and the first ringsbegin to assemble. c: At P17, VMAT2-IR granules show increaseddensity and are more clearly organized into rings compared to P15. d:

At P19, clear rings of VMAT2-IR granules are evident, but theirdensity is low. e: At P21, most VMAT2-IR granules are organized intorings, but the density of such rings is still lower than seen in the adultretina, as illustrated in (f). Perikarya showing VMAT2-IR are indi-cated by arrows in panels c, d, and f. Scale bar in a � 20 �m appliesto all panels.

358 P. WITKOVSKY ET AL.

Fig. 7. VMAT2-IR rings surround AII amacrine cells. Flat-mountviews of VMAT2-IR rings (red) and AII amacrine cells immunostainedwith parvalbumin (green). a: At P13, the axonal rings are not formedand VMAT2-IR is seen in scattered granules. b: At P15, a few imper-

fectly formed rings of VMAT2-IR granules are noted. c: At P17, mostVMAT2-IR granules are associated with rings that surround the AIIamacrines. d: At P21, the rings of VMAT2-IR granules are complete.Scale bar in d � 10 �m applies to all panels.

Fig. 8. Colocalization of SNAP-25 and TH in dopaminergic neu-rons. a,b: Conventional fluorescence micrographs of sectioned retina.a: At P11, SNAP25-IR (red) is prominent in the proximal half of the ipl(full extent of ipl indicated by arrow), but does not colocalize withTH-IR processes (green) which are restricted to the distal border ofthe ipl. b: At P15, although SNAP-25-IR extends throughout the ipl(arrow), there is no colocalization with TH-IR. c: Merged confocal

images from an oblique section. At P19, SNAP25-IR and TH-IR areseen to colocalize. d: Merged confocal images of adult retina, showingcolocalization of SNAP25-IR and TH-IR in dopaminergic axonal rings.Each image was first obtained by averaging 20 confocal scans. Theaveraged images were further enhanced by subtracting computednoise using Auto-Quant software (see Materials and Methods). Scalebar in d � 10 �m; 30 �m for a–c.

dicating that they are sites of dopamine release. Thus, thedevelopment of distally directed processes occurs contem-poraneously with the development of the axonal ring sys-tem. This fact, together with the presence of VMAT2 insome distally directed processes (Witkovsky et al., 2004),indicates that such processes are axonal in nature.

DISCUSSION

Temporal separation of somatodendriticversus axonal maturation

The two main points of this study were, first, to providemorphological and immunocytochemical bases for distin-guishing axons and dendrites of DA neurons in the devel-oping rat retina. Second, applying those criteria, we doc-umented a temporal separation between maturation of thesomatodendritic and axonal compartments. We noted thatdopaminergic perikarya are evident at P5 and undergo asteady increase in size and an elaboration of dendriticprocesses that appears to be complete by P15. These find-ings match those reported for mouse (Wulle andSchnitzer, 1989) and rabbit (Casini and Brecha, 1992a,b)retinas. DA axons are evident at P11 based on criteria ofsize and the presence of TTX-sensitive Na channels, but atthis age there is as yet no evidence of synaptic termina-tions, indicated by rings. Also, two proteins characteristicof dopamine synaptic release sites, SNAP-25 and VMAT2,are absent in DA neurons at age P11. During the thirdpostnatal week axonal terminals undergo maturation, in-cluding the formation and increasing density of rings, andthe increasing presence and organization of VMAT2-containing rings around their primary target, the AII am-acrine cell (Kolb et al., 1990; Hampson et al., 1992; Continiand Raviola, 2003). This period corresponds to the age atwhich rodents open their eyes, and it is associated with anincreased accumulation of dopamine in the retina (Lam etal., 1981; Parkinson and Rando, 1984; Wulle andSchnitzer, 1989).

Presynaptic proteins at DA output synapses

A large number of specialized proteins are associatedwith the presynaptic terminal, of which some are incorpo-rated into the synaptic vesicle, others with bringing the

vesicle to the active zone, and still others with the activezone itself. We used antibodies against proteins in each ofthe above categories in an attempt to characterize thepresynaptic terminal of the DA neurons. Most of the syn-aptic proteins we tested for failed to show colocalizationwith the DA neuron, including the putative calcium sen-sor, synaptotagmin I, which is abundant in vertebrateretinas (Heidelberger et al., 2003). Synaptogyrin and syn-aptophysin, both associated with the synaptic vesicle,were absent in DA neurons, although the synaptic vesicleprotein 2A did colocalize with the DA neuron. SNAP-25 isassociated with the presynaptic membrane, as part of thefusion complex in the active zone, where it associates withother proteins, including syntaxin, to regulate release ofneurotransmitter. Although SNAP-25 colocalized with theDA neuron, syntaxin did not. Our findings indicate thatthe DA neuron release sites are differently organized thanother conventional synapses in the retina and brain. How-ever, we did not concentrate on synapse organization, butrather used one presynaptic protein, SNAP-25, as a meansfor studying the time course of synapse maturation. It isinteresting that synaptophysin, synapsin I, and syntaxinall appear within the first postnatal week of rat retinaldevelopment (Dhingra et al., 1997), and that the majorityof ipl synapses are formed before the end of the secondpostnatal week. These data emphasize that DA axon ter-minal synapses are among the last to be generated in theinner retina.

Possible environmental influences onmaturation of retinal DA neurons

The delayed development of DA axon terminals untilthe third postnatal week, and the associated increase indopamine production and release (Lam et al., 1981), sug-gest that environmental factors play an important role inthe onset of DA function in the retina. In the developingrat retina, the ability of light to increase dopamine syn-thesis and turnover appears only after eye opening, i.e.,during the third postnatal week (Morgan and Kamp,1982). The effect of light on dopaminergic neurons de-pends on maturation of retinal synaptic circuits, but thereis as yet no information on the ontogeny of the light-induced electrical responses of the retinal DA cells. It has

Fig. 9. Maturation of distal dopaminergic processes; all photoswere obtained from whole-mount preparations. a,b: At P13, distallydirected TH-IR processes are just beginning to emerge from theperikarya (arrows in b). Such processes terminate in small varicosi-ties (a). c: A montage from a P15 retina showing a process (arrow-

heads) which terminates in a growth cone (arrow). d: At P17, distalprocesses extend up to 100 � from the perikaryon and contain multi-ple varicosities. e: At P21, some TH-IR neurons give rise to distalarborizations consisting of multiple beaded processes. Scale bar in e �20 �m applies to all panels.

360 P. WITKOVSKY ET AL.

been shown, however, that developing retinal DA neuronspossess TrkB receptors (Cellerino and Kohler, 1997),which respond to brain derived neurotrophic factor(BDNF), by increased process growth (Cellerino et al.,1998). In mice homozygous for a TrkB knockout (Rohrerand Ogilvie, 2003), DA neurons fail to develop completely(or possibly are present but completely lack immunoreac-tivity to tyrosine hydroxylase). Similarly, in cultures ofchick retinal neurons, stimulation of cAMP productionincreases both the number of DA neurons and their pro-cess growth (Guimaraes et al., 2001). These data suggestthat the onset of visual function following eye openingmay induce maturation of DA neurons through activationof second messenger systems triggered by synaptic inputs.

On the other hand, the systems which mediate a re-sponse to dopamine, including postsynaptic dopamine re-ceptors (Koulen, 1999), dopamine-stimulated adenylatecyclase (Parkinson and Rando, 1984), and calcium/calmodulin kinase together with ionotropic glutamate re-ceptors (Xue et al., 2002) are already in place by thesecond postnatal week. This indicates that once light be-gins to regulate dopamine release, the neurons with do-pamine receptors (most or all retinal neurons have dopa-mine receptors; Nguyen-Legros et al., 1999) can respondappropriately.

A goal of our study was to clarify the distinction be-tween DA axons and dendrites, because in spite of Dacey’sstudy (1990) in which clear anatomical distinctions be-tween DA dendrites and axons were drawn, there is atendency in the literature to refer to DA dendrites andaxons as “processes.” The confusion is not merely seman-tic, because the rings, which are the primary sites ofdopamine release, are axonal, not dendritic. The ringslack synaptic inputs (Pourcho, 1982; Kolb et al., 1990;Dacey, 1990), but possess TTX-sensitive Na channels(Witkovsky et al., 2004), suggesting that dopamine releasefrom rings is regulated by the spike rate, rather than bydirect synaptic inputs to those rings. This is not to say thatsynaptic inputs to DA neurons are not important, butrather to point out that those synapses impinge on thesomatodendritic compartment, where presumably theyplay an important role in regulating spike firing rate(Gustincich et al., 1997; Puopolo et al., 2001).

As a cautionary point in relation to synaptic inputs, wenote that retinal DA neurons have a very wide spatialextent and receive sparse synaptic input; there has beenno serial section analysis of the synapses they receive. Theconclusion, therefore, that DA axons receive no synapticinput is provisional. Moreover, although DA dendrites andaxons are concentrated in ipl sublamina 1, some DA pro-cesses extend into more proximal portions of the ipl, some-times as a part of their trajectory before returning tosublamina 1 (Kolb et al., 1991). The possibility that theymight receive synaptic inputs in more proximal regions ofthe IPL has not been examined in detail. As a final point,the functional data available so far derive from exogenousapplication of transmitter-related drugs to DA cells inculture; data on modulation of spiking behavior of DA cellsby neurotransmitter-related substances in an intact mam-malian retina are still lacking.

Another important difference between somatodendriticand axonal DA compartments is in relation to their re-spective calcium channel populations. A pharmacologicalstudy (Tamura et al., 1995) showed that dopamine releasein the rat retina is governed by more than one calcium

channel, and an immunocytochemical study (Xu et al.,2003) found L-type channels in the DA perikarya but Nand P/Q channels in the processes. This distribution isconsistent with the data of Puopolo et al. (2001) thatdopamine release from cultured DA amacrine cells de-pended on a dihydropyridine-sensitive Ca channel. In asuperfused eyecup preparation, blockage of L-type chan-nels reduced total dopamine release by less than 10%,whereas N and P/Q blockers inhibited DA release by morethan 50% (Tamura et al., 1996), indicating differences inthe regulation of DA release from somatodendritic andaxonal compartments, similar to that which has been es-tablished for DA neurons in the nigrostriatal system(Chen and Rice, 2001).

In conclusion, our results show that the axonal compo-nent of DA neurons in a mammalian retina attains theadult configuration only after the eyes open and the ani-mal begins to function visually. In comparison with thesomatodendritic compartment of DA cells, which maturesin advance of eye opening, the delayed development of theaxons suggests that the processing of light information bythe retina may play an important role in effecting axonaldevelopmental changes.

LITERATURE CITED

Bjelke B, Goldstein M, Tinner B, Andersson C, Sesack SR, SteinbuschHWM, Lew YU, He X, Watson S, Tengroth B, Fuxe K. 1996. Dopami-nergic transmission in the rat retina: evidence for volume transmis-sion. J Chem Neuroanat 16:37–50.

Bloomfield SA, Dacheux R. 2001. Rod vision: pathways and processing inthe mammalian retina. Prog Ret Eye Res 20:351–384.

Boiko T, VanWart A, Caldwell HF, Levinson SR, Trimmer JS, Matthews G.2003. Functional specialization of the axon initial segment by isoform-specific sodium channel targeting. J Neurosci 23:2306–2313.

Casini G, Brecha NC. 1992a. Postnatal development of tyrosine hydroxy-lase immunoreactive amacrine cells in the rabbit retina. I. Morpholog-ical characterization. J Comp Neurol 326:283–301.

Casini G, Brecha NC. 1992b. Postnatal development of tyrosine hydroxy-lase immunoreactive amacrine cells in the rabbit retina. II. Quantita-tive analysis. J Comp Neurol 326:302–313.

Cellerino A, Kohler K. 1997. Brain-derived neurotrophic factor/neurotrophin-4 receptor TrkB is localized on ganglion cells and dopa-minergic amacrine cells in the vertebrate retina. J Comp Neurol 386:149–160.

Cellerino A, Pinzon-Duarte G, Carroll P, Kohler K. 1998. Brain-derivedneurotrophic factor modulates the development of the dopaminergicnetwork in the rodent retina. J Neurosci 18:3351–3362.

Chen BT, Rice ME. 2001. Novel Ca2� dependence and time course ofsomatodendritic dopamine release: substantia nigra vs striatum.J Neurosci 21:7841–7847.

Contini M, Raviola E. 2003. GABAergic synapses made by a retinal dopa-minergic neuron. Proc Natl Acad Sci U S A 100:1358–1363.

Cooper JR, Bloom FE, Roth RH. 1986. The biochemical basis of neurophar-macology. New York: Oxford University Press.

Dacey DM. 1988. Dopamine-accumulating amacrine cells revealed by invitro catecholamine-like fluorescence display a unique morphology.Science 240:1196–1198.

Dacey DM. 1990. The dopaminergic amacrine cell. J Comp Neurol 301:461–489.

Dhingra NK, Ramamohan Y, Raju TR. 1997. Developmental expression ofsynaptophysin, synapsin I and syntaxin in the rat retina. Dev BrainRes 102:267–273.

Dos Santos RM, Gardino PF. 1998. Differential distribution of a secondtype of tyrosine hydroxylase immunoreactive amacrine cell in the chickretina. J Neurocytol 27:33–43.

Grace AA, Bunney BS. 1983. Intracellular and extracellular electrophysi-ology of nigral dopaminergic neurons. 1. Identification and character-ization. Neuroscience 10:301–315.

Guimaraes MZP, Hokoc JN, Duvoisin R, Reis RAM, Garcia de Mello F.

361DEVELOPMENT OF RAT RETINAL DA NEURONS

2001. Dopaminergic retinal cell differentiation in culture: modulationby forskolin and dopamine. Eur J Neurosci 13:1931–1937.

Gustincich S, Feigenspan S, Wu KD, Koopman LJ, Raviola E. 1997. Con-trol of dopamine release in the retina: a transgenic approach to neuralnetworks. Neuron 18:723–736.

Hampson ECGM, Vaney DI, Weiler R. 1992. Dopaminergic modulation ofgap junction permeability between amacrine cells in mammalian ret-ina. J Neurosci 12:4911–4922.

Haycock JW, Becker L, Ang L, Furukawa Y, Hornykiewicz O, Kish SJ.2003. Marked disparity between age-related changes in dopamine andother pre-synaptic dopaminergic markers in striatum. J Neurochem87:574–585.

Heidelberger R, Wang MM, Sherry DM. 2003. Differential distribution ofsynaptotagmin immunoreactivity among synapses in the goldfish,salamander, and mouse retina. Vis Neurosci 20:37–49.

Hoffman BJ, Hansson SR, Mezey E, Palkovits M. 1998. Localization anddynamic regulation of biogenic amine transporters in the mammaliancentral nervous system. Front Neuroendocrinol 19:187–231.

Iuvone PM, Rauch AL, Marshburn PB, Glass DB, Neff NH. 1982. Activa-tion of retinal tyrosine hydroxylase in vitro by cyclic AMP-dependentprotein kinase: characterization and comparison to activation in vivoby photic stimulation. J Neurochem 39:1632–1640.

Kolb H, Cuenca N, DeKorver L. 1991. Postembedding immunocytochem-istry for GABA and glycine reveals the synaptic relationships of thedopaminergic amacrine cell of the cat retina. J Comp Neurol 310:267–284.

Koontz MA, Hendrickson AE. 1993. Comparison of immunolocalizationpatterns for the synaptic vesicle proteins p65 and synapsin I in ma-caque monkey retina. Synapse 14:268–282.

Koulen P. 1999. Postnatal development of dopamine D1 receptor immuno-reactivity in the rat retina. J Neurosci Res 56:397–404.

Lam DMK, Fung SC, Kong YC. 1981. Postnatal development of dopami-nergic neurons in the rabbit retina. J Neurosci 1:1117–1132.

Mariani AP, Hokoc JN. 1988. Two types of tyrosine hydroxylase-immunoreactive amacrine cells in the rhesus monkey retina. J CompNeurol 276:81–91.

Morgan WW, Kamp CW. 1982. Postnatal development of the light responseof the dopaminergic neurons in the rat retina. J Neurochem 39:283–285.

Nguyen-Legros J, Berger B, Vigny A, Alvarez C. 1982. Presence of inter-plexiform dopaminergic neurons in the rat retina. Brain Res Bull9:379–381.

Nguyen-Legros J, Vigny A, Gay M. 1983. Post-natal development of TH-like immunoreactivity in the rat retina. Exp Eye Res 37:23–32.

Nguyen-Legros J, Botteri-Versaux C, Vernier P. 1999. Dopamine receptorlocalization in the mammalian retina. Mol Neurobiol 19:181–204.

Parkinson D, Rando RR. 1984. Ontogenesis of dopaminergic neurons in thepost-natal rabbit retina: pre- and post-synaptic elements. Dev BrainRes 13:207–217.

Pourcho RG. 1982. Dopaminergic amacrine cells in the cat retina. BrainRes 252:101–109.

Puopolo M, Hochstetler SE, Gustincich S, Wightman RM, Raviola E. 2001.Extrasynaptic release of dopamine in a retinal neuron: activity depen-dence and transmitter modulation. Neuron 30:211–235.

Rohrer B, Ogilvie JM. 2003. Retarded outer segment development in TrkBknockout mouse retina organ culture. Mol Vis 9:18–23.

Tamura N, Yokotani K, Okuma Y, Okada M, Ueno H, Osumi Y. 1995.Properties of the voltage-gated calcium channels mediating dopamineand acetylcholine release from the isolated rat retina. Brain Res 767:363–370.

Waessle H, Grunert U, Rohrenbeck J. 1993. Immunocytochemical stainingof AII-amacrine cells in the rat retina with antibodies against parval-bumin. J Comp Neurol 322:407–421.

Wang HH, Cuenca N, Kolb H. 1990. Development of morphological typesand distribution patterns of amacrine cells immunoreactive to tyrosinehydroxylase in the cat retina. Vis Neurosci 4:159–175.

Witkovsky P. 2004. Dopamine and retinal function. Doc Ophthal 108:17–40.

Witkovsky P, Schuette M. 1991. The organization of dopaminergic neuronsin vertebrate retinas. Vis Neurosci 7:113–124.

Witkovsky P, Veisenberger E, Haycock JW, Akopian A, Garcia-Espana A,Meller E. 2004. Activity-dependent phosphorylation of tyrosine hydrox-ylase in dopaminergic neurons of the rat retina. J Neurosci 24:4242–4249.

Wulle I, Schnitzer J. 1989. Distribution and morphology of tyrosinehydroxylase-immunoreactive neurons in the developing mouse retina.Dev Brain Res 48:59–72.

Xu HP, Zhao JW, Yang XL. 2003. Cholinergic and dopaminergic amacrinecells differentially express calcium channel subunits in the rat retina.Neuroscience 118:763–768.

Xue J, Li G, Bharucha E, Cooper NGF. 2002. Developmentally regulatedexpression of CAMKII and iGluRs in the rat retina. Dev Brain Res138:61–70.

362 P. WITKOVSKY ET AL.