The retina is a part of the central nervous system (CNS) that originates from the neural ectoderm during development. The retina develops from an apparently homogeneous collection of cells in a single-layered neuroepithelium. During maturation, it develops into a laminated tissue with five major neuronal cell classes (photoreceptors, bipolar, horizontal, amacrine, and ganglion cells), each of which occupies a characteristic position within the retina. Retinal cell differentiation proceeds in a highly conserved histogenic order. Ganglion cells are generated first, followed in overlapping phases by horizontal cells, cone photoreceptors, amacrine cells, rod photoreceptors, bipolar cells, and finally, Müller cells [1,2,3]. A number of transcription factors or homeodomain proteins are known to be involved in retinal cell fate determination and differentiation [4,5]. During this process, two synaptic layers of the retina are formed and retinal cells are accurately positioned. The inner plexiform layer (IPL) appears as ganglion cell dendrites grow, creating a demarcating border between amacrine cells in the inner nuclear layer (INL) and ganglion cells in the ganglion cell layer (GCL). The outer plexiform layer (OPL) is formed when horizontal cell processes are elaborated laterally, generating a demarcating border between photoreceptors in the outer nuclear layer (ONL) and bipolar cells in the INL [6]. In this way, the formation of the plexiform layer clearly delineates the position of the main retinal cell classes, although the position of the retinal cell classes is primarily thought to be programmed by intrinsic factors [7]. However, within the INL, bipolar cells are located in the distal half, while amacrine cells are located in the proximal half, without any demarcating border between them.

Previously, Kang and Chung [8] reported development of the substance P (SP) (so called neurokinin 1) receptor-immunoreactive amacrine cells in the rat retina. A distinct plexiform-like structure was found within the INL in developing retina. In this study, we demonstrate that this plexiform layer-like structure, the transient intermediate plexiform layer (TIPL) temporarily exists in the middle of the INL, the border between the bipolar and amacrine sublayers in the INL, in the rat retina from postnatal day (PD) 6 to PD 12, during retinal development and further characterize it.

In consistent with a previous report [8], a distinct type of cells showing SP receptor immunoreactivity was observed in the developing rat retina (Fig. 1A, B). This type of labelled cell was located in the middle of the INL. They had horizontal processes emerging from their primary dendrites, and these horizontal processes formed plexuses in the plexiform layer-like structure within the INL. We called this structure the TIPL, because it was temporarily observed during retinal development. The TIPL appeared as a dotted line (Fig. 1C), suggesting that the mature retina is composed of columnar units of cells that are clonally derived [15,16]. They were only observed at postnatal day PD 6 to PD 12, frequently PD 7 to PD 10, as reported [8].

In order to explore the TIPL, we carefully examined vertical sections of developing rat retinas stained with toluidine blue (Fig. 1D, 2). As shown in Fig. 1D, the TIPL was located at the border between the bipolar and amacrine cell zone in PD 7 retinas (arrows), which corresponded to the location at which horizontal processes of one type of SP receptor-immunoreactive cells ramified and formed a plexus (Fig. 1C), and appeared to be simply a patch-shaped extracellular matrix. However, electron microscopic observation clearly demonstrated that this structure was composed of numerous cell processes (Fig. 1E, F), like the IPL and the OPL.

During the retinal development of the mouse and rat, the IPL appears before birth and the OPL is formed at about PD 5. At PD 5, the OPL becomes visible as a thin, pale stained line formed by processes arising from horizontal cells (Fig. 2). The OPL separates the rods into two groups. Rod nuclei situated on the inner side of that layer immediately move across it to enter the ONL, but numerous cells die during retinal migration [17,18]. Thus, the OPL seems to function as a filter or barrier of migration and thus, struggling inner rods seem to be destined to be broken down. Similarly, the TIPL and a distinct type of SP receptor-immunoreactive cells, of which horizontal processes constitute the TIPL, may play a role in dividing the INL into bipolar and amacrine cell sublayers, by filtering the inappropriately localised bipolar and amacrine cells, or by interrupting their axonal growth onto the IPL and the formation of synapses. This idea may be supported by the finding that the TIPL was observed frequently at PD 7 to PD 10, but not at PD 13 (Fig. 2), similar to the time-course of SP receptor-immunoreactive cells (Fig. 2 in Kang and Chung [8]). The period of the presence of this structure coincides with that of the apoptosis of bipolar and amacrine cells in the INL [19].

To test the above working hypothesis, we performed triple-labeling experiments including TUNEL staining, to determine the relationship between the TIPL and apoptotic bipolar and/or amacrine cells. However, some bipolar cells develop and differentiate late, and thus, their markers are not generally expressed at PD 10 [20]. We indirectly tested this using glutamine synthase (GS), a specific marker for Müller glial cells [10]. As shown in Fig. 3A and B, TUNEL staining revealed that numerous apoptotic cells were present in the middle portion of the INL. In particular, apoptotic cells observed at this time were found near the TIPL (Fig. 3C~E). In addition, the majority of TUNEL-positive cells (96/98) did not show GS immunoreactivity (Fig. 3F~H), suggesting that apoptotic cells may be immature neuronal cells, that is, immature bipolar and/or amacrine cells.

Furthermore, we attempted to ablate TIPL or SP receptor-immunoreactive cells selectively using SP-SAP [11,21] to demonstrate structural changes of the INL of the retina after the disruption of the TIPL, expecting that bipolar and amacrine cells would be intermingled in the middle of the IPL without distinct sublayers. Unexpectedly, however, SP receptor-immunoreactive cells and the TIPL remained and could be clearly observed apparently at PD 14. Moreover, processes of some amacrine cells and axons of some recoverin-immunoreactive bipolar cells ramified and terminated in the TIPL (Fig. 4). These results could be interpreted as demonstrating plasticity in development due to the application of toxin to the developing retina.

A previous report by Johansson and Ehinger [22] was related to these unexpected results described above. In their study, the TIPL, or a similar structure, was observed in the middle of the INL in explanted neonatal rat retina after 11 days in vitro. They described it as an ectopic neuropil area. In addition, they found immunoreactivity to synaptophysin (a presynaptic marker) in the area. Taken together, these findings indicate that the TIPL exists in the INL during retinal development.

In summary, we show that the TIPL, constituted of a morphologically distinct type of cells, including an SP receptor-immunoreactive cell type, temporarily exists in the INL during retinal development, and they appear to be associated with neuronal cell removal via apoptosis in the INL. Our results suggest that they contribute to dividing the INL into bipolar and amacrine cell sublayers. Further investigations will be needed to determine whether the TIPL is a general structure appearing during development of all retinas, including human, and to elucidate the role of the cell type that form this structure.