Unbiased analysis of bulk axonal segregation patterns

Abstract

The projection of retinal ganglion cell axons to the dorsal lateral geniculate nucleus of the thalamus (dLGN) is organized into eye-specific layers, which are macroscopic structures that reflect the bulk organization of thousands of axons. The processes that underlie the formation of these layers is the focus of research in several laboratories. The recent advent of fluorescently tagged tracers allows for the simultaneous visualization of axons from both eyes in the same dLGN section and therefore the analysis of axonal segregation patterns. However, the techniques traditionally used to quantify eye-specific segregation are far from standardized. Here we present an analysis method that objectively quantifies the extent of segregation. We apply this analyses to dLGN images from mice with normal retinogeniculate projection patterns and genetically altered mice with dramatically altered projection patterns. In addition, we compare dLGN images acquired at different optical resolutions to measure the spatial scale over which we can determine segregation unambiguously.

Introduction

In binocular animals, retinal ganglion cell (RGC) axons from the two eyes project to both the right and left lateral geniculate nucleus of the thalamus (dLGN) and segregate into distinct bands called eye-specific layers. Retinogeniculate projections can be made visible by intraocular injection of fluorescently labeled tracers that fill axons in bulk and concentrate in synaptic terminals. Fluorescence images of the contralateral and ipsilateral retinogeniculate termination sites (which we will call the contralateral and ipsilateral projections) have been acquired by injecting a different fluorophore into each eye. Ideally, eye-specific segregation could then be quantified by computing the fraction of the total area of the dLGN that contains fluorescence in either the ipsilateral or contralateral channels but not both.

Eye-specific layers emerge during development from an initially overlapping distribution of ipsilateral and contralateral projecting RGC axons (Godement et al., 1984, Sretavan and Shatz, 1986). Several labs have used transgenic mice to study the role that neural activity and other signaling processes play in this segregation process (Godement et al., 1984, Feldheim et al., 1998, Upton et al., 1999, Huh et al., 2000, Lyckman et al., 2001, Pham et al., 2001, Rossi et al., 2001, Muir-Robinson et al., 2002, Ravary et al., 2003). Despite a small ipsilateral projection (roughly 5% of RGC axons project ipsilaterally), mouse retinogeniculate projections segregate into eye-specific regions. By analogy to ferrets, cats, and monkeys, we will refer to these regions in mice as “layers” even though they are not layer-like in appearance.

There are two factors which make the analysis of eye-specific segregation in mice more difficult than in these other species. First, there exists no cellular lamination of mouse LGN neurons to clearly delineate eye-specific regions (e.g. in ferrets, see Hutchins and Casagrande, 1990). Second, the contralateral retinogeniculate axons that branch from the optic tract and terminate in the medial part of the dLGN traverse the region where ipsilateral axons terminate (Fig. 1A, top row). Hence, the ipsilateral layer also contains contralateral axons. Given these two factors, quantification of segregation depends critically on the choice of a “threshold” below which fluorescence intensities are considered “background” and above which they are considered “signal.” If high thresholds are chosen, there is extremely strong eye-specific segregation (Fig. 1A, second row). If low thresholds are chosen, there is weak segregation (Fig. 1A, bottom). Hence, comparing the extent of eye-specific segregation across mice requires either a consistent method for defining a threshold that distinguishes the fluorescence in terminals from that measured in axons or a threshold-independent quantification.

Here we first describe the pattern of mouse retinogeniculate projections and the limitations of using a threshold-dependent method for quantifying eye-specific segregation as a single scalar value. Second, we present a general representation of the fluorescence intensities of retinal projections to the dLGN. By comparing images acquired with conventional light microscopy to those acquired with a confocal microscope, we estimate the spatial scale over which we can measure segregation. Third, we briefly describe a method we previously published that represents segregation not as single quantity, but rather as a function of different thresholds for the contralateral image (Muir-Robinson et al., 2002). Last, we describe a novel, threshold-independent method for analyzing eye-specific segregation. We demonstrate the utility of this new method by comparing results for WT mice and mutant mice that have significantly altered retinogeniculate projections.