LINEAGE STUDIES OF CHICK RETINAL HORIZONTAL CELLS AND OPTIC NERVE/RETINAL GLIA
Santiago Rompani (Graduate Student)
We
showed that retinal progenitor cells from rat, chick, and mice are multipotent,
able to produce many different types of neurons throughout development. In
addition, late in development they can produce both neurons and glia, even in a
single division. Similarly, many other structures of the developing central
nervous system including the spinal cord, diencephalon, and cerebral cortex have
been found to contain multipotent progenitor cells. However, the early lineage
studies conducted in these systems did not address whether these multipotent
progenitor cells were biased in their production of particular neuronal
subtypes. This question is relevant since neuronal subtypes are the
physiological functional units from which neural circuits are composed. For
instance, ganglion cells are a single neuronal cell type, as defined by it
receiving input from bipolars and amacrines and projecting to the brain.
However, ganglion cells are composed of at least 12 distinct physiological
subtypes, each with a stereotypical physiological function and morphology.
In
order to determine if progenitor cells are also multipotent to produce all the
neural subtypes, we have labeled single retinal progenitors in the chick with a
replication-incompetent lentivirus expressing membrane-bound GFP (Rompani and Cepko, 2008). The retina was then
allowed to develop in ovo and the neuronal subtype composition of each clone was
quantified. Clones obtained were easily distinguished, as demonstrated by
co-infection with a virus expressing tdtomato (Figure 1).
Figure 1. Chick
retinal clones from E1.5. The lentiviruses, FUmGW expressing mGFP from a UbC
promoter, and FUtdTW, expressing tdTomato from a UbC promoter were used to
infect the E1.5 chick retina. At E
18, the boundaries of two clones marked with FUmGW and one with FUtdTW can be
seen on the whole mount retina.
For the purposes of this
study, we looked at the horizontal cell, which in the chick has only 3
subtypes, the H1, H2, and H3. These subtypes are easily distinguishable by
morphology and are easier to image than the neuronal subtypes with processes in
the other process-rich region of the retina, thus allowing unambiguous
quantification of all the subtypes of this cell type in every clone (Figure 2).
Figure 2. HC
subtypes. (a) H1 cell with a characteristic long axon with exuberant axon
terminal (red arrowhead) and bushy dendrites (blue arrowhead). (b) H2 cell with
sparse, longer processes. (c) H3 cell with short, thin processes terminating in
larger boutons. Scale bars = 10um.
We
found that while large clones had multiple horizontal cell subtypes, smaller
clones with only 2 horizontal cells were overwhelmingly composed of either a
pair of H1 or a pair of H3 cells, while clones with only 1 horizontal cell were
enriched for H2 cells (Figure 3).
Figure 3. Clones
containing 2 HCs. The left panel shows a clone containing 2 H1 HCs (red arrows)
and the right shows a clone containing 2 H3 HCs. (blue arrows). White
arrowheads = photoreceptors. Scale bar = 20um. Asterisk = clonally related
columns.
Statistical analysis of
larger clones showed that H1 and H3 subtypes were commonly observed in even
numbers, while H2's were found in even and odd numbers equally. Furthermore, a
small number of clones were seen composed entirely of H1 cells. The ratio of
all three cell types is uniform across the retina and every clone has
region-independent subtype composition, which suggests regional subtype
variation or viral inactivation is not causing such observations. These data suggest a model whereby some
cycling progenitor cells can give rise to other cycling progenitor cells, as well
as a transient progenitor that is determined to produce a pair of horizontal
cells of either the H1 or the H3 subtype (Figure 4).
Figure 4. A model for the bias of progenitor
cells to produce particular subtypes of HCs. A multipotent progenitor cell produces a determined daughter
cell that is specified to produce a single HC subtype (either H1 or H3) after a
terminal division. Earlier divisions in the depicted clone would lead to
production of progenitor cells that produce other HC subtypes as larger clones
contain all 3 HCs. H2 cells are not produced by a committed progenitor cell.
This is in contrast to our
previous observations, where we did not see the products of these apparently
restricted progenitor cells. Many clones observed previously had several cell
types, even 2 cell clones, where for example, one cell might be a rod and one
might be a bipolar neuron. The previous observations still hold true for the
majority of cell types, and even holds true for the H2 horizontal cell. H2
cells appear as single cells in small clones, rather than as pairs, as is seen
for H1 and H3 cells. We interpret these observations to mean that different
neuronal subtypes might arise from either of the two mechanisms, production by
a mitotic restricted progenitor cell, or production by a multipotent progenitor
cell. Interestingly, the H2 is the scarcest subtype of horizontal cell,
accounting for only about 15% of the horizontal cells in the chick, while the
H1 and H3 cells account for 55% and 30% respectively. Thus, this newly
uncovered mechanism of subtype specification may be a means to amplify the
numbers of some subtypes over others in a way that bypasses the need to affect
cell cycle kinetics. In order to appreciate whether other neuronal subtypes are
produced by restricted progenitor cells, we are examining clones for the
details of the cellŐs morphology to allow an analysis such as conducted for the
horizontal cell subtypes.
ASTROCYTES AND
OLIGODENDROCYTES HAVE A COMMON PROGENITOR CELL
While
much is known about the multipotency of neural progenitors to produce many
distinct types of neurons in many different regions of the developing CNS, it
is currently unclear if these same neural progenitors give rise to both
astrocyte and oligodendrocytes, or are restricted to producing either
astrocytes or oligodendrocytes. While seminal studies in rat optic nerve have
found that progenitors are multipotent in vitro, such findings were not
investigated definitively in vivo. More recent studies have found that in
certain circumstances in vivo neural progenitors can be multipotent. One such
case is in the adult SVZ, where the astrocyte-like adult stem cells give rise
to both neurons and oligodendrocytes, as well as producing more astrocyte-like
adult stem cells (Menn et al. 2006).
Furthermore, glia positive for Ng2, which in vitro give rise to
oligodendrocytes only, in the adult mouse have been found to be capable of
producing oligodendrocytes and a subpopulation of protoplasmic astrocytes (Zhu
et. all 2008). However, a separate study found that Ng2 glia positive for
PDGFRalpha give rise to oligodendrocytes and neurons but not astrocytes in vivo
(Rivers et al. 2008).
We
are determining whether progenitors are multipotent for the two glial cell
types. We are infecting early chick and mouse embryos with replication-incompetent
retroviruses driving a fluorescent or histochemical reporter gene (Figure 1). From an early infection of
the chick neural tube, we labeled progenitors that gave rise to clones of glia
that form "highways" across the mature chick retina. These highways
are overwhelmingly composed of both astrocytes and oligodendrocytes (121/126
clones) and have very clear clonal boundaries, indicating each highway was
derived from a single virally-infected progenitor (Figure 5).
Figure 5.
Retroviruses expressing either mGFP. Tdtomato, or PLAP were coinjected into the
neural tube to label clones deriving from early progenitor cells. Examination
of the tissue later for labeled glial cells was carried out to determine
lineage relatiohships among oligodendrocytes and astocytes in the retina
(chick) or optic nerve (mice and chicks).
The cells in these highways
also express molecular markers consistent with their morphological
identification as astrocytes and oligodendrocytes. Clones in the mouse optic
nerve derived from an ultrasound-guided embryonic injection of neural tube are
composed of either astrocytes and oligodendrocytes or just astrocytes (Figure
6).
Figure 6. The
early neural tube of a chick embryo was infected with two retroviruses, one
encoding GFP and one encoding nlstdtomato. The retina was then examined as a
whole mount from the ganglion cell surface. Highways of glia were observed,
presumed to be clones, as indicated by the outlines on the figure. Most such
clones exhibited cells with the morphologies shown on the right, and were
composed of oligodendrocytes and astrocytes.
Unlike the chick retina,
however, the clonal boundaries in the mouse optic nerve are unclear, so we are
using a virus containing a degenerate oligonucleotide tag such that each viral
particle would contain a unique DNA "barcode" in addition to GFP or
PLAP (see Lineage Analysis: Intro and Methods).
Thus, cells can be processed for the histological or fluorescent reporter and
their specific tag identified by picking the cell, performing PCR for the
conserved portion of the tag, and sequencing the unique tag. Cells that are
derived from a single infected progenitor would thus have the same tag, while
cells derived from different progenitors would have different tags.
Figure 7. High
magnification view of cells in the optic nerve of the mouse following infection
of an embryonic mouse neural tube at the optic vesicle stage.