The number and calibre of axons in transgenic optic nerves
I found th at optic nerves from the bcl-2 transgenic m ouse line 73 (M artinou et al., 1994) contained 80% m ore axons th an w ild-type nerves. This w as the result of a decrease in the norm al p o stn atal loss of RGC axons: in w ild-type optic nerves 60% of the axons are lost betw een b irth and P14, w hereas only 35% are lost in the transgenic nerves.
The expression of Bcl-2 varies in different neuronal pop u latio n s in this m ouse an d w h ereas som e n e u ro n a l p o p u la tio n s are in creased com pared to w ild-type anim als, others are not. All RGCs p resen t in the transgenic retina expressed h u m an bcl-2 transgene ab o u t equally, as assessed by im m unofluoresence staining of new born retinal sections (not shown). Bcl-2 expression did not therefore guarantee survival, since not all Bcl-2+ RGCs present at birth survived to adulthood. W hat determ ines how m any RGC neurons survive in this m ouse? One factor m ay be the size of the target field. M ouse RGC axons project to the lateral geniculate nucleus (LGN) via the superior colliculus (SC, L inden an d Perry, 1983). D u rin g n o rm al m ouse d ev elo p m en t ab o u t 30% of th e d o rsa l LGN n eu ro n s die (H eum ann an d Rabinow icz, 1980). Since h u m a n Bcl-2 is expressed in LGN neurons in line 73 mice (M artinou et al., 1994), there m ay be a greater num ber of LGN neurons in the transgenic m ouse due to a suppression of norm al cell deaths, and the sam e m ay be true of the SC. A greater num ber of target neurons could supply a greater am ount of trophic survival factor than in w ild-type anim als and thereby su p p o rt a greater num ber of RGCs. The size of the LGN and SC in this transgenic m ouse has yet to be determ ined.
I found th at there w as no difference in the calibre of in d iv id u al axons in transgenic and w ild-type optic nerves. The size of the superior colliculus could also be im portant in this context. It has been show n in the peripheral nervous system th at an increase or decrease in target size will cause a c o rre sp o n d in g change in the calibre of in n e rv a tin g axons (Voyvodic, 1989). My results w ould predict a proportional increase in the size of the target field in the transgenic m ouse if the sam e rules apply in the CNS.
Oligodendrocytes
The m ajority of axons in the bcl-2 tran sg en ic optic n erv es are m yelinated (Table 2.1), suggesting th at the oligodendrocyte population has eith er in creased in absolute n u m b ers or each o lig o d e n d ro cy te h ad increased the num ber of axons it m yelinated. I find th at the form er is the case (Table 2.2). Using intracellular dye injection and confocal microscopy, it has been show n th at m ature oligodendrocytes have 15-20 longitudinal processes (Butt et al., 1994). The spatial distribution of axons m yelinated by an oligodendrocyte is tightly constrained to a radial field w ith a diam eter of approxim ately 30 pm. It w ould be interesting to establish w hether these restrictions ap p ly w hen oligodendrocytes are p resen ted w ith a greater num ber of axons than norm al as in the transgenic m ouse, b u t this has not been determ ined.
It has been show n in the developing rat optic nerve th at axons are required for oligodendrocyte survival (Barres et al., 1993b), and th at 50% of the oligodendrocytes produced in the rat optic nerve norm ally die w ithin 2-3 days after they develop (Barres et al., 1992). It w as suggested th at this m assive death reflects a com petition for axon-dependent survival signals (Barres et al., 1992). Forcing newly-form ed oligodendrocytes to com pete for
lim iting am ounts of axon-dependent signals m ay help m atch the final num ber of oligodendrocytes to the num ber of axons requiring m yelination (Barres et al., 1992). A prediction of this hypothesis is th at an increase in axon num bers should decrease oligodendrocyte death.
My findings are consistent w ith this prediction. The n u m b er of oligodendrocytes in the transgenic optic nerve is tw ice th at in the w ild- type optic nerve, and m uch of the increase seems to result from a decrease in norm al oligodendrocyte death. D uring the first three p o stn atal weeks, w hen oligodendrocyte death w ould be expected to be m axim al (Barres et al., 1992), the p ro p o rtio n of d ead cells in tran sg en ic optic n erv es is significantly less than in nontransgenic nerves. It w as show n th at 90% of the dead cells in the postnatal rat optic nerve are oligodendrocytes and 10% are oligodendrocyte precursor cells (Barres et al., 1993b). A lthough I have not directly identified the dead cells in the m ouse optic nerve, it seem s likely that m ost of them belong to the oligodendrocyte lineage, as none of them express the astrocyte m arker GFAP.
T hree lines of evidence su g g est th a t the d ecrease in n o rm al oligodendrocyte death that occurs in the transgenic optic nerve is m ainly secondary to the increase in axons, astrocytes, or both, rather th an directly related to the expression of the bcl-2 transgene in oligodendrocytes. First, transection of the trangenic nerve at PIS leads to increased glial cell death, w hich is proportionally even greater than th at seen follow ing transection of w ild-type nerves. As none of the dead cells in the cut nerve are GFAP+, and, in the rat, the increased cell d eath in tran sected optic n erves is confined to oligodendrocytes (Barres et al., 1993b), it seem s likely th at the increased cell d e ath in the c u t m ouse optic n erv e is also m ain ly oligodendrocyte death. Thus the expression of the bcl-2 tran sg en e in oligodendrocytes seem s n o t to p ro tect them from PCD w h en th ey are
d ep riv ed of axonal signals in vivo. Second, w h en d ep riv ed of survival signals in vitro, transgenic oligodendrocytes do n o t survive b etter th an w ild-type oligodendrocytes. Third, a different bcl-2 transgenic m ouse line (line 71), in w hich the transgene is not expressed in RGCs (M artinou et al., 1994) b u t is expressed in 30% of the oligodendrocytes an d 15% of the astrocytes in the optic nerve, has a norm al n u m ber of optic nerve glial cells (assessed at various ages) and a norm al num ber of dead cells (assessed at P14). M oreover, the p ro p o rtio n of oligodendrocytes exp ressin g the transgene in this line does n o t increase from P8 to P I 4, suggesting th at expression of the transgene does n o t confer a selective survival advantage on these olig o d en d ro cy tes. As th e a m o u n t of h u m a n Bcl-2 p ro te in expressed in individual glial cells is at least as great in line 71 as in line 73, it seem s u n lik e ly th a t th e ex p ressio n of th e b cl-2 tra n s g e n e in oligodendrocytes in line 73 protects these cells from norm al cell death.
The assessm ent of cell proliferation of oligodendrocyte p recu rso r cells in the developing optic nerves is com plicated by the fact th at unlike cell death, w here only oligodendrocyte lineage cells are involved bo th astrocyte and oligodendrocyte lineage cells proliferate in the developing nerve. An investigation of the p roportion of astrocytes th at incorporated BrdU at P5 revealed that m ore w ere in S-phase in transgenic nerves at this age w hen com pared to w ild-type (Table 3.1). This m eans th at d u rin g the first postnatal w eek m ore astrocytes were being generated in the transgenic nerves. As m ost of the increase in astrocyte n u m b ers in ro d e n t optic nerves occurs earlier than the increase in oligodendrocyte num bers (Barres et al., 1992; Skoff, 1990), m any of the BrdU-labelled cells at later ages w ould be expected to be oligodendrocyte p recursor cells. If one considers the proportion of all cells that incorporated BrdU in transgenic and w ild-type nerves at P8 and P14 no differences w ere revealed (Figure 2.4b). The
implication of this observation is that, unlike for astrocytes at P5, there was no increase in the p roportion of oligodendrocyte lineage cells dividing at any of the times examined.
A xons n o rm a lly p ro m o te th e p ro life ra tio n a n d su rv iv a l of oligodendrocyte precursor cells in the developing optic nerve (Barres and Raff, 1993), b u t m y findings suggest th at the rate at w hich these cells proliferate is not increased by the presence of a greater n u m b er of axons. However, there are greater num bers of oligodendrocyte progenitor cells in ad u lt transgenic optic nerves than in a d u lt w ild-type nerves (Table 2.2), raising the possibility th at the norm al death of oligodendrocyte precursor cells (Barres et al., 1992) is decreased in the tran sg en ic optic nerve. Consistent w ith this possibility, it w as previously show n th at transection of the neonatal rat optic nerve results in a decrease in the nu m b ers of oligodendrocyte and their precursors (David et al., 1984), w ith o u t a change in the proportions of the cells that incorporated tritiated thym idine. These fin d in g s a n d m ine su g g e st th a t axons p ro m o te th e s u rv iv a l of oligodendrocyte precursor cells in the developing optic nerve.
Microglia
Microglia also seem to be increased about 2-fold in the transgenic com pared to w ild-type optic nerves. The m echanism (s) u n d erly in g this apparent increase is unknow n as little is know n about how the num ber of m icroglial cells in the nerve is norm ally controlled. These cells m igrate into the developing nerve via the blood vessels and pia-arachnoid, w hich are presum ably increased in the transgenic nerves; this increase by itself m ight be enough to account for the increase in m icroglial cells. As c u ltu red astrocytes p ro d u ce factors th a t are m itogenic for c u ltu re d m icroglial cells (Shafit-Zagardo et al., 1993; Lee et al., 1994), it is also
possible th at the increase in m icroglia is secondary to the increase in astrocytes.