By microinjecting rhodamine-conjugated porcine tubulin into pea epider
mis we recently showed how cortical microtubules reorientate from tran
sverse to longitudinal in living cells (Yuan et nl,, 1994, Proc. Nat.
Acad. Sci, USA, 91, 6050-6053). In the present paper we compare this r
eorientation with the contrary longitudinal to transverse realignment
induced by adding gibberellic acid to preinjected cells on the microsc
ope slide, Both kinds of reorientation are initiated by the appearance
of 'discordant' microtubules which do not share the existing alignmen
t but anticipate the new direction. These increase in number as the ex
isting microtubules depolymerize, one alignment apparently replacing t
he other in a continuous process, By rotating stacks of confocal secti
ons by computer methods we have previously shown that microtubules at
the outer tangential cell wall do not necessarily have the same orient
ation as microtubules at the adjoining anticlinal walls of the same ce
ll (Yuan et al., 1995. Plant J. 7, 17-23), This suggests that microtub
ule reorientations epidermal cells occur mainly (or, at least, first)
at the outer wall, indicating that the array may not reorientate as a
whole, Collectively, these data emphasize the discontinuous nature of
the realignment process, the importance of new microtubule polymerizat
ion, and the special property of the outer epidermal surface as a sens
itive domain.