9
The most powerful measure of a microscopy platform to resolve intracellular details is the ability
196
to discriminate between very proximal structures. For this reason, it was attempted to quantify
197
the composition of cortical microtubule bundles, taking advantage of the extensive bundling
198
observed in hypocotyl epidermal cells of the mpk4 mutant. The cortical microtubule bundle
199
composition was approached in both wild type and mpk4 hypocotyl epidermal cells labeled with
200
GFP-MBD microtubule marker, by two alternative ways: intensity profiling and determination of
201
the Rayleigh criterion.
202
First, bundles were quantified by means of additive fluorescence intensity. Ideally, this means
203
that the maximum fluorescence intensity of a given bundle would increase linearly upon the
204
successive addition of microtubules. However, it might be expected that according to the
205
specifications of the microscopy used, fluorescence intensities may become saturated after a
206
certain point making it impossible to further quantify microtubule bundles. Thus the second
207
criterion in this approach of quantitation of microtubule bundles was to determine the maximum
208
number of microtubules per bundle before fluorescence intensity was saturated.
209
By means of absolute fluorescence intensity, bundles visualized by SIM (Figs. 2A to C; Fig.
210
S4A; Figs. 2J, K), were resolvable with very good linear correlation (Figs. 2B, C; n=119, 99, 33,
211
and 26 measurements for one, two, three, and four microtubules, respectively; all p values
212
comparing absolute fluorescence intensity between 1 and 2, 2 and 3, and 3 and 4 microtubules
213
were <0.001). By contrast the linear correlation between microtubule numbers and bundle
214
fluorescence intensity was inferior with WF (Figs. 2D to F; Fig. S4B; Fig. 2K) unable to
215
discriminate between 3 and 4 microtubules (Figs. 2E, F; n=119, 99, 33, and 26 measurements for
216
one, two, three, and four microtubules, respectively; p=0.158 between 3 and 4 microtubules), and
217
such correlations deteriorated even further with CLSM showing saturation after 2 microtubules
218
(Figs. 2G to I; Fig. S4C; Fig. 2L, n=119, 99, 33, and 26 measurements for one, two, three, and
219
four microtubules, respectively; p=0.057 between 2 and 3 microtubules and p=0.051 between 3
220
and 4 microtubules). Given the broad variability and the frequent saturation of fluorescence
221
intensity of microtubule bundles observed by WF and CLSM, profiles of such bundles were
222
generated only in cells that were previously observed with SIM.
223
www.plantphysiol.org
on July 21, 2018 - Published by
Downloaded from
Copyright © 2014 American Society of Plant Biologists. All rights reserved.
10
In fluorescence microscopy, the application of the Rayleigh criterion deems two proximal
224
structures separable when the minimum distance between them (peak-to-peak separation) allows
225
fluorescence intensity drop of ca. 25% of the maximum intensity of either one, provided that the
226
two maximum intensities are nearly equal. In principle, at fluorescence intensity profiles of SIM
227
showing pronounced peak separation (Fig. 3A, D; Fig. S5A) proximal microtubules were
228
indiscriminable by WF (Figs. 3B, E; Fig. S5B) and CLSM (Figs. 3C, F; Fig. S5C). The
229
application of the Rayleigh criterion proved to be particularly cumbersome, especially in the case
230
of WF and CLSM as it was difficult to locate proximal microtubules with nearly equal maximum
231
fluorescence intensities. By applying the Rayleigh criterion comparatively, SIM showed the best
232
discrimination capacity being able to deem adjacent microtubules as separate ones at 131
±8 nm
233
(mean
±SD; n=32; Figs. 3G, J, M; Fig. S5D), while this distance was 236±17 nm for WF
234
(mean
±SD; n=32; Figs. 3H, K, M; Fig. S5E) and 243±16 nm for CLSM (mean±SD; n=32; Figs.
235
3I, L, M; Fig. S5F).
236
Individual microtubule dynamics
237
Although SIM provides greatly improved resolution with respect
to commonly used descriptive
238
microscopies such as CLSM and WF there are other superresolution approaches including
239
PALM/STORM and STED which can apparently resolve subcellular structures at nearly
240
molecular level. However, PALM/STORM reconstructs superresolution images from raw
241
acquisitions in a time scale of minutes (Cox and Jones, 2013). STED, although much faster, has
242
limitations to time-lapsed imaging of very small fields of view (Westphal et al., 2008), while
243
additionally being phototoxic. Thus the main challenge for SIM would be to evaluate its
244
potential for time-lapsed imaging. It is noteworthy that only few studies of this kind (Kner et al.,
245
2009, Shao et al., 2011, Fiolka et al., 2012) have been published during the 14 year period from
246
first introduction of SIM (Gustaffsson, 2000). Although SIM has been previously demonstrated
247
in plants (e.g., Fitzgibbon et al., 2010, 2013; Linnik et al., 2013; Liesche et al., 2013) its capacity
248
for time-lapsed recordings necessary for quantitative dynamic plant
cell studies was never shown
249
before.
250
To test the efficiency of SIM for time-lapsed imaging, acquisition settings were adjusted to
251
achieve the best possible compromise between spatial and temporal resolution
for SIM resulting
252
www.plantphysiol.org
on July 21, 2018 - Published by
Downloaded from
Copyright © 2014 American Society of Plant Biologists. All rights reserved.