assembly (Fig. 3A), and we monitored the re-
action by using total internal reflection fluores-
cence (TIRF) microscopy.With either guanosine
triphosphate or guanosine-5′-[(a,b)-methyleno]
triphosphate (GMPCPP) in solution, we observed
the usual elongation of template A-microtubules
at both tips. However, we observed nucleation
and elongation of patches of fluorescent rho-
damine signal on the A-microtubule only in
the presence of GMPCPP (movies S2 and S3),
thesameconditionusedinourcryo-EMexperi-
ments. We thus interpreted these patches as
B-microtubules (Fig. 3B and fig. S4A). This
result suggests that MTD formation requires
a certain level of stabilization, mediated in
our experiments by GMPCPP and in vivo pos-
sibly by the presence of MIPs. Investigating the
growth rates of A- and B-microtubule tips showedthat, unlike the plus and minus tips of the
A-microtubules, which are known to grow
at different rates ( 17 ), B-microtubules grow
atthesamerateinbothdirections(Fig.3,
C and D, and fig. S4, A to D). The isotropic
B-microtubule growth rate was faster than the
growth rate of the plus tip of the A-microtubule
(Fig. 3D), and this rate correlated with increas-
ing tubulin concentration (Fig. 3D and fig. S4D).
Thus, B-microtubules are dynamic structures
nucleating on the lattice of the A-microtubule
and elongating in both directions without ap-
parent anisotropy. To estimate the protofilament
number in B-microtubules, we compared the
rhodamine fluorescent signal in the center of
the B-microtubules with the steady-state rho-
damine fluorescent signal at the tips of the
template A-microtubules, which are formed
by 14 protofilaments because of the presence
of GMPCPP ( 18 ). This quantification suggested
that after 40 min, B-microtubules were assembled,
with on average (±SD) 5.7 ± 2.6 and 13.8 ± 3.8
protofilaments at the free tubulin concentrations
of 1 and 2mM, respectively (fig. S4E and movies
S2 and S4). Finally, we repeated the experiment
with A-microtubules treated with decreasing
subtilisin:tubulin ratios (1:1, 1:50; 1:100, and
1:1000), leading to predominantb-tubulin
C-terminal tail removal (fig. S5) ( 13 ). We found
that MTD nucleation decreased markedly (fig. S5
and movie S5), suggesting that the removal of
C-terminal tails from botha- andb-tubulin is
necessary for the MTD nucleation.
In vivo, MTDs are composed of heterodimers
ofa-andb-tubulin and dozens different MIPs.
Our work establishes that the C-terminal tail
of tubulin exhibits an inhibitory effect that,
in vivo, may prevent uncontrolled MTD forma-
tion. Molecular simulations suggested that the
C-terminal tails of one specific protofilament
hinder the attachment of protofilament B1 at
theinternalsideoftheOJ.Weproposethat
in vivo, specific MIPs bind and displace the
C-terminal tails of A11 and allow for the for-
mation of a B-microtubule that elongates bi-
directionally (Fig. 3E). Moreover, such proteins
may be needed to precisely position the MTD
branching to a specific protofilament on the
A-microtubule, as well as to stabilize the entire
MTD. The requirement for such protein is al-
leviated in our in vitro minimal system by pro-
viding GMPCPP.
In summary, our work highlights the crucial
role of tubulin C-terminal tails in regulating
MTDs, which are key to the assembly and func-
tion of centrioles, cilia, and flagella.REFERENCES AND NOTES- J. J. Malicki, C. A. Johnson,Trends Cell Biol. 27 , 126– 140
(2017). - S. C. Goetz, K. V. Anderson,Nat. Rev. Genet. 11 , 331– 344
(2010). - L. Stepanek, G. Pigino,Science 352 , 721–724 (2016).
- J. L. Rosenbaum, G. B. Witman,Nat. Rev. Mol. Cell Biol. 3 ,
813 – 825 (2002). - D. Nicastroet al.,Science 313 , 944–948 (2006).
- R. V. Dippell,Proc. Natl. Acad. Sci. U.S.A. 61 , 461–468 (1968).
- I. V. Nechipurenko, C. Berciu, P. Sengupta, D. Nicastro,eLife 6 ,
e25686 (2017).
Schmidt-Cernohorskaet al.,Science 363 , 285–288 (2019) 18 January 2019 3of4
050100150200250
ns
ABC DE SeamMT MTD initiation MTD elongationA10B1
A11Fig. 3. Dynamics of MTD assembly.(A) Protocol to visualize MTD assembly by using TIRF
microscopy. (B) Montage showing MTDs. Subtilisin-treated A-microtubules are in green;
A-microtubule tips and B-microtubules formed by rhodamine-labeled tubulin are depicted in
magenta. White arrowheads indicate the tip elongation of the A-microtubule. Yellow arrowheads point
to the B-microtubule assembling on the surface of the A-microtubule. Scale bar, 1mm. (C) Montage
showing MTDs and the corresponding multichannel kymograph. Scale bars: horizontal, 5mm;
vertical, 15 min. (D) Polymerization rate of B-microtubules at 2mM free tubulin. ns, not significant,
*P< 0.0001, determined by the Mann-Whitney test. Plus and minus tips of A-microtubules
polymerize at 53.93 ± 11.56 nm/min (n= 25 tips) and 37.13 ± 13.13 nm/min (n= 22 tips),
respectively. The polymerization rates of B-microtubules toward plus and minus tips of the
A-microtubules are 83.51 ± 30.44 nm/min (n= 53 tips) and 96.52 ± 45.87 nm/min (n= 52 tips)
[values are averages (represented by red lines) ± SD]. (E) Model of MTD formation. In vitro, MTD
assembly initiates on the surface of an A-microtubule (green) deprived of tubulin C-termini
(red tails) by the addition of a protofilament owing to a noncanonical surface-to-side tubulin
interaction. Protofilaments in the B-microtubule (purple) continue to assemble to ultimately
lead to a near-complete MTD.
RESEARCH | REPORT
on January 17, 2019^http://science.sciencemag.org/Downloaded from