no longer had NUMA at the spindle poles, and
the poles became fully defocused (Fig. 1, G to
I). Thus, NUMA is required for pole focusing
in human oocytes.
aMTOC-free mouse oocytes assemble spindles
that mimic the morphology and organization of
spindle poles in human oocytes
Next, we investigated NUMA localization in
bovine and porcine oocytes, which, like human
oocytes, naturally lack aMTOCs. NUMA was
similarly enriched at the microtubule minus
ends in these nonrodent mammalian oocytes
(fig. S1, B to E). By contrast, NUMA was hardly
enriched at the spindle poles in mouse oocytes
but weakly stained aMTOCs (fig. S1, F to I). To
confirm that NUMA also marks microtubule
minus ends in mouse oocytes, we tracked indi-
vidual microtubules after depleting specific
subpopulations of microtubules to facilitate
analysis of the otherwise densely packed spin-
dles. Kinetochore and interpolar microtubules
account for most microtubules within the spin-
dle. To enrich for kinetochore and interpolar
microtubules, we performed cold treatment
to selectively depolymerize non–kinetochore-
bound microtubules ( 57 )andacuteNDC80/
HEC1 depletion to depolymerize kinetochore-
bound microtubules ( 58 ). Both kinetochore
and interpolar microtubules ended at NUMA
and aMTOCs (fig. S2 and movies S2 and
S3), consistent with the model that NUMA
anchorsmicrotubulestoaMTOCsinmouse
oocytes ( 33 ).
Immature human oocytes are not available
in large quantities, which makes it challenging
to gain mechanistic insights into the organi-
zation of spindle poles and the causes of spin-
dle instability during meiosis I. Mouse oocytes
with spindles that more closely mimic the mor-
phology and organization of spindle poles in
human would be a useful model of human
oocytes because mouse oocytes are readily
available, progress through meiosis synchro-
nously, and can be manipulated with genetic
tools such as follicle RNAi ( 55 ). One main dif-
ference between mouse oocytes and other
mammalian oocytes, including human, is the
presence of aMTOCs ( 2 ). To ablate aMTOCs in
immature mouse oocytes, we used Trim-Away
to deplete PCNT (pericentrin), an essential
scaffolding component of aMTOCs ( 59 , 60 )
(fig. S3, A to D). In PCNT-depleted mouse
oocytes, aMTOC components such as CEP192
(centrosomal protein of 192 kDa), CDK5RAP2
(CDK5 regulatory subunit-associated protein 2),
andg-tubulin were dispersed in the cytoplasm
and no longer formed distinct foci at the spin-
dle poles (fig. S3D), consistent with the success-
ful ablation of aMTOCs (for simplicity, hereafter
referred to as aMTOC-free mouse oocytes).
We used complementary approaches to con-
firm that the spindles assembled in aMTOC-
free mouse oocytes mimic the spindles in
human oocytes. Spindle assembly in human
oocytes is mediated by chromosomes and the
small guanosine triphosphatase Ran ( 2 ). In this
pathway, a chromosome-centered gradient of
Ran–guanosine triphosphate (GTP) promotes
microtubule nucleation in the proximity of
chromosomes by locally releasing spindle
assembly factors from inhibitory binding to
importins ( 61 ). We found that microtubule
nucleation occurred around chromosomes in
aMTOC-free mouse oocytes (fig. S3E), similar
to human oocytes ( 2 ). Dominant negative inhi-
bition with RanT24N prevented microtubule
nucleation and spindle assembly in aMTOC-
free mouse oocytes (fig. S3, F and G). This is in
contrast to wild-type mouse oocytes, in which
RanT24N did not prevent microtubule nucle-
ation and spindle assembly (fig. S3, H and I).
Thus, spindle assembly in aMTOC-free mouse
oocytes, as in human oocytes, is dependent on
the Ran pathway. Furthermore, the localization
patterns of several minus end–binding proteins
and spindle pole–associated proteins were
similar in human oocytes and in aMTOC-free
mouse oocytes (figs. S1A and S3J). In particu-
lar, NUMA was highly enriched at microtubule
minus ends in aMTOC-free mouse oocytes
(Fig. 2A and fig. S4, A to C), as confirmed by
cold treatment, acute NDC80 depletion, and
immuno–electron microscopy (Fig. 2B; fig. S4,
D to F; and movies S4 to S6). NUMA was
enriched at the spindle poles at similar levels
in human oocytes and in aMTOC-free mouse
oocytes (fig. S4G), and in live imaging, NUMA
was localized in a similar pattern in aMTOC-
free mouse oocytes to that in human oocytes
throughout meiosis (Fig. 1, A and C; fig. S4,
H to J; and movie S7). Finally, we used Trim-
Away to test whether NUMA is required to
organize the spindle poles in aMTOC-free
mouse oocytes. In wild-type mouse oocytes,
perturbation of NUMA results in hyperfocused
spindle poles, which coincides with the ag-
gregation of aMTOCs ( 32 , 33 ). By contrast, in
aMTOC-free mouse oocytes, depletion of NUMA
caused defocused spindle poles (Fig. 2, C to
E; fig. S5, A to C; and movie S8), similar to
those in human oocytes (Fig. 1H). As a result
of spindle pole defocusing, the loose packing
of spindle microtubules was also reflected by
a significant decrease in the microtubule pack-
ing index (Fig. 2F), which measures how well
microtubules fit into the space of a minimum
bounding box surrounding the spindle. Thus,
the spindles assembled in aMTOC-free mouse
oocytes better mimic the spindles in human
oocytes than those in wild-type mouse oocytes.NUMA forms a stable scaffold at aMTOC-free
spindle poles
NUMA anchors spindle pole microtubules to
centrosomes in somatic cells ( 44 , 62 ) and to
aMTOCs in wild-type mouse oocytes ( 33 ), but
these stable scaffolds are not present in humanoocytes and in aMTOC-free mouse oocytes.
Because NUMA can self-assemble into stable
oligomers in vitro ( 63 , 64 ), we hypothesized that
the NUMA enriched at the spindle poles could
form a stable scaffold in the absence of aMTOCs.
This hypothesis predicts that NUMA would
have a lower turnover in aMTOC-free mouse
oocytes compared with wild-type mouse oocytes.
To examine the dynamics of NUMA, we photo-
activated NUMA at the pole-proximal, pole-
distal, and cytoplasmic regions in wild-type
and aMTOC-free mouse oocytes, similar to a
previous study ofg-tubulin dynamics at the
spindle poles in mitotic cells ( 65 ). Photoac-
tivation of the pole-proximal region revealed a
long-lived NUMA population in aMTOC-free
mouse oocytes but not in wild-type mouse
oocytes (fig. S5, D to G). By contrast, photo-
activation of the pole-distal region revealed a
distinct, short-lived NUMA population in both
wild-type and aMTOC-free mouse oocytes
(fig. S5, D to G). Because NUMA was weakly
detected along kinetochore microtubules in
both wild-type and aMTOC-free mouse oocytes
(figs. S2, D and E; and S4, E and F), we propose
that this short-lived population corresponds
to the transient, dynamic NUMA cross-links
recently reported on kinetochore microtubules
( 66 , 67 ). Furthermore, photoactivated NUMA
in the cytoplasm of aMTOC-free mouse oocytes
was only incorporated in the pole-distal region
(fig. S5, D to G), providing further support for
the low turnover of NUMA at aMTOC-free
spindle poles.
In mitotic cells, NUMA at the spindle poles
forms insoluble aggregates upon microtubule
depolymerization ( 68 – 72 ). To test whether
NUMA forms similar cytoplasmic aggregates in
aMTOC-free mouse oocytes, we depolymerized
spindle microtubules acutely with nocodazole
and prolonged cold treatment. However, NUMA
did not persist as cytoplasmic foci upon micro-
tubule depolymerization (fig. S5, H and I),
suggesting that the associations between
NUMA at the spindle poles are predominantly
microtubule dependent. Thus, NUMA forms a
stable scaffold at aMTOC-free spindle poles in
a microtubule-dependent manner.NUMA and dynein-dynactin-LIS1 cluster
microtubule minus ends at aMTOC-free
spindle poles
We exploited aMTOC-free mouse oocytes to
further decipher how the stably associated
NUMA organizes the spindle poles. NUMA
has been proposed to organize the spindle
poles by different mechanisms in different
systems, either as a microtubule–cross-linking
cargo that is transported to the poles by dynein-
dynactin or as an adaptor that recruits dynein-
dynactin to microtubule minus ends ( 25 , 73 – 78 ).
To understand how NUMA drives pole fo-
cusing in the absence of aMTOCs, we depleted
endogenous NUMA using follicle RNAi andSoet al.,Science 375 , eabj3944 (2022) 11 February 2022 3 of 19
RESEARCH | RESEARCH ARTICLE