Home » Drugs » Dual-spindle formation in zygotes
keeps parental genomes apart
in early mammalian embryos

Dual-spindle formation in zygotes
keeps parental genomes apart
in early mammalian embryos

Judith Reichmann1, Bianca Nijmeijer1, M. Julius Hossain1, Manuel Eguren1,
Isabell Schneider1, Antonio Z. Politi1, M. Julia Roberti1, Lars Hufnagel1,
Takashi Hiiragi2, Jan Ellenberg1*
At the beginning of mammalian life, the genetic material from each parent meets when the
fertilized egg divides. It was previously thought that a single microtubule spindle is
responsible for spatially combining the two genomes and then segregating them to create
the two-cell embryo. We used light-sheet microscopy to show that two bipolar spindles
form in the zygote and then independently congress the maternal and paternal genomes.
These two spindles aligned their poles before anaphase but kept the parental genomes
apart during the first cleavage. This spindle assembly mechanism provides a potential
rationale for erroneous divisions into more than two blastomeric nuclei observed in
mammalian zygotes and reveals the mechanism behind the observation that parental
genomes occupy separate nuclear compartments in the two-cell embryo.
After fertilization, the haploid genomes of egg
and sperm come together to form the genome
ofa newdiploid organism,amoment
that is of fundamental biological importance.
In mammals, parental chromosomes
meet for the first time upon entry into the first
zygotic mitosis after nuclear envelope breakdown
(NEBD). It has been assumed that, as observed
in oocytes, a single bipolar microtubule system
would self-assemble around both parental genomes
in zygotes (1–6). However, owing to the
extreme light sensitivity of the mammalian embryo,
the details of the dynamic process of zygotic
spindle assembly remain unclear.
To examine how parental genomes join for the
first time, we imaged live embryos in which the
maternal and paternal centromeres were differentially
labeled (7) using our recently developed
inverted light-sheet microscope, which allows
fast three-dimensional (3D) imaging of embryonic
development owing to its low phototoxicity
(8). This revealed that the two genomes remain
spatially separate throughout the first mitosis
(fig. S1 and movie S1). To understand why the
genomes are not mixed, we next imaged spindle
assembly using fluorescently labeled microtubule
organizing centers (MTOCs) and spindle microtubules
(Fig. 1A and movie S2). Newly nucleated
microtubules self-organized into two separate
bipolar spindles after NEBD, attracting a subset
of the cytoplasmic MTOCs that had accumulated
around each pronucleus to their poles (Fig. 1, A
and B). Subsequently, the two spindles aligned
and came into close apposition to form a compound
barrel–shaped system. This structure typically
had two clusters ofMTOCs at at least one
of its poles, suggesting that the two spindles were
aligned closely but not completelymerged (Fig. 1,
A and B, andmovie S2). To probe this further, we
performed 3D immunofluorescence analysis of
zygotes, visualizing endogenous spindle poles,
microtubules, kinetochores, and DNA. In early
and mid pro-metaphase, two separate bipolar
spindles were formed in in vivo–developed zygotes
(Fig. 1, C and D). Given the delayed and partial
association ofMTOCs with themicrotubule mass,
we hypothesized that dual-spindle formation
might be driven by self-assembly ofmicrotubules
nucleated by chromosomes. To test this, we assayed
microtubule regrowth after washing out
themicrotubule-depolymerizing drugNocodazole.
Indeed, a large proportion of microtubules was
nucleated on chromosomes, particularly at kinetochores
(fig. S2), whereas MTOCs became
associated with microtubules only later. This observation
prompted us to investigate the organization
of K-fibers in zygotic pro-metaphase by
means of high-resolution immunofluorescence
of zygotes fixed after brief cold treatment in
order to highlight stable microtubules (Fig. 1D
and fig. S3). Two bipolar arrays of K-fibers started
assembling in early pro-metaphase and were
stably organized in mid pro-metaphase, and their
remnants were clearly recognizable by their slightly
offset centers and split poles even after parallelization
in metaphase, during which the two
spindles were no longer separated by a large gap.
To characterize the kinetics of zygotic spindle
assembly in live embryos, we next imaged maternal
and paternal centromeres in relation to
growing spindle microtubule tips and observed
three phases of zygotic spindle assembly (Fig. 2A).
A transient first phase (~3min; 10.3 ± 3.5 to 13. 4 ±
4min after NEBD), characterized by the clustering
of growingmicrotubules around the two pronuclei,
was followed by phase 2 (~16min; 14.5 ± 4 to 30.7 ±
6.5min after NEBD), in which individual bipolar
spindles assembled around each parental genome,
and subsequently, phase 3 (~83 min; 46.7 ± 17 to
129.2 ± 16.5 min after NEBD), when the two
spindles aligned and combined into a compound
barrel–shaped structure.
To test whether the two zygotic spindles are
functionally independent, we measured the timing
and direction of maternal and paternal
chromosome congression (fig. S4, A and B, and
supplementary materials, materials and methods).
Congression started in pro-metaphase (phase 2),
while the spindles were clearly separated (fig. S4,
A and B). Parental chromosome congression was
not correlated in time until shortly before anaphase,
suggesting that they are moved by different
microtubule systems (fig. S4C). Furthermore,
the parental genomes were congressed in different
directions along separate spindle axes, as
evidenced by the large difference between the
angles of the two forming metaphase plates,
which became parallel only later during dual
spindle alignment in phase 3 (Fig. 2, B to D, and
fig. S4, D and E). Tracking of growing microtubule
tips showed two major directions of
microtubule flow during phase 2 and one main
direction during phase 3, as indicated by the
corresponding kymograph profiles (fig. S5 and
movies S3 to S8). The independent congression
frequently led to an offset in the final bioriented
position of the paternal and maternal metaphase
plates at the end of phase 3 before and during
segregation (Fig. 2A, arrows). Thus, each of the
two spindles around the parental genomes functions
independently for chromosome congression
and may even function separately in chromosome
In in vitro fertilization (IVF) clinics, the zygotic
division is error prone and often leads to embryos
with blastomeres that contain two nuclei
(9–13). We hypothesized that failure to align the
two parental spindles before anaphase could explain
this enigmatic phenotype. To test this, we
increased the distance between the two pronuclei
with transient treatment with Nocodazole, which
led to a larger gap between the two self-assembling
spindles (Fig. 3 andmovies S9 to S11). Indeed, such
embryos frequently failed to fully align the parental
spindles at one or both poles. This did not
delay anaphase but resulted in chromosome segregation
by two spindles into different directions,
leading to two cell embryos with one or
two binucleated blastomeres (Fig. 3 and movies
S10 and S11). By contrast, embryos that did align
the two spindles parallel to each other before anaphase
cleaved into two blastomeres with single
nuclei (movie S9). Thus, failure to align the two
zygotic spindles gives rise to multinucleated
two-cell embryos, phenocopying frequently
observed errors in human embryonic development
in IVF clinics.
Dual-spindle assembly in the mammalian zygote
would also offer a mechanistic explanation
for the long-standing observation that the parental
genomes occupy separate compartments
inside the nuclei of two- and four-cell embryo
Reichmann et al., Science 361, 189–193 (2018) 13 July 2018 1 of 5
1Cell Biology and Biophysics Unit, European Molecular
Biology Laboratory, Meyerhofstrasse 1, 69117 Heidelberg,
Germany. 2Developmental Biology Unit, European
Molecular Biology Laboratory, Meyerhofstrasse 1, 69117
Heidelberg, Germany.
*Corresponding author. Email: jan.ellenberg@embl.de
Downloaded from http://science.sciencemag.org/ on August 6, 2018
blastomeres (14, 15). If dual-spindle assembly
around two pronuclei was responsible for genome
compartmentalization, parental genomes should
mix in subsequent divisions, in which only one
nucleus is present per cell. Imaging of themetaphase
plate of live hybrid mouse embryos from
the zygote to the eight-cell stage showed that
parental genomes were separated in zygotes but
rapidly became mixed in the subsequent developmental
stages, as predicted (fig. S6, A to D).
This loss of genome compartmentalization was
also seen in in vivo–developed isogenic embryos
(fig. S6, E to I). Thus, parental genomes are kept
separate by two spindles only during the first
mitosis but then mix during subsequent divisions,
driven by a single common spindle.
If dual-spindle assembly is the mechanism for
parental genome compartmentalization (fig. S7A),
formation of a single spindle around both genomes
in the zygote should already mix them in the first
division. To test this prediction, we redirected
spindle assembly with two small-molecule inhibitors
of microtubule polymerization (Nocodazole)
and the motor protein Eg5 (Monastrol). Transient
treatmentwithMonastrol collected both genomes
in a single microtubule aster, and subsequent
depolymerization of microtubules withNocodazole
Reichmann et al., Science 361, 189–193 (2018) 13 July 2018 2 of 5
Fig. 1. Individual bipolar
spindle formation
around each pronucleus.
(A) Time-lapse imaging of
Mus musculus x Mus
musculus (MMU x MMU)
zygotes expressing EB3-
mCherry (marker for
microtubules; green) and
tdEos-Cep192 (marker for
MTOCs; magenta). Scale
bar, 10 mm. In 10 out of
13 zygotes, both or at
least one of the dualspindle
poles remained
clearly split after the
spindles had parallelized.
(B) Schematic diagram
showing progression of
dual-spindle formation
based on the data
presented in (A) and data
shown subsequently in
the manuscript. Microtubules,
gray; MTOCs,
magenta. (C) Immunofluorescence
staining of
MMU x MMU zygotes
fixed at consecutive
stages of development.
Maximum z projections of
confocal sections of
zygotes at prophase, early
pro-metaphase, late prometaphase,
and early
metaphase.White arrowheads
indicate poles.
(D) Immunofluorescence
staining of cold-treated
MMU x MMU prometaphase
Maximum z projections of
confocal sections. Microtubules,
a-Tubulin (green);
MTOCs, Pericentrin
(magenta); kinetochores,
Crest (white); and DNA,
Hoechst (blue). Scale
bars, 5 mm in (C) and (D).
Downloaded from http://science.sciencemag.org/ on August 6, 2018
followed by regrowth after washout then resulted
in one bipolar spindle around both genomes
(figs. S7B and S8, MoNoc-treated zygotes). Such
MoNoc-treated embryos captured and congressed
chromosomes within a single spindle and showed
a high degree of parental genome mixing in the
firstmitotic metaphase (Fig. 4). This was substantially
different from untreated or control zygotes
in which the order of drug treatments is reversed
(NocMo-treated zygotes), which maintained dualspindle
formation and genome separation (Fig. 4
and figs. S7C and S8). Thus, dual-spindle formation
in the zygote is required for parental genome
separation in mammals.
Having this experimental method to induce
mixing of the parental genomes in hand allowed
us to demonstrate that genome separation was
not required for parental genome epigenetic
asymmetry and its resolution (16–19), as proposed
previously (figs. S9 and S10) (15, 20–22).
Here, we showed that two spindles form
around pronuclei in mammalian zygotes, which
individually collect the parental genomes and
then position them next to each other before the
Reichmann et al., Science 361, 189–193 (2018) 13 July 2018 3 of 5
Fig. 2. Spindle assembly and chromosome dynamics in the zygote.
(A) Time-lapse imaging of Mus musculus x Mus Spretus (MMU x MSP)
zygotes expressing EB3-mCherry and fluorescent TALEs to label maternal
and paternal chromosomes though distinction of Major satellites (MajSat)
and Minor satellites (MinSat). Phase 1: Microtubule ball formation around
pronuclei. Phase 2: Bipolarization of maternal and paternal spindle. Phase 3:
Formation of single barrel–shaped spindle. (Top and top middle) 3D-rotated
images of the whole spindle volume. Maternal chromosomes, MajSat
(magenta); paternal chromosomes, MinSat (cyan); microtubules, EB3-
mCherry (white). (Bottom middle and bottom) Segmentation of maternal
(MatSpd;magenta) and paternal (PatSpd; cyan) spindles in phase 1 and phase
2 and single bipolar spindle in phase 3 (CompoundSpd; gray). Offset between
maternal and paternal chromosomes at metaphase and anaphase is
indicated with white arrows. (B) Schematic of measurements on maternal
(MatChr) and paternal chromatin masses (PatChr) in phases 2 and 3.
(C and D) Angle between maternal and paternal chromosome axis over time
for (C) a single embryo and (D) averaged for 12 embryos (mean ± SD) is
shown. Phase 1, blue; phase 2, red; phase 3, green.
Downloaded from http://science.sciencemag.org/ on August 6, 2018
first anaphase by aligning the two spindles parallel
to each other. Our data explain how parental
genome separation is achieved in mammalian
embryos. To date, the formation of physically
distinct mitotic spindles around the two pronuclei
has been thought to be specific to certain
arthropod species (23, 24). Our finding that this
occurs also during mammalian pro-metaphase
suggests that two zygotic spindles might be
characteristic for many species that maintain
separate pronuclei after fertilization. Failure to
align the two spindles produced errors in the
zygotic division that closely resemble clinical
phenotypes of human embryos in IVF procedures,
suggesting that a similar mechanism of dual
Reichmann et al., Science 361, 189–193 (2018) 13 July 2018 4 of 5
Fig. 3. Proximity dependency of bipolar spindle fusion. (A) Time-lapse
imaging of MMU x MMU zygotes expressing H2B-mCherry (chromatin;
magenta) and aTubulin–enhanced green fluorescent protein (EGFP)
(microtubules; green). Shown is spindle morphology from pro-metaphase
to postmitosis in three representative zygotes treated with Nocodazole
for >10 hours. Maximum z projections are of pro-metaphase, metaphase,
anaphase, telophase, and postmitosis. Arrowheads and “PB” indicate
nuclei and polar body, respectively. Scale bar, 10 mm. In the absence of
NEBD as a timing reference, anaphase onset was set at 90 min (average
time from NEBD to anaphase in MMU x MMU zygotes), and the other
times were calculated accordingly. (B) Initial distance of pro-nuclei
(n = 19). Statistics, Student’s t test.
Fig. 4. Distribution of parental centromeres in control, NocMo-treated,
and MoNoc-treated zygotes. (A) Differential labeling of maternal (MajSat;
magenta) and paternal (MinSat; cyan) centromeres through distinction
of single-nucleotide polymorphisms by means of fluorescent TALEs.
Mitotic spindle is labeled with EB3-mCherry (white). Representative
z-projected images of parental chromosome distribution in untreated,
MoNoc, and NocMo MMU x MSP zygotes. Scale bar, 10 mm. (B) Degree
of overlap between 3D convex hulls of parental chromosomes for
untreated (n = 31), MoNoc (n = 16), and NocMo (n = 12) zygotes
and embryos with in silico randomized distribution (n = 40) (fig. S1
and supplementary materials, materials and methods). Statistics,
Student’s t test.
Downloaded from http://science.sciencemag.org/ on August 6, 2018
zygotic spindle assembly also occurs in humans.
This view is supported by the spatial separation
of parental chromosomes reported in human
zygotes (25) and divisions of human zygotes into
multinucleated blastomeres, as well as by reports
that identified a mix of paternal, maternal, and
diploid cells in eight-cell cattle embryos. A dual
zygotic spindle would provide a mechanistic
basis for this parental genome segregation
(9–13, 26, 27). These severe and relatively frequent
zygotic division errors in human and agriculturally
used mammals thus find their likely
mechanistic explanation in a failure of the close
alignment of the two zygotic spindles before anaphase.
If a similar mechanism of microtubuledriven
parental genome separation indeed occurs
in human zygotes, it would be important from an
ethical and legal perspective because “pronuclear
fusion,” a process that strictly speaking does not
occur in mouse zygotes, is used to define the
beginning of embryonic life as protected by
law in several countries (for example, Germany,
§ 8 Abs. 1 Embryonenschutzgesetz).

  1. A. Courtois, M. Schuh, J. Ellenberg, T. Hiiragi, J. Cell Biol. 198,
    357–370 (2012).
  2. G. Fitzharris, Development 136, 2111–2119 (2009).
  3. P. A. Coelho et al., Dev. Cell 27, 586–597 (2013).
  4. M. Schuh, J. Ellenberg, Cell 130, 484–498 (2007).
  5. T. S. Kitajima, M. Ohsugi, J. Ellenberg, Cell 146, 568–581
  6. Z. Holubcová, M. Blayney, K. Elder, M. Schuh, Science 348,
    1143–1147 (2015).
  7. Y. Miyanari, C. Ziegler-Birling, M. E. Torres-Padilla, Nat. Struct.
    Mol. Biol. 20, 1321–1324 (2013).
  8. P. Strnad et al., Nat. Methods 13, 139–142 (2016).
  9. B. Kalatova, R. Jesenska, D. Hlinka, M. Dudas, Acta Histochem.
    117, 111–125 (2015).
  10. H. Balakier, K. Cadesky, Hum. Reprod. 12, 800–804 (1997).
  11. J. Meriano, C. Clark, K. Cadesky, C. A. Laskin, Reprod. Biomed.
    Online 9, 511–520 (2004).
  12. A. Egashira et al., J. Reprod. Dev. 61, 595–600 (2015).
  13. I. Kligman, C. Benadiva, M. Alikani, S. Munne, Hum. Reprod. 11,
    1492–1498 (1996).
  14. W. Mayer, A. Niveleau, J. Walter, R. Fundele, T. Haaf, Nature
    403, 501–502 (2000).
  15. W. Mayer, A. Smith, R. Fundele, T. Haaf, J. Cell Biol. 148,
    629–634 (2000).
  16. M. Puschendorf et al., Nat. Genet. 40, 411–420 (2008).
  17. K. Iqbal, S. G. Jin, G. P. Pfeifer, P. E. Szabó, Proc. Natl. Acad.
    Sci. U.S.A. 108, 3642–3647 (2011).
  18. A. Inoue, Y. Zhang, Science 334, 194 (2011).
  19. M. Wossidlo et al., Nat. Commun. 2, 241 (2011).
  20. A. V. Probst, G. Almouzni, Differentiation 76, 15–23 (2008).
  21. R. Duffié, D. Bourc’his, Curr. Top. Dev. Biol. 104, 293–328
  22. R. De La Fuente, C. Baumann, M. M. Viveiros, Development
    142, 1806–1817 (2015).
  23. R. R. Snook, D. J. Hosken, T. L. Karr, Reproduction 142,
    779–792 (2011).
  24. N. Kawamura, Dev. Growth Differ. 43, 343–349 (2001).
  25. C. van de Werken et al., Nat. Commun. 5, 5868 (2014).
  26. A. Destouni et al., Genome Res. 26, 567–578 (2016).
  27. A. Destouni, J. R. Vermeesch, BioEssays 39, 1600226 (2017).
    We thank N. Daigle for cloning the EB3-mCherry plasmid; P. Neveu
    and M. Schuh for providing tdiRFP670 and tdEos-Cep192,
    respectively; and N. Galjart for providing full-length Homo sapiens
    EB3 cDNA (NM_001303050.1). We thank EMBL’s laboratory of
    animal resources for support, P. Strnad for development and
    assistance of the inverted light-sheet microscope, A. Rybina for
    assistance with the Zeiss LSM 880 Airy microscope, Arivis for
    support in image analysis, and the EMBL Advanced Light
    Microscopy Facility for support in image acquisition and analysis. We
    thank J. Reddington and S. Alexander for critical reading of the
    manuscript. Funding: This work was supported by funds from the
    European Research Council (ERC Advanced Grant “Corema,” grant
    agreement 694236) to J.E. and by the European Molecular Biology
    Laboratory (all authors). J.R. was further supported by the EMBL
    Interdisciplinary Postdoc Programme (EIPOD) under Marie Curie
    Actions COFUND; M.E. by the EMBO long-term postdoctoral
    fellowship and EC Marie Slodowska-Curie postdoctoral fellowship;
    I.S. by a Boehringer Ingelheim Fonds Ph.D. fellowship; and
    M.J.R. by a Humboldt Foundation postdoctoral fellowship. I.S. is
    a candidate for a joint Ph.D. between EMBL and Heidelberg
    University, Faculty of Biosciences. Author contributions: J.E.
    and J.R. conceived the project and designed the experiments. J.R.,
    B.N., M.E., and I.S. performed the experiments. M.J.R. supported
    the mouse EDU experiments. J.R., M.J.H., and A.Z.P. analyzed the
    data. T.H. and L.H. contributed to conception and design of the
    work. J.E. and J.R. wrote the manuscript. All authors contributed to
    the interpretation of the data and read and approved the final
    manuscript. Competing interests: L.H. and J.E. are scientific
    cofounders and advisers of Luxendo GmbH (part of Bruker), which
    makes light-sheet–based microscopes commercially available.
    Data and materials availability: All images processed in this
    study are available in the Image Data Resource (IDR), accession
    number idr0045 (https://idr.openmicroscopy.org/webclient/
    ?show=project-405). All code is available from EMBL’s git
    depository (https://git.embl.de/grpellenberg/dual_zygotic_
    Materials and Methods
    Figs. S1 to S10
    Table S1
    Movies S1 to S11
    References (28–34)
    15 December 2017; accepted 8 June 2018