Histological sections through limb, body wall,
and diaphragm musculature revealed an absence
of multinucleated myofibers in Myomixer KO
embryos (Fig. 3B). Instead, presumptive muscle-forming regions were populated by mononucleated
cells that stained for myosin expression (Fig. 3C).
Tissues other than muscle appeared normal in
these embryos. The cutting sites of the two
sgRNAs were 122 base pairs apart in the third
exon of the Myomixer gene. PCR with primers
amplifying this region showed mutant mice
with deletions that ranged from 163 to 470 bp,
reflecting different indels (fig. S3, A and B).
Western blot of hindlimbs confirmed the absence of Myomixer protein in the KO embryos
The muscle phenotype of Myomixer mutant
mice is reminiscent of that seen in mice lacking
Myomaker, a fusogenic transmembrane muscle
protein (18, 19), suggesting a functional relationship between these two regulators of myoblast
fusion. To assess the functional relationship between Myomixer and Myomaker in myoblast
fusion, we retrovirally expressed each protein in
C2C12 myoblasts and monitored myotube formation. As shown in Fig. 4A, Myomixer markedly
enhanced the fusion of C2C12 cells. Moreover,
when Myomixer and Myomaker were overexpressed together in C2C12 cells, they promoted
the formation of massive multinucleated myotubes, which is indicative of their synergistic
activity (Fig. 4A and fig. S4).
Myomaker can induce fusion of 10T1/2 mouse
fibroblasts with myoblasts (18, 19). In contrast,
Myomixer expressing 10T1/2 cells did not fuse
with C2C12 cells (Fig. 4B). To investigate whether
Myomixer can synergize with Myomaker to promote heterologous cell fusion, we infected green
fluorescent protein (GFP)–labeled 10T1/2 cells
with Myomaker and Myomixer retroviruses and
mixed them with mCherry-labeled C2C12 cells
(fig. S5A). We found that expression of Myomixer
and Myomaker together in 10T1/2 cells induced
dramatic fusion to C2C12 myotubes so that only a
few mononucleated GFP-labeled fibroblasts remained in the cultures (Fig. 4B). Moreover, when
Myomaker and Myomixer were coexpressed in
two populations of 10T1/2 fibroblasts labeled
with GFP or mCherry and the cells were mixed,
fibroblast-fibroblast fusion was observed (Fig. 4C
and fig. S5B). Multinucleated fibroblasts coexpressing GFP and mCherry, which appeared
yellow, often resembled bird nests filled with
eggs (Fig. 4C). Western blot confirmed expression
of Myomaker and Myomixer in the appropriate
cells (fig. S5C). We observed no effect of either
protein on the level of expression of the other,
indicating that their synergy does not reflect an
effect on stability of either protein. A summary of
the effects of Myomixer and Myomaker on fusion
is shown in fig. S5D.
To further test the functional dependency of
Myomaker on Myomixer for cell fusion, we mixed
Myomaker-expressing 10T1/2 cells with WT or
Myomixer-KO C2C12 myoblasts and switched them
to DM for 1 week, after which we immunostained
for myosin. Although Myomaker-expressing 10T1/2
cells fused with WT C2C12 cells, they were unable
to fuse with Myomixer KO C2C12 cells (fig. S5E).
This implies that Myomaker relies on Myomixer
for cell fusion in trans. Indeed, Myomaker was
expressed normally in Myomixer KO myoblasts and
embryos, suggesting a dependency of Myomaker
on Myomixer for normal muscle fusion and devel-
opment (fig. S6, A and B).
To begin to define the mechanistic basis of the
cooperativity between Myomixer and Myomaker,
we examined whether the two proteins physically
interact. Indeed, we found that in coimmuno-
precipitation assays, FLAG-tagged Myomaker co-
immunoprecipitated with Myomixer in 10T1/2
fibroblasts and differentiated C2C12 cells (Fig. 4D).
Replacement of arginines at amino acid posi-
tion 34, 38, or 46 with glutamic acid residues
within the charged segment of Myomixer dimin-
ished association with Myomaker but did not af-
fect membrane association (Fig. 4E and fig. S7A).
The EEEAA mutant failed to synergize with
Myomaker to stimulate fusion of C2C12 myo-
blasts (Fig. 4F and fig. S7B), as well as het-
erologous fusion of C2C12 myoblasts with 10T1/2
fibroblasts (fig. S7C), indicating a correlation be-
tween the association of Myomixer with Myomaker
and their synergistic fusogenic activity. Although
mutation of cysteine 52 to alanine in the second
hydrophobic region disrupted the fusogenic ac-
tivity of Myomixer (Fig. 4F and fig. S7B), this
mutation did not prevent membrane localiza-
tion (fig. S7A) or interaction with Myomaker in
coimmunoprecipitation experiments (Fig. 4E).
This suggests that the second hydrophobic do-
main of Myomixer may mediate the fusogenic
function after its binding with Myomaker. A sum-
mary of the effects of Myomixer mutations is
shown in fig. S7D. Given the cross-species hom-
ology of Myomixer orthologs, we tested wheth-
er the ortholog from zebrafish (Danio rerio)
could also promote heterologous fusion of cells.
As shown in fig. S7E, this zebrafish peptide of
only 75 amino acids induced cell fusion when
overexpressed with mouse Myomaker. Consist-
ent with this finding, deletion of the C terminus
of mouse Myomixer (MyomixerDC) did not abol-
ish its fusogenic function (fig. S7E). Thus, we
conclude that Myomixer is an evolutionarily con-
served regulator of myoblast fusion and that the
C-terminal region that is specific to higher ver-
tebrates is dispensable for function, at least in
The requirement of Myomixer for fusion of
myoblasts in vivo and in vitro, its ability to
synergistically promote fusion together with
Myomaker, and the physical and functional
interaction between these proteins indicate that
they govern the critical step of multinucleation
of skeletal muscle. We hypothesize that Myo-
mixer activates the fusogenic activity of Myomaker
to drive membrane mixing, perhaps by estab-
lishing a fusion pore.
The small size of Myomixer places it in the
category of micropeptides, characterized by un-
processed ORFs of less than 100 amino acids
(20). A majority of micropeptides identified to
date are embedded in membranes (21, 22). There
are some striking similarities between Myomixer
and the heart-specific micropeptide DWORF
(Dwarf open reading frame), which localizes to
the sarcoplasmic reticulum of cardiomyocytes
where it associates with the sarco/endoplasmic
reticulum Ca2+-ATPase (SERCA) calcium adeno-
sine triphosphatase (ATPase) and stimulates its
activity (23). We speculate that the activities of
many membrane proteins may be governed by
association with micropeptides that are as yet
REFERENCES AND NOTES
1. C. F. Bentzinger, Y. X. Wang, M. A. Rudnicki,
Cold Spring Harb. Perspect. Biol. 4, a008342
2. M. Buckingham, P. W. J. Rigby, Dev. Cell 28, 225–238
3. K. Rochlin, S. Yu, S. Roy, M. K. Baylies, Dev. Biol. 341, 66–83
4. A. R. Demonbreun, B. H. Biersmith, E. M. McNally, Semin.
Cell Dev. Biol. 45, 48–56 (2015).
5. R. S. Krauss, G. A. Joseph, A. J. Goel, Cold Spring Harb.
Perspect. Biol. 9, a029298 (2017).
6. A. Simionescu, G. K. Pavlath, Adv. Exp. Med. Biol. 713, 113–135
7. S. M. Abmayr, G. K. Pavlath, Development 139, 641–656
8. H. Yin, F. Price, M. A. Rudnicki, Physiol. Rev. 93, 23–67
9. J. D. Doles, B. B. Olwin, Curr. Opin. Genet. Dev. 34, 24–28
10. A. S. Brack, T. A. Rando, Cell Stem Cell 10, 504–514
11. E. H. Chen, E. N. Olson, Science 308, 369–373
12. N. E. Sanjana, O. Shalem, F. Zhang, Nat. Methods 11, 783–784
13. I. H. B. Chen, M. Huber, T. Guan, A. Bubeck, L. Gerace,
BMC Cell Biol. 7, 38 (2006).
14. N. Liu et al., Nat. Cell Biol. 19, 202–213 (2017).
15. F. Yue et al., Nature 515, 355–364 (2014).
16. A. P. Fong et al., Dev. Cell 22, 721–735 (2012).
17. Y. M. Chen, Z. W. Du, Z. Yao, Acta Biochim. Biophys. Sin.
(Shanghai) 37, 789–796 (2005).
18. D. P. Millay et al., Proc. Natl. Acad. Sci. U.S.A. 113, 2116–2121
19. D. P. Millay et al., Nature 499, 301–305 (2013).
20. B. R. Nelson, D. M. Anderson, E. N. Olson, Circ. Res. 114, 18–20
21. D. M. Anderson et al., Cell 160, 595–606 (2015).
22. D. M. Anderson et al., Sci. Signal. 9, ra119 (2016).
23. B. R. Nelson et al., Science 351, 271–275 (2016).
We are grateful to J. Cabrera for graphics, B. Chen for
CRISPR library data processing, W. Ye for chromatin
immunoprecipitation sequencing data analysis, and N. Grishin
and J. Pei for bioinformatic analysis of Myomixer orthologs. We
thank K. White in the Histopathology Core for technical
assistance. We also thank H. Zhou, L. Amoasii, A. Bookout,
Z. Wang, and N. Liu for biological materials and other members
of the Olson laboratory for technical advice. This work was
supported by grants from the NIH (AR-067294, HL-130253,
DK-099653, and HD-087351) and the Robert A. Welch
Foundation (grant 1-0025 to E.N.O.).
Materials and Methods
Figs. S1 to S7
7 February 2017; accepted 26 March 2017
Published online 6 April 2017