on the mesodermal side of the first macro-
phage’s protrusion (fig. S6). Thus, macrophages
appear to enter next to the ectodermal cells and
are separated from the mesoderm by the ECM.
Next, we examined whether ectodermal di-
vision is required for macrophage entry. We
injected embryos with the drug dinaciclib,
which inhibits cyclin-dependent kinase 1 (Cdk1)
and other cyclin-dependent kinases and effec-
tively stops the cell cycle (fig. S7, A and B)
without affecting morphogenetic tissue move-
ments (fig. S7C) or total macrophage numbers
(fig. S7D). The density of rounded ectodermal
cells was 83% lower after dinaciclib injection
(Fig. 2A). Macrophages remained motile and
moved directionally (fig. S7, E and F) toward
the entry point; however, they almost com-
pletely failed to invade (Fig. 2, B and B′, and
fig. S7, G and G′). We quantified macrophages
inside the GB ~60 min after macrophages
arrived at the entry point and normalized for
the total macrophage number. In control
embryos, ~20% of total macrophages had
moved inside the GB by this time. However, in
dinaciclib-treated embryos, only a few macro-
phages entered the GB (Fig. 2C). Live imaging
revealed that when a few macrophages en-
tered, they moved in at the usual site next to
a round ectodermal cell that did not progress
to complete division for a long period of time
(fig.S7,HtoK,andmovieS3).Thus,phar-
macological inhibition of cell division resulted
in macrophages not invading at all, or enter-
ing next to the remaining mitotically rounded
ectodermal cells (Fig. 2D).
To confirm these results, we inhibited divi-
sion only locally, by expressing RNA inter-
ference (RNAi) constructs in the ectoderm
against mRNA encoding a positive regulator
of mitosis (fig. S7A), Cdc25 (Stg). This knock-
down (KD) reduced the ectodermal frequency
of rounded cells by ~30% (Fig. 2E). Between
25 and 40% fewer macrophages penetrated
the GB upon two differentstgKDs compared
with the control (Fig. 2F and fig. S8, A and
A′). In embryos expressing these RNAis, in
which no divisions occurred in the GB edge
for a long time, no macrophages entered (fig.
S8B and movie S4, left). When we detected
entry in other embryos, it was always adjacent
to a remaining mitotic ectodermal cell at the
normal location (fig. S8C and movie S4, right).
Similar results were observed upon ectoder-
mal overexpression of p53 (fig. S8, D and E).
Therefore, reducing the frequency of ectoder-
mal cell division decreases macrophage entry,while maintaining the entry-mitotic rounding
correlation (Fig. 2G and fig. S8F).
To test whether increasing the rate of mitosis
in the ectoderm would cause the opposite re-
sult, we knocked down a negative regulator of
mitosis, Tribbles (Trbl), or overexpressed the
G1 progression regulators Cyclin D (CycD) and
Cdk4 in the ectoderm (fig. S7A). These treat-
ments increased the density of ectodermal
rounded cells before macrophage entry by 60%
(Fig. 2E′) and led to a 25 to 50% increase in
macrophages in the GB compared with the con-
trol (Fig. 2F′and fig. S8, G and G′). Macrophages
always entered at their usual position and next
to a mitotic ectodermal cell (Fig. 2G′, fig. S8H,
and movie S5). Total macrophage numbers
were unchanged compared with controls (fig.
S8, I and J). These results strongly suggest that
the timing of ectodermal divisions is a rate-
limiting factor for macrophage invasion.
To examine whether macrophages stimu-
late division to facilitate invasion, we expressed
Hid in macrophages, inducing their apoptosis
before reaching the GB (fig. S9A). However,
the density of dividing cells in the GB was un-
altered in these embryos, and the timing and
speed of GB retraction remained unchanged
(fig. S9, B and C, and movie S6).SCIENCEscience.org 22 APRIL 2022•VOL 376 ISSUE 6591 395
Fig. 3. Macrophages enter when FAs disassemble
during ectodermal mitotic rounding.(A) Scheme of
ectodermal FAs. (B) FAs in the ectoderm–mesoderm
interface visualized by vinculin::mCherry (magenta)
with ectodermal cells’membranes (green). DE-Cad,
DrosophilaE-cadherin. (C) Time-lapse imaging of
ectodermal FAs during mitotic rounding and subsequent
macrophage entry att= 0 (yellow and white stars).
White arrows indicate large FA that disassembles.
(D) Vinculin::mCherry on the basal side of mitotic cell,
cutout from (C). (E) Vinculin::mCherry intensity along
the basal side of rounding cell in (C) and (D) from
point x to point y (white arrowheads) over time. Green
circle indicates first macrophage nucleus. a.u., arbitrary
units. (F) Amplitude of FA peak over time. Mean ± SD
(shading). Scale bars, 5mm [(B) and (C)] and 2mm (D).
Fig. 2. Macrophage entry requires ectodermal
mitotic rounding.(A) Rounded cell density in
the GB tissue at macrophage entry time and (Band
C) percent of all macrophages that were found
inside the GB as GB retraction started in control
dimethyl sulfoxide–injected embryos (Ctrl) and
dinaciclib-injected embryos (Dina). (B) and (B′)
show dorsal view of maximum intensity projections
of stage 12 embryos. Dashed line indicates the
GB edge. Scale bars, 20mm. mac-nuclei, macrophage
nuclei; ecto-memb, ectoderm membrane. **P<
0.0001 [(A) and (C)]. (D) Quantification of macrophage
entry timing. (EandE′) Rounded cell density in the
GB altered by cell cycle regulators. P= 0.0021; **P< 0.0001. (FandF′) Percent of all macrophages that were found inside the GB. P= 0.0067, **P< 0.0001 (F);
(left to right) *P= 0.0007, **P< 0.0001, *P= 0.0009 (F′). (GandG′) Quantification of macrophage entry timing. Mean ± SEM. Ten planes per embryo analyzed in
(A), 5 to 10 in (E) and (E′). Mann-Whitney tests [(A) and (C)], unpaired two-tailedttests [(E) to (F′)].
RESEARCH | REPORTS