transfer was observed from B cells to cDCs (fig.
S3D). MZ B cells were more trogocytic than
their follicular (FO) counterparts (Fig. 3C). Fur-
thermore, B cells did not trogocytoseMarch1–/–
macrophage, neutrophil, or T cell membranes
(fig. S3E). Thus, B cells—particularly MZ B
cells—specifically acquire cDC plasma mem-
brane and surface proteins via trogocytosis, and
MARCH1 deficiency makes cDC membranes
more susceptible to trogocytic transfer.
MARCH1 controls the amount of C3 that
accumulates on the surface of cDCs
Trogocytosis is mediated by surface receptors
( 9 ), so we hypothesized that MARCH1 regu-
lates the expression of a receptor that medi-
ates trogocytosis. The only receptors known to
increase in expression inMarch1–/–cDCs are
MHC II and CD86 (fig. S2E) ( 2 ), but a recent
plasma membrane proteome analysis showed
that the surface ofMarch1–/–cDCs is also highly
enriched in C3 ( 12 ). Because C3 is inaccessible to
cytosolic ubiquitination by MARCH1, we sought
to characterize the mechanism that caused its
accumulation on the cell surface before inves-
tigating its potential role in trogocytosis.
First, we confirmed that C3 is present on
wild-type cDC1s and cDC2s and is overex-
pressed on theirMarch1–/–counterparts (Fig. 4A
and fig. S4A). Analysis of C3-deficient mice
demonstrated the specificity of this detection
(Fig. 4A). MARCH1 is active in all hemato-
poietic APCs, where it keeps surface MHC II
expression at intermediate to low levels ( 12 ).
We observed an increase in C3 deposition on
both professional and“atypical”APCs from
the spleen, lymph nodes, and thymus but not
on T cells (Fig. 4B). C3 binding to cDCs was
not caused by enzymatic tissue digestion be-
cause it was also observed in cell suspensions
prepared by mechanical disruption (fig. S4B).
Furthermore, when wild-type andMarch1–/–×
C3–/–spleens were pooled before cell purifica-
tion, the mutant cDCs remained negative for
C3 expression (fig. S4C).
In mixed–bone marrow (BM) chimeric mice
where 1:1 wild-type andMarch1–/–BM was
used to reconstituteC3–/–recipients, neither
wild-type norMarch1–/–cDCs displayed C3
(Fig. 4C). This indicated that C3 was captured
from the extracellular environment, as expected,
because it is produced mainly by liver cells ( 13 ).
If the recipient chimeric mice were wild-type,
the cDCs generated from theMarch1–/–BM
displayed higher surface expression of C3 than
their wild-type counterparts (Fig. 4C), which
implies that the effect of the mutation was cell-
intrinsic. Thus, C3 is secreted into circulation by
nonhematopoietic cells and is deposited on all
APCs.IfthecellsdonotexpressMARCH1,C3depositionincreasesbyasmuchasafactorof
20inthecaseofcDC1s.C3 activated by tickover binds to MHC II
Proteomic analysis of theMarch1–/–cDC plasma
membrane ( 12 ) (table S2) indicated that the
C3 species found on the cDC surface is one
or both of its inactivated forms (i.e., C3dg or
C3d) (fig. S1B and table S1). Analysis of wild-
type,March1–/–, andC3–/–cDC lysates by im-
munoblot identified several molecular species
of C3 that were only detected inMarch1–/–
cDCs (Fig. 5A, fig. S1B, and table S1). The most
prominent species had a molecular weight of
~70 kDa and was absent from the serum. We
hypothesized that this species corresponded to
C3dg or C3d covalently bound to MHC IIaorb.
The cDCs of mice deficient in both MARCH1
and MHC II (March1–/–×H2-Aa–/–) showed
elevated CD86 expression (fig. S4D), indicat-
ing that the absence of MHC II did not prevent
surface accumulation of MARCH1 substrates.
However, C3 was barely detectable on the sur-
face of these cells (Fig. 5B) or on cDCs that only
lacked MHC II expression (fig. S4E). Further-
more, the cDCs of knock-in mice—in which
the only MHC II ubiquitination site, Lys^225
of thebchain, has been replaced with Arg
(MHC IIKRKI/KImice) ( 3 , 14 )—expressed sim-
ilarly high levels of C3 relative toMarch1–/–Schrieket al.,Science 375 , eabf7470 (2022) 11 February 2022 2 of 12
Fig. 1. Mice deficient in
MARCH1 E3 ubiquitin ligase
have reduced numbers of
splenic cDCs.(AandB) Numbers
of the indicated wild-type (WT)
andMarch1–/–cell types in whole
splenocytes (A) or low-density
splenocyte preparations (B)
before (left) and after (right)
depletion of non-cDCs. Graphs
display data pooled from three
independent experiments, with
each symbol representing an
individual mouse (n= 2 or 3 per
experiment); bars denote mean ±
SD. *P< 0.0002, **P<
0.0001 [independent-samples
ttest with Welch’s correction
(no assumption of equal varian-
ces), two-tailedPvalue (95% CI)];
ns, not significant.
A whole splenocytes0306090120
ns05101520 ns051015 ns03060901201501800204060cDC1 cDC2 pDC B cell CD4+ T cell CD8+ T cell07142128^35 ns
number(×1(^40)
)
number
(×10
4 )
number
(×1
(^40)
)
March1
−/−
WT
number
(×1
(^60)
)
number
(×1
(^60)
)
number
(×1
(^60)
)
B low-density splenocytes low-density splenocytes
neg. selection
- number
(×10
4 )
number
(×10
4 )
number (×10
4 )
0
10
20
30
0
20
40
60
80
100
number
(×10
4 )
number
(×1
(^40)
)
cDC1 cDC2 pDC
0
10
20
30
40
(^50)
0
30
60
90
120
(^150) *
0
7
14
21
28
35 ns
cDC1 cDC2
March1
−/−
WT
March1
−/−
WT
March1
−/−
WT
March1
−/−
WT
March1
−/−
WT
March1
−/−
WT
March1
−/−
WT
March1
−/−
WT
March1
−/−
WT
March1
−/−
WT
RESEARCH | RESEARCH ARTICLE