Lee et al. eLife 2013;2:e00291. DOI: 10.7554/eLife.00291 1 of 22
UNC93B1 mediates differential trafficking
of endosomal TLRs
Bettina L Lee1, Joanne E Moon1, Jeffrey H Shu1, Lin Yuan2, Zachary R Newman1,
Randy Schekman2,3, Gregory M Barton1,2*
1Division of Immunology and Pathogenesis, Department of Molecular and Cell
Biology, University of California, Berkeley, Berkeley, United States; 2Division of Cell
and Developmental Biology, Department of Molecular and Cell Biology, University of
California, Berkeley, Berkeley, United States; 3Howard Hughes Medical Institute,
University of California, Berkeley, Berkeley, United States
Abstract UNC93B1, a multipass transmembrane protein required for TLR3, TLR7, TLR9, TLR11,
TLR12, and TLR13 function, controls trafficking of TLRs from the endoplasmic reticulum (ER) to
endolysosomes. The mechanisms by which UNC93B1 mediates these regulatory effects remain
unclear. Here, we demonstrate that UNC93B1 enters the secretory pathway and directly controls
the packaging of TLRs into COPII vesicles that bud from the ER. Unlike other COPII loading factors,
UNC93B1 remains associated with the TLRs through post-Golgi sorting steps. Unexpectedly, these
steps are different among endosomal TLRs. TLR9 requires UNC93B1-mediated recruitment of
adaptor protein complex 2 (AP-2) for delivery to endolysosomes while TLR7, TLR11, TLR12, and
TLR13 utilize alternative trafficking pathways. Thus, our study describes a mechanism for differential
sorting of endosomal TLRs by UNC93B1, which may explain the distinct roles played by these
receptors in certain autoimmune diseases.
Toll-like receptors (TLRs) recognize conserved microbial features and initiate signals critical for induc-
tion of immune responses to infection. A subset of TLRs (TLR3, TLR7, TLR8, and TLR9) recognizes
forms of nucleic acids, including double-stranded RNA, single-stranded RNA, and DNA (Barbalat et al.,
2011). This specificity facilitates recognition of a broad array of microbes but introduces the potential
for recognition of self-nucleic acids. TLR7 and TLR9 recognition of self-RNA and self-DNA, respect-
ively, contributes to autoimmune diseases such as systemic lupus erythematosus (SLE) (Marshak-
Rothstein, 2006; Christensen and Shlomchik, 2007).
Discrimination between self and microbial nucleic acids cannot be achieved solely through recogni-
tion of distinct features but instead relies on differential delivery of these potential ligands to TLRs
(Barton and Kagan, 2009). All of the TLRs capable of nucleic acid recognition localize within endo-
somal compartments which sequesters these receptors away from self nucleic acids in the extracellular
space (Barton and Kagan, 2009). Our previous studies as well as work from other groups indicate that
a requirement for ectodomain cleavage of intracellular TLRs further restricts receptor activation to
protease-rich acidic compartments (Ewald et al., 2008, 2011; Park et al., 2008; Garcia-Cattaneo
et al., 2012). Bypassing this requirement enables responses to extracellular self nucleic acid and
leads to fatal inflammatory disease in mice (Mouchess et al., 2011). Moreover, the system appears
carefully balanced as simply overexpressing TLR7 in mice causes responses to self-RNA and develop-
ment of an SLE-like disease (Pisitkun et al., 2006; Subramanian et al., 2006; Deane et al., 2007).
Thus, defining the regulatory steps that control TLR localization and influence the threshold of recep-
tor activation has important implications for self/non-self discrimination.
*For correspondence: barton@
Competing interests: See page 19
Funding: See page 19
Received: 28 September 2012
Accepted: 08 January 2013
Published: 19 February 2013
Reviewing editor: Ruslan
Medzhitov, Yale University,
Copyright Lee et al. This
article is distributed under the
terms of the Creative Commons
Attribution License, which
permits unrestricted use and
redistribution provided that the
original author and source are
Cell biology | Immunology
Lee et al. eLife 2013;2:e00291. DOI: 10.7554/eLife.00291 2 of 22
TLR9 and other intracellular TLRs must traffic from the endoplasmic reticulum (ER) to endolyso-
somes before responding to ligands. UNC93B1, a multi-pass transmembrane protein localized to the ER,
appears to facilitate this trafficking (Brinkmann et al., 2007; Kim et al., 2008). Mice homozygous for
a nonfunctional Unc93b1 (H412R) allele (Unc93b13d/3d) fail to respond to TLR3, TLR7, or TLR9 ligands,
and mice and humans deficient in UNC93B1 are highly susceptible to viral infection (Casrouge et al.,
2006; Tabeta et al., 2006; Lafaille et al., 2012). More recently, UNC93B1 has been implicated in the
function of TLR11, TLR12, and TLR13 (Pifer et al., 2011; Shi et al., 2011; Koblansky et al., 2012;
Oldenburg et al., 2012). UNC93B1 is not required for responses by surface localized TLRs such as
TLR2 and TLR4 (Tabeta et al., 2006). UNC93B1 associates with endosomal TLRs, and in cells with
defective UNC93B1, TLR9 and TLR7 fail to leave the ER (Brinkmann et al., 2007; Kim et al., 2008).
However, the mechanism by which UNC93B1 facilitates TLR trafficking to endosomal compart-
ments remains enigmatic, especially considering its reported direct translocation from the ER to
endolysosomes (Kim et al., 2008). This pathway is inconsistent with our findings that TLR9 and TLR7
traffic through the general secretory pathway en route to endosomes (Ewald et al., 2008, 2011). In
addition, mice expressing an aspartic acid to alanine mutation at amino acid position 34 in UNC93B1
(Unc93b1D34A/D34A) were recently shown to develop spontaneous autoimmunity due to enhanced TLR7
responses and diminished TLR9 responses (Fukui et al., 2009, 2011). These findings suggest that
regulation of TLRs by UNC93B1 can influence the relative thresholds of receptor activation. For these
reasons, we have sought to define the molecular basis by which UNC93B1 controls endosomal TLR
trafficking and function.
Beyond the implied role for UNC93B1 discussed above, little is known about the molecular mech-
anisms that mediate proper localization of endosomal TLRs and no other factor required specifically for
endosomal TLR trafficking has been identified. Nevertheless, several reports suggest that endosomal
TLR trafficking may be influenced at both ER and post-Golgi trafficking steps. Gp96 functions as an ER
folding chaperone for many TLRs, including TLR9, and PRAT4A has been implicated in TLR trafficking
from the ER (Takahashi et al., 2007; Yang et al., 2007; Lee et al., 2012). Additionally, the HRS/ESCRT
pathway is involved in post-Golgi trafficking by sorting ubiquitinated TLR7 and TLR9 to endosomal
compartments (Chiang et al., 2012), and the adaptor protein-3 (AP-3) has been reported to target
TLR9 and TLR7 to lysosome related organelles specialized for type I IFN induction (Honda et al., 2005;
eLife digest Toll-like receptors (TLRs) are proteins that are responsible for recognizing specific
molecules associated with invading pathogens, known as pathogen-associated molecular patterns.
Upon detecting these signals, TLRs activate the body’s immune response, which fights the infection.
A subset of TLRs recognizes nucleic acids, including DNA and RNA, enabling the immune system
to respond to foreign material from a diverse range of bacteria and viruses. However, some of the
body’s own DNA and RNA is also found outside cells (e.g., in the bloodstream) and TLRs must be
able to discriminate between these nucleic acids and those belonging to pathogens, because failure
to tell the difference between the two could result in autoimmune disease. To reduce this risk, TLRs
are sequestered inside the cell within membrane-bound compartments known as endosomes.
UNC93B1 is a transmembrane protein that is known to control the movement of TLRs from the
endoplasmic reticulum—where TLRs are assembled—to endosomes. However, the exact
mechanisms by which this protein controls TLR trafficking were unclear. Now Lee et al. reveal that it
directly controls the packaging of at least six TLRs at the endoplasmic reticulum: it helps to load
these TLRs into vesicles, which are in turn processed by the Golgi apparatus—the organelle wherein
proteins are sorted and packaged en route to their final destinations. Surprisingly, UNC93B1
remains associated with the TLRs even after Golgi processing.
Lee et al. also reveal that specific endosomal TLRs are subject to distinct post-Golgi trafficking
mechanisms. In order for TLR9 to be delivered to the endosome, UNC93B1 must recruit an adaptor
protein called AP-2, whereas other TLRs appear to require different actions by UNC93B1. By
defining the mechanisms that underlie the differential trafficking of endosomal TLRs, Lee et al.
suggest that we may learn how to manipulate distinct aspects of TLR activation, and also gain
insights into the causes of certain autoimmune diseases.
Cell biology | Immunology
Lee et al. eLife 2013;2:e00291. DOI: 10.7554/eLife.00291 3 of 22
Blasius et al., 2010; Sasai et al., 2010). Interestingly, UNC93B1 trafficking to these compartments is
also impaired in AP-3 deficient cells (Sasai et al., 2010). Whether UNC93B1 interacts with other com-
ponents implicated in trafficking of endosomal TLRs remains to be determined.
In this study, we report that UNC93B1 is required for multiple steps of TLR trafficking. UNC93B1
plays a direct role in facilitating exit of TLRs from the ER as well as a later role in recruitment of adaptor
protein-2 (AP-2) to facilitate endocytosis of TLR9 from the plasma membrane. Surprisingly, TLR7 does
not have the same requirements for UNC93B1 and utilizes distinct trafficking machinery to reach
endolysosomes. Thus, our results describe how UNC93B1 controls endosomal TLR trafficking and
provide the first mechanistic basis for differential regulation of these receptors.
UNC93B1 traffics to phagosomes via the Golgi compartment
UNC93B1 has been described as an ER-resident trafficking chaperone that translocates TLRs directly
from the ER to endolysosomes upon TLR activation (Kim et al., 2008). This model is based in part
on the observation that UNC93B1 never acquires Endoglycosidase H (EndoH)-resistant glycans
(Brinkmann et al., 2007), which are acquired only when proteins traffic through the medial Golgi.
Because this proposed function for UNC93B1 conflicts with our model of TLR9 trafficking (Ewald et al.,
2008, 2011), we first examined whether UNC93B1 is present in endolysosomal compartments in
unstimulated cells. Wildtype (WT) UNC93B1 but not the nonfunctional (H412R) mutant was detect-
able in phagosomes purified from unstimulated RAW264 cells (Figure 1A). Moreover, a portion of
UNC93B1-WT gained EndoH-resistance in multiple cell types, while UNC93B1-H412R was entirely
EndoH-sensitive (Figure 1B–F). These results agree with a previous report that the H412R mutant
fails to leave the ER (Kim et al., 2008). To formally demonstrate that the increased molecular weight
of UNC93B1-WT is due to N-linked glycans, we mutated Asn-251, which is within a consensus
N-glycosylation site, and this mutant failed to acquire EndoH-resistant glycans (Figure 1G). Based on
the acquisition of EndoH-resistant glycans by UNC93B1, we examined whether UNC93B1 is detecta-
ble within COPII vesicles, which mediate transport of cargo between the ER and Golgi (Zanetti et al.,
2012). Using an in vitro COPII budding assay (Kim et al., 2005; Merte et al., 2010), we compared
levels of UNC93B1-WT and UNC93B1-H412R in purified vesicles. UNC93B1-WT, but not H412R, was
clearly present within the vesicles, further supporting a model in which UNC93B1 exits the ER through
the general secretory pathway (Figure 1H). Altogether, these data indicate that a pool of UNC93B1
protein exits the ER and traffics through the Golgi in unstimulated cells. Moreover, transit from the ER
to the Golgi may be important for UNC93B1 function, as the nonfunctional UNC93B1-H412R mutant
fails to enter COPII vesicles and does not reach the medial Golgi.
UNC93B1 facilitates TLR9 loading into COPII vesicles
Our previous work reported that three species of TLR9 can be detected within macrophages, rep-
resenting distinct maturation stages: an initial 150-kDa species with EndoH-sensitive glycans cor-
responding to the ER-resident protein (TLR9-ER), a larger species with EndoH-resistant glycans
corresponding to full-length receptor that has passed through the Golgi (TLR9-Precursor), and a
80-kDa band with EndoH-resistant glycans corresponding to the mature, cleaved receptor within
endolysosomes (TLR9-Cleaved) (Figure 2A, lane 1) (Ewald et al., 2008, 2011). To examine how
UNC93B1 function impacts TLR9 localization, we compared these three forms of TLR9 in immortalized
macrophages derived from Unc93b13d/3d mice and complemented with UNC93B1-WT or UNC93B1-
H412R. While all three bands were present in macrophages with functional UNC93B1, only the
ER-resident form was detectable in cells expressing the UNC93B1-H412R mutant (Figure 2A), consist-
ent with our previous analysis of TLR9 in UNC93B1 shRNA knockdown cells (Ewald et al., 2008).
These data indicate that TLR9 does not reach the medial Golgi in the absence of functional UNC93B1.
Because a pool of UNC93B1 can traffic from ER to Golgi by entering COPII vesicles (Figure 1B), we
considered whether UNC93B1 regulates this aspect of TLR9 trafficking. Indeed, analysis of TLR9 load-
ing into COPII vesicles revealed that TLR9 was only detectable in the presence of functional UNC93B1
whereas a control traffic protein, ERGIC/p58, was packaged independently of UNC93B1 (Figure 2B).
Quality control mechanisms ensure that only properly folded proteins can exit the ER, and one
mechanism by which UNC93B1 could regulate ER exit of TLR9 is through regulation of TLR9 folding,
as has been reported for gp96 (Yang et al., 2007). To address whether UNC93B1 serves as a folding
Cell biology | Immunology
Lee et al. eLife 2013;2:e00291. DOI: 10.7554/eLife.00291 4 of 22
IB: ERGIC p58
Pre PHG Pre PHG Pre PHG
Figure 1. UNC93B1 traffics to the Golgi en route to endolysosomes. (A) UNC93B1 is present in phagolysosomes of unstimulated cells. Phagosomes (PHG)
isolated by flotation from RAW264 cells only (Ø) or expressing GFP tagged UNC93B1-WT or UNC93B1-H412R and cells prior to isolation (Pre) were
separated by SDS-PAGE, and immunoblotted with anti-GFP, anti-LAMP1 (lysosome marker), and anti-calnexin (ER marker). (B) A portion of UNC93B1
protein traffics to the Golgi apparatus. Wildtype UNC93B1 (WT) or H412R, each with a C-terminal 3× FLAG tag, were expressed in HEK293Ts by transient
transfection. The immunoprecipitated proteins were treated with EndoH (E), PNGaseF (P) or left untreated (−), separated by SDS-PAGE, and visualized by
immunoblot with anti-FLAG antibody. Bands representing EndoH-sensitive (white arrow) and resistant (black arrow) forms of UNC93B1 are indicated.
(C)–(F) UNC93B1 acquires EndoH-resistant modifications. UNC93B1 tagged with GFP (C) or myc-His (D) from transiently transfected HEK293Ts, and FLAG
tagged UNC93B1 expressed in MEFs (E) or 3d iMac cells (F) were analyzed for the presence of EndoH-resistant glycans. Lysates were separated by SDS-PAGE
and immunoblotted with the indicated antibodies. EndoH-sensitive (white arrow) and EndoH-resistant (black arrow) forms are indicated. (G) Mutation
of UNC93B1 glycosylation sites abolishes EndoH resistant forms. Lysates from HEK293Ts transiently transfected with FLAG tagged UNC93B1-WT, -N251A
or -N251A/N272A were separated by SDS-PAGE and immunoblotted with anti-FLAG antibody. (H) UNC93B1 is loaded into COPII vesicles. Digitonin-
permeabilized COS7 cells expressing 3× FLAG-tagged UNC93B1-WT or UNC93B1-H412R, or no cells (Ø) were incubated with ATP regenerating system,
GTP, and rat liver cytosol, as indicated, in an in vitro COPII budding assay. Vesicles purified by ultracentrifugation were analyzed by SDS-PAGE and
immunoblot using the indicated antibodies. 20% of the COS7 cells prior to the budding reaction serves as a loading control (20% donor). ERGIC/p58
serves as a positive control for the formation of COPII vesicles. Results are representative of at least three experiments (A–G) or two experiments (H).
chaperone, we tested whether a chimeric CD4-TLR9 protein, consisting of the ectodomain of CD4
fused to the transmembrane and cytosolic regions of TLR9 (Figure 2C, left), required UNC93B1 func-
tion. Because trafficking of CD4 is not UNC93B1-dependent (Figure 2D), this chimera can be used to
test whether TLR9 requires UNC93B1 to ensure correct folding of the TLR9 ectodomain. CD4-TLR9
Cell biology | Immunology
Lee et al. eLife 2013;2:e00291. DOI: 10.7554/eLife.00291 5 of 22
Figure 2. UNC93B1 controls ER exit of TLRs 3, 7, 9, 11, and 13. (A) TLR9 fails to exit the ER in cells lacking functional UNC93B1. Lysates from 3d iMac
cells complemented with either UNC93B1-WT or UNC93B1-H412R and expressing TLR9-HA were analyzed by SDS-PAGE and immunoblotted with the
indicated antibodies. The precursor (black arrow), ER (white arrow) and cleaved (grey arrow) forms of TLR9-HA are indicated. (B) UNC93B1 is required for
TLR9 loading into COPII vesicles. RAW264 macrophages stably transduced with retroviruses encoding control or Unc93b1-directed shRNA and express-
ing TLR9-HA were used in an in vitro COPII budding assay as described in (Figure 1H). Lysates of purified vesicles or donor membranes were probed
with the indicated antibodies. (C) The transmembrane and cytosolic domain of TLR9 is sufficient to confer UNC93B1-dependence. (Left) schematic of
TLR9 and the CD4-TLR9 chimera. Transmembrane (TM), ectodomain (Ecto) and cytosolic domain (Cyto) are indicated. (Right) CD4-TLR9 was expressed in
HEK293Ts together with FLAG-tagged UNC93B1-WT or UNC93B1-H412R. Total lysates were analyzed by SDS-PAGE and immuno blotted with anti-CD4
and anti-FLAG antibodies. EndoH-sensitive (white arrow) and resistant (black arrow) forms are indicated. (D) CD4 trafficking to the cell surface is normal
in Unc93b13d/3d cells. Splenocytes from C57BL/6 (blue line) or Unc93b13d/3d (red line) mice were stained with anti-CD4 and analyzed by flow cytometry.
(E) CD4-TLR chimeric proteins for each of the indicated TLRs were expressed in HEK293Ts together with FLAG-tagged UNC93B1-WT or UNC93B1-H412R.
Lysates were separated by SDS-PAGE and visualized by immunoblot with anti-HA and anti-FLAG antibodies. EndoH-sensitive (white arrows) and resistant
(black arrows) forms are indicated. The chimeras were constructed as shown in Figure 1E, except with the addition of a C-terminal HA tag. Results are
representative of at least three experiments (A, C, and E) or two experiments (B and D).
acquired EndoH-resistant glycans when expressed with UNC93B1-WT but not when expressed with
mutant UNC93B1-H412R (Figure 2C, right). Thus, the requirement for UNC93B1 is not based on TLR9
ectodomain folding. Furthermore, the transmembrane domain and cytosolic regions of TLR9 are suf-
ficient to mediate UNC93B1-dependent trafficking.
Taken together, these data indicate that UNC93B1 regulates ER to Golgi transport of TLR9. While
we cannot rule out that a pool of UNC93B1 bypasses the Golgi en route to endosomes as suggested