You are on page 1of 17

Arteriosclerosis, Thrombosis, and Vascular Biology

BASIC SCIENCES

Endothelial HMGB1 Is a Critical Regulator of LDL


Transcytosis via an SREBP2–SR-BI Axis
Siavash Ghaffari, Erika Jang,* Farnoosh Naderinabi,* Rajiv Sanwal, Negar Khosraviani, Changsen Wang,
Benjamin E. Steinberg, Neil M. Goldenberg, Jiro Ikeda, Warren L. Lee

OBJECTIVE: LDL (low-density lipoprotein) transcytosis across the endothelium is performed by the SR-BI (scavenger receptor
class B type 1) receptor and contributes to atherosclerosis. HMGB1 (high mobility group box 1) is a structural protein in
the nucleus that is released by cells during inflammation; extracellular HMGB1 has been implicated in advanced disease.
Whether intracellular HMGB1 regulates LDL transcytosis through its nuclear functions is unknown.

APPROACH AND RESULTS: HMGB1 was depleted by siRNA in human coronary artery endothelial cells, and transcytosis of LDL was
measured by total internal reflection fluorescence microscopy. Knockdown of HMGB1 attenuated LDL transcytosis without
affecting albumin transcytosis. Loss of HMGB1 resulted in reduction in SR-BI levels and depletion of SREBP2 (sterol regulatory
element-binding protein 2)—a transcription factor upstream of SR-BI. The effect of HMGB1 depletion on LDL transcytosis
required SR-BI and SREBP2. Overexpression of HMGB1 caused an increase in LDL transcytosis that was unaffected by
inhibition of extracellular HMGB1 or depletion of RAGE (receptor for advanced glycation endproducts)—a cell surface receptor
for HMGB1. The effect of HMGB1 overexpression on LDL transcytosis was prevented by knockdown of SREBP2. Loss of
HMGB1 caused a reduction in the half-life of SREBP2; incubation with LDL caused a significant increase in nuclear localization
of HMGB1 that was dependent on SR-BI. Animals lacking endothelial HMGB1 exhibited less acute accumulation of LDL in the
aorta 30 minutes after injection and when fed a high-fat diet developed fewer fatty streaks and less atherosclerosis.
Downloaded from http://ahajournals.org by on November 17, 2022

CONCLUSIONS: Endothelial HMGB1 regulates LDL transcytosis by prolonging the half-life of SREBP2, enhancing SR-BI
expression. Translocation of HMGB1 to the nucleus in response to LDL requires SR-BI.

GRAPHIC ABSTRACT: A graphic abstract is available for this article.

Key Words: atherosclerosis ◼ cell nucleus ◼ cytoplasm ◼ endothelial cells ◼ humans

A
therosclerosis is a disease remarkable for its
See accompanying editorial on page 217
latency, beginning in the arteries of young adults
yet manifesting decades later. The accumulation of
LDL (low-density lipoprotein) cholesterol in the intimal circulation. We and others have observed that the
layer of arteries represents a first and pivotal step1 and scavenger receptor SR-BI (scavenger receptor class
implies that circulating LDL has crossed the endothelial B type 1)4 and the TGFβ family receptor ALK1 (activin
barrier. Early on in the disease, the overlying endothelium receptor-like kinase 1) mediate transcytosis,5 while the
is healthy2 and precludes the passage of LDL between high-affinity LDLR (LDL receptor) does not. The patho-
cells; thus LDL passes through individual endothelial logical importance of SR-BI–mediated LDL transcyto-
cells by transcytosis.3 sis to atherosclerosis has now also been established
Until recently, almost nothing was known about using murine models of the disease.6 Nonetheless, it
the mechanisms of LDL transcytosis in the systemic remains largely unclear how the process is regulated.

Correspondence to: Warren L. Lee, MD, PhD, Keenan Centre for Biomedical Research, St. Michael’s Hospital, 30 Bond St, Toronto, Ontario, Canada. Email warren.
lee@unityhealth.to
*These authors contributed equally to this article.
The Data Supplement is available with this article at https://www.ahajournals.org/doi/suppl/10.1161/ATVBAHA.120.314557.
For Sources of Funding and Disclosures, see page 215.
© 2020 American Heart Association, Inc.
Arterioscler Thromb Vasc Biol is available at www.ahajournals.org/journal/atvb

200   January 2021 Arterioscler Thromb Vasc Biol. 2021;41:200–216. DOI: 10.1161/ATVBAHA.120.314557
Ghaffari et al HMGB1 Is a Critical Regulator of LDL Transcytosis

Nonstandard Abbreviations and Acronyms Highlights

Basic Sciences - AL
HCAEC human coronary artery endothelial cell • Intracellular HMGB1 (high mobility group box 1)
HMGB1 high mobility group box 1 regulates LDL (low-density lipoprotein) transcyto-
LDL low-density lipoprotein sis via SREBP2 (sterol regulatory element-binding
protein 2) and SR-BI (scavenger receptor class B
LDLR low-density lipoprotein receptor
type 1).
ORO oil red O • Nuclear HMGB1 stabilizes SREBP2 mRNA.
PCR polymerase chain reaction • Mice deficient in endothelial HMGB1 develop fewer
RAGE receptor for advanced glycation fatty streaks and are protected from atherosclerosis.
endproducts
SR-BI scavenger receptor class B type 1
SREBP2 sterol regulatory element-binding protein 2 MATERIALS AND METHODS
TBST Tris-buffered saline-0.1% Tween-20 Data for the study can be obtained upon reasonable request
TIRF total internal reflection fluorescence to the corresponding author. Additional information on the
Methods can be found in the Major Resources Table in the
Data Supplement.
The alarmin HMGB1 (high mobility group box
1)—the most abundant nonhistone structural protein Cell Culture
in the nucleus—is expressed abundantly in the endo- Primary HCAECs were purchased from PromoCell (Heidelberg,
thelial cells of normal arteries and is found in high Germany) and maintained in EGM-2 (endothelial cell growth
amounts extracellularly in atherosclerotic lesions.7 It is medium-2) media with supplements (Lonza, Canada). Cells were
ubiquitously expressed, including by endothelial cells from 5 donors, 3 men (aged 27, 48, and 58 years) and 2 women
(aged 32 and 40 years). For all experiments, HCAECs were used
and leukocytes from which it is released by a poorly
at passage 5 or 6. All cells were incubated under humidified air
characterized secretory pathway8,9; it is also released containing 5% CO2 at 37 °C.
passively by dying cells. Neutralization of HMGB1
by antibodies attenuates atherosclerosis in ApoE−/− Isolation and Labeling of Human LDL
Downloaded from http://ahajournals.org by on November 17, 2022

mice,10 suggesting that a role for HMGB1 in advanced Blood was drawn from human volunteers after obtaining informed
atherosclerosis is likely. Whether HMGB1 is involved in consent and in accordance with a protocol approved by the
early disease—a period in which transcytosis of LDL is Institutional Research Ethics Board (19–314). Plasma was
critical—is unknown. obtained by centrifuging the blood at 2000g for 10 minutes at room
Another limitation of the existing literature is the temperature. EDTA (EDT001; BioShop) and KBr (POB301.1;
BioShop) were added to the plasma at a final concentration of 1
lack of distinction between intracellular and extracel-
mmol/L and 1.2 g/mL, respectively, and the mixture was mixed
lular HMGB1. While less studied, intracellular HMGB1 on a shaker until dissolved. The plasma mixture was overlaid with
has nuclear structural functions and regulates gene 0.3 mmol/L EDTA in PBS and centrifuged at 65 000 rpm for
expression.11 This is potentially important since intra- 1.5 hours at 15 °C in vacuum. After centrifugation, the orange
cellular HMGB1 has been reported to bind to and LDL layer was kept. LDL was concentrated by filtration through
a 10k Molecular Weight cutoff filter (UFC901008; Millipore) and
enhance the function of SREBP (sterol regulatory
centrifuging at 4000g for 1 hour at 4 °C. The concentrated LDL
element-binding protein) 1 and 2, transcription fac- was dialyzed in 0.3 mmol/L EDTA in PBS and filtered through a
tors involved in lipogenic and cholesterologenic gene 0.45-μm pore filter. Concentration was measured using the BCA
expression.12 Because of this, we hypothesized that in assay (No. 23227; ThermoFisher). LDL was stored in the dark at
early disease, intracellular HMGB1 in endothelial cells 4 °C for up to 3 months. To generate DiI-LDL, human LDL (2 mg)
might play an important role in regulating transcyto- was mixed with 300 μg DiI (No. 145311; Abcam) and 4 mL of
lipoprotein-deficient serum. The mixture was incubated at 37 °C
sis, over and above the contributions of extracellular
for 15 to 16 hours. After incubation, 0.5 g of KBr was added and
HMGB1, which may be derived from endothelial cells, mixed until dissolved. KBr solution (1.063 g/mL in 0.3 mmol/L
smooth muscle cells, or leukocytes.7 EDTA in PBS) was overlaid on top of the LDL mixture, and tubes
In this article, we report that nuclear HMGB1 enhances were centrifuged at 35 000 rpm for 24 hours at 4 °C in vacuum.
SREBP2 stability and SR-BI expression, thereby promot- The pink DiI-LDL layer was collected and dialyzed against 0.3
mmol/L EDTA in PBS and filtered through a 0.45-μm pore filter;
ing LDL transcytosis by primary human coronary artery
the final concentration was measured using the BCA assay. To
endothelial cells (HCAECs). Mice deficient in endothelial generate Alexa Fluor 568–labeled LDL, human LDL (2 mg in 1
HMGB1 develop significantly fewer fatty streaks and are mL of 0.3 mmol/L EDTA/PBS) was mixed with 100 μL of 1 M
protected from atherosclerosis. NaHCO3 and 100 μg of A568 Dye (Alexa Fluor 568 Succinimidyl

Arterioscler Thromb Vasc Biol. 2021;41:200–216. DOI: 10.1161/ATVBAHA.120.314557 January 2021   201
Ghaffari et al HMGB1 Is a Critical Regulator of LDL Transcytosis

Ester, No. A20003; ThermoFisher) in accordance with the manu- 37 °C for 2 minutes before initial image acquisition. Confluent
facturer’s instructions; this was followed by mixing in the dark at regions of the monolayer are selected by viewing the number
Basic Sciences - AL

room temperature for 1 hour. The mixture was dialyzed against of nuclei in the DAPI (4′,6-diamidino-2-phenylindole) field of
0.3 mmol/L EDTA/PBS. view after staining with NucBlue Live ReadyProbes Reagent
(ThermoFisher), and TIRF microscopy of the basal membrane
is performed to visualize exocytosis. For each coverslip, 10 to
Transfection 15 videos of 150 frames (100 ms exposure) are captured.
For siRNA transfections, a master mix was made containing A single-particle tracking algorithm developed in MATLAB
150 μL Opti-MEM (ThermoFisher, Canada), 4 μL Lipofectamine (R2014b; Mathworks) was used to quantify the number of
RNAiMAX Reagent (ThermoFisher), and 32 nmol/L siRNA. transcytosis events by analyzing the videos.4,13
siRNAs were Hs_SREBF2 (NM_004599; Qiagen, Canada), Glycyrrhizin—a direct inhibitor of extracellular HMGB114—
Hs_SCARB1 (NM_005505; Qiagen), and Hs_AGER was purchased from Sigma Canada and was diluted in DMSO.
(NM_001206934; Qiagen) and nontargeting control siRNA Cells were incubated with glycyrrhizin for 24 hours at 100
(1027310; Qiagen). A FlexiTube GeneSolution was used to µmol/L in EGM-2 before measurement of transcytosis.
knock down HMGB1 (GS3146, 1027416; Qiagen) and con- Similar to LDL, albumin transcytosis events were acquired
tains 4 different oligos: Hs_HMGB1_6 (S103650374), Hs_ using TIRF microscopy as described previously.15 Cells were
HMGB1_4 (S102627842), Hs_HMGB1_3 (S102627835), treated with 10 μg/mL Alexa Fluor 488 (AF488)–conjugated
and Hs_HMGB1_2 (S102627828). HCAECs were transfected albumin (ThermoFisher Scientific).
at 70% confluency in complete serum EGM-2, and media was
changed 24 hours after transfection. All experiments on cells
transfected with siRNA were performed 36 to 72 hours after Immunoblotting
transfection when the cells reached 100% confluency. Cell Cell lysates were collected by lysing cells in SDS-containing
lysates were collected after the experiment to determine the lysis buffer with 2 mmol/L sodium orthovanadate (Sigma,
degree of knockdown by immunoblot. In some experiments, Canada), 1 mmol/L PMSF (Sigma), 1X protease inhibitor cock-
total RNA was isolated from similar cells that were seeded on tail (Roche, Canada), and 1X PhosSTOP (Sigma). Lysates were
another plate and treated identically. resolved on 10% polyacrylamide gel and transferred onto a
The green fluorescent protein-HMGB1 wild-type construct nitrocellulose membrane and blocked for 1 hour in 5% BSA in
was a generous gift from Dr Brian Ratliff. Transfection of plas- Tris-buffered saline-0.1% Tween-20 (TBST). Primary antibod-
mid DNA was performed on HCAECs that were 80% confluent ies were diluted in 0.5% BSA in TBST with the working concen-
using Lipofectamine-3000 (ThermoFisher). A mix containing 1 tration of 1 µg/mL and incubated at 4 °C overnight on a rocker.
μg/mL plasmid DNA was added to Opti-MEM and then added Primary antibodies were mouse anti–β-actin (sc-47778; Santa
Downloaded from http://ahajournals.org by on November 17, 2022

to HCAECs in serum-free EGM-2. After 2 hours, media was Cruz, CA), rabbit anti–SR-BI (NB400-104; Novus Bio, Canada),
changed to complete serum EGM-2. Experiments involving rabbit anti–α-actinin (No. 3134; Cell Signaling, Danvers, MA),
plasmid-transfected cells were performed 24 hours after trans- rabbit anti-HMGB1 (3935; Cell Signalling, Canada), mouse
fection and at 100% confluency. anti-SREBP2 (NBP1-54446; Novus Bio), rabbit anti-CD36
(NB400-144; Novus Bio), and rabbit anti- caveolin1 (sc-894;
Santa Cruz, CA). After washing several times in TBST, mem-
LDL and Albumin Transcytosis Assay branes were incubated with HRP-conjugated secondary anti-
Total internal reflection fluorescence (TIRF) microscopy uses bodies (1 μg/mL) for 1 hour at room temperature followed
an evanescent wave to illuminate just the proximal 100 nm or by detection by enhanced chemiluminescence (Clarity ECL;
so of the cell; thus, we can selectively image the basal mem- BioRad, Canada) and imaging on a ChemiDoc Imaging System
brane of a live endothelial cell with minimal confounding from (BioRad). Protein abundance was quantified by densitometry
the overlying cytoplasm and apical surface. This technique obvi- (ImageJ) after normalization to loading control.
ates confounding from paracellular gaps and does not require
a high transfection efficiency.4 Essentially, confluent endothelial
monolayers are exposed to a fluorophore-tagged ligand added Immunofluorescence
to the apical cell surface while the basal membrane of the cell HCAECs were fixed/permeabilized in 100% ice-cold metha-
is imaged by TIRF. Cytoplasmic vesicles undergoing exocytosis nol at −20 °C for 10 minutes followed by washing 3× with
with the basal membrane are directly visualized and quantified. PBS. Cells were blocked with 5% BSA in PBS for 30 min-
TIRF microscopy was performed on a Leica DMi8 microscope utes at room temperature followed by immunodetection
with 63×/1.47 (O) objectives; 405, 488, 561, and 637 nm with primary antibodies diluted in PBS with the working
laser lines; 450/50, 525/50, 600/50, 610/75, and 700/75 concentration of 10 µg/mL. Primary antibodies were rabbit
emission filters; and run with Quorum acquisition software anti-HMGB1 (3935; Cell Signalling, Canada) and rabbit anti-
(Quorum, Canada). Microscope settings were kept constant caveolin1 (sc-894; Santa Cruz, CA). After a 1-hour incuba-
between conditions. tion at room temperature, cells were washed 3× in PBS and
Briefly, cells at 100% confluency were placed in a live treated with species-specific, chromophore-conjugated sec-
cell imaging chamber and treated with 20 μg/mL DiI-LDL in ondary antibody diluted 1:1000 (Jackson ImmunoResearch
cold HPMI (RMPI growth medium with HEPES) media for 10 Laboratories, West Grove, PA). Cells were then washed
minutes at 4 °C to allow apical membrane binding. Following and mounted on slides with DAPI-containing (1:5000)
membrane binding, cells are washed twice with cold PBS+ Dako Fluorescence Mounting Medium (Agilent, Canada).
to remove unbound ligand, and room temperature HPMI is Negative controls (no primary antibody) were included in all
added. Cells are incubated on the live cell imaging stage at experiments. Slides were imaged by spinning disk confocal

202   January 2021 Arterioscler Thromb Vasc Biol. 2021;41:200–216. DOI: 10.1161/ATVBAHA.120.314557
Ghaffari et al HMGB1 Is a Critical Regulator of LDL Transcytosis

microscopy performed on a Leica DMi8 microscope with set- Fluorescence Recovery After Photobleaching
tings kept constant between conditions.

Basic Sciences - AL
For measurement of HMGB1 translocation, cells were pre-
Analysis
HCAECs were seeded on 25-mm glass coverslips and trans-
treated with serum-free EGM-2 for 2 hours. Then unlabeled
fected with HMGB1-green fluorescent protein 2 days before
LDL (40 µg/mL) was added to cells, and cells were incubated
the experiment. On the day of experiment, unlabeled LDL (40
at 37 °C with 5% CO2 for 15 minutes. A MATLAB script was
µg/mL) was added to cells that were then incubated at 37 °C
used to quantify nuclear/cytoplasmic intensity. Briefly, the
with 5% CO2 for 15 minutes. Fluorescence recovery after pho-
image of nuclei in the DAPI channel was thresholded to obtain
tobleaching analyses was performed with cells maintained in
a binary image, and the whole field of view then divided into
EBM-2 growth medium at 37 °C with 5% CO2 supplied to the
nuclear and cytoplasmic area. Subsequently, the intensity of the
microscope chamber. Images were captured on a ZEISS LSM
HMGB1 channel was quantified in each region, and nuclear/
700 confocal microscope equipped with an environmental
cytoplasmic intensity was obtained. chamber (37 °C, 5% CO2) using a 63× 1.4 NA objective. ZEN
lite software was used to measure the intensity of fluorescence
RNA Isolation and Quantitative Polymerase in the bleached region of interest and in the whole cell at each
time point. Two prebleach images were captured with excitation
Chain Reaction
at 488 nm before photobleaching. The whole nucleus was then
Total RNA was isolated from 100% confluent HCAECs seeded
photobleached (100 ms, 100% laser power). Any remaining
on 6-well plates. For measurement of siRNA-mediated knock- fluorescence in the bleached area after the bleach was normal-
down efficiency by quantitative polymerase chain reaction ized to zero. Fluorescence recovery was recorded at 1-second
(PCR), RNA was isolated 48 hours after transfection. RNA iso- intervals for the period of 24 seconds. Error bars in the recovery
lation was performed with the Qiagen RNeasy RNA Isolation curves were drawn showing SEM. The mobile fractions were
Kit according to manufacturer’s instructions. cDNA synthesis calculated from the end point of these curves.
was performed on 1 μg total RNA using Applied Biosystems
High Capacity cDNA Reverse Transcription Kit. Quantitative
PCR was performed on an Applied Biosystems QuantStudio mRNA Stability Assay
7 Flex Real-Time PCR system: (1) 95 °C for 3 minutes, (2) 95 HCAECs were transfected with the HMGB1 siRNA or scram-
°C for 15 seconds, (3) 58 °C for 30 seconds, (4) repeat from bled negative control siRNA. After 24 hours, cells were treated
2 an additional 39×, (5) melt curve from 65 °C to 95 °C at with 5 μg/mL actinomycin D (Bishop, Canada) diluted in the
0.5 °C increments. mRNA abundance is presented relative to media to inhibit transcription. Total RNA was isolated just after
18S, which is a reference gene for our experiment. Primers for treatment with actinomycin D and once every 2 hours for 8
Downloaded from http://ahajournals.org by on November 17, 2022

quantitative PCR are listed in Table 1, and the sequences were hours. Subsequently, cDNA was synthesized, and real-time
provided by PrimerBank. quantitative PCR was performed. The data were fitted to a lin-
ear regression, and the regression was used to calculate the
half-life of the mRNA.
Table 1. Primers for Quantitative Polymerase Chain Reaction

cDNA Forward primer (5′–3′) Reverse primer (5′–3′) Protein Half-Life Analysis
18S GATGGAAAATACAGCCAG- TTCTTCAGTCGCTCCAG- HCAECs were transfected with the HMGB1 siRNA or scram-
GTCCTA GTCTT bled negative control siRNA. After 48 hours, cells were treated
HMGB1 TATGGCAAAAGCGGA- CTTCGCAACATCAC- with 50 μg/mL cycloheximide (Bishop) to inhibit protein syn-
CAAGG CAATGGA thesis. Proteins were harvested at different times, and Western
ALK1 CGAGGGATGAA- GTCATGTCTGAGGCGAT- blot analysis was performed as described above. The data were
CAGTCCTGG GAAG fitted to a linear regression, and the regression was used to
SR-BI CTGTGGGTGAGATCAT- GCCAGAAGTCAACCTT- calculate the half-life of the protein.
GTGG GCTC
LDLR ACGGCGTCTCTTCCTAT- CCCTTGGTATCCGCAA-
GACA CAGA Mice
CD36 TCTTTCCTGCAGCCCAATG AGCCTCTGTTCCAACT-
Male mice with an endothelial cell–specific deletion of HMGB1
GATAGTGA (EC-HMGB1−/−; Fl/FlHMGB1 Tie2Cre+) were generated by
ICAM-1 ATGCCCAGACATCTGT- GGGGTCTCTATGCCCAA-
crossing C57BL/6J mice expressing Cre under the Tie2 pro-
GTCC CAA moter with C57BL/6J mice containing floxed HMGB1 (the
CAV-1 GCGACCCTAAACACCT- ATGCCGTCAAAACTGTGT-
latter a gift from Tadatsugu Taniguchi, University of Tokyo).
CAAC GTC HMGB1 floxed C57BL/6J mice not expressing Cre were
RAGE GTGTCCTTCCCAACGGCTC ATTGCCTGGCACCGGAAAA
used as controls. To generate HMGB1 knockout mice on an
LDLR−/− background, male EC-HMGB1−/− mice were crossed
SREBP2 AACGGTCATTCACCCAG- GGCTGAAGAATAGGAGTT-
GTC GCC
with female LDLR−/−, and the progeny were then crossed to
obtain EC-HMGB1−/−LDLR−/− mice. LDLR−/− mice with floxed
ALK1 indicates activin receptor-like kinase 1; CAV-1, caveolin-1; CD36, clus- HMGB1 not expressing Cre were used as controls. Genotyping
ter of differentiation 36; HMGB1, high mobility group box 1; ICAM-1, intercellular
adhesion molecule 1; LDLR, low-density lipoprotein receptor; RAGE, receptor for
was performed by ear notching and PCR using the KAPA
advanced glycation endproducts; SR-BI, scavenger receptor class B type 1; and Mouse Genotyping Kit (No. KK7301; Kapa Biosystems) and
SREBP2, sterol regulatory element-binding protein 2. primers for floxed HMGB1, LDLR, and Tie2Cre (Table 2).

Arterioscler Thromb Vasc Biol. 2021;41:200–216. DOI: 10.1161/ATVBAHA.120.314557 January 2021   203
Ghaffari et al HMGB1 Is a Critical Regulator of LDL Transcytosis

Table 2. Primers for Floxed HMGB1, LDLR, and Tie2Cre face submerged in PBS. Photographs were taken with a Nikon
camera mounted to a dissecting microscope. Quantification
Basic Sciences - AL

DNA Forward primer (5′–3′) Reverse primer (5′–3′)


was performed on the images in ImageJ by setting the same
HMGB1 TGT CAT GCC ACC CTG TGT GCT CCT CCC GGC
color threshold to all images and then quantifying the area of
AGC AGT T AAG TT
positive pixels. Plaque area of knockout mice was normalized to
CRE GCT CGA CCA GTT TAG TCG CGA TTA TCT TCT ATA
plaque area of cage mate control mice.
TTA CCC TCT TCA G
LDLR TAT GCA TCC CCA GTC CTA CCC AAC CAG CCC
TTT GG CTT AC (wild-type) ORO Staining for Aortic Roots
ATA GAT TCG CCC TTG TGT After fixation, aortic roots were frozen in OCT on dry ice then
CC (mutant) stored at −80 °C. Ten-micrometer cryosections were mounted
CRE indicates Cre recombinase; HMGB1, high mobility group box 1; LDLR, onto Superfrost Plus (12-550-15, Fisherbrand) microscope
low-density lipoprotein receptor; and Tie2Cre, Cre recombinase expressed under glass slides at 70-μm intervals starting from the bottom of the
the Tie2 promoter. cusps until 8 sections were obtained per slide. Sections were
fixed in 10% formalin for 10 minutes and washed in ddH2O for
Starting at 5 weeks old, EC-HMGB1−/− mice and their con- 5 min. Sections were incubated in 60% isopropanol for 1 min-
trols were fed the Paigen high-fat, high-cholesterol diet con- ute, stained in ORO (No. O0625; Sigma-Aldrich) working solu-
taining 15% cocoa butter, 1% corn oil, 1.25% cholesterol, and tion for 12 minutes, destained in 60% isopropanol for 1 minute,
0.5% cholic acid (Dyets 611695) for 18 weeks.16 Starting at and washed in ddH2O for 30 seconds. Hematoxylin was used
10 weeks old, EC-HMGB1−/−; LDLR−/− mice and their controls as a counterstain for 1 minute, followed by washing in ddH2O
were fed a high-fat, high-cholesterol diet (D12108C; Research for 10 minutes. Coverslips were mounted on slides with aqua-
Diets) for 5 weeks. Mice were maintained in the St. Michael’s tex aqueous mounting media (Merck). Slides were imaged on
Hospital Vivarium on a standard light-dark cycle and were pro- the Zeiss Axioscan.Z1 microscope with the 20× 0.8 NA objec-
vided access to food and water ad libitum. Experiments were tive lens with the same settings for all samples. Images were
performed in accordance with the Animal Care Committee quantified using Halo analysis software (Indica Labs) with set-
guidelines at St. Michael’s Hospital and followed approved ani- tings kept constant between groups.
mal protocols (ACC899) and the American Heart Association
guidelines for experimental atherosclerosis studies.17 Although
endothelial cells from both male and female donors were
Evans Blue Dye Assay
included in the study, the animal experiments were limited to At experimental end point, EC-HMGB1−/− mice and controls
male mice due to cost. Follow-up work will include the study on the high-fat diet were injected via retro-orbital vein with
1% Evans blue dye (206334; Millipore Sigma) dissolved in
Downloaded from http://ahajournals.org by on November 17, 2022

of female mice.
sterile PBS and then euthanized after 30 minutes by exsan-
guination. The mice were perfused and aortic arches were
Blood Collection cleaned as described above. Evans blue dye was extracted
Plasma lipid levels were measured at 3 time points: before from aortic arches by incubating in formamide at 56 °C for 4
starting the high-fat diet, during the diet, and at the time of days before measuring absorbance at 620 nm. Absorbance
euthanasia. Before blood collection, mice were fasted for 16 measurements from EC-HMGB1−/− mice were normalized to
hours. Blood was collected via the saphenous vein and mixed that of cage mate control mice.
in tubes containing K2EDTA (BD Microtainer with K2EDTA, No.
365974). Plasma was obtained by centrifuging whole blood at
5000×g at 8 °C, for 10 minutes. Plasma was stored at −80 °C
Analysis of LDL Transcytosis In Vivo
before being sent to The Centre for Phenogenomics (Toronto, To measure LDL transcytosis in vivo, we adapted a published
Ontario) for analysis of total cholesterol, LDL, and triglycerides. method.5 Briefly, EC-HMGB1−/− mice and EC-HMGB1−/−;
LDLR−/− mice or littermate controls on a standard laboratory
diet (Teklad 2918) were anesthetized and injected via the
Oil Red O Staining retro-orbital vein with 200 μg of Alexa Fluor 568-LDL; Alexa-
At experimental end point, mice on the high-fat diet were euth- conjugated LDL was used as it is fixable and the fluorophore
anized by exsanguination. A syringe was inserted into the left is conjugated covalently to LDL. After 30 minutes, mice were
ventricle, and the mouse was perfused with PBS and then 10% euthanized by exsanguination; the short time point was chosen
buffered formalin. The aorta attached to the heart was fixed in to focus on endothelial permeability to LDL and to minimize
10% formalin for 1 hour. Using a dissecting microscope, the confounding from leukocytes. A syringe was inserted into the
adventitial tissue was removed, and the arch was separated left ventricle, and the mouse was perfused with PBS and then
from the aortic root, the descending aorta, and the aortic arch 10% buffered formalin. The aorta attached to the heart was
branches. Oil red O (ORO; No.O0625; Sigma-Aldrich) stock fixed in 10% formalin for 1 hour. Using a dissecting micro-
solution was prepared by dissolving ORO (2.58 g) in 500 mL of scope, the adventitial tissue was removed, and the arch was
isopropanol. The working solution was prepared by diluting the separated from the aortic root, the descending aorta, and the
stock solution in double-distilled water (3:2) and then centrifug- aortic arch branches. Aortas were permeabilized for 5 minutes
ing at 2000×g for 10 minutes. The arch was stained with ORO at room temperature in 0.5% Triton X-100 (TRX506.500;
working solution for 30 minutes, destained in 60% isopropanol BioShop) and 10% DMSO (DMS666.100; BioShop) in PBS.
for 5 minutes twice, and then washed in PBS. The arch was cut They were washed in PBS 3× for 5 minutes. Aortas were incu-
longitudinally along the greater curvature and pinned open en bated with 1:100 (10 μg/mL) rabbit anti-mouse CD31 primary

204   January 2021 Arterioscler Thromb Vasc Biol. 2021;41:200–216. DOI: 10.1161/ATVBAHA.120.314557
Ghaffari et al HMGB1 Is a Critical Regulator of LDL Transcytosis

antibody (NB100-2284; Novus) in 5% BSA in PBS overnight Tween-20, pH 6.0) for antigen retrieval. Slides were washed
at 4 °C. After washing, they were incubated with 1:200 (7.5 μg/ in TBST twice for 5 minutes, followed by permeabilization with

Basic Sciences - AL
mL) anti-rabbit Alexa Fluor 488 secondary antibody (Jackson PBS-1% Triton X-100 for 30 minutes at room temperature.
ImmunoResearch) in 5% BSA in PBS for 1 hour at room tem- Samples were circled using a hydrophobic pen and blocked
perature. After washing, aortas were incubated in DAPI (1:400 using 10% goat serum with 1% BSA in 1X TBS for 1 hour.
in PBS; No. 62248; Thermo Scientific) for 20 minutes at room
Slides were washed 3× for 2 minutes, followed by blocking with
temperature. Aortas were washed in PBS 5× for 3 minutes
AffiniPure Fab Fragment Goat Anti-Mouse IgG (115-007-003;
and then mounted on slides with Dako fluorescence mounting
medium (S3023; Agilent). Slides were imaged with a ZEISS Jackson ImmunoResearch) for 1 hour at room temperature.
LSM700 confocal microscope using a 20× 0.8 NA objective Slides were washed once with TBST and incubated overnight
lens and acquired with the ZEN Black software; settings were at 4° with primary antibody: 1:50 (4 μg/mL) rabbit anti-mouse
kept the same between groups. Using ImageJ, background Tie-2 (H-176, sc-9026; Santa Cruz Biotechnology), 1:250 (4
fluorescence was subtracted, and the integrated density of the μg/mL) rabbit anti-mouse AQP5 (ab78486; Abcam), or 1:200
Alexa Fluor 568-LDL signal was quantified and normalized to (5 μg/mL) mouse anti-HMGB1 (ab190377; Abcam). Slides
tissue area and then to control mice. were washed twice with TBST for 5 minutes and incubated
with conjugated secondary antibody Cy3-conjugated goat anti-
Flow Cytometry and Histology mouse IgG (115-165-003; Jackson ImmunoResearch) and
Mice on a standard laboratory diet (Teklad 2918) ranging from Alexa Fluor-647–conjugated goat anti-rabbit IgG (111-605-
5 to 10 weeks of age were euthanized via exsanguination by 003; Jackson ImmunoResearch) for 1 hour at room tempera-
cardiac puncture. Blood was collected in tubes containing a final ture. Slides were counterstained with DAPI.
concentration of 10% v/v of 0.5M EDTA, pH 8. Blood was incu- For heart sections, hearts were flushed with PBS and flash
bated with 1× red blood cell lysis buffer (1:10 blood to RBC lysis frozen using isopentane for 20 seconds before being stored
buffer; 10X RBC lysis buffer: 0.155 M NH4Cl; 0.01 M KHCO3; at −80 °C for further processing. Serial 5-μm cryosections
0.1 mmol/L EDTA; dissolved in ddH2O and filter sterilized) on
were mounted onto Superfrost microscope glass slides and
a shaker at room temperature for 5 minutes. Blood was cen-
were allowed to dry for 10 minutes. Sections were fixed and
trifuged at 300g for 5 minutes at 4 °C. Supernatant was dis-
carded. Cells were washed in 2% FBS in PBS and spun at 500g permeabilized with methanol at −20 °C for 20 minutes and
for 5 minutes at 4 °C. Cells were counted on a hemocytometer then washed with PBS +0.1% Tween-20 3× for 5 minutes.
and resuspended at a concentration of 1 to 2×106 cells/mL. After blocking and washing, slides were incubated with primary
Cells were stained with 1:100 (5 μg/mL) Alexa Fluor 488 anti- antibody overnight at 4 °C 1:100 (10 μg/mL) mouse anti-
Downloaded from http://ahajournals.org by on November 17, 2022

CD115 (No. 135511; BioLegend), 1:100 (5 μg/mL) Alexa Fluor HMGB1 (ab190377; Abcam). Excess primary antibody was
647 anti-Ly6G (No. 127609; BioLegend), and 1:1000 Fixable washed off and secondary antibody applied as with lung sec-
Viability stain 780 (No. 565388; BD Bioscience) in 2% FBS in tions. After washing, the next primary antibodies were applied
PBS for 30 minutes on ice. After washing, cells were fixed with for 1 hour at room temperature: 4 μg/mL rabbit anti-mouse
2% PFA (No. 15710; Electron Microscopy Sciences) in PBS for Tie-2 (H-176; Santa Cruz Biotechnology) or 1:100 (4 μg/mL)
15 minutes on ice. After washing, cells were permeabilized with
rabbit anti-mouse cardiac troponin I (ab47003; Abcam). These
0.1% Triton X-100 in PBS for 15 minutes on ice. After washing,
were labeled with appropriate secondary antibody for 1 hour.
cells were stained with 1:25 (36 μg/mL) Alexa Fluor 405 anti-
HMGB1 antibody (NB100-2322AF405; Novus) in 2% FBS in Finally, the slides were coated with 20 μL per section of DAPI-
PBS for 15 minutes on ice. Cells were washed and then filtered containing (1:5000) Dako fluorescence mounting medium
through a 0.70-μm filter. Single-stain controls were prepared for (S3023; Agilent), and coverslips were added. For both lung and
each antibody with the UltraComp eBeads Plus Compensation heart sections, a no-primary-antibody control was performed
beads (01-3333-42; Invitrogen), and the single stain for the (ie, incubation with secondary antibody only).
viability dye was prepared with cells that were half dead and
half live. Fluorescence-minus-one controls were prepared for
each group. Samples were analyzed by the Sick Kids/University Statistics
Health Network Flow Cytometry Facility in Toronto, Canada. Cells Unless indicated otherwise, each experiment was performed at
were gated for single cells in the forward scatter and then side least 4× independently. Statistical tests were performed using
scatter channels, then gated for live cells using the gating from a GraphPad Prism software, version 6.0 (La Jolla, CA); data were
live/dead control sample. Positive CD115 or Ly6G was graphed tested for normality and equal variance to confirm appropri-
against forward scatter. The positive populations were then ateness of parametric tests. For experiments with 2 indepen-
graphed as HMGB1 versus forward scatter to obtain double- dent groups, P values were determined by paired or unpaired
positive CD115+/HMGB1+ or Ly6G+/HMGB1+ populations.
Student t tests where appropriate. P<0.05 was deemed statis-
For lung histology, mice were euthanized via cardiac punc-
tically significant. For experiments with >2 independent groups,
ture. Lungs were perfused with 600 μL of 10% formalin, dis-
sected and fixed in 10% formalin for 2 days, followed by tissue P values were determined by 1-way ANOVA with significance
processing and embedding in paraffin wax. The tissue was cut set at P<0.05; individual comparisons between groups were
in serial 5-μm-thick sections and mounted on glass slide. The determined by Tukey post hoc test. For the time point experi-
sections were deparaffinized and then boiled for 20 minutes ments, P values were corrected for multiple comparisons using
in sodium citrate buffer (10 mmol/L sodium citrate, 0.05% the Holm-Sidak method.

Arterioscler Thromb Vasc Biol. 2021;41:200–216. DOI: 10.1161/ATVBAHA.120.314557 January 2021   205
Ghaffari et al HMGB1 Is a Critical Regulator of LDL Transcytosis
Basic Sciences - AL
Downloaded from http://ahajournals.org by on November 17, 2022

Figure 1. HMGB1 is required for LDL transcytosis by human coronary artery endothelial cells.
A, HMGB1 (high mobility group box 1) was knocked down by siRNA in human coronary artery endothelial cells (HCAECs) achieving a
reduction of 80% to 90% in protein levels. Representative blots (left); quantification (right). n=4 independent experiments; ***P<0.001 by
t test. Error bars represent SEM. B, Transcytosis of DiI-tagged LDL (low-density lipoprotein) across confluent HCAEC from male donors is
measured in real time using total internal reflection fluorescence microscopy. Knockdown of HMGB1 led to a decrease in LDL transcytosis;
each point represents 1 cell. n=4 experiments; ***P<0.001 by t test; error bars represent SD. C, Knockdown of HMGB1 led to a decrease in
LDL transcytosis across confluent cells from female donors; each point represents 1 cell. n=4 experiments; ***P<0.001 by t test; error bars
represent SD. D, Knockdown of HMGB1 had no effect on albumin transcytosis. Transcytosis of Alexa 488–albumin across confluent cells from
a male donor was measured; each point represents 1 cell. n=4 experiments; error bars represent SD. E, Knockdown of HMGB1 in human
coronary artery endothelial cells had no effect on transendothelial electrical resistance (ER); PAF (platelet-activating factor) is a positive control.
n=4; ***P<0.001 by 1-way ANOVA; ns indicates nonsignificant by Tukey multiple comparison test. SCRM indicates non-targeting control RNA.

206   January 2021 Arterioscler Thromb Vasc Biol. 2021;41:200–216. DOI: 10.1161/ATVBAHA.120.314557
Ghaffari et al HMGB1 Is a Critical Regulator of LDL Transcytosis

RESULTS HMGB1 Regulates LDL Transcytosis via

Basic Sciences - AL
SREBP2
HMGB1 Is Required for LDL Transcytosis by
Given the role of HMGB1 as a nuclear structural protein,
HCAECs
we wondered whether the connection between intracel-
Primary HCAECs were depleted of HMGB1 by siRNA, lular HMGB1 and SR-BI expression reflected transcrip-
resulting in ≈80% to 90% reduction in protein levels (Fig- tional regulation of the receptor. Knockdown of HMGB1
ure 1A). Loss of HMGB1 caused a significant inhibition
caused a significant reduction in protein levels of both
of LDL transcytosis by cells from both male and female
mature and immature (precursor) SREBP2 protein (Fig-
donors (Figure 1B and 1C). The effect was not due to
ure 3A), as well as mRNA (Figure 2B). SREBP2 is a
a reduction in cell viability or generalized impairment in
critical transcription factor that regulates expression of
vesicular traffic as albumin transcytosis was unaffected
cholesterologenic genes, including LDLR and SR-BI.19
(Figure 1D). Finally, depletion of HMGB1 had no effect
While depletion of either HMGB1 or of SREBP2 signifi-
on endothelial barrier function in vitro as measured by
cantly attenuated LDL transcytosis, combinatorial knock-
transendothelial electrical resistance (Figure 1E).
down had no additional effect (Figure 3B and 3C). In
addition, the effect of HMGB1 overexpression to induce
Deficiency of HMGB1 Leads to Downregulation LDL transcytosis required SREBP2 as it was prevented
of SR-BI But Not ALK1 by knockdown of the transcription factor (Figure 3D).
The inhibition of LDL but not albumin transcytosis by
depletion of HMGB1 made us consider whether the HMGB1 Translocates to the Nucleus in an LDL-
mechanism involved a receptor. Knockdown of HMGB1
and SR-BI–Dependent Manner and Regulates
resulted in a significant reduction in SR-BI expression
at both the protein (Figure 2A) and the mRNA levels SREBP2 Half-Life
(Figure 2B). However, levels of ALK1, caveolin-1, and We examined the cellular localization of HMGB1 by immu-
CD36 were unaffected (Figure 2B and 2C), and the sub- nofluorescence before and after coronary artery endothe-
cellular distribution of caveolin-1 was also unchanged lial cells were incubated with LDL. As expected, at baseline,
Downloaded from http://ahajournals.org by on November 17, 2022

(Figure I in the Data Supplement). Interestingly, mRNA most HMGB1 was localized to the nucleus. However, upon
for SREBP2 and LDLR were significantly reduced in addition of LDL, we observed a significant increase in the
HMGB1-depleted cells. ratio of nuclear-to-cytoplasmic HMGB1 (Figure 4A). This
These data suggested that the effect of HMGB1 deple- effect was partially inhibited by knockdown of SR-BI (Fig-
tion on LDL transcytosis might be through regulation of ure 4A) but not of LDLR (not shown). To verify this finding
SR-BI. To test this notion, we depleted both HMGB1 and by a complementary approach, we transfected endothe-
SR-BI in coronary artery endothelial cells. While knock- lial cells with green fluorescent protein–tagged HMGB1
down of either SR-BI or HMGB1 significantly inhibited
and performed fluorescence recovery after photobleach-
LDL transcytosis, combinatorial depletion of both proteins
ing experiments of nuclear HMGB1 in the presence or
did not achieve any further reduction, consistent with a
absence of LDL in the media. The addition of LDL sig-
shared pathway (Figure 2D). To explore the contribution of
nificantly increased the recovery of nuclear HMGB1 after
HMGB1 to LDL transcytosis by an alternative approach,
photobleaching (Figure 4B). Theorizing that HMGB1 lev-
we overexpressed HMGB1 in coronary endothelial cells
els in endothelial cells might be regulated by sterols, we
by transfection. Transfected cells displayed a significant
loaded endothelial cells with LDL and measured HMGB1
increase in LDL transcytosis (Figure 2D). To determine
whether this effect was mediated by intracellular versus protein levels 24 hours later. Incubation with LDL caused
extracellular (ie, secreted) HMGB1, we treated trans- a significant increase in total HMGB1 expression (Fig-
fected cells with glycyrrhizin14—an inhibitor of HMGB1 ure 4C). We then speculated that one of the functions of
and measured LDL transcytosis. We also treated trans- nuclear HMGB1 might be to stabilize SREBP2, prolonging
fected cells with siRNA to RAGE (receptor for advanced its half-life and, therefore, its activity upstream of SR-BI.
glycation endproducts)—the main cell surface receptor for We measured the half-life of SREBP2 mRNA and protein
HMGB1.18 Neither of these interventions attenuated the in control cells and in cells depleted of HMGB1, using acti-
effect of HMGB1 overexpression on LDL transcytosis nomycin to inhibit transcription and cycloheximide to inhibit
(Figure 2E). Because of this, we decided to focus on iden- protein synthesis. In cells depleted of HMGB1, the half-
tifying a potential intracellular pathway by which endothe- life of SREBP2 mRNA and protein was reduced by ≈50%
lial HMGB1 might regulate LDL transcytosis. (Figure 4D and 4E).

Arterioscler Thromb Vasc Biol. 2021;41:200–216. DOI: 10.1161/ATVBAHA.120.314557 January 2021   207
Ghaffari et al HMGB1 Is a Critical Regulator of LDL Transcytosis
Basic Sciences - AL
Downloaded from http://ahajournals.org by on November 17, 2022

Figure 2. Deficiency of HMGB1 leads to down-regulation of SR-BI but not ALK1.


A, Knockdown of HMGB1 (high mobility group box 1) in human coronary artery endothelial cells by siRNA led to a decrease in SR-BI
(scavenger receptor class B type 1) protein levels. n=4; *P<0.05 by t test; error bars represent SEM. B, mRNA level by quantitative
polymerase chain reaction in HMGB1-depleted cells. n=4; *P<0.05 for LDLR (LDL [low-density lipoprotein] receptor), **P<0.01 for SR-BI and
SREBP2 (sterol regulatory element-binding protein 2); error bars represent SEM. C, Depletion of HMGB1 by siRNA had no effect on levels
of CAV-1 (caveolin-1) or CD36. Blot is representative of 4 experiments, and quantification is normalized to cells receiving nontargeting control
RNA (SCRM). D, Left, Knockdown of either HMGB1 or SR-BI significantly inhibited LDL transcytosis, but depletion of both did not achieve
any further reduction; each point represents 1 cell. n=4 experiments; ***P<0.001 by 1-way ANOVA. Right, Cells were transfected (Continued )

208   January 2021 Arterioscler Thromb Vasc Biol. 2021;41:200–216. DOI: 10.1161/ATVBAHA.120.314557
Ghaffari et al HMGB1 Is a Critical Regulator of LDL Transcytosis

Mice Deficient in Endothelial HMGB1 Develop acid. As cholate can be hepatotoxic,23 this may confound
interpretation of the results. To study the contribution of

Basic Sciences - AL
Fewer Fatty Streaks on a High-Fat Diet
endothelial HMGB1 to more advanced atherosclerotic dis-
To validate our data from cell culture, we generated EC- ease and without the confounder of dietary cholic acid, we
HMGB1−/− mice using a Tie2Cre promoter.20 By histology generated EC-HMGB1−/− mice on a LDLR−/− background.
of lung and heart sections, we observed complete loss of As before, we injected these double knockout mice and
endothelial HMGB1 but retention of the protein in the adja- LDLR−/− littermate controls with Alexa-568–tagged LDL
cent epithelium and myocardium. Expression of HMGB1 and measured its accumulation in the inner curvature of
in circulating monocytes and neutrophils (as well as the the aortic arch 30 minutes later. Far less Alexa-685–LDL
absolute number of these leukocytes) was not significantly entered the aorta of EC-HMGB1−/−:LDLR−/− mice than
reduced in the knockout animals (Figure II in the Data LDLR−/− controls (Figure 6A). Double knockout mice and
Supplement) despite literature describing expression of LDLR−/− littermate controls were then fed a high-fat diet
Tie2 by hematopoietic cells.21,22 To determine whether loss for 5 weeks. EC-HMGB1−/−:LDLR−/− mice developed sig-
of endothelial HMGB1 reduced LDL transcytosis in vivo, nificantly less disease than controls (Figure 6B and 6C)
we injected EC-HMGB1−/− and littermate control animals despite similar or even slightly higher circulating LDL and
with fluorophore-tagged (Alexa-568) LDL and measured triglyceride levels (Figure IV in the Data Supplement).
its accumulation 30 minutes later in the inner curvature
of the aortic arch.5 We observed a significant reduction in
LDL accumulation in the knockout animals (Figure 5A); the DISCUSSION
reduction over such a short period of time is consistent
Although the existence of endothelial transcytosis has
with impairment of endothelial transcytosis of LDL.
been postulated for decades24 until recently, its mecha-
We next assessed the effect of HMGB1 deletion on
nisms and pathological importance have remained con-
the earliest stage of atherosclerosis—the formation of
troversial. Technical limitations have hindered its study:
fatty streaks. Mice deficient in endothelial HMGB1 and morphological experiments using electron microscopy
wild-type littermates were fed a high-fat diet for 18 weeks provided elegant images but were not feasible for mech-
followed by staining of the inner curvature of the aorta anistic studies. Experiments using animals were limited
with ORO. As expected, not all mice (≈50%) developed by gaps in our understanding of the regulation of trans-
fatty streaks, and these were modest in size.16 However, cytosis, as targets for gene knockout were unclear.
EC-HMGB1−/− mice displayed significantly smaller fatty
Downloaded from http://ahajournals.org by on November 17, 2022

Recent advances in live cell imaging using TIRF


streaks in the aortic arch compared with controls (Fig- microscopy4 and sensitive studies using radionuclear iso-
ure 5B), whereas there was no difference in the aortic topes25 have now elucidated some of the earliest stages
root (Figure 5C). This reduction in neutral lipid accumu- of transcytosis. Two receptors have been identified (SR-
lation in the aortic arch and the decrease in fluorescent BI and ALK15), and there is a proven pathological role for
LDL uptake in the aorta (ie, Figure 5A) were not due to endothelial SR-BI–mediated LDL transcytosis in athero-
enhanced endothelial barrier function in EC-HMGB1−/− genesis.6 Determining the regulation of SR-BI–mediated
animals as permeability to albumin (a smaller molecule transcytosis is now a priority.
than LDL15) was not significantly altered (Figure 5D). Our data demonstrate that endothelial HMGB1 plays
The effect was not attributable to lower cholesterol levels an unsuspected role upstream of SR-BI. While HMGB1
as deficiency of endothelial HMGB1 did not significantly is best studied as an alarmin—a protein with cytokine-
affect total cholesterol or LDL cholesterol at baseline or like properties that is secreted or released by dying
on the high-fat diet. Triglyceride levels were also similar cells26,27—it is also an abundant and ubiquitous nuclear
between groups (Figure III in the Data Supplement). protein. Here, we show that HMGB1 translocates to
the nucleus in an LDL- and SR-BI–dependent manner
Mice Lacking Endothelial HMGB1 in the where it regulates the half-life of SREPB2. This in turn
regulates SR-BI expression and LDL transcytosis. The
Setting of LDLR Deficiency Are Protected From
effect of HMGB1 depletion on LDL transcytosis was not
Atherosclerosis due to nonspecific impairment of cell health or vesicu-
Although informative about the earliest stages of dis- lar traffic as albumin transcytosis was unaffected. Fur-
ease, a drawback of using wild-type mice is that they only thermore, while it is well established that caveolin-1 is a
develop lesions on a high-fat diet that contains cholic critical regulator of LDL (and albumin) transcytosis,15,28,29

Figure 2 Continued. with GFP alone or GFP-HMGB1. Overexpression of HMGB1 significantly increased LDL transcytosis; each point
represents 1 cell. n=4 experiments; ***P<0.001 by t test; error bars represent SD. E, Neither glycyrrhizin (GCZ) nor RAGE (receptor for
advanced glycation endproducts) siRNA blocked the effect of HMGB1 overexpression on LDL transcytosis. Cells were transfected with GFP
alone or GFP-HMGB1 before treatment with GCZ; each point represents 1 cell. n=4 experiments; ***P<0.001 by 1-way ANOVA; error bars
represent SD; ns indicates nonsignificant by Tukey multiple comparison test. GFP indicates green fluorescent protein.

Arterioscler Thromb Vasc Biol. 2021;41:200–216. DOI: 10.1161/ATVBAHA.120.314557 January 2021   209
Ghaffari et al HMGB1 Is a Critical Regulator of LDL Transcytosis
Basic Sciences - AL
Downloaded from http://ahajournals.org by on November 17, 2022

Figure 3. HMGB1 regulates LDL transcytosis via SREBP2.


A, HMGB1 (high mobility group box 1) knockdown in human coronary artery endothelial cells by siRNA led to a decrease in both mature (m)
and precursor (p) forms of SREBP2 (sterol regulatory element-binding protein 2) by Western blot. Representative blot (left); quantification
(right). n=4; *P<0.05 by t test; error bars represent SEM. B, Knockdown of either HMGB1 or SREBP2 significantly inhibited LDL (low-density
lipoprotein) transcytosis, but depletion of both did not achieve any further reduction; each point represents 1 cell. n=4 experiments; ***P<0.001
by 1-way ANOVA; error bars represent SD. C, SREBP2 was knocked down by siRNA achieving a reduction in protein levels of around 80%
(quantification is of precursor form). n=4; *P<0.05 by t test; error bars represent SEM. D, Depletion of SREBP2 prevented induction of LDL
transcytosis caused by HMGB1 overexpression. Each point represents 1 cell. n=4 experiments; ***P<0.001 by 1-way ANOVA; error bars
represent SD; ns indicates nonsignificant by Tukey multiple comparison test.

cells depleted of HMGB1 displayed no change in the these proteins functioning in the same pathway. Over-
total levels or the subcellular distribution of caveolin-1. expression of HMGB1 stimulated LDL transcytosis in
Instead, combinatorial siRNA knockdown of HMGB1 an SREBP2-dependent manner and was unaffected by
and SR-BI, as well as SR-BI and SREBP2, supports inhibition of extracellular HMGB1; conversely incubation

210   January 2021 Arterioscler Thromb Vasc Biol. 2021;41:200–216. DOI: 10.1161/ATVBAHA.120.314557
Ghaffari et al HMGB1 Is a Critical Regulator of LDL Transcytosis

Basic Sciences - AL
Downloaded from http://ahajournals.org by on November 17, 2022

Figure 4. HMGB1 translocates to the nucleus in an LDL and SR-BI-dependent manner and regulates SREBP2 half-life.
A, Human coronary artery endothelial cells (HCAECs) were immunostained for HMGB1 (high mobility group box 1; red) before and after incubation
with LDL (low-density lipoprotein; 40 μg/mL) for 15 min. LDL significantly increased the nuclear-to-cytoplasmic ratio, and SR-BI (scavenger receptor
class B type 1) depletion by siRNA partially inhibited this effect. Representative images (left); quantification (right). n=4; ***P<0.001 by 1-way
ANOVA, ***P<0.001 and *P<0.05 by Tukey multiple comparison test; error bars represent SEM. Scale bars=30 µm. B, FRAP of nuclear HMGB1
in HCAECs expressing GFP-HMGB1 in the presence or absence of 40 µg/mL unlabeled LDL; ***P<0.001 by Holm-Sidak multiple comparisons
test for each time point. Error bars represent SD. C, Incubation of serum-starved endothelial cells with LDL (40 μg/mL, 24 h) induced HMGB1
expression. Western blot is representative of 4 independent experiments; **P<0.01 by paired t test with data normalized to control. D, The stability
of SREBP2 (sterol regulatory element-binding protein 2) mRNA was assessed in HCAECs. The half-life of SREBP2 mRNA in control cells and cells
depleted of HMGB1 by siRNA was measured in the presence of actinomycin. E, The half-life of SREBP2 protein was assessed in cells depleted
of HMGB1 by siRNA; cycloheximide (CHX) was used to inhibit protein synthesis for the indicated number of hours (h). *P<0.05 and **P<0.01 by
Holm-Sidak multiple comparisons test for each time point. n=4 experiments; error bars represent SEM. M indicates mature; and p, precursor.

Arterioscler Thromb Vasc Biol. 2021;41:200–216. DOI: 10.1161/ATVBAHA.120.314557 January 2021   211
Ghaffari et al HMGB1 Is a Critical Regulator of LDL Transcytosis
Basic Sciences - AL
Downloaded from http://ahajournals.org by on November 17, 2022

Figure 5. Mice deficient in endothelial HMGB1 develop fewer fatty streaks on a high-fat diet.
A, Mice deficient in endothelial HMGB1 (high mobility group box 1) and wild-type littermates were injected retro-orbitally with 200 μg of AlexaFluor
(AF) 568–conjugated LDL (low-density lipoprotein) and euthanized 30 min later. The accumulation of LDL was measured and normalized to the
imaged area and to the paired control animal. Representative images of the inner curvature of the aorta en face (left); quantification with each
point representing 1 animal (right); *P<0.05 by paired, nonparametric t test. B, Mice deficient in endothelial HMGB1 and wild-type littermates
were fed a high-fat diet for 18 wk followed by staining of the inner curvature of the aorta with oil red O (ORO). Of 12 wild-type mice, 6 developed
visible plaque compared with 5 (of 11) knockout mice. Of those displaying plaque, endothelial cell (EC)–HMGB1−/− mice showed significantly
smaller fatty streaks than littermate controls (scatterplot). Representative images (left); quantification (right); each point represents 1 animal;
**P<0.01 by t test. C, Lipid accumulation at the aortic root was also measured (representative image at left) and quantified (right); data are from
7 knockout animals and 9 controls. D, EC-HMGB1−/− mice and controls were injected via retro-orbital vein with 1% Evans blue dye and then
euthanized after 30 min. Permeability to Evans blue dye was not significantly altered. n=4 to 5. ns indicates nonsignificant.

212   January 2021 Arterioscler Thromb Vasc Biol. 2021;41:200–216. DOI: 10.1161/ATVBAHA.120.314557
Ghaffari et al HMGB1 Is a Critical Regulator of LDL Transcytosis

Basic Sciences - AL
Downloaded from http://ahajournals.org by on November 17, 2022

Figure 6. Mice lacking endothelial HMGB1 in the setting of LDLR deficiency are protected from atherosclerosis.
A, Endothelial cell (EC)–HMGB1−/−:LDLR−/− and LDLR−/− littermate controls were injected retro-orbitally with 200 μg of AlexaFluor (AF)
568–conjugated LDL (low-density lipoprotein) and euthanized 30 min later. The accumulation of LDL was measured and normalized to the
imaged area and to the paired control animal. Representative images of the inner curvature of the aorta en face (left); quantification with each
point representing 1 animal (right); *P<0.05 by paired, nonparametric t test. B, Double knockout mice (EC-HMGB1−/−:LDLR−/−) and LDLR−/−
littermate controls were fed a high-fat diet for 5 wk followed by staining of the inner curvature of the aorta with oil red O (ORO). EC-HMGB1−
/−
:LDLR−/− mice showed significantly smaller fatty streaks than controls. Representative images (left), quantification (right). n=15 to 17 (each
point represents 1 mouse); *P<0.05 by t test. C, Lipid accumulation at the aortic root was also measured (representative image at left) and
quantified (right); data are from 15 knockout animals and 18 controls; P<0.05 by unpaired t test. HMGB1 indicates high mobility group box 1;
and LDLR, low-density lipoprotein receptor.

Arterioscler Thromb Vasc Biol. 2021;41:200–216. DOI: 10.1161/ATVBAHA.120.314557 January 2021   213
Ghaffari et al HMGB1 Is a Critical Regulator of LDL Transcytosis

with LDL induced HMGB1 expression. Finally, mice knockout animals were similar; yet knockout mice had a
deficient in endothelial HMGB1 developed smaller fatty significant reduction in fatty streaks. Despite the incom-
Basic Sciences - AL

streaks and (when on a LDLR−/− background) signifi- plete specificity of the Tie2Cre promoter for endothe-
cantly less atherosclerosis. lial cells, we found no significant reduction in HMGB1
While HMGB1 has been implicated previously in ath- expression in monocytes or neutrophils. Furthermore,
erosclerosis, prior literature has focused on its function the attenuation in LDL accumulation in the aortic arch of
extracellularly where major sources include leukocytes7 knockout animals within 30 minutes of injection is most
and smooth muscle cells.30 Our data indicate an important consistent with an impairment of endothelial LDL trans-
role for endothelial HMGB1 that occurs far earlier in the cytosis rather than a leukocyte-driven phenotype.
disease and is independent of its paracrine functions. We obtained similar results in a double knockout
Under low sterol conditions, SREPB2 is translocated to mouse model in which endothelial HMGB1 was deleted
the nucleus; otherwise, it resides in its immature form in the in a LDLR−/− background. Unlike the ApoE−/− mouse
endoplasmic reticulum.31,32 The enhanced recruitment of model, LDLR-deficient animals develop elevated LDL
HMGB1 to the nucleus upon addition of LDL may, there- levels appropriate for studying LDL transcytosis. How-
fore, seem paradoxical. Our data suggest that a function ever, the contribution of HMGB1 to LDL transcytosis by
of HMGB1 is to ensure the stability of transcription fac- the endothelium is harder to interpret in advanced athero-
tors, including SREBP2. We speculate that this would per- sclerotic disease given the protein’s established role in
mit continuing expression of SR-BI and other downstream endothelial activation and leukocyte recruitment, as well
genes in response to elevated circulating cholesterol. At as the presence of extracellular HMGB1. This makes it
this point, it is unknown whether this system is specific to impossible to attribute the reduction in atherosclerosis to
LDL transcytosis or whether it might also have the effect impaired transcytosis alone. As with the single knockout
of enhancing HDL transcytosis (and potentially reverse animals, double knockout mice displayed a significant
cholesterol transport33). While it is known that HMGB1 reduction in the accumulation of LDL in the aorta within
can enhance the function of other transcription factors 30 minutes of injection. As these mice and their controls
(eg, p5334; glucocorticoid receptor35), in our experiments, both lack LDLR, this observation supports the notion that
the effect of HMGB1 overexpression and HMGB1 knock- SR-BI–mediated LDL transcytosis is reduced by deletion
down on LDL transcytosis required SR-BI and SREBP2. of HMGB1 in the endothelium.
This supports the notion of a specific axis linking HMGB1– In contrast to its proinflammatory characteristics once
SREBP2–SR-BI that regulates LDL transcytosis (Sche- secreted or released, HMGB1 that is intracellular has been
Downloaded from http://ahajournals.org by on November 17, 2022

matic). The mechanism by which nuclear HMGB1 regulates reported to be anti-inflammatory by promoting autophagy
the stability of SREBP2 mRNA remains unknown as does and preventing apoptosis.37,38 These studies have focused
its specificity; the effect may be via RNA binding or through on the role of the protein when in the cytosol rather than
the recruitment of other proteins.36 on its nuclear functions. Given that the vast majority of
The translocation of HMGB1 from the nucleus to HMGB1 in cells is localized to the nucleus, the reduction
the cytoplasm requires posttranslational modifications in LDL transcytosis caused by siRNA to HMGB1 likely
such as hyperacetylation and phosphorylation.11 Further reflects the loss of nuclear rather than cytoplasmic pro-
research will be needed to determine the mechanisms by tein. As autophagy suppresses LDL transcytosis,39,40 loss
which HMGB1 accumulates in the endothelial nucleus of cytoplasmic HMGB1 would be expected to enhance
in response to LDL, for instance, whether they are sim- rather than inhibit LDL transcytosis. While our data impli-
ply reciprocal to what has been reported (deacetylation, cate nuclear HMGB1 as a positive regulator of transcyto-
dephosphorylation) or whether specific and additional sis, our work leaves open the possibility that extracellular
modifications are required. Our data also raise the ques- and intracellular HMGB1 may work additively or synergis-
tion of how the endothelial cells sense the addition of LDL; tically to accelerate atherogenesis.
while depletion of SR-BI significantly attenuated HMGB1 In closing, our findings add to the growing body of litera-
translocation, the effect was only partial. This suggests ture showing that LDL transcytosis is a regulated (and thus
the potential involvement of another receptor or sensor. potentially therapeutically tractable) process.41 For instance,
Our preliminary experiments suggest that LDLR is not we recently reported that estrogen significantly attenuates
required, and we are currently examining other candidates. LDL transcytosis by endothelial cells from male donors13;
Our experiments in mice deficient in endothelial interestingly, the mechanism of this effect was also through
HMGB1 support a role for the protein in regulating early SR-BI. Further understanding of the regulation of SR-BI–
disease: we observed a significant reduction in accu- mediated LDL transcytosis is likely to provide insights into
mulation of LDL in the aortic arch of knockout animals the therapy and prevention of atherosclerosis.
within 30 minutes of injection into the circulation. This
supports the notion that LDL transcytosis was reduced
in vivo, consistent with our in vitro data. On a high-fat ARTICLE INFORMATION
diet, circulating cholesterol levels in the wild-type and Received April 23, 2020; accepted October 5, 2020.

214   January 2021 Arterioscler Thromb Vasc Biol. 2021;41:200–216. DOI: 10.1161/ATVBAHA.120.314557
Ghaffari et al HMGB1 Is a Critical Regulator of LDL Transcytosis

Affiliations 14. Mollica L, De Marchis F, Spitaleri A, Dallacosta C, Pennacchini D, Zamai M,


Agresti A, Trisciuoglio L, Musco G, Bianchi ME. Glycyrrhizin binds to high-

Basic Sciences - AL
Keenan Centre for Biomedical Research, St. Michael’s Hospital, Toronto, Canada
(S.G., F.N.N., R.S., N.K., C.W., W.L.L.). Department of Laboratory Medicine and Patho- mobility group box 1 protein and inhibits its cytokine activities. Chem Biol.
biology (E.J., R.S., W.L.L.) and Department of Biochemistry (W.L.L.), University of 2007;14:431–441. doi: 10.1016/j.chembiol.2007.03.007
Toronto, Canada. Hospital for Sick Children, Toronto, Canada (B.E.S., N.M.G.). To- 15. Raheel H, Ghaffari S, Khosraviani N, Mintsopoulos V, Auyeung D, Wang C,
ronto General Hospital Research Institute, University Health Network, Canada (J.I.). Kim YH, Mullen B, Sung HK, Ho M, et al. Cd36 mediates albumin transcyto-
sis by dermal but not lung microvascular endothelial cells - role in fatty acid
Acknowledgments delivery. Am J Physiol Lung Cell Mol Physiol. 2019;316:L740–L750. doi:
We thank Dr Monika Lodyga from the Research Core Facility of the Keenan 10.1152/ajplung.00127.2018
Research Centre for assistance with flow cytometry. 16. Paigen B, Morrow A, Holmes PA, Mitchell D, Williams RA. Quantita-
tive assessment of atherosclerotic lesions in mice. Atherosclerosis.
Sources of Funding 1987;68:231–240. doi: 10.1016/0021-9150(87)90202-4
17. Daugherty A, Tall AR, Daemen MJAP, Falk E, Fisher EA, García-Cardeña G,
This work was funded by grants from the Heart and Stroke Foundation of Canada
(G-16-00013521) and the Canadian Institutes of Health Research (CIHR; PJT Lusis AJ, Owens AP 3rd, Rosenfeld ME, Virmani R; American Heart Asso-
168947), both to W.L. Lee. S. Ghaffari was supported by a postdoctoral schol- ciation Council on Arteriosclerosis, Thrombosis and Vascular Biology; and
arship from the Research Training Centre of the Keenan Research Centre for Council on Basic Cardiovascular Sciences. Recommendation on design,
Biomedical Science; E. Jang was supported by a Canada Graduate Scholarship– execution, and reporting of animal atherosclerosis studies: a scientific state-
Master’s Award from the CIHR. W.L. Lee is supported by a Canada Research ment from the American Heart Association. Arterioscler Thromb Vasc Biol.
Chair in Mechanisms of Endothelial Permeability. 2017;37:e131–e157. doi: 10.1161/ATV.0000000000000062
18. Huebener P, Pradere JP, Hernandez C, Gwak GY, Caviglia JM, Mu X,
Disclosures Loike JD, Schwabe RF. The HMGB1/RAGE axis triggers neutrophil-medi-
None. ated injury amplification following necrosis. J Clin Invest. 2015;125:539–
550. doi: 10.1172/JCI76887
19. Tréguier M, Doucet C, Moreau M, Dachet C, Thillet J, Chapman MJ, Huby T.
Transcription factor sterol regulatory element binding protein 2 regulates
REFERENCES scavenger receptor Cla-1 gene expression. Arterioscler Thromb Vasc Biol.
1. Tabas I, Williams KJ, Borén J. Subendothelial lipoprotein retention as the initiat- 2004;24:2358–2364. doi: 10.1161/01.ATV.0000147896.69299.85
ing process in atherosclerosis: update and therapeutic implications. Circulation. 20. Wong AL, Haroon ZA, Werner S, Dewhirst MW, Greenberg CS, Peters KG.
2007;116:1832–1844. doi: 10.1161/CIRCULATIONAHA.106.676890 Tie2 expression and phosphorylation in angiogenic and quiescent adult tis-
2. Williams KJ, Tabas I. The response-to-retention hypothesis of early atherogenesis. sues. Circ Res. 1997;81:567–574. doi: 10.1161/01.res.81.4.567
Arterioscler Thromb Vasc Biol. 1995;15:551–561. doi: 10.1161/01.atv.15.5.551 21. Westerterp M, Tsuchiya K, Tattersall IW, Fotakis P, Bochem AE, Molusky
3. Fung KYY, Fairn GD, Lee WL. Transcellular vesicular transport in epithelial MM, Ntonga V, Abramowicz S, Parks JS, Welch CL, et al. Deficiency of ATP-
and endothelial cells: challenges and opportunities. Traffic. 2018;19:5–18. binding cassette transporters A1 and G1 in endothelial cells accelerates
doi: 10.1111/tra.12533 atherosclerosis in mice. Arterioscler Thromb Vasc Biol. 2016;36:1328–
4. Armstrong SM, Sugiyama MG, Fung KY, Gao Y, Wang C, Levy AS, Azizi P, 1337. doi: 10.1161/ATVBAHA.115.306670
Roufaiel M, Zhu SN, Neculai D, et al. A novel assay uncovers an unexpected 22. Payne S, De Val S, Neal A. Endothelial-specific cre mouse models. Arte-
role for SR-BI in LDL transcytosis. Cardiovasc Res. 2015;108:268–277. doi: rioscler Thromb Vasc Biol. 2018;38:2550–2561. doi: 10.1161/ATVBAHA.
Downloaded from http://ahajournals.org by on November 17, 2022

10.1093/cvr/cvv218 118.309669
5. Kraehling JR, Chidlow JH, Rajagopal C, Sugiyama MG, Fowler JW, Lee MY, 23. Getz GS, Reardon CA. Diet and murine atherosclerosis. Arterioscler Thromb
Zhang X, Ramírez CM, Park EJ, Tao B, et al. Genome-wide RNAi screen Vasc Biol. 2006;26:242–249. doi: 10.1161/01.ATV.0000201071.49029.17
reveals ALK1 mediates LDL uptake and transcytosis in endothelial cells. 24. PALADE GE. Blood capillaries of the heart and other organs. Circulation.
Nat Commun. 2016;7:13516. doi: 10.1038/ncomms13516 1961;24:368–388. doi: 10.1161/01.cir.24.2.368
6. Huang L, Chambliss KL, Gao X, Yuhanna IS, Behling-Kelly E, Bergaya S, 25. Rohrer L, Ohnsorg PM, Lehner M, Landolt F, Rinninger F, von Eckardstein A.
Ahmed M, Michaely P, Luby-Phelps K, Darehshouri A, et al. SR-B1 dri- High-density lipoprotein transport through aortic endothelial cells involves
ves0020endothelial cell LDL transcytosis via DOCK4 to promote athero- scavenger receptor BI and ATP-binding cassette transporter G1. Circ Res.
sclerosis. Nature. 2019;569:565–569. doi: 10.1038/s41586-019-1140-4 2009;104:1142–1150. doi: 10.1161/CIRCRESAHA.108.190587
7. Kalinina N, Agrotis A, Antropova Y, DiVitto G, Kanellakis P, Kostolias G, 26. Wang H, Yang H, Czura CJ, Sama AE, Tracey KJ. HMGB1 as a late mediator
Ilyinskaya O, Tararak E, Bobik A. Increased expression of the DNA-binding of lethal systemic inflammation. Am J Respir Crit Care Med. 2001;164(10 pt
cytokine HMGB1 in human atherosclerotic lesions: role of activated macro- 1):1768–1773. doi: 10.1164/ajrccm.164.10.2106117
phages and cytokines. Arterioscler Thromb Vasc Biol. 2004;24:2320–2325. 27. Czura CJ, Wang H, Tracey KJ. Dual roles for HMGB1: DNA binding and cytokine.
doi: 10.1161/01.ATV.0000145573.36113.8a J Endotoxin Res. 2001;7:315–321. doi: 10.1177/09680519010070041401
8. Wang H, Bloom O, Zhang M, Vishnubhakat JM, Ombrellino M, Che J, 28. Fernández-Hernando C, Yu J, Suárez Y, Rahner C, Dávalos A, Lasunción MA,
Frazier A, Yang H, Ivanova S, Borovikova L, et al. HMG-1 as a late media- Sessa WC. Genetic evidence supporting a critical role of endothelial caveo-
tor of endotoxin lethality in mice. Science. 1999;285:248–251. doi: lin-1 during the progression of atherosclerosis. Cell Metab. 2009;10:48–54.
10.1126/science.285.5425.248 doi: 10.1016/j.cmet.2009.06.003
9. Lee WL, Grinstein S. Immunology. The tangled webs that neutrophils weave. 29. Fernández-Hernando C, Yu J, Dávalos A, Prendergast J, Sessa WC. Endo-
Science. 2004;303:1477–1478. doi: 10.1126/science.1095484 thelial-specific overexpression of caveolin-1 accelerates atherosclerosis in
10. Kanellakis P, Agrotis A, Kyaw TS, Koulis C, Ahrens I, Mori S, Takahashi HK, apolipoprotein E-deficient mice. Am J Pathol. 2010;177:998–1003. doi:
Liu K, Peter K, Nishibori M, et al. High-mobility group box protein 1 neutral- 10.2353/ajpath.2010.091287
ization reduces development of diet-induced atherosclerosis in apolipopro- 30. Porto A, Palumbo R, Pieroni M, Aprigliano G, Chiesa R, Sanvito F, Maseri A,
tein e-deficient mice. Arterioscler Thromb Vasc Biol. 2011;31:313–319. doi: Bianchi ME. Smooth muscle cells in human atherosclerotic plaques secrete
10.1161/ATVBAHA.110.218669 and proliferate in response to high mobility group box 1 protein. FASEB J.
11. Bertheloot D, Latz E. Hmgb1, il-1alpha, il-33 and s100 proteins: dual-func- 2006;20:2565–2566. doi: 10.1096/fj.06-5867fje
tion alarmins. Cell Mol Immunol. 2017;14:43–64. doi: 10.1038/cmi.2016.34 31. Sakai J, Duncan EA, Rawson RB, Hua X, Brown MS, Goldstein JL. Sterol-
12. Najima Y, Yahagi N, Takeuchi Y, Matsuzaka T, Sekiya M, Nakagawa Y, regulated release of SREBP-2 from cell membranes requires two sequential
Amemiya-Kudo M, Okazaki H, Okazaki S, Tamura Y, et al. High mobility cleavages, one within a transmembrane segment. Cell. 1996;85:1037–
group protein-B1 interacts with sterol regulatory element-binding proteins 1046. doi: 10.1016/s0092-8674(00)81304-5
to enhance their DNA binding. J Biol Chem. 2005;280:27523–27532. doi: 32. Wang X, Sato R, Brown MS, Hua X, Goldstein JL. SREBP-1, a membrane-
10.1074/jbc.m414549200 bound transcription factor released by sterol-regulated proteolysis. Cell.
13. Ghaffari S, Naderi Nabi F, Sugiyama MG, Lee WL. Estrogen inhibits ldl 1994;77:53–62. doi: 10.1016/0092-8674(94)90234-8
transcytosis by human coronary artery endothelial cells via gper and 33. Martel C, Randolph GJ. Atherosclerosis and transit of HDL through
SR-BI. Arterioscler Thromb Vasc Biol. 2018;38:2283–2294. doi: 10.1161/ the lymphatic vasculature. Curr Atheroscler Rep. 2013;15:354. doi:
ATVBAHA.118.310792 10.1007/s11883-013-0354-4

Arterioscler Thromb Vasc Biol. 2021;41:200–216. DOI: 10.1161/ATVBAHA.120.314557 January 2021   215
Ghaffari et al HMGB1 Is a Critical Regulator of LDL Transcytosis

34. Stros M, Ozaki T, Bacikova A, Kageyama H, Nakagawara A. HMGB1 and 38. Tang D, Kang R, Livesey KM, Cheh CW, Farkas A, Loughran P, Hoppe G,
HMGB2 cell-specifically down-regulate the p53- and p73-dependent Bianchi ME, Tracey KJ, Zeh HJ 3rd, et al. Endogenous HMGB1 regulates
Basic Sciences - AL

sequence-specific transactivation from the human Bax gene promoter. J autophagy. J Cell Biol. 2010;190:881–892. doi: 10.1083/jcb.200911078
Biol Chem. 2002;277:7157–7164. doi: 10.1074/jbc.M110233200 39. Torisu K, Singh KK, Torisu T, Lovren F, Liu J, Pan Y, Quan A, Ramadan A,
35. Calogero S, Grassi F, Aguzzi A, Voigtländer T, Ferrier P, Ferrari S, Bianchi ME. Al-Omran M, Pankova N, et al. Intact endothelial autophagy is required to
The lack of chromosomal protein Hmg1 does not disrupt cell growth but maintain vascular lipid homeostasis. Aging Cell. 2016;15:187–191. doi:
causes lethal hypoglycaemia in newborn mice. Nat Genet. 1999;22:276–
10.1111/acel.12423
280. doi: 10.1038/10338
40. Bai X, Yang X, Jia X, Rong Y, Chen L, Zeng T, Deng X, Li W, Wu G,
36. Matveeva EA, Mathbout LF, Fondufe-Mittendorf YN. PARP1 is a versatile
Wang L, et al. Cav1-cavin1-lc3b-mediated autophagy regulates high glu-
factor in the regulation of mRNA stability and decay. Sci Rep. 2019;9:3722.
doi: 10.1038/s41598-019-39969-7 cose-stimulated ldl transcytosis. Autophagy. 2019;16:1111–1129. doi:
37. Zhu X, Messer JS, Wang Y, Lin F, Cham CM, Chang J, Billiar TR, Lotze MT, 10.1080/15548627.2019.1659613
Boone DL, Chang EB. Cytosolic HMGB1 controls the cellular autophagy/ 41. Bian F, Yang XY, Xu G, Zheng T, Jin S. CRP-induced NLRP3 inflamma-
apoptosis checkpoint during inflammation. J Clin Invest. 2015;125:1098– some activation increases LDL transcytosis across endothelial cells. Front
1110. doi: 10.1172/JCI76344 Pharmacol. 2019;10:40. doi: 10.3389/fphar.2019.00040
Downloaded from http://ahajournals.org by on November 17, 2022

216   January 2021 Arterioscler Thromb Vasc Biol. 2021;41:200–216. DOI: 10.1161/ATVBAHA.120.314557

You might also like