Pancreatic ductal adenocarcinoma (PDA) was responsible for ~44,000 deaths in the USA
in 2018, and is the epitome of a recalcitrant cancer driven by a pharmacologically
intractable oncoprotein, KRAS [1–4]. Downstream of KRAS, the RAF→MEK→ERK signaling
pathway plays a central role in pancreatic carcinogenesis . However, paradoxically,
inhibition of this pathway has provided no clinical benefit to PDA patients . Here
we show that inhibition of KRAS→RAF→MEK→ERK signaling elicits autophagy, a process
of cellular recycling that protects PDA cells from the cytotoxic effects of KRAS pathway
inhibition. Mechanistically, inhibition of MEK1/2 leads to activation of the LKB1→AMPK→ULK1
signaling axis, a key regulator of autophagy. Furthermore, combined inhibition of
MEK1/2 plus autophagy displays synergistic anti-proliferative effects against PDA
cell lines in vitro, and promotes regression of xenografted patient-derived PDA tumors
in mice. The observed effect of combination trametinib plus chloroquine was not restricted
to PDA as other tumors, including patient-derived xenografts (PDX) of NRAS-mutated
melanoma and BRAF-mutated colorectal cancer displayed similar responses. Finally,
treatment of a PDA patient with the combination of trametinib plus hydroxychloroquine
resulted in a partial, but nonetheless striking disease response. These data suggest
that this combination therapy may represent a novel strategy to target RAS-driven
Increased autophagic flux in response to inhibition of KRAS→RAF→MEK→ERK signaling.
To test the hypothesis that RAF→MEK→ERK signaling may regulate autophagic flux in
PDA cells, we tested the consequences of targeted inhibition of this pathway in MIA-PaCa2
(KRASG12C), BxPC3 (BRAFΔV487-P492) and PDX220 (KRASG12V) PDA cells, the last derived
from a KRAS-mutated PDA PDX. PDA cells were engineered to express a chimaeric autophagic
flux reporter protein consisting of: mCherry, GFP and LC3 (AFR, Fig. 1a) . The
LC3 component targets the chimaera to the autophagosome, the mCherry component contributes
a pH insensitive red fluorescence, and the GFP component contributes a pH sensitive
green fluorescence that is diminished in the low pH (≤5) environment of the autophagosome
and lysosome. Hence, the ratio of mCherry:GFP fluorescence is a measure of autophagic
flux in these cells (Fig. 1a, Ext. Fig 1) .
Treatment of Mia-PaCa2AFR cells with temsirolimus, an mTORC1 inhibitor, led to the
expected increase in the mCherry:GFP fluorescence ratio (Fig. 1b, Ext. Fig. 1d). By
contrast, treatment of Mia-PaCa2AFR cells with either chloroquine (CQ) or SAR-405,
an inhibitor of the class III PI3’-kinase VPS34, led to the expected decrease in the
mCherry:GFP fluorescence ratio (Fig. 1b, Ext. Figs. 1b-c) [9, 10]. Next, Mia-PaCa2AFR
cells were treated with inhibitors of KRASG12C→RAF→MEK→ERK signaling including: ARS-853
(covalent inhibitor of KRASG12C), trametinib or cobimetinib (MEK1/2 inhibitors), or
SCH772984 (ERK1/2 inhibitor) (Figs. 1c-f) [11–15]. All of these inhbitors increased
the mCherry:GFP fluorescence ratio indicating that blockade of multiple nodes of KRASG12C→RAF→MEK→ERK
signaling led to increased autophagic flux. Confirmation of increased autophagic flux
was obtained by immunoblotting of extracts of trametinib-treated Mia-PaCa2 cells for
the degradation p62 and the conversion of LC3-I to LC3-II by the covalent conjugation
of phosphatidylethanolamine, (Ext. Figs. 2a,c). To extend these observations BxPC3AFR
and PDX220AFR cells were treated with trametinib, which also led to a readily detected
increase in autophagic flux (Figs. 1g-h, Ext. Fig 2b).
Trametinib-induced autophagic flux is mediated by the LKB1→AMPK→ULK1/ATG1 signaling
To determine the mechanism(s) by which inhibition of KRASG12C→RAF→MEK→ERK signaling
promotes autophagic flux, Mia-PaCa2 cells were treated with different concentrations
of trametinib for 48 hours or PDX220 cells were treated with trametinib (100nM) over
a time course with the expression or phosphorylation of potential downstream mediators
of autophagy assessed by immunoblotting (Figs. 1i-j). Previous work indicated that
ERK1/2 can inhibit LKB1 through phosphorylation of serine 428 (pS428) . LKB1 in
turn acts upstream of the AMPK→ULK1/ATG1 signaling axis to regulate autophagy [17,
18]. Consistent with this, inhibition of MEK→ERK signaling in Mia-PaCa2 or PDX220
cells led to decreased phosphorylation of pS428-LKB1 and increased phosphorylation
of AMPK (pT172) and ULK1 (pS555). Consistent with these observations, either shRNA-mediated
inhibition of LKB1 expression or ectopic expression of dominant-negative AMPKK45R
or ULK1M92A significantly attenuated, but did not fully abrogate, trametinib-induced
autophagy (Figs. 1l-m, Ext. Figs. 2d &e). Hence, these data are consistent with the
hypothesis that trametinib-induced autophagy in PDA cell lines is mediated, at least
in part, by increased flux through the LKB1→AMPK→ULK1/ATG1 signaling axis (Fig. 1k)
Trametinib and chloroquine are synergistically cytotoxic to PDA cell lines in vitro.
Despite the central role of the RAF→MEK→ERK MAP kinase signaling in PDA, MEK1/2 inhibitors
have failed to display clinical benefit in PDA patients [9, 10]. Hence, we hypothesized
that trametinib-induced autophagic flux may serve as a protective mechanism for the
survival of PDA cells in the face of RAF→MEK→ERK pathway inhibition. To test this,
Mia-PaCa2, BxPC3 or PDX220 cells were treated with different concentrations of trametinib
or chloroquine, either alone or in combination, with drug synergy/antagonism assessed
by the Loewe Additivity method (Fig. 2a) . Consistent with our hypothesis, we
observed synergistic anti-proliferative effects at chloroquine concentrations in the
range of 12.5–25μM when combined with trametinib in the range of 8–200nM (Fig. 2a).
Additionally, treatment with trametinib plus chloroquine resulted in increased caspase
3/7 activation and increased cumulative cell death compared to the single agents,
suggesting cooperative activation of apoptotic cell death (Figs. 2b-d). These data
are consistent with our model that trametinib-induced autophagic flux serves to protect
PDA cells from the potentially pro-apoptotic effects of RAF→MEK→ERK pathway inhibition.
Dominant-negative ATG4BDN combined with trametinib promotes regression of established
4-aminoquinolones such as chloroquine and hydroxychloroquine are pleiotropic such
that, in addition to inhibiting autophagy, they have effects on macro- and micropinocytosis,
mitochondrial function and other processes [21, 22]. We therefore wished to determine
whether the cooperative effects of trametinib plus chloroquine could be ascribed,
at least in part, to autophagy inhibition. To address this, we expressed a dominant-negative
(DN) form of ATG4B (ATG4BC74A, ATG4BDN) under the control of a tetracycline-regulated
promoter in Mia-PaCa2AFR cells (Mia-PaCa2AFR-ATG4BDN cells). Consistent with its ability
to inhibit autophagy [23, 24], expression of ATG4BDN inhibited the trametinib-induced
autophagic flux observed in Mia-PaCa2AFR cells as assessed by flow cytometry or immunoblotting
for p62 abundance or LC3 processing (Figs. 3a-b). Next, tumors generated by xenografting
Mia-PaCa2AFR-ATG4BDN cells into NOD/SCID mice were treated with: 1. vehicle control;
2. Doxycycline (to induce ATG4BDN); 3. trametinib (1mg/kg, q.d.) or; 4. the combination
of doxycycline plus trametinib. Whereas trametinib treatment had a modest cytostatic
effect, expression of ATG4BDN had no detectable effect on tumor growth. However, expression
of ATG4BDN in the presence of trametinib led to regression of established tumors (Fig.
3c). Immunohistochemical analysis of tumor sections revealed decreased pERK1/2 in
tumors from trametinib treated mice and elevated expression of ATG4BDN in tumors from
doxycycline treated mice. Importantly, tumors from trametinib treated mice displayed
reduced abundance of p62, consistent with increased autophagic flux. However, p62
expression was greatly increased when autophagic flux was inhibited by ATG4BDN expression
(Fig. 3d). These data indicate that, in a tumor cell autonomous manner, ATG4BDN-mediated
inhibition of autophagic flux in trametinib treated mice can elicit regression of
established MIA-PaCa2 xenografts.
Trametinib plus chloroquine/hydroxychloroquine promote regression of RAS→RAF→MEK→ERK
To determine if the anti-neoplastic effects of combined trametinib plus autophagy
inhibition with either chloroquine or ATG4BDN observed in vitro or in vivo respectively
might translate more broadly into additional tumor models, tumors generated by xenografting
MIA-PaCa2 or BxPC3 cells into NOD/SCID mice were treated with vehicle control (control),
trametinib, chloroquine or the combination of both trametinib plus chloroquine (Figs.
3e & f). Whereas chloroquine treatment had no effect on MIA-PaCa2 tumors, trametinib
elicited a modest reduction in tumor growth (Fig. 3e). Similarly, single agent trametinib
or chloroquine had only modest inhibitory effects on the growth of BxPC3 tumors. By
contrast, the combination of trametinib plus chloroquine elicited striking regression
of established Mia-PaCa2 or BxPC3 tumors (Figs. 3e & f). These observations were subjected
to further scrutiny using mice xenografted with two KRAS-mutated PDA PDX models: PDX220
or PDX227, which were then treated as described above. In parallel, a cohort of PDX220
or PDX227 PDX-bearing mice were treated with a regimen of gemcitabine plus nab-paclitaxel
that approximates the standard-of-care for a subset of human PDA patients (Figs. 3g
& h) . In this experiment, the combination of trametinib plus chloroquine/hydroxychloroquine
not only resulted in tumor regression but was superior to gemcitabine plus nab-paclitaxel.
Consistent with our treatment regimen, pERK1/2 was decreased and the abundance of
p62 increased in PDX227 tumors derived from mice treated with trametinib plus chloroquine
(Ext. Fig. 3). To determine whether orthotopically engrafted tumors would respond
to treatment, PDX220 fragments were implanted into the pancreata of NOD/SCID mice
and treated 21 days later with vehicle control, trametinib, hydroxychloroquine or
trametinib plus hydroxychloroquine. As observed previously, the growth of these tumors
was substantially inhibited by the combination of trametinib plus chloroquine but
not by either of the single agents (Ext. Figs. 4a-c). Moreover, mice treated with
the combination of trametinib plus hydroxychloroquine demonstrated inhibition of [18F]-deoxyglucose
(FDG) uptake as assessed by PET/CT imaging. By contrast, vehicle- or hydroxychloroquine-treated
mice demonstrated continuous tumor growth and trametinib-treated mice demonstrated
only a partial response.
Next, we tested whether the combination of trametinib plus chloroquine would promote
regression of other tumor types driven either by mutationally activated RAS or BRAF.
To that end, we employed PDX models of either NRAS-driven melanoma (HCI-Mel002 & NCI515677)
or BRAFV600E-driven colorectal cancer (HCI-CRC004). As before, PDX tumors were treated
with vehicle control, trametinib, chloroquine or trametinib plus chloroquine (Figs.
3i-j & Ext. Fig. 5a). Under the conditions of this experiment, only the combination
of trametinib plus chloroquine led to regression of all three PDX models. Importantly,
mice treated with the combination therapy displayed no weight loss (Ext. Fig. 5b-e),
however, side-effects of facial rash and hair loss were noted, although these were
mitigated by reducing the dose of chloroquine to 25mg/kg, which remained effective
in combination with trametinib (Ext. Fig. 5a). To further investigate the potential
role of autophagy in the response of RAS mutated cancer cells to MEK1/2 inhibition
we employed two KRASG12D/TP53Null-driven mouse lung cancer cell lines (SC196 & SC274)
derived from suitably manipulated Kras
mice . Whereas MEK1/2 inhibition in SC274 cells led to increased autophagic flux,
similar treatment of SC196 cells did not induce autophagic flux for reasons that are
unclear (Ext. Figs. 6a, c, d). When assessed in vitro, we detected synergy between
trametinib and chloroquine in SC274 cells but not in SC196 cells (Ext. Fig. 6b). Moreover,
when tested in xenografted tumors in mice, only xenografted SC274 tumors displayed
regression in response to the combination of trametinib plus chloroquine, whereas
SC196 cell derived tumors failed to respond to this combination of agents (Ext. Figs.
6e-f). These data indicate that the ability of trametinib to promote autophagy in
cultured KRASG12D/TP53Null-driven lung cancer cell lines is predictive of their response,
or lack thereof, to the combination of trametinib plus chloroquine in mice. Furthermore,
these data are broadly consistent with the hypothesis that the in vitro and in vivo
inhibitory effects of combined treatment with trametinib plus chloroquine is due to
a tumor cell autonomous induction of protective autophagy by MEK1/2 inhibition that
is abrogated by autophagy inhibitors such as chloroquine that convert an otherwise
cytostatic response into a cytotoxic one. These data suggest that the combination
of MEK1/2 inhibition plus chloroquine may promote regression of several tumor types
in which RAS→RAF→MEK→ERK signaling is constitutively activated.
Partial response of a refractory pancreatic cancer patient to trametinib plus hydroxychloroquine.
We encountered a patient with metastatic pancreatic cancer in our GI malignancies
clinic, who was refractory to all standard-of-care therapy options. The patient, a
68 year-old man, had been pre-treated with neo-adjuvant mFOLFIRINOX, adjuvant gemcitabine/capecitabine
and with palliative gemcitabine/abraxane/cisplatin. The patient’s best response was
stable disease with the first two drug regimens and disease progression with the last.
Moreover, the patient was displaying signs of PDA recurrence as evidenced by the development
of celiac plexus pain and a rapid increase in the level of the PDA blood-borne cancer
antigen 19–9 (CA19–9).
Given our compelling preclinical data, compassionate treatment of this patient was
initiated on off-label, off-trial trametinib plus hydroxychloroquine (T/HCQ) starting
at 2mg of trametinib and 400mg hydroxychloroquine daily in compliance with all relevant
ethical regulations. Keeping the trametinib dose unchanged, the hydroxychloroquine
was then escalated to 800mg daily and then to 600mg twice daily. After initiation
of 2mg of trametinib plus 800mg of hydroxychloroquine the patient reported resolution
of his celiac plexus pain. However, the patient’s CA19–9 continued to rise from ~17,000
to ~33,000 during the first two weeks of treatment. However, once the patient began
receiving 2mg of trametinib plus 1200mg of hydroxychloroquine daily, his CA19–9 levels
declined precipitously by ~95% over the ensuing 2 months indicative of response (Fig.
4a). Moreover, CT imaging four months following initiation of T/HCQ therapy (2mg T/1200mg
HCQ per day) indicated a 50% reduction in tumor burden by RECIST 1.1 criteria indicating
a partial response (Figs. 4b-e). During the first 60 days that the patient received
T/HCQ therapy, he experienced grade 1 rash and grade 1 fatigue. Moreover, since both
trametinib and hydroxychloroquine have noted ocular and cardiac toxicities, we conducted
monthly ophthalmologic exams and weekly electrocardiograms but without evidence of
Results presented here are consistent with previous observations that autophagy serves
as an adaptive and protective response to inhibition of RAS→RAF→MEK→ERK signaling
in cancer [27–29]. Moreover, they are consistent with a companion manuscript that
describes similar phenomena in PDA cells treated with ERK1/2 inhibitors (Bryant et
al., Nature Medicine, Submitted). Similar observations have been reported in BRAFV600E-driven
melanoma, consistent with our analysis of NRAS-mutated melanoma PDX models (Fig. 3i,
Ext. Fig. 5) . Although we demonstrate a link between MEK1/2 inhibition and activation
of the ULK1→AMPK→LKB1 axis leading to autophagy induction in PDA, it is likely other
pathways are involved in autophagy induction in response to inhibition of RAS→RAF→MEK→ERK
Recently, it has been demonstrated that ATG4BDN-mediated inhibition of autophagy promotes
regression of KRASG12D/TP53R172H-driven tumors in the KPC GEM model of PDA, but in
a gene dosage dependent manner . Whereas our data are consistent with these observations,
our data also suggest that the dependence of pancreatic (and possibly other RAS mutated)
cancer cells on autophagy becomes more acute in the face of pathway-targeted inhibition
of RAF→MEK→ERK signaling. Indeed, in our pre-clinical models, tumors were relatively
resistant to single agent trametinib or chloroquine/hydroxychloroquine, but were exquisitely
sensitive to the combination, unlike the situation with BRAFV600E-driven brain tumors
or melanoma which are initially sensitive to single agent BRAFV600E inhibition. Although
previous work suggests that macroautophagy is dispensable for growth of KRAS mutated
tumors and for the efficacy of chloroquine, even when used in combination with other
agents, this work did not test the combination of MEK inhibitors plus chloroquine
The status of TP53 has been reported to determine whether autophagy inhibition can
either inhibit or promote the progression of PDA in GEM models. In this case it was
reported that PDA arising due to concomitant expression of KRASG12D and silencing
of TP53 in the pancreas was accelerated either by genetic (ATG5Null or ATG7Null) or
pharmacological (chloroquine) inhibition of autophagy . By contrast, our data
fail to support a role for TP53 in the response of PDA cells to combined inhibition
of MEK1/2 plus autophagy. First, both MIA-PaCa2 cells (TP53R248W) and the SC274 lung
cancer cells (TP53Null) are sensitive to combined inhibition of MEK1/2 plus autophagy.
Furthermore, genetic analysis Patient 1, who responded to T/HCQ therapy, indicated
mutational alteration of TP53 (data not shown). Hence, in our research, TP53 status
does not obviously diminish the anti-tumor effects of the T/HCQ combination (Ext.
Figs. 6b, e-f).
It has previously been demonstrated that autophagic flux in cells of the pancreatic
cancer microenvironment (e.g. stellate cells or macrophages) can contribute to tumor
maintenance [24, 33]. Here, using tumor cell specific expression of ATG4BDN in MIA-PaCa2
cells, we demonstrate a tumor cell autonomous role for autophagy to protect cells
from MEK1/2 inhibition. Although the anti-tumor effects of this regimen may be further
enhanced by systemic inhibition of autophagy within the tumor microenvironment, combined
pharmacological blockade of MEK1/2 and autophagy in the malignant cell appears sufficient
for tumor regression. Moreover, there are newer and more specific inhibitors of
autophagy that target the VPS34 class III PI3’-kinase or the ULK1/ATG1 protein kinase
that may warrant testing in combination with inhibitors of RAS→RAF→MEK→ERK signaling
[10, 35, 36].
Finally, since both trametinib and hydroxychloroquine are orally administered, FDA-approved
drugs [37–39], these observations were translated to the clinic for a single, heavily
pre-treated PDA patient. Remarkably, the T/HCQ combination resulted in substantial
reduction in this patient’s overall tumor burden, CA19–9 tumor marker, and resolution
of debilitating cancer pain. Moreover, the safety and tolerability of the T/HCQ combination
is likely to be superior to traditional cytotoxic chemotherapy for PDA patients. However,
caution must be exercised in interpreting and extrapolating from the response of a
single patient such that we urge that the potential benefits of T/HCQ therapy be tested
in PDA patients only in the context of suitably designed clinical trials. However,
the combination of compelling preclinical data and the striking response of the first
patient to be treated with the T/HCQ combination provides a compelling impetus to
conduct a rigorous clinical trial to test T/HCQ therapy on overall response rate and
measures of survival in PDA patients. Furthermore our pre-clinical data suggests that
the testing of this combination of agents may eventually be warranted in patients
with other malignancies driven by mutationally activated RAS/BRAF genes such as melanoma,
colon or lung cancer.
MiaPaCa2 and BxPC3 cell lines were originally obtained from ATCC and maintained in
1:1 DMEM/F-12 (Gibco) with 10% fetal bovine serum (FBS). The PDX220 cell line was
derived from a pancreatic cancer patient derived xenograft by mechanical dissociation
of tumor tissue followed by culture in 1:1 DMEM/F-12 (Gibco) with 10% FBS. SC196 and
SC274 cell lines were derived from tumors initiated by intra-tracheal infection of
mice with an adenoviral vector expressing FLP recombinase. Cell lines were established
from lungs harvested from mice 6–12 weeks after tumor initiation by enzymatic and
mechanical dissociation. Cell lines were periodically tested for mycoplasma contamination
and discarded if positive.
Autophagic Flux Assay.
pBabePuro: mCherry-GFP-LC3 was obtained from Addgene (a gift from Jayanta Debnath;
plasmid # 22418) and the mCherry-GFP-LC3 cDNA was introduced into the lentiviral construct
pUltra-Hot resulting in pUltra-Auto. Lentivirus derived from pUltra-Auto was transduced
into cell lines resulting in AFR cell lines. AFR cell lines were subjected to various
treatments, trypsinized and then resuspended for analysis of of GFP and mCherry fluorescence
by flow cytometry using a BD FACS-Canto II. Cells were co-stained with DAPI at 1ug/ml
to exclude dead cells. mCherry/GFP ratio was generated using FACSDIVA v8.0.1 software
and gates were set for low/intermediate/high populations and maintained throughout
the experiment. All experiments were conducted in triplicate and significance of difference
in autophagic flux was calculated using a two-tailed t-test.
pUltra-Auto and TetR-Flag-ATG4B DN (a gift from Andrew Thorburn) lentiviral constructs
were used to express mCherry-GFP-LC3 and ATG4B DN respectively in cell lines. To general
lentivirus 3×106 HEK293T cells per 10cm dish in 6 mL of 1:1 DMEM/F12 with 10%(v/v)
FBS were plated 6 hours prior to transfection. Transfection of vector DNA (3μg), psPAX2
(3μg), and CMV-VSVG (1.5μg) were combined in 600μL of sterile PBS and 18μL of FugeneHD
(Promega) was added to the mixture per 10 cm dish. The transfection mixture was incubated
for 15 minutes at room temperature then added dropwise to the HEK293T cells. Media
was exchanged for fresh 1:1 DMEM/F12 with 10% FBS the next day. After 48 hours, supernatants
were harvested and filtered through 0.45uM filters, which were then added fresh to
target cell lines or flash frozen for storage in liquid nitrogen for later use. When
performing lentiviral transduction 8μg/mL of polybrene was added. After 6 hours viral
transduction media was exchanged for fresh media. After 48 hours cells were selected
via FACS for mCherry/GFP expression or puromycin at 10μg/mL for pUltra-Auto orTetR-Flag-ATG4BDN
AMPK and ULK1 Dominant Negative Expression.
Mia-PaCa2AFR were transiently transfected with pcDNA3 (Vector), or vectors encoding
AMPKα (WT or K45R, gift from Morris Birnbaum; Addgene plasmids #15991 and 15992) or
myc-mULK1 (WT or M92A, gift from Do-Hyung Kim; Addgene plasmids # 31961 and # 31962)
utilizing Lipofectamine 3000 (Invitrogen). For confirmation of expression, cells lysates
were harvested 48 hours after transfection for analysis by immunoblotting. To test
effects on autophagic flux, AFR cells transiently transfected with the various plasmid
constructs were treated with trametinib beginning 24 hours after transfection and
analyzed by flow sytomrtery 48 hours following trametinib addition.
Cells were washed three times with ice cold PBS, detached by adding ice cold PBS with
5mM EDTA, pelleted by centrifugation at 250xg for 5 minutes, and then solubilized
using RIPA buffer containing phosphatase and protease inhibitors (Thermo) at 4oC for
one hour. Detergent insoluble material was removed by centrifugation at 15,000xg for
10 minutes at 4oC. Protein concentrations were determined by BCA Protein Assay (Thermo).
Membranes were blocked in Odyssey Blocking Buffer (LI-COR) for 1 hour then immunoblotted
with the following primary antibodies overnight in Odyssey Blocking Buffer: phospho-ERK
1:1000 T202/Y204 (CST D13.14.4E), total ERK1/2 1:1000 (CST), p62 1:500 (Progen p62-C),
LC3A/B 1:500 (CST), phospho-LKB1 S428 1:500 (Abcam Ab63473), LKB1 1:500 (CST D60C5),
phospho-AMPK T172 1:500 (CST 40H9), AMPK 1:500 (CST D5A2), phospho-ULK1 S555 1:500
(CST D1H4), ULK1 1:500 (CST D8H5), and ATG4B 1:500 (CST D162R). Standard immunoblotting
procedures were then followed with Alexa 680 and 800 conjugated species specific secondary
antibodies. Immunoblotting was visualized with a LI-COR CLx infrared scanner.
In Vitro Synergy Assay.
To evaluate synergy in vitro, cells were seeded into 384-well plates in complete medium,
cultured overnight, and then treated in quadruplicate with trametinib or chloroquine,
either alone or in various combinations in 20%(v/v) medium in EBSS. At end-point,
medium was removed and cells were assayed using ATPlite 1step (Perkin Elmer) according
to the manufacturer’s protocol. Luminescence was quantified using a Perkin Elmer Envision
plate reader, normalized to control, and analyzed with Combenefit software (Loewe
In Vitro Incuycte Caspase 3/7 and cell death assays.
Cell lines were seeded at 4,000–10,000 cells/well in the wells of a 96-well plate
(100μL RPMI/10% FBS/1% Pen/Strep per well). After 24 hours, cells were treated with
DMSO control, trametinib, chloroquine, or trametinib plus chloroquine in at least
triplicate. To detect apoptosis, 5nM Incucyte Caspase-3/7 Green Apoptosis Assay Reagent
(Essen Bioscience, Cat#: 4440) was added to each well. To detect cell death, 250nM
Incucyte Cytotox Red Reagent (Essen Bioscience, Cat#: 4632) was added to each well.
Cells were imaged every 2 hours using the IncuCyte live cell anlysis system for 48
hours with data collection by the Incucyte ZOOM 2016B accompanying software. GFP-positive
(Caspase 3/7 positive cells) or RFP-positive (dead cells) and total confluence was
recorded at each time point. One-way ANOVA was used to determine statistical significance
and significance for each treatment was compared to DMSO control.
NOD/SCID mice were bred and maintained in a pathogen-free facility by the Pre-clinical
Research Resource (PRR) at the Huntsman Cancer Institute. All animal experiments were
performed in accordance with protocols approved by the University of Utah Institutional
Animal Care and Use Committees and we have complied with all relevant ethical regulations.
Xenografted tumors were established by sub-cutaneous injection of 2×106 MIA-PaCa2
or BxPC3 cells resuspended in 100μL of Matrigel into NOD/SCID mice and allowed to
establish. Treatment was then initiated with vehicle control (corn oil), trametinib
at 1mg/kg, chloroquine at 25–50mg/kg or the combination of trametinib plus chloroquine
at the aforementioned dosages via oral gavage daily. In the case of Mia-PaCa2 cells
expressing the TetR-Flag-ATG4BDN construct, mice were treated with vehicle control
(corn oil), trametinib at 1mg/kg via oral gavage daily and either standard chow or
doxycycline chow (625mg/kg). Tumors were measured twice weekly via calipers and tumor
volume was calculated by volume = 4/3*π* (length+width)/2)/2)3. Significance of difference
in tumor size was calculated by a two-tailed t-test.
Patient Derived Xenograft Assays.
Tumor tissue was obtained from patients who provided written informed consent according
to a tissue collection protocol (University of Utah IRB 89989 and 10924) approved
by the Huntsman Cancer Institute Institutional Review Board and subcutaneously implanted
into NOD/SCID mice for generation of PDA PDX and into NSG mice for the generation
of melanoma or colorectal PDX. PDX220 was derived from a neck metastasis from a 53
year old woman who had received prior treatment with FOLFIRINOX, Gem/Abraxane, FOLFOX,
and 5-FU/Cisplatin with KRAS
p16INK4a Q50*, and TGFBR2
mutations. PDX227 was derived from a 75 year old man from a primary resection sample
that had squamous differentiation prior to any treatment with a BRCA2
mutation. HCI-Mel002 was derived from a cutaneous biopsy of an NRAS-mutated (NRASQ61R)
melanoma isolated from a previously untreated 85 year old woman. HCI-CRC004 was derived
from a resection specimen of a BRAF-mutated (BRAFV600E) colorectal cancer from a previously
untreated 63 year old woman. NCI 516677 (515677–202-R, Passage 3) was obtained from
the public NCI PDX bank and information regarding patient demographics, site, prior
treatment and genotype are publically available. These tumors were propagated and
expanded. Upon experiment initiation 50–70mg tumor fragments were implanted bilaterally
into the flanks of NOD/SCID mice (PDA PDX) or NSG mice (melanoma and colorectal PDX).
When established tumors were measurable, treatment was initiated with vehicle control
(corn oil), trametinib (1mg/kg), chloroquine (25–50mg/kg), hydroxychloroquine (40mg/kg)
or the combination of trametinib and chloroquine/hydroxychloroquine at the aforementioned
single agent dosages via oral gavage. For gemcitabine/nab-paclitaxel treatment, 100mg/kg
of gemcitabine and 10mg/kg of nab-paclitaxel was infused via tail vein weekly for
three weeks with one week off approximating the dosing schedule for pancreatic cancer
patients. Tumors were measured and tumor volumes calculated as previously described.
Mice were anesthestized with 1.5–2% sevoflurane prior to injecting approximately 0.5mCi
of [18F]-fluorodeoxyglucose (FDG). CT imaging was performed using a NanoScan™ SPECT/CT
scanner followed by PET and MRI imaging using a NanoScan™ PET/MRI scanner (Mediso
Medical Imaging, Budapest). The animal remained anesthetized and immoblized in a common
MultiCell™ animal chamber to provide intrinsic spatial co-registration of CT, MRI,
and PET images. T1-weighted Gradient Echo (GRE) images and T2-weighted 2D Fast Spin
Echo (FSE) images were acquired prior to initiating a 20-minute PET emission scan
at 60 minutes post-injection of FDG. Quantitative analysis was performed using VivoQuant
(inviCRO, Boston, MA). Metabolic Tumor Volumes (MTV) were defined semi-automatically
using a minimum threshold of the Standardized Uptake Value (SUV). Total Lesion Glycolysis
(TLG) was then calculated as the MTV x SUV mean. For each mouse, the optimal SUV threshold
was defined on the baseline images and applied consistently to the post-treatment
images. Changes in TLG following treatment was then calculated for each mouse relative
to the pre-treatment baseline TLG.
Tumor-bearing mice were euthanized and tumor tissues were harvested and fixed in 10%(v/v)
formalin overnight. Tissues were transferred to 70%(v/v) ethanol, embedded in paraffin,
and four-micron sections were cut. Immunohistochemistry (IHC) was performed manually
on Sequenza slide staining racks (Thermo). Sections were treated with Bloxall (Vector
labs) followed by horse serum (Vector labs), primary antibody for phospho-ERK (CST
D13.14.4E) 1:600, p62 (Progen GP62-C) 1:200 and ATG4B 1:200 (CST D162R), then anti-Guinea
Pig (Vectastain) or anti-Rabbit (Vector Labs) HRP-polymer. The slides were developed
with DAB (Vector) and counterstained with hematoxylin.
Two-sided t-testing was used for all autophagy flux reporter assays comparing control
high (red) versus experimental high autophagic flux data. Two-sided t-testing was
also used for all in vivo tumor growth assay data compared at the days noted in the
graphs. One-way ANOVA testing was used to compare groups for all Incucyte experiments.
DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding
author upon reasonable request.
Ext. Figure 1.
Flow cytometry analysis of autophagic flux reporter with autophagy inhibitors and
a-e: Autophagic flux was assessed by flow cytometry in Mia-PaCa2AFR cells following
48 hours treatment with control, chloroquine (CQ), SAR-405, temsirolimus, or trametinib.
Experiments were repeated three times with similar results.
f: Autophagic flux was assessed by fluorescent imaging in Mia-PaCa2AFR cells following
48 hours treatment with control, chloroquine (CQ), VPS34i (SAR-405), or trametinib.
Experiments were repeated three times with similar results.
Ext. Figure 2.
Inhibition of RAS→RAF→MEK→ERK signaling pathway induces autophagic flux (AF) as seen
by p62 degradation and LC3 conversion in pancreatic cancer cells.
a & b: Cell lysates prepared from Mia-PaCa2 (a) or BxPC3 (b) cells treated with 0.1–100
nM of trametinib for 48 hours were analyzed by immunoblotting for the phosphorylation
(p) or total (t) abundance of ERK1/2, p62, LC3, or actin as indicated. Experiments
were repeated three times with similar results.
c: Cell lysates prepared from Mia-PaCa2 cells treated with ARS-853 (KRASG12Ci), SCH772984
(ERKi), or cobimetinib (MEKi) for 48 hours were analyzed by immunoblotting for the
phosphorylation (p) or total (t) abundance of ERK1/2, p62, LC3, or actin as indicated.
Experiments were repeated three times with similar results.
d: Cell lysates prepared from Mia-PaCa2AFR cells transiently expressing exogenous
ULK1 WT, ULKM92A (dominant negative), AMPK WT, or AMPKK45R (dominant negative) were
analyzed by immunoblotting for ULK1, AMPK, or actin as indicated. Experiments were
repeated three times with similar results.
e: Cell lysates prepared from Mia-PaCa2AFR cells lentivirally transduced with shRNAs
targeting LKB1 or scrambled control were analyzed by immunoblotting for LKB1 or actin
as indicated. Experiments were repeated three times with similar results.
Ext. Figure 3.
Trametinib and chloroquine are synergistically cytotoxic in vitro.
Mia-PaCa2 cells, BxPC3 and PDX220 cells were treated for 48–96 as indicated with trametinib
and chloroquine and analyzed for cell viability by ATPlite assay. Synergy scores were
generated utilizing Combenefit Software. Experiments were repeated four times with
Ext. Figure 4.
Treatment of pancreatic tumors with trametinib and chloroquine results in decreased
pERK and increased p62 abundance respectively.
Representative images of immunohistochemical analysis of sections of PDX 227 tumors
that were treated with 1. vehicle (Control), 2. trametinib; 3. chloroquine or; 4.
the combination of both agents. Sections were stained with H&E or with antisera against
pERK1/2 or p62 as indicated. Experiments were repeated four times with similar results.
Scale bar is 500 μM located in the bottom right of the upper left panel and is consistent
for all images.
Ext. Figure 5.
Treatment of orthopically xenografted pancreatic tumors with trametinib and hydroxychloroquine
demonstrates regression consistent with subcutaneous xenografts.
a: PDX220 tumors were orthotopically transplanted and after 3 weeks were imaged via
FDG-PET/CT for baseline. They were then treated with trametinib, hydroxychloroquine
or trametinib plus hydroxychloroquine for 2 weeks prior to re-imaging. n=3 for control;
n=2 for HCQ; n=3 for trametinib; n=2 for trametinib+hydroxychloroquine.
b & c: Quantification of total lesion glycolysis (b) and % change (c) for individual
tumors within each treatment group.
Ext. Figure 6.
Regression of established NRAS driven melanoma tumors by combined inhibition of MEK1/2
a. The growth of NRAS-mutated melanoma (NCI515677) PDX was assessed over 21 days in
mice treated with: 1. vehicle (Control), 2. trametinib (1mg/kg), 3. chloroquine (25mg/kg)
or; 4. the combination of both agents at the aforementioned doses as indicated. n=4
for all treatment groups except combination of both agents n=5. Center values are
the mean; statistical testing was performed by two-sided t-test; ***p<0.001 vs. control;
tttp<0.001 vs. trametinib. Error bars represent SD.
b-e. The percentage weight change of HCI-Mel002 NRAS-mutated PDX was assessed over
21 days in mice treated with: b. vehicle (Control), c. trametinib (1mg/kg), d. chloroquine
(50mg/kg) or; e. the combination of both agents at the aforementioned doses as indicated.
However, side-effects of facial rash and hair loss were noted.
Ext. Figure 7.
Lack of autophagy induction by MEK1/2 inhibition results in resistance to combined
trametinib and chloroquine treatment.
a: Cell lysates prepared from two suitably manipulated KRASG12D/TP53Null mouse lung
cancer-derived cell lines, SC196 or SC274 treated with 100nM of trametinib were analyzed
by immunoblotting for the phosphorylation (p) or total (t) abundance of ERK1/2, p62,
LC3, or actin as indicated. Experiments were repeated three times with similar results.
b: SC196 and SC274 KRASG12D/TP53Null mouse lung cancer cells were treated for 48 hours
respectively with trametinib and chloroquine and analyzed for cell viability by ATPlite
assay. Synergy scores were generated utilizing Combenefit Software. Experiments were
repeated four times with similar results.
c & d: Autophagic flux was measured in SC196AFR (c) or SC274AFR (d) following treatment
with 0.1–1000nM trametinib for 48 hours.. n=3; center values are the mean; statistical
testing was performed by two-sided t-test of control high (red) versus experimental
high; ***p<0.001 vs. control. Error bars represent SD.
e & f: The growth of SC196 (e) or SC274 (f) KRASG12D/TP53Null mouse lung cancer derived
tumors in xenografted mice treated with: 1. vehicle (Control), 2. trametinib (1mg/kg),
3. chloroquine (50mg/kg) or; 4. the combination of both agents was assessed over ~15
days as indicated. n=10 for all treatment groups.. Center values are the mean; statistical
testing was performed by two-sided t-test;***p<0.001 vs. control; tttp<0.001 vs. trametinib.
Error bars represent SD.