Background
The inhibition of integral components of the Ras/Raf/mitogen-activated protein kinase kinase (MEK)/extracellular signal-regulated kinase (ERK) signaling pathway has been proven clinically effective in
B-raf melanoma and suggested to be effective in some RAS-activated tumors [
1],[
2]. MEK has a critical position in this pathway with some direct upstream activators (e.g. Raf) and some downstream targets (e.g. ERK) and has been identified as a promising target for selective inhibition of
K-ras and
B-raf-mutated tumor types [
3],[
4]. Many MEK inhibitors have been developed and have entered preclinical efficacy testing. Consequently, the need for new translational pharmacodynamic (PD) biomarkers for MEK inhibition efficacy in cancer treatment is increasing.
Non-invasive positron emission tomography (PET) imaging with the fluorine-18 labeled glucose analog 2-fluoro-2-deoxy-D-glucose ([
18F] FDG) is being used as a functional endpoint in phase I to III clinical trials in oncology for assessing therapeutic response, in addition to conventional endpoints such as toxicity in tumor size. Fluorine-18-labeled glucose analog 2-fluoro-2-deoxy-D-glucose-positron emission tomography ([
18F] FDG-PET) imaging has been used for monitoring neoadjuvant chemotherapy in patients with locally advanced breast cancer [
5], non-small cell lung cancer [
6], and lymphomas [
7], for an early prediction of response to imatinib mesylate (Glivec
TM) in patients with advanced soft tissue sarcoma [
8] and in the evaluation of vemurafenib (Zelboraf
TM), an oral
B-raf inhibitor [
9]. Several recent phase I dose-escalation clinical trials have also included [
18F] FDG-PET, for instance in assessing MEK inhibition as a monotherapy or in combination with other drugs in patients with advanced solid tumors [
10]-[
12].
PET imaging of small animals is increasingly used as a translational efficacy biomarker in anti-cancer drug development. Despite the inclusion of [
18F] FDG-PET as a possible clinical efficacy biomarker for MEK inhibition, very few preclinical studies have yet been performed to evaluate the nature and extent of its pharmacodynamic and/or predictive capability. In addition, it is critical for interpreting clinical data that the biological mechanisms related to changes in [
18F] FDG uptake are fully elucidated, a task for which pre-clinical experimentations is particularly well-suited. [
18F] FDG-PET/CT and MRI were used to show the synergistic effects of NVP-BEZ235, a dual PI3K/mTOR inhibitor, with the MEK inhibitor selumetinib in
K-ras-mutated lung adenocarcinoma xenografts [
10],[
13] and of the combination of PI3K/mTOR inhibitor PE-04691502 and MEK inhibitor PD0325901 in a mouse model of ovarian cancer [
14]. Metabolic [
18F] FDG-PET responses could also monitor combination therapy with docetaxel and the MEK inhibitor selumetinib in mice with lung cancer [
15]. Recently, [
18F] FDG-PET was used as an early PD biomarker of the efficacy of MEK inhibitor RO5126766 in human colorectal cancer xenografts with
K-ras and
B-raf mutations [
16]. It was shown to be a sensitive biomarker not only for predicting efficacy but also acquired resistance in
B-raf
V600
mutant melanoma xenografts in mice treated with vemurafenib alone and in combination with the MEK inhibitor GDC-0973 [
17].
Reverse phase protein array (RPPA) analysis is a high throughput antibody-based technology for large-scale analyses and quantitative assessments of activated signaling pathways and identification of biomarkers [
18],[
19]. Multiplex analyses of PD biomarkers by RPPA have been performed
in vitro in cancer cell lines [
20],[
21] as well as in cancer tissue and blood samples [
22]-[
24] in order to assess response to target inhibition.
The MEK inhibitor RO4987655/CH4987655 (Hoffmann-La Roche/Chugai Pharmaceutical, Tokyo, Japan) is an orally active small molecule targeting mitogen-activated protein kinase (MAP2K or MEK) with potent anti-tumor activity [
25]. RO4987655 binds to and inhibits MEK1, which results in the inhibition of MEK-dependent cell signaling and tumor cell proliferation. A phase I study of RO4987655 demonstrated its promising anti-tumor activity, which was further investigated in specific populations of patients with RAS and/or RAF mutation-driven tumors. [
18F] FDG-PET imaging was included in the clinical efficacy assessments. Reduction in tumor [
18F] FDG uptake between baseline and day 15 was observed in 79.4% of patients with advanced solid tumors [
10].
To examine the molecular basis for using [
18F] FDG as a PD biomarker for an early response to MEK inhibition, we have here performed longitudinal [
18F] FDG-PET imaging of metabolic responses to RO4987655 therapy in human lung carcinoma xenografts. The expression levels of the glucose transporter (GLUT1) and the activity of hexokinase 1, which have been shown to be related to [
18F] FDG uptake in human cancer [
26], were examined using semi-quantitative fluorescent immunohistochemistry (fIHC). RPPA was used to assess RO4987655 effects on the MAPK/PI3K pathway components in order to make correlations with the metabolic changes determined in the tumors by [
18F] FDG-PET.
Methods
Cell culture and reagents
The human lung adenocarcinoma cell line NCI-H2122 was purchased from the American Type Culture Collection (ATCC). All cells were maintained in the designated media and indicated concentrations of heat-inactivated fetal bovine serum (Gibco, Langley, OK, USA) and L-glutamine (Sigma, St. Louis, MO, USA) according to the ATCC recommendations. Cells were grown at 37°C in an atmosphere of 5%CO
2. RO4987655 (CH4987655), 3,4-difluoro-2-(2-fluoro-4-iodophenylamino)-5-(3-oxo-1,2-oxazinan-2-ylmethyl)benzohydroxamic acid 2-hydroxyethyl ester, was synthesized by Chugai Pharmaceuticals Co., Ltd. The RO4987655 chemical structure is provided in Additional file
1. For
in vivo use, the drug was dissolved in 50% ethanol/50% Cremophor® EL (Sigma) and stored at ‒20°C. The vehicle and RO4987655 stock solutions for
in vivo experiments were diluted fivefold with distilled water on each dosing day.
Cell proliferation and assay and Western blotting
Cells were treated with various concentrations of RO4987655 for 72 h in 96-well plates and viable cells were quantified with Cell Counting Kit-8 (Dojindo Molecular Technologies, Inc, Rockville, MD, USA). For Western blotting, cells were treated with RO4987655 for indicated periods and lysed with cell lysis buffer (Cell Signaling Technology, Beverly, MA, USA) containing a protease inhibitor cocktail (cOmplete, EDTA-free, Roche, Deutschland, Germany), phosphatase inhibitor cocktails 2 and 3 (Sigma), and 1 mM PMSF (Sigma). For detection of protein bands, the following were used as primary antibodies: pEGFR (Y1068, Cell Signaling Technology, 20234S), EGFR (Cell Signaling Technology, 20234S), pMKK4 (S257/T261, Cell Signaling Technology, #9156), MKK4 (SEK1(5C10), Cell Signaling Technology, #3346), pAKT (S473, Cell Signaling Technology, 9271), AKT (Cell Signaling Technology, 9272), pERK (T202/Y204, Cell Signaling Technology, 9101), ERK (Cell Signaling Technology, 9102), pMEK1/2 (S217/S221, Cell Signaling Technology, 9121), MEK (Cell Signaling Technology, 9122), Cyclin D1 (NeoMarkers, RB-010-P0), and actin (Santa Cruz, sc-1616). All protein bands were visualized with secondary antibodies labeled with HRP and ECL system (GE Healthcare, Wilmington, MA, USA) by using ImageQuant LAS 4000 (GE Healthcare).
Synthesis of [18F] FDG
[18F] FDG was obtained as an aliquot from daily clinical production at the Karolinska University Hospital and was subjected to all quality tests performed for administration in humans. Identity and radiochemical purity were determined using radio-thin layer chromatography and radio-liquid chromatography analyses prior to release.
Xenograft tumor models
Female athymic nude mice Balb nu/nu, age 5 to 6 weeks (18 to 22 g) were purchased from NOVA/SCB (Sollentuna, Sweden) and Charles River Lab International, Inc. (New York, NY, USA). Animal care, handling, and health monitoring were carried out in accordance with the Guidelines for Accommodation and Care of Laboratory Animals. All animal experiments were performed in accordance with protocols approved by the institutional animal care committee. NCI-H2122 cells (4 × 106/mouse) were inoculated subcutaneously in the right flank of Balb-nu/nu mice. Once tumors were established (100 to 200 mm3), mice were randomized into groups with similar mean tumor volumes at the start of the study. Tumor volume and body weight were measured on days 0 (baseline), 1, 2, 3, and 9 of [18F] FDG-PET imaging. Tumor growth inhibition was calculated using the following formula: TGI = [1‒(T‒T0) / (C‒C0)] × 100, where T and T0 are the mean tumor volumes on specific experimental day and on the first day of the treatment, respectively, for the experimental groups and likewise, where C and C0 are the mean tumor volumes for the control groups. The daily administration was selected on preliminary screening studies and was lower than the maximal tolerated dose (MTD), defined as the maximum dose associated with <20% weight loss and no toxic deaths.
PET imaging
Further details on the PET imaging are provided in Additional file
2. A time interval of 20 to 24 h was used between daily RO4987655 administration and completion of PET imaging for each tumor-bearing mouse and for each PET imaging time point (day 0, 1, 3 and 9). Mice were fasted for 6 to 8 h prior to start of the imaging session. [
18F] FDG (7 to 8 MBq per mouse, maximum volume of 200 µL) was administered to awake, warmed (37°C) mice by a bolus injection via the tail vein. Forty to sixty minutes after the tracer injection, the mice were administered with isoflurane, controlled by an E-Z anesthesia vaporizer (5% initially and then 1.5% to maintain anesthesia, blended with 7:3 air/O
2 and delivered through a Microflex non-rebreathing mask from Euthanex Corporation, Palmer, PA, USA). The mice were placed on a heated pad (37°C) on the camera bed, with most of the body volume in the field of view (7.68 cm). Emission data were collected for 20 min in list mode with a microPET Focus 120 scanner (CTI Concorde Microsystems). Maximum standardized uptake values (SUV
max) of [
18F] FDG uptake in the tumor were calculated and normalized to the administered activity (MBq/body weight, g). The drug effect on tumor metabolism was estimated as%SUV
max change to day 0 (baseline).
Fluorescent immunohistochemistry (fIHC)
Further details on the fluorescent immunohistochemistry (fIHC) are provided in Additional file
2. Tumors were collected for fIHC analyses 20–22 h after daily RO4987655 administration for each [
18F] FDG-PET time point (days 0, 1, 3 and 9). For fIHC, tumor samples were fixed by immersion for 24 h in a solution of 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4) containing phosphatase inhibitors. Tissue blocks containing 4 to5 tumor samples of different treatment groups were snap frozen, cryosectioned at 14-µm thickness, and thaw-mounted on Superfrost Plus slides. Single- and multi-labeling experiments were performed as described previously [
27]. Briefly, sections were incubated with rabbit anti-GLUT1 (Abcam, AB653, dilution 1:300) and anti-hexokinase 1 (Abcam, AB65069, dilution 1:200) antibodies and were visualized by incubation with Alexa488 conjugated donkey anti-rabbit antibody. Whole slide images were captured on a Metasystems Vslide system (Newton, MA, USA) equipped with appropriate filter sets using a × 10 objective. Channel grey scale images were analyzed using ImageJ software (1.45p NIH, Bethesda, MD, USA) and GLUT1 and hexokinase 1 fluorescence intensity were measured from 50 randomly selected spots within each tumor sample. Supporting quantification figure is available as an additional information (Additional file
3).
Reverse phase protein arrays
Tumors were collected for RPPA analyses 20 to 22 h after daily RO4987655 administration for each [18F] FDG-PET time point (days 0, 1, 3, and 9). Histological examination of 15-µm thick frozen sections revealed good morphology of tissues, with 75% to 90% vital tumor content (10% to 25% fibrosis, no necrosis). Tumor extracts of all samples were prepared using 100 µL of lysis buffer CLB1 for 30 min at room temperature. Protein concentrations of lysates were determined by Bradford Assay (Coomassie Plus-The better Bradford® Assay, Thermo Scientific, Waltham, MA, USA). For the production of RPPA chips, the tumor lysates were adjusted to 3 mg/ml protein concentration in CLB1 buffer and subsequently diluted tenfold with printing buffer CSBL1. The lysates were printed onto Zeptosens chips (Bayer Technology Services GmbH, Leverkusen, Germany) using a Gesim NP2.0 Nanoplotter (GeSiM, Grosserkmannsdorf, Germany) and single droplet deposition of lysate (0.4 nL per sample spot). Lysate samples were printed at four serial dilutions (start concentration 0.3 mg/mL, plus 1.6-fold dilutions), each dilution as duplicate spots (in total eight spots per sample). Array images and data were analyzed with the software ZeptoVIEW 3.1 (Zeptosens, Witterswil, Switzerland).
The list of signaling proteins representing MAPK/PI3K/glycolic pathways and included in the RPPA analyses is provided in Additional file
4 for day 1 and in Additional file
5 for day 3 of the RO4987655 treatment. RPPA analyses were performed at the NMI Natural and Medical Sciences Institute, Reutlingen, Germany. The Ingenuity pathway analysis, a web-based software application, was used for comparison of MAPK/PI3K pathway modulations on days 1 and 3 of RO4987655 treatment.
Statistical analysis
For the comparison of [18F] FDG-PET data, a pair-wise t test with Tukey-Kramer correction for multiple testing was applied using the software JMP (Version 10, © 2012 SAS Institute Inc., Cary, NC, USA). For the analysis of the relationship of dose level and biomarker data from both RPPA and fIHC, a statistical model based on a least squares fit including the factors dose and dose2 was implemented using the software JMP (Version 10, © 2012 SAS Institute Inc.).
Discussion
We have recently reported the use of [
18F] FDG-PET as a PD biomarker for the efficacy of a first-in-class dual MEK/Raf inhibitor, RO5126766, in preclinical
K-ras or
B-raf mutant tumor xenografts models [
16]. RO5126766 was designed to inhibit ERK signaling outputs more effectively than standard MEK inhibitors [
28]. The preclinical [
18F] FDG-PET results in tumor xenografts were found to parallel the [
18F] FDG-PET results obtained in a phase I dose escalation study in humans [
11]. The novel allosteric MEK inhibitor RO4987655 studied here binds and inhibits MEK, resulting in the suppression of MEK-dependent cell signaling [
25]. Both RO5126766 and RO4987655 have been assessed with [
18F] FDG-PET in phase I dose-escalation clinical trials and have shown promising anti-tumor activities with strong decreases in tumor metabolic responses [
10],[
11]. Interestingly, the [
18F] FDG-PET data collected during these independent studies led to different recommendations in the doses of RO5126766 and RO4987655 for phase II [
29].
To investigate further whether preclinical [
18F] FDG-PET imaging would also mirror the clinical observations obtained for RO4987655, we performed longitudinal PET scans of metabolic responses in
K-ras-mutated NCI-H2122 tumor xenografts in mice. Unlike the gradual reductions in tumor [
18F] FDG uptake by RO5126766 over the treatment time, RO4987655 induced significant decreases in tumor metabolic activities already by day 1. However, this decrease was followed by rapid dose-dependent rebounds in the [
18F] FDG uptake at day 3, even though the drug was administered daily. To investigate the mechanism for the [
18F] FDG-PET response despite continued TGI, we first examined whether the tumor metabolic feedback activation in response to MEK inhibition on day 3 correlated with GLUT1 and hexokinase 1, using semi-quantitative fIHC. We found that the increase in [
18F] FDG uptake on day 3 was associated with elevated hexokinase 1, which is consistent with studies reporting that increased levels of hexokinases enhance [
18F] FDG intracellular trapping [
30],[
31]. Furthermore, our observation was in agreement with several other studies that have shown correlations of [
18F] FDG uptake and hexokinase expression in response to MEK inhibition [
17]. However, the hexokinase changes observed here were not statistically significant nor did changes in GLUT1 transporter expression significantly correlate with the observed [
18F] FDG changes. These results are mostly likely due to the high intra- and inter-tumoral variability of GLUT1 and hexokinase 1 expression in our IHC. No other significant treatment effects were identifiable with this technique. Previous studies have demonstrated that correlations between GLUT expressions and hexokinase activities and [
18F] FDG uptake are more pronounced at the cellular level and have proposed that this could serve as a good
in vitro screening for testing the feasibility of cells to be used in xenograft cancer models for PET imaging [
16],[
32]. However, as discussed in [
33]-[
35] and also observed in our study, these relationships
in vivo were not as predictive as expected, most likely because [
18F] FDG uptake
in vivo will depend on many factors: not only glucose transporters, hexokinases, and glucose-6-phosphatase activities but also the intra-tumoral cell density, blood supply, fraction of hypoxic tissue, and viable cell numbers.
We also analyzed the effect of RO4987655 on specific cellular signaling components. We showed
in vitro (Figure
1) that RO4987655 decreased pERK1/2 activities in tumor cells followed by a pERK1/2 increase and up-regulated phosphorylation of MEK, MAP2K4 (MKK4), and EGFR at later time points. Also, a modest increase of pAKT, which is a component of the alternative PI3K pathway involved in cell survival signaling as well as glucose homeostasis regulation, was observed. These results might indicate that NCI-H2122 cells can be affected by a negative feedback upon the MEK kinase inhibition.
RPPA proteomics technology was used to investigate the molecular changes of the signaling status in vivo that would influence the [18F] FDG uptake in NCI-H2122 tumors upon treatment since it permits multiplex, highly sensitive, and reproducible quantitative analysis of protein expression and phosphorylation levels. To our knowledge, there are no previous reports in the literature that have combined preclinical [18F] FDG-PET with RPPA for evaluating MEK inhibition efficacy.
In the xenograft experiment at 1.0, 2.5, and 5.0 mg/kg, the significant pERK1/2 down-regulations revealed by RPPA, both on day 1 and day 3 in all dose groups, confirmed that this MEK inhibitor clearly hits its target in the xenografted tumors. These results support previous publications proposing the inhibition of ERK phosphorylation in tumors as a PD biomarker of MEK inhibition [
28],[
36],[
37]. RPPA also revealed significant reductions of levels of pMKK4, a direct activator of MAP kinases in response to various environmental stresses of mitogenic stimuli [
38], and of pEGFR, an upstream Ras activator, on both treatment days (Tables
1 and
2).
In a recently published study, MEK inhibition in cancers by RO4987655 was demonstrated by decreases in ERK1/2 phosphorylation [
10]. Prior reports using RO4987655 with
B-raf-mutated HT-29 human colon cancer xenografts also showed a marked reduction of pERK [
25]. In addition, an apparent relationship between [
18F] FDG-PET data and the degree of pERK suppression by RO4987655 in peripheral blood mononuclear cells of melanoma patients was reported [
29]. In this study, we observed different pERK1/2 responses on day 3 in the
in vivo and
in vitro settings. In our
in vitro study at 0.1 µM of RO4987655, we observed pERK1/2 up-regulation after initial down-regulation. This is well known as a relief from a negative feedback on RAF by MEK inhibition and is apparently due to feedback regulation since there is no drug clearance. In contrast, in our
in vivo study, as revealed by RPPA, pERK1/2 was clearly suppressed even on day 3 at all doses. In our previous study with
B-raf-mutated HT-29 xenografts [
25], tumor pERK1/2 returned to the basal level after 24 h from 6.25 mg/kg RO4987655 administration due to the clearance of the drug from blood in animals. In the current study, we administered the drug daily and we resected the tumors just after the drug administration. Therefore, pERK1/2 down-regulation was observed even at day 3.
We also observed an up-regulation of pC-Raf and pMEK1/2 on day 3 (Table
2), which is in agreement with other reports [
39] in which MEK inhibition led to an increase in pMEK through feedback-mediated Ras activation. Interestingly, RO4987655 caused an activation of cyclin D1, which was observed as early as day 1 of the treatment. This increase can only be partly explained by an activation of alternative pathways since the re-activation of the MAPK pathway seems to be fully established at day 3 that is 2 days later than the up-regulation of cyclin D1 could be observed. As described by Rexer et al. [
40], the activation of alternative PI3K/AKT pathways limits the anti-tumor activity of MEK inhibitors and down-regulation of cyclin D1 or reduced tumor cell proliferation is not necessarily expected. Increased cyclin D1 expression in cells that showed resistance to MAPK inhibitors has been reported [
41]. Thus, the activation of cyclin D1 during the RO4987655 treatment may limit the inhibitory effects of MEK inhibition on tumor growth.
In addition, as described above, the current study found significantly increased phosphorylation of AKT on day 3 of the RO4987655 treatment. Up-regulation of pAKT may lead to the activation of the downstream signals that regulate glucose metabolism and cell survival. Thus, there are two potential implications of the [18F] FDG-PET rebound on day 3 of RO4987655 treatment: up-regulating glucose uptake by activating AKT signal and/or stimulating cell growth by reactivating the ERK/MAPK pathway signal.
Concerning the direct up-regulation of glucose uptake by activating AKT signaling, our initial investigation of the expression levels of GLUT1 and hexokinase 1 in the tumors with fIHC failed to identify any statistically significant correlations between the [18F] FDG-PET rebound and protein expression. The expression profiling of these proteins as well as other GLUT-family members (for example, GLUT3) and hexokinase 2 in combination with a more accurate quantitative approach needs to be undertaken to clearly understand the correlations between [18F] FDG uptake and AKT signal activation by MEK inhibition in K-ras-mutated tumors.
Concerning the indirect up-regulation of glucose uptake by stimulating cell survival via AKT signaling, we hypothesize that the rebound effect in response to MEK inhibition by RO4987655 seen in the [
18F] FDG-PET analysis is due to activation of the PI3K/AKT mediated cell survival pathway. This is supported by studies with other MEK inhibitors, such as AZD6244 and HER2-driven cancers [
42], PD325901 in prostate cancer xenografts [
43], and U0126 in a Wilms tumor model in mice [
44]. Interestingly, it was shown that dual blockage of the compensatory PI3K/AKT/mTOR and RAS/MEK/ERK pathways can synergistically inhibit tumor cell growth
in vitro and
in vivo in different cancers [
45]-[
47]; combination treatment with MEK inhibitor AZD6244 and AKT inhibitor MK2206 was more effective than each drug alone in lung cancer patients [
48].
An indirect up-regulation of glucose uptake by stimulating cell growth via reactivated ERK signaling is supported by a recent study with CH5126766 [
28]. An enormous suppression of ERK signal achieved maximal tumor growth inhibition in both
B-raf and
K-ras xenograft models. Thus, reactivation of ERK signal by relief from the negative feedback to RAF may limit the anti-tumor activity as well as [
18F] FDG uptake at later time points.
These results are important for the use of [18F] FDG-PET imaging in clinical drug development and for understanding the mechanisms behind changes in [18F] FDG uptake induced by MEK inhibitors. In a clinical setting, a similar rebound in [18F] FDG uptake observed in patients treated with MEK inhibitors may be useful for detecting the development of drug resistance long before increases in tumor size become detectable.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
TT, LL, and SSE designed and carried out the FDG-PET imaging, data acquisition, and analysis; TT and SSE drafted the manuscript. NM and JM performed immunohistochemical studies and statistical analysis; MV contributed to the concept of the paper and the selection of the technical platform (for RPPAs). ER, MV, and MP designed the RPPA studies and carried out RPPA analyses and helped in interpreting the RPPA data. SB and CHO performed the statistical and bioinformatics analysis. JT and VM were involved in planning, designing, and interpretation of the FDG-PET imaging studies. NI, SA, and YY designed and carried out the efficacy studies. NI also supported interpreting the data and drafting the manuscript. All authors read and approved the final manuscript.