European Journal of Cancer (2013) 49, 264– 271

A v a i l a b l e a t w w w . s c i e nc e d i r e c t . c o m

journa l homepage : www.e j cancer . in fo

Defining hypoxic microenvironments by non-invasivefunctional optical imaging

Pablo Iglesias a, Maximo Fraga b, Jose A. Costoya a,⇑

a Molecular Oncology Laboratory MOL, Departamento de Fisioloxia, Facultade de Medicina, Universidade de Santiago de Compostela,

Santiago de Compostela, Galicia, Spainb Departamento de Anatomia Patoloxica e Ciencias Forenses, Universidade de Santiago de Compostela, Santiago de Compostela, Galicia, Spain

Available online 3 July 2012

09

ht

⇑

KEYWORDS

HypoxiaIn vivo optical imagingBRETFluorescenceBioluminescenceCancer

59-8049/$ - see front matter

tp://dx.doi.org/10.1016/j.ejca.

Corresponding author: Tel.:E-mail address: josea.costoy

� 2012 E

2012.06.0

+34 647a@usc.e

Abstract Functional imaging has become an important tool in oncology by informing aboutlocalisation and size of the tumour as well as the pathophysiological features of tumoural cells.One of the most characteristic features of some tumour types is the activation of the neoan-giogenic programme which is specifically mediated by the transcription factor hypoxia-induc-ible factor (HIF)-1a, an important player in regulating this process and a prognostic marker oftumoural aggressiveness. Here we report a non-invasive in vivo detection of lung micrometas-tases in a mouse model of breast cancer using self-illuminating genetically encoded tracersresponsive to intracellular HIF-1a levels and a preliminary analysis of the contribution ofthe tumoural masses to the metastatic niche. This model lays the foundations for novelhypoxia sensing probes able to detect micrometastatic disease with high sensitivity and spec-ificity. Thus, optical functional imaging shows promise in the understanding of disease, drugdevelopment or image-guided therapy.� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

The monitoring of biological processes in a func-tional non-invasive way has been a major milestone inbasic and clinical research in recent years, and especiallyin oncology. Functional imaging gives more detailedinformation about tumoural physiological processessuch as oxygenation rate, perfusion and alterations ofblood flow. As an example, a variation of the magnetic

lsevier Ltd. All rights reserved.

01

344082; fax: +34 881 812432.s (J.A. Costoya).

resonance imaging (MRI) technique, functional MRI(fMRI) is currently used to map areas of the cerebralcortex in relation to brain cancers.1 Optical imaging alsoallows the possibility of shifting from structural to func-tional imaging by revealing pathophysiological featureswithin certain tissue, because it is able to translate spe-cific molecular and cellular processes associated to a dis-ease to changes in light emission. Besides, these methodsdo not display the harming effects of ionising radiationson living organisms or require high-budget equipment tomonitor the overall process as MRI does.2 However,although fluorescence imaging (FLI) is known to face

P. Iglesias et al. / European Journal of Cancer 49 (2013) 264–271 265

problems when it comes to autofluorescence in whole-body imaging this can be circumvented by using probesable to emit light in the near infrared part of the spec-trum or even higher. And while bioluminescence imag-ing (BLI) suffers from signal scattering and poorspatial accuracy, new advances in reconstructive tomog-raphy intend to overcome these technical difficultiesenhancing BLI spatial resolution and quantification.3,4

In cancer biology, hypoxia/hypoxia-inducible factor(HIF)-1 pathway seems to play a key role in invasionand metastasis. HIF-1 is a central regulator of hypoxicgene expression and involved in the restoration of cellu-lar oxygen homoeostasis. HIF-1 is a HIF-1a/HIF-1bheterodimer that binds the hypoxia response elements(HREs) of target genes under hypoxic conditions. WhileHIF-1b is constitutively expressed, the expression andtranscriptional activity of HIF-1a is precisely regulatedby intra-cellular O2 concentrations.5 Under normoxia,prolyl hydroxylases modify Pro-402 and Pro-564 ofHIF-1a in a reaction that uses O2 as a substrate. Then,hydroxylated HIF-1a interacts with von Hippel–Lindau(VHL), which is part of an E3 ubiquitin ligase complextargeting HIF-1a for 26S proteasomal degradation. Onthe other hand, under hypoxic conditions, HIF-1a is sta-bilised because of the lack of O2 and dimerises withHIF-1b to bind to the HREs. In a coordinated fashionwith coactivator CBP/p300, HIF-1 activates the tran-scription of target genes involved in such diverse pro-grammes as glucose transportation and glycolysis,angiogenesis, survival and proliferation, or invasionand metastasis. Mutations in tumour suppressor genes,such as PTEN and VHL, hyperactivation of oncogenicpathways (i.e. Ras/mitogen-activated protein kinase(MAPK) and PI3K/AKT pathways) or the presence ofreactive oxygen species (ROS) can also regulate HIF-1a in an oxygen-independent manner. Accordingly,HIF-1a overexpression is usually linked with poor prog-nosis in several cancer types since it correlates with highmetastatic potential of these tumoural cells.6,7

Another kind of transcription factors, the signaltransducer and activator of transcription (STAT) pro-tein family, is also regulated by hypoxia and oxidativestress, besides other cytokines and growth factors. Thesetranscription factors exist in an unphosphorylated latentform in the cytoplasm until undergoing phosphorylationby a number of tyrosine kinases. When activated, theseproteins play a dual role by acting as signal transducersin the cytoplasm and transcriptional activators in thenucleus.8 Aberrant activation of epidermal growth fac-tor receptor kinases, especially HER1/erbB-1 andHER2/neu, and overexpression of EGF ligands arelinked with breast carcinoma progression,9,10 and resultin downstream hyper-activation of STAT proteins.Although STAT proteins are upregulated in tumouralcells, only STAT3 and STAT5 are thought to have a rel-evant role in oncogenic development.11 In keeping withthis, constitutive STAT3 activation occurs frequently in

a variety of human tumour cell lines, namely breast can-cer,12,13 leukaemia,14 prostate cancer,15 melanoma16 andmyeloma.17 It also reveals the fact that this member ofthe JAK-STAT pathway maintains a crosstalk withthe hypoxia response, as several reports suggest thatSTAT3 is required conjointly with HIF-1a for maxi-mum induction of vascular endothelial growth factor(VEGF) under hypoxic conditions or via Src.18

2. Methods

2.1. Establishment and characterisation of cell sublines

Human breast cancer cells MDA-MB 231 were cul-tured in Dulbecco’s modified Eagle medium, DMEM(Sigma–Aldrich, MO, USA), supplemented with 10%foetal bovine serum (Fisher Scientific, PA, USA). Cellcultures were maintained at 37 �C and 5% CO2. DNAtransfections were carried out using the Effectene trans-fection reagent (Qiagen, CA, USA). Cells were seeded in12-well plates at a cellular density of 5 � 104 cells/plate.Each well was transfected with 10:1 or 5:1 M ratios of E-M-H-mCherry-Luc and pMC1neo-polyA (Stratagene,CA, USA). Cell cultures were maintained under antibi-otic selection at a concentration of 1 lg/mL of G418(Sigma–Aldrich, MO, USA) for 2 weeks. Cells fromeach subline (ECL-A5, ECL-A10, ECL-B5, ECL-B10)were seeded at a density of 1 � 106 cells/plate and inorder to induce an artificial hypoxic environment eachpool was treated with CoCl2 (Sigma–Aldrich, MO,USA) at a concentration of 500 lM. Also, L-Mimosine(Sigma–Aldrich, MO, USA), another hypoxia-mimeticagent, was added to each pool up to a final concentra-tion of 800 lM. All cells were trypsinised 24 h later, col-lected by centrifugation, washed thoroughly with PBS toeliminate traces of medium and resuspended in 1� PBS.Fluorescence acquisition was performed as previouslyreported,19 using an IVIS Spectrum (Caliper LS, CA,USA). Cells from ECL-B10 subline were serially dilutedin 96-well plates and D-Luciferin was subsequentlyadded to each plate at a final concentration of 150 lg/mL before imaging. Imaging time was 1 min/plate. Allexperiments were carried out three times at least.

2.2. Cell implantation and fluorescence-bioluminescencein vivo assays

Cells from ECL-B10 cell line were seeded and trypsin-ised 48 h later with Trypsin (Fisher Scientific, PA,USA), collected by centrifugation and resuspended insterile 1� PBS at a concentration of 2 � 103 cells/lL.Seven BALB/c nude mice were injected intravenously(i.v.) each with 2 � 105 cells (100 lL). Upon sedationwith isoflurane (2%) using the XGI-8 Gas AnaesthesiaUnit fluorescence and bioluminescence data were regis-tered with an IVIS Spectrum�. Fluorescence images

266 P. Iglesias et al. / European Journal of Cancer 49 (2013) 264–271

were collected in a single 10-s exposition while biolumi-nescence images were collected during at least 20 min in60-s expositions to ensure the peak emission wasincluded. Upon imaging, the highest emission was cho-sen as the most representative of the series (Fig. 1b).For bioluminescence and bioluminescence resonantenergy transfer (BRET) assays each mouse (CharlesRiver Laboratories, Paris, France) was injected intraper-itoneally with a single dose (150 lg/kg) of D-Luciferin(L9504, Sigma–Aldrich, MO, USA) dissolved in sterile1� PBS. Fluorescence and bioluminescent 3D tomo-graphic reconstructions were performed using LivingImage� software. Anatomic co-registration was per-formed using the Xenogen digital mouse atlas includedin the IVIS software suite. All mice were euthanisedand tissue samples collected for biochemical and histo-logical analysis. Mice were housed in specific patho-gen-free (SPF) conditions, following FELASA(Federation of European Laboratory Animal ScienceAssociations) guidelines for animal housing and accord-ing to the USC Bioethic Board regulation, passed on23rd September, 2003.

2.3. Immunoblot and immunohistochemistry analysis

Tissue samples were homogenised in a conical mortarand lysed in ice-cold RIPA buffer (1� PBS, 1% NonidetP-40, 0.5% sodium deoxycholate, 0.1% SDS, 10 mg/mlPMSF, 40 lg of aprotinin/ml, 100 mM orthovanadate).Cell lysates from each pool were collected upon treat-ment with 20 lM MG-132 (Calbiochem, Merck-Milli-pore, Germany), a potent proteasome inhibitor.

Fig. 1. (a) Optical performance (right fluorescence imaging (FLI), left biosubmitted to hypoxic conditions upon treatment with 500 lM of CoCl2 (lowMDA-MB 231 pools (MDA-MB 231, ECL-A5, ECL-A10, ECL-B5 and Ecollected with Dulbecco’s modified Eagle medium (DMEM) without phenplated in triplicate and treated with 500 lM of CoCl2. D-Luciferin substraWells containing medium only and cells only were also included as contro

Analysis of protein levels was carried out by immuno-blot analysis using a monoclonal antibody againstHIF-1a (H1a) (sc-53546), polyclonal antibody againstSTAT-3 (C-20) (sc-482, Santa Cruz, CA, USA), phos-phorylated STAT-3 (B-7) (sc-8059, Santa Cruz, CA,USA) and a monoclonal a-tubulin antibody (T-5168,Sigma–Aldrich, MO, USA). As for the immunohisto-chemical analysis, primary antibodies for Caspase-3,Histone H3 (both from Cell Signaling Technology,MA, USA) and a biotinylated polyclonal antibodyagainst red fluorescent protein (RFP) (ab34771, Abcam,UK) were employed.

3. Results

3.1. Establishment and characterisation of light producing

cell lines

We have previously described the design and in vitro

and in vivo characterisations of a genetically encodedbiosensor (ECL) able to detect the transcription factorHIF-1a,19 which is a key regulator of neo-angiogenesisand metastasis in the tumoural cells.20,21 Moreover, inour biosensor the HIF-1a specificity gives a great advan-tage when compared to other constructs harbouringentire promoter regions of target genes regulated by thistranscription factor, e.g. VEGF. Numerous physiologi-cal and pathological processes regulate VEGF expres-sion by means of cytokines and growth factors bindingto the VEGF receptor and modulating its expression,interleukin (IL)-lb, interleukin (IL)-6, platelet derivedgrowth factor (PDGF)-BB, transforming growth factor

luminescence imaging (BLI)) of MDA-MB-231 pools transfected ander row) or control (upper); (b) biochemical analysis of HIF-1a levels inCL-B10); (c) in vitro bioluminescence of ECL-B10 subline. Cells wereol red (D5030, Sigma–Aldrich), serially diluted up to 62 cells per well,te was added and bioluminescent pseudocolour images were obtained.ls.

P. Iglesias et al. / European Journal of Cancer 49 (2013) 264–271 267

(TGF)-b, basic fibroblast growth factor (bFGF), epider-mal growth factor (EGF) and hepatocyte growth factor(HGF) being some of them.22 Accordingly, the promis-cuity of VEGF promoter raises doubts about the speci-ficity of these constructs to efficiently inform us aboutneo-angiogenic activity within the cell under hypoxicconditions.

Additionally, the fusion of a red fluorescent protein(mCherry) and the firefly luciferase (FLuc) yielded anadditional and interesting feature known as BRET. Thisphenomenon occurs between a bioluminescent donor(FLuc) and a near infrared acceptor (mCherry), whichcould be an advantage in in vivo environments whereexternal excitation sources would be heavily hampered.Thus, the existence of a self-illuminating tracer capableto overcome technical problems, such as tissue autofluo-rescence and weak tissue penetration of wavelength exci-tation light, is an important feature that allows its use asan in vivo diagnostic tool. Although nanodevices basedon quantum dots capable of BRET have been reportedbefore, their light emissions were not directly associatedto any biological mechanism.23

As we reported before,19 this genetically encoded bio-sensor comprises a regulatory moiety formed by a chi-meric enhancer able to bind specifically HIF-1a,24 anda fusion protein formed by one of the most known nearinfrared-emitting fluorescent protein known as mFruits,mCherry.24 Thus, optical measuring of this transcriptionfactor allows for an in vivo and non-invasive way oflearning about the tumour biology since high concentra-tions of HIF-1a correlate directly with tumour aggres-sivity and the ability of these cells to metastasise toneighbouring locations.3

To this purpose, we have modified a breast cancer cellline (MDA-MB 231), which is known to have a signifi-cant ability to metastasise and does not have affectedtranscription factor HIF-1a.25–27 We transfected cellsfrom this line with our biosensor ECL E-M-H-mCher-ry-Luciferase (ECL) besides a plasmid carrying the neo-mycin resistance gene (pMC1neo-polyA). Uponchecking the efficiency of the transfection using fluores-cence microscopy (data not shown) we maintained thecells under antibiotic selection for at least two weeks.

We have chosen to use pools rather than clonesbecause they better recapitulate the in vivo selection oforgan-specific metastatic variants from human cancercells.28,29 Four cell pools were harvested upon selectionand re-seeded to test their hypoxia sensitivity by submit-ting them to an artificially induced hypoxic environmentadding CoCl2 to culture media. Cobalt chloride is a well-known hypoxia-mimetic agent that inhibits prolylhydroxylation in HIF-1a yielding the stabilisation ofthe transcription factor and thus enhancing its transcrip-tional activity.30 As Fig. 1a shows, the B10 cell line(ECL-B10) resulted to display the brightest signal,attending either to its fluorescent or bioluminescent per-formance. We next wanted to determine the limit of

detection of this subline so we performed several serialdilutions of cells from the ECL-B10 pool, and submittedthem to the same CoCl2 treatment (500 lM) for 24 h.We next added D-Luciferin substrate and acquired bio-luminescent pseudocolour images. Also, and to furthercorroborate these findings we employed anotherhypoxia-mimetic agent such as L-Mimosine with eachone of the pools; as Supplementary Fig. 1 displays, weobtained a similar outcome as with CoCl2.31

We next wanted to correlate the emission of fluores-cence and bioluminescence with stabilised HIF-1a levelsunder hypoxic conditions. With this in mind, we per-formed a biochemical analysis of HIF-1a levels withor without CoCl2 in each pool (Fig. 1b). As expected,we observed a high correlation with the induced levelsof the biosensor and HIF-1a levels of each pool. Finally,and with the intention to determine the minimum detect-able number of cells we performed a serial dilution ofour brightest pool and registered the bioluminescentactivities of each dilution, as Fig. 1c shows.

3.2. In vivo model of breast cancer metastasis

Once characterised in vitro, we wondered if we couldreplicate these results in vivo, injecting cells from thissubline in immunodeficient nude mice to develop abreast cancer model of metastasis. We subsequentlyinjected 2 � 105 cells from the ECL-B10 subline in thetail vein of immunodeficient nude BALB/c mice and fol-lowed up the tumoural mass formation by periodic fluo-rescent and bioluminescent acquisitions. Cells fromMDA-MB 231 are known to display a moderate meta-static potential, primarily targeting the lungs.25 Eightweeks after injection, several micrometastases with highbioluminescent activity were detected in the lungs of themice injected with ECL-B10, indicating high concentra-tions of HIF-1a. Acquisitions of fluorescence and biolu-minescence light emission of tumoural masses in thelungs were collected and can be seen in Fig. 2a inpseudocolour images. Whole-body BRET images werealso collected (Supplementary Fig. 2). To locate thesources of light emission before carrying out the necrop-sies and check whether or not they spatially correlated,we also performed tomographical reconstructions fromboth fluorescence (FLIT) and bioluminescence (BLIT)signals. We observed that both techniques, as seen inFig. 2b (and also Supplementary Video 1 and Supple-mentary Video 2), located a tumoural mass in the leftlobe displaying an intense signal of fluorescence andbioluminescence.

3.3. Validation of hypoxic conditions in tumour

microenvironment

To validate these results, we subsequently sacrificedthe animals and collected FLI, BLI and BRET ex vivo

acquisitions directly from the lungs (Fig. 3a). We

Fig. 2. (a) In vivo acquisitions of lung tumoural masses’ signal from mCherry fluorescence (left) and firefly luciferase (FLuc) bioluminescence (right)in an immunodeficient mouse (#2576); (b) 3D reconstruction of FLIT (fluorescence tomography) light source (tumoural mass appears highlightedin red) and 3D reconstruction of BLIT (bioluminescence tomography) light source, tumoural mass appears highlighted in yellow. See alsoSupplementary Video 1 (FLIT) and Supplementary Video 2 (BLIT). Fluorescence and bioluminescent 3D tomographic reconstructions wereperformed using Living Image� software. (For interpretation of the references to colour in this figure legend, the reader is referred to the webversion of this article.)

268 P. Iglesias et al. / European Journal of Cancer 49 (2013) 264–271

observed that these masses displayed differential opticalactivity, probably because of the different HIF-1a levelsin those foci. In order to investigate whether these differ-ences were real, we collected tissue samples from thesemasses and performed a simple biochemical analysis.We picked several of them from different mice and clas-sified them into two groups, either ‘low emitters’ (solidarrowhead) or ‘high emitters’, according to their respec-tive fluorescent brightness (white arrows), Fig. 3b (upperpanel). We obtained tissue extracts from these metasta-ses and analysed their HIF-1a levels by immunoblotting.As the lower panel shows, higher HIF-1a levels faith-fully correlate with metastases labelled as ‘high-emit-ters’. We also quantified the total flux in photons persecond of the fluorescent light emission of all of themasses analysed and grouped into these two classes.The result of the statistical analysis of the fluxes shownin Fig. 3c demonstrates a significant difference betweenthose labelled as ‘high emitters’ and ‘low emitters’. Inaddition, we managed to culture one of the tumouralmasses and carried out the same induction assay thatwe performed previously with the pools. Results areshown in Supplementary Fig. 3.

In keeping with this, we wanted to investigate if otherbiochemical markers corroborated these differencesobserved in HIF-1a activity. Thus, we next wanted toanalyse the levels of STAT3 and its phosphorylatedactive form given that it is known for its important role

in the aberrant hyperactivation of the JAK-STAT signal-ling pathway in breast cancer,12,13 and its cooperation inthe angiogenic response together with HIF-1a.18 Weobtained whole cell lysates from several masses of differ-ent optical activities, i.e. HIF-1a levels, in addition tocontrol tissue from surrounding lung tissue without opti-cal activity. As we show in Fig. 3d, what we found is thatalthough in ‘low-emitters’ masses STAT3 and p-STAT3appear upregulated as expected, this is not also true inmasses where HIF-1a is also strongly expressed (‘high-emitters’). This apparent contradiction might beexplained taking into account that the HIF-1a levelsobserved correspond solely to cells with high hypoxic lev-els within the tumour and therefore express high levels ofthe optical tracer, while these registered STAT3 levels areobserved in the whole tumoural mass comprised of light-emitting and non-emitting cells, depending on the intra-tumoural heterogeneity of HIF-1a expression.32

To further investigate these findings, we next per-formed a histological analysis of the tumoural massesresected. Attending to the histological sections of thelungs, tumoural populations can be easily identified, withcells with bigger nuclei than that of lung epithelial cellsinvading bronchial cavities, and a high number of themsurrounding new vessels, as seen in Fig. 4a. Cells formingthese masses did not present apoptosis (Caspase-3 stain-ing, data not shown) with some of them displaying aber-rant nuclei due to defective mitosis, and showed a strong

Fig. 3. (a) Ex vivo acquisitions of fluorescence imaging (FLI) (mCherry), BLI (firefly luciferase (FLuc)) and bioluminescence resonant energytransfer (BRET) (FLuc excites mCherry) on the whole affected organ (lung); (b) upper panel, white arrowheads highlight the optically activetumoural masses from where tissue samples were collected. From those three, and attending to their respective optical activities two of them wereidentified as ‘high emitters’ (denoted in the figure with open arrows) and the one with a weaker fluorescence signal as ‘low emitters’ (solidarrowhead); lower panel, immunoblot of hypoxia-inducible factor (HIF)-1a levels in several metastases labelled as high-emitters and low-emitters.a-Tubulin was used as loading control. (c) Quantification of fluorescence imaging signal intensity of tumoural masses grouped either as ‘low-emitter’ (n = 12) or ‘high-emitter’ (n = 7) metastasis. Quantified values are shown in total flux (photons per second). ***P < 0.0005; (d) analysis ofprotein levels by immunoblot of STAT-3 and its activated phosphorylated form in control tissue, low-emitter and high-emitter masses.

P. Iglesias et al. / European Journal of Cancer 49 (2013) 264–271 269

positive staining to the proliferative marker histone H3(data not shown), consistent with the high proliferativerate of the tumoural masses. In order to identify the cellsharbouring the fusion protein elicited by HIF-1a we alsoperformed immunostainings with a RFP antibody, giventhat mCherry was evolved from the sequence of RFPfrom Discosoma sp.33 The positive immunostainings seenin Fig. 4b demonstrate positive RFP staining, which cor-relates with mCherry expression, in tumoural cells thatare surrounded by negative vascular walls (upper pic-ture) or stroma (lower picture).

4. Discussion

Taken together, these results demonstrate that wesuccessfully developed a novel breast cancer cell line thatcarries a genetically encoded hypoxia biosensor able toexpress a self-illuminating tracer. In a metastatic modelof human breast carcinoma, where we have injected sys-temically breast cancer cells into nude mice, we identi-fied tumoural masses in vivo using both FLI and BLIand performed tomography reconstruction for bothtechniques. In addition, the differential optical activities

observed correlate with actual differences in endogenousHIF-1a levels, which is a good prognostic marker ofaggressiveness and metastatic potential in a number oftumours.34,35

We also tried to validate the biological significance ofthese high HIF-1a levels with other biochemical markersof tumoural progression such as STAT3. We found thatthis transcriptional activator is markedly upregulated intumoural masses that moderately express HIF-1a, butcuriously enough we did not observe the same correla-tion as HIF-1a levels increased. It is also worth to con-sider that these differences may be attributable to thedifferent microenvironments surrounding the tumouralmasses. Cells with higher optical activities are morelikely to be located in the core of these masses, whilecells on the outer face of the mass are able to alleviatetheir hypoxic condition in an easier way. Several reportsconfirm that high levels of hypoxia seem to promotetumour progression by regulating key cellular pro-grammes that can limit the clonogenic survival of hyp-oxic breast cancer cells in vitro36 or induce theexpression of tumour suppressors.37–39 Therefore, sur-vival of the tumoural cell in hypoxic conditions relies

Fig. 4. (a) Haematoxylin and eosin stainings of selected lung cross-sections affected by tumoural masses (20�) and close-ups of the tumoural cellssurrounding neo-vessels (40�); Scale bars 50 m (20�) and 25 m (40�); (b) Red fluorescent protein (RFP) immunostaining of lung cross-sectionsaffected with MDA-MB-231 tumoural cells. mCherry is a derivative of red fluorescent protein (RFP from Discosoma sp.), 33 and shares an 80%sequence homology with its parental protein. Scale bars 50 m (20�) and 25 m (40�).

270 P. Iglesias et al. / European Journal of Cancer 49 (2013) 264–271

on the balance between the pro-tumoural and tumoursuppressive effects,32 which fits in our case where a mod-erate HIF-1a expression correlates with a strong induc-tion of STAT3, indicating tumour progression, but notconversely where higher levels of this hypoxic factorcould hinder the viability of the tumoural cell.

In summary, we have devised and tested in vivo a met-astatic model using an optical biosensor inducible byhypoxia and the invasive breast cancer cell line MDA-MB 231. We were able to detect several optically activemasses with high HIF-1alpha activity and validated thebiological relevance of these data with an additionalmarker of tumour progression in breast cancer such asSTAT3. Although this biosensor cannot be directlytranslated into the clinical environment it serves as avaluable proof of concept and test benchmark for futurehypoxia sensing probes based on small molecules ornanodevices. These hypothetical BRET-powered tracerswould inform us about tumoural populations with highmetastatic potential, informing not only about the loca-tion, size and shape of the tumour, but also giving a dee-per view on biology of the tumour microenvironment.

Conflict of interest statement

The authors declare no competing financial interests.

Acknowledgements

We thank the members of Molecular Oncology Lab-oratory (MOL) for helpful discussions. We also thankProf. R.Y. Tsien and Dr. W.H. Suh for kindly providingus with some of the reagents used in this study. This

study was supported by the Spanish Ministry of Educa-tion and Science SAF2008-00543 and SAF2009-08629(JAC).

Appendix A. Supplementary data

Supplementary data associated with this article canbe found, in the online version, at http://dx.doi.org/10.1016/j.ejca.2012.06.001.

References

1. Torigian DA, Huang SS, Houseini M, Alavi A. Functionalimaging of cancer with emphasis on molecular techniques. CA

Cancer J Clin 2007;57:206–24.2. Sampath L et al.. Near infrared fluorescent optical imaging for

nodal staging. J Biomed Opt 2008;13:041312.3. Zhang Q et al.. Quantitative bioluminescence tomography guided

by diffuse optical tomography. Opt Express 2008;16:1481–6.4. Virostko J, Jansen ED, et al. Validation of bioluminescent imaging

techniques. Methods Mol Biol 2009;574:15–23.5. Wang GL, Jiang BH, Rue EA, Semenza GL. Hypoxia-inducible

factor 1 is a basic-helix-loop-helix-PAS heterodimer regulated bycellular O2 tension. Proc Natl Acad Sci USA 1995;92:5510–4.

6. Semenza GL. Targeting HIF-1 for cancer therapy. Nat Rev Cancer

2003;3:721–32.7. Zhong H, De Marzo AM, Laughner E, et al. Overexpression of

hypoxia-inducible factor 1 alpha in common human cancers andtheir metastases. Cancer Res 1999;59:5830–5.

8. Horvath CM, Wen Z, Darnell Jr JE. A STAT protein domain thatdetermines DNA sequence recognition suggests a novel DNA-binding domain. Genes Dev 1995;15:984–94.

9. Khazaie K, Schirrmacher V, Lichtner RB. EGF receptor inneoplasia and metastasis. Cancer Metastasis Rev 1993;12:255–74.

10. Ottenhoff-Kalff AE, Rijksen G, van Beurden EA, et al. Charac-terization of protein tyrosine kinases from human breast cancer:involvement of the c-src oncogene product. Cancer Res

1992;52:4773–8.

P. Iglesias et al. / European Journal of Cancer 49 (2013) 264–271 271

11. Buettner R, Mora LB, Jove R. Activated STAT signaling inhuman tumors provides novel molecular targets for therapeuticintervention. Clin Cancer Res 2002;8:945–54.

12. Garcia R, Bowman TL, Niu G, et al. Constitutive activation ofSTAT3 by the Src and JAK tyrosine kinases participates in growthregulation of human breast carcinoma cells. Oncogene

2001;20:2499–513.13. Watson CJ, Miller WR. Elevated levels of members of the STAT

family of transcription factors in breast carcinoma nuclearextracts. Br J Cancer 1995;71:840–4.

14. Gouilleux-Gruart V, Gouilleux F, Desaint C, et al. STAT-relatedtranscription factors are constitutively activated in peripheralblood cells from acute leukemia patients. Blood 1996;87:1692–7.

15. Ni Z, Lou W, Leman ES, Gao AC. Inhibition of constitutivelyactivated STAT3 signaling pathway suppresses growth of prostatecancer cells. Cancer Res 2000;60:1225–8.

16. Niu G, Bowman T, Huang M, et al. Roles of activated Src andStat3 signaling in melanoma tumor cell growth. Oncogene

2002;21:7001–10.17. Catlett-Falcone R, Landowski TH, Oshiro MM, et al. Constitu-

tive activation of Stat3 signaling confers resistance to apoptosis inhuman U266 myeloma cells. Immunity 1999;10:105–15.

18. Gray MJ, Zhang J, Ellis LM, et al. HIF-1alpha, STAT3, CBP/p300 and Ref-1/APE are components of a transcriptional complexthat regulates Src-dependent hypoxia-induced expression ofVEGF in pancreatic and prostate carcinomas. Oncogene

2005;24:3110–20.19. Iglesias P, Costoya JA. A novel BRET-based genetically encoded

biosensor for functional imaging of hypoxia. Biosens Bioelec

2009;10:3126–30.20. Bardos JI, Ashcroft M. Negative and positive regulation of HIF-1:

a complex network. BioEssays 2004;2004(26):262–9.21. Liao D, Johnson RS. Hypoxia: a key regulator of angiogenesis in

cancer. Cancer Metastasis Rev 2007;26:281–90.22. Akagi Y, Liu W, Zebrowski B, Xie K, Ellis LM. Regulation of

vascular endothelial growth factor expression in human coloncancer by insulin-like growth factor-I. Cancer Res

1998;58:4008–14.23. So MK, Xu C, Loening AM. Self-illuminating quantum dot

conjugates for in vivo imaging. Nat Biotechnol 2006;24:339–43.24. Lee JY, Lee YS, Kim KL, et al. A novel chimeric promoter that is

highly responsive to hypoxia and metals. Gene Ther

2006;13:857–68.25. Shaner NC, Campbell RE, Steinbach PA, et al. Improved

monomeric red, orange and yellow fluorescent proteins derivedfrom Discosoma sp. red fluorescent protein. Nat Biotechnol

2004;22:1567–72.

26. Price JE, Polyzos A, Zhang RD, Daniels LM. Tumorigenicity andmetastasis of human breast carcinoma cell lines in nude mice.Cancer Res 1990;50:717–21.

27. Indelicato M, Pucci B, Schito L, et al. Role of hypoxia andautophagy in MDA-MB-231 invasiveness. J Cell Physiol

2010;223:359–68.28. Kim MY, Oskarsson T, Acharyya S, et al. Tumor self-seeding by

circulating cancer cells. Cell 2009;139(7):1315–26.29. Bos PD, Nguyen DX, Massague J. Modeling metastasis in the

mouse. Curr Opin Pharmacol 2010;10:571–7.30. Fu OY, Hou MF, Yang SF, Huang SC, Lee WY. Cobalt chloride-

induced hypoxia modulates the invasive potential and matrixmetalloproteinases of primary and metastatic breast cancer cells.Anticancer Res 2009;29(8):3131–8.

31. Warnecke C, Griethe W, Weidemann A, et al. Activation of thehypoxia-inducible factor-pathway and stimulation of angiogenesisby application of prolyl hydroxylase inhibitors. FASEB J

2003;7(9):1186–8.32. Jubb AM, Buffa FM, Harris AL. Assessment of tumour hypoxia

for prediction of response to therapy and cancer prognosis. J Cell

Med 2010;14:18–29.33. Shaner NC, Steinbach PA, Tsien RY. A guide to choosing

fluorescent proteins. Nature 2005;2:905–9.34. Evens AM, Sehn LH, Farinha P, et al. Hypoxia-inducible factor-1

alpha expression predicts superior survival in patients with diffuselarge B-cell lymphoma treated with R-CHOP. J Clin Oncol

2010;28:1017–24.35. Schindl M, Schoppmann SF, Samonigg H, et al. Overexpression of

hypoxia-inducible factor 1alpha is associated with an unfavorableprognosis in lymph node-positive breast cancer. Clin Cancer Res

2002;8:1831–7.36. Blancher C, Moore JW, Talks KL, et al. Relationship of hypoxia-

inducible factor (HIF)-1alpha and HIF-2alpha expression tovascular endothelial growth factor induction and hypoxia survivalin human breast cancer cell lines. Cancer Res 2000;60:7106–13.

37. Boyd JM, Gallo GJ, Elangovan B, et al. Bik, a novel death-inducing protein shares a distinct sequence motif with Bcl-2 familyproteins and interacts with viral and cellular survival-promotingproteins. Oncogene 1995;11:1921–8.

38. Ivanova AV, Ivanov SV, Danilkovitch-Miagkova A, et al. Reg-ulation of STRA13 by the von Hippel–Lindau tumor suppressorprotein, hypoxia, and the UBC9/ubiquitin proteasome degrada-tion pathway. J Biol Chem 2001;276:15306–15.

39. Nishiwaki T, Daigo Y, Kawasoe T, et al. Isolation and mutationalanalysis of a novel human cDNA, DEC1 (deleted in esophagealcancer 1), derived from the tumor suppressor locus in 9q32. Genes

Chromosomes Cancer 2000;27:169–76.