Hypoxia alters gene expression in humanneuroblastoma cells toward an immatureand neural crest-like phenotypeAnnika Jögi*, Ingrid Øra*†, Helén Nilsson*, Åsa Lindeheim*, Yuichi Makino‡, Lorenz Poellinger‡, Håkan Axelson*,and Sven Påhlman*§

*Department of Laboratory Medicine, Division of Molecular Medicine, Lund University, University Hospital MAS, S-205 02 Malmö, Sweden;†Department of Pediatrics, Oncology–Hematology Section, Lund University Hospital, S-221 85 Lund, Sweden; and ‡Department of Cell andMolecular Biology, Medical Nobel Institute, Karolinska Institute, S-171 77 Stockholm, Sweden

Edited by George Klein, Karolinska Institute, Stockholm, Sweden, and approved March 26, 2002 (received for review December 11, 2001)

Insufficient oxygen and nutrient supply often restrain solid tumorgrowth, and the hypoxia-inducible factors (HIF) 1� and HIF-2� arekey transcription regulators of phenotypic adaptation to lowoxygen levels. Moreover, mouse gene disruption studies haveimplicated HIF-2� in embryonic regulation of tyrosine hydroxylase,a hallmark gene of the sympathetic nervous system. Neuroblas-toma tumors originate from immature sympathetic cells, andtherefore we investigated the effect of hypoxia on the differen-tiation status of human neuroblastoma cells. Hypoxia stabilizedHIF-1� and HIF-2� proteins and activated the expression of knownhypoxia-induced genes, such as vascular endothelial growth factorand tyrosine hydroxylase. These changes in gene expression alsooccurred in hypoxic regions of experimental neuroblastoma xeno-grafts grown in mice. In contrast, hypoxia decreased the expres-sion of several neuronal�neuroendocrine marker genes but in-duced genes expressed in neural crest sympathetic progenitors, forinstance c-kit and Notch-1. Thus, hypoxia apparently causes de-differentiation both in vitro and in vivo. These findings suggest anovel mechanism for selection of highly malignant tumor cells withstem-cell characteristics.

Solid tumors often have areas in which circulation is compro-mised because of structurally disorganized blood vessels andtumor cells that grow faster than the developing tumor capillarynetwork (1). The poor circulation results in selection of tumorcells that can grow or survive under conditions of hypoxia, poornutrient supply, low pH, and high intratumor pressure (2). Thismicroenvironment has prognostic implications, because cells inhypoxic areas are less vulnerable to ionizing radiation andcytotoxic drugs, and tumors with substantial hypoxia metastasizemore efficiently (3, 4). In an adaptive response to hypoxicconditions, cells alter their gene expression program, primarilyby action of the hypoxia-inducible factors (HIFs) 1� and HIF-2�(the latter also designated EPAS-1) (5–8). Hypoxia stabilizesthese two transcription factors against degradation (9–12) andthereby induces expression of several target genes involved inmaintaining homeostasis, for instance vascular endothelialgrowth factor (VEGF) and erythropoietin (5, 13, 14). HIF-1� andHIF-2� are essential for embryonic survival and proper vascu-larization (7, 14–16). HIF-2� is also necessary for the developingsympathetic nervous system (SNS) and shows a transient em-bryonic expression pattern confined to sympathetic ganglia andparaganglia (7). HIF-2�-deficient mice have been reported tolack sympathetic tyrosine hydroxylase (TH) expression and die ofcatecholamine shortage (7). The SNS is neural crest-derived andis composed of two major cell types, neurons and neuroendo-crine (chromaffin) cells (17), the latter forming the adrenalmedulla and the paraganglia. In the developing embryo andfetus, paraganglia are the main source of catecholamines thatregulate heart rate and blood pressure (18).

Neuroblastoma (NB) is a childhood malignancy originatingfrom the developing SNS (19–21). The tumor cells vary regard-

ing differentiation stage, with immature cells forming moreaggressive tumors (22). Most NBs exhibit characteristics ofimmature sympathetic neuroblasts, often with remaining neuralcrest traits (19, 21, 23). A phenotypically distinct subset of NBscontains a mixture of neuroblastic and neuroendocrine cell typesthat are organized in lobular structures with a central necroticzone (19). A spontaneous neuronal-to-neuroendocrine lineageshift occurs toward the necrotic zone (20). Therefore, we re-cently hypothesized that low oxygen levels and compromisednutrient supply may cause spontaneous neuroendocrine differ-entiation (24). Here, we examined whether oxygen deprivationaffects the differentiation status of cultured human NB cells.Unexpectedly, we found that hypoxic cells down-regulated SNSmarker genes, including the lineage-specific transcription factorsHASH-1 and dHAND. The cells further up-regulated genesexpressed during early neural crest development, hence thehypoxic cells seemed to adopt an immature, neural crest-likephenotype.

Materials and MethodsCell Culture. Cells were maintained as monolayers in standardmedia with 10% FCS at 37°C in 5% CO2 and 95% air. Hypoxicconditions were created by flushing 5% CO2 and 95% N2through a humidified chamber at 37°C, until an atmospherecontaining 1% or 5% O2 was achieved, and measured with aMiniOX1 oxygen meter (Mine Safety Appliances Company,Pittsburgh). Anoxia (0% O2, 5–10% CO2 and 90–95% N2) wasestablished in an anaerobic workstation (Electrotek, Keighleg,U.K.). Culture media were preequilibrated at the indicatedoxygen levels.

Immunoblot Analysis. Cells were lysed in 1% Nonidet P-40, 10%glycerol, 20 mM Tris�HCl (pH 8.0), 137 mM NaCl, and Completeprotease inhibitor mixture (Roche Molecular Biochemicals).Western blotting was performed with 60 �g of total protein perlane. The following primary Abs were used: HIF-2�, affinity-purified rabbit antiserum directed against amino acids 138–152of human HIF-2� (custom-made by Innovagen, Lund, Sweden);HIF-1�, mAb (Novus Biologicals, Littleton, CO); TH, mAb(Roche Molecular Biochemicals); chromogranin, rabbit anti-serum (Euro-Diagnostica, Malmö, Sweden); and Notch-1, goatpolyclonal IgG (Santa Cruz Biotechnology). Super Signal sub-strate (Pierce) was used for chemiluminescence detection of thesecondary Abs.

This paper was submitted directly (Track II) to the PNAS office.

Abbreviations: HIF, hypoxia-inducible factor; NB, neuroblastoma; SNS, sympathetic ner-vous system.

§To whom reprint requests should be addressed. E-mail: sven.pahlman@molmed.mas.lu.se.

The publication costs of this article were defrayed in part by page charge payment. Thisarticle must therefore be hereby marked “advertisement” in accordance with 18 U.S.C.§1734 solely to indicate this fact.

www.pnas.org�cgi�doi�10.1073�pnas.102660199 PNAS � May 14, 2002 � vol. 99 � no. 10 � 7021–7026

MED

ICA

LSC

IEN

CES

Dow

nloa

ded

by g

uest

on

Apr

il 8,

202

1

Northern Blot Hybridizations and Semiquantitative Reverse Transcrip-tion (RT)-PCR. Total RNA was prepared with Trizol reagent (LifeTechnologies, Rockville, MD), and 15 �g of total RNA per lanewas analyzed with probes as described (23, 25, 26). The inhibitorof DNA binding 2 (Id2) probe covered nucleotides 111–515.Expression of c-kit was analyzed by RT-PCR with �2-microglobulin as an internal loading control as described (27).Total RNA was treated with DNase I (Appligene, Strasbourg,France), followed by cDNA synthesis and PCR amplificationaccording to standard procedures (denaturation 30 s at 95°C;annealing 30 s at 55°C; extension 1 min at 72°C). The PCRproducts were analyzed by electrophoresis, stained with SYBRGreen I (Roche Diagnostics), and evaluated with LAS-1000CCD imaging equipment (Fuji).

Experimental Tumors, Immunohistochemistry, and in Situ Hybridiza-tion. SK-N-BE(2) cells were injected s.c. into female athymicmice of the NMRI strain (nu�nu) as described (28). All proce-dures were approved by the regional ethical committee foranimal research (Dnr.M101-99). Tumors were recovered, fixedin diethyl pyrocarbonate-treated 4% buffered formaldehyde,and embedded in paraffin. Immunohistochemistry and in situhybridization were done as described (19, 20, 23).

ResultsHIF-2� Expression in Embryonal Mouse Paraganglia. Of the twopublished HIF-2� gene targeting studies, only one reports aneffect on the SNS (7, 16). Therefore, we examined abdominalsections of embryonic day (E) 14.5 embryonic mice for HIF-2�expression in the SNS. In situ hybridization showed distinctHIF-2� expression in paraganglia cells of the organ of Zucker-kandl, identified morphologically and by high TH levels (Fig. 1).At this developmental stage, the adjacent sympathetic ganglion,which weakly expressed TH (Fig. 1) and GAP-43 (not shown),did not express HIF-2� in accordance with published data (7).

HIF-1� and HIF-2� Are Accumulated in Hypoxic NB Cells. We subse-quently studied the impact of 4 h of exposure to normoxic (21%

O2) and hypoxic (5%, 1%, and 0% O2) conditions on stabiliza-tion of HIF-1� and HIF-2� protein in cultured human NB cellswith HeLa cells as a positive control (29). Considering theextensive homology between HIF-1� and HIF-2�, anti-HIF Abspecificities were verified with HIF proteins translated in vitro(Fig. 2A). Hypoxic stress resulted in accumulation of HIF-1� inall five analyzed NB cell lines, whereas HIF-2� was detectable inthree cell lines (Fig. 2 A). Neither HIF-1� nor HIF-2� wasdetected in cells grown under normoxic conditions or at 5%oxygen, which approximates tissue levels (30). There was nocorrelation between the extent of hypoxia-induced accumulationof HIF-1� and HIF-2�, as exemplified by comparison of IMR-32and SH-SY5Y cells (Fig. 2 A). At 1% oxygen, all tested cellssurvived for at least 72 h with no apparent effect on cellmorphology or increase in cell death (not shown). Lower oxygenlevels (approaching 0%) had no effect on cell morphology after4 h, but most of the cells died within 72 h. Therefore, subsequentstudies were performed at 1% oxygen.

Expression of TH and Other Known Hypoxia-Inducible Genes. MostNB tumors and cell lines produce catecholamines, hence theyalso express TH. One function of the SNS is to produce andexcrete noradrenaline and adrenaline into the blood stream toregulate heart rate and blood pressure, for example in responseto low oxygenation, and it is well established that low oxygenlevels can induce TH expression (31). In six of seven NB cell lines,TH expression was induced after growth in 1% oxygen for 72 h(Fig. 2B). Both normoxic and hypoxic TH levels varied consid-erably between the cell lines, probably reflecting that thesetumor cells are arrested at different stages of maturation or thatexpression levels are low because of a mixed cholinergic�noradrenergic phenotype, as exemplified by the LA-N-2 cells.The sympathetic catecholaminergic lineage specificity of hy-poxia-dependent TH accumulation was demonstrated by thelack of TH expression in HeLa and SK-N-MC neuroepitheliomacells, irrespective of oxygen levels (Fig. 2B).

We also analyzed VEGF expression as part of a more detailedcharacterization of the hypoxic NB phenotype. At 1% oxygen, allfour tested cell lines showed pronounced up-regulation of VEGFwithin 4 h, and hypoxia-induced expression of this gene persisted3 days later (Fig. 2C). Initially, VEGF expression was notdetected in cells grown under normoxic conditions, except forweak expression in KCN-69n cells (Fig. 2C, 4-h time points), butafter 72 h, VEGF was expressed in some of the cell lines, showingthat VEGF can be transcribed in a nonhypoxic environment. Weanalyzed the expression of two additional hypoxia-inducedgenes: insulin-like growth factor 2 (IGF-2) (32) and the glycolyticenzyme and housekeeping gene glyceraldehyde-3-phosphate de-hydrogenase (GAPDH) (33). IGF-2 was not expressed in NB cellsexposed to normoxia or hypoxia for 4 h but it was distinctlyexpressed in three of four cell lines after 3 days of hypoxia, andGAPDH expression was augmented in all NB cell lines grown in1% oxygen, although with some variation in the expressionkinetics (Fig. 2C).

Down-Regulation of SNS Marker Genes in Hypoxic NB Cells. Althoughelevated expression levels of TH and IGF-2 are known indicatorsof a hypoxic phenotype, such a rise can also be the result ofinduced sympathetic neuroendocrine differentiation (19).Therefore, we examined the effects of oxygen deprivation on thedifferentiation status of NB cells by investigating the expressionpatterns of the neuroendocrine chromaffin marker protein,chromogranin, and the sympathetic neuronal peptide neuro-transmitter gene, neuropeptide tyrosine (NPY) (19, 20). All fivetested NB cell lines expressed chromogranin, and under hypoxicconditions the level decreased in four of these (Fig. 3A). In thefifth cell line, SH-SY5Y, hypoxia did not alter the chromograninlevel. The same samples were analyzed for TH expression,

Fig. 1. Expression of TH and HIF-2� in mouse embryonal (E14.5) paragan-glion (PG) cells of the organ of Zuckerkandl. TH expression is strong in PG cellsand weak in the neuroblasts of the adjacent sympathetic ganglion (SG)(Upper). HIF-2� is expressed in the PG but not in the sympathetic neuroblastsat this developmental stage (Lower). (Bar � 90 �m.)

7022 � www.pnas.org�cgi�doi�10.1073�pnas.102660199 Jögi et al.

Dow

nloa

ded

by g

uest

on

Apr

il 8,

202

1

confirming increased TH levels in the hypoxic cells. NPY mRNAwas abundant in SK-N-BE(2), KCN-69n, and SH-SY5Y cells butabsent in IMR-32 cells (Fig. 3B). Under hypoxic conditions, NPYwas down-regulated within 4 h in SK-N-BE(2) and KCN-69ncells, and the differences between hypoxic and control sampleswere more pronounced after 3 days. In contrast, hypoxia did notdown-regulate NPY in SH-SY5Y cells (Fig. 3B), which is note-worthy because these cells differ from the other cell lines in atleast one important aspect: N-myc is not amplified in SH-SY5Ycells (25). We therefore tested two additional NB cell lineswithout N-myc amplification, SK-N-F1 and SK-N-RA. TH ex-pression increased at 1% oxygen, whereas chromogranin andNPY levels decreased (Fig. 3 A and C), suggesting that the weakresponse to hypoxia in SH-SY5Y cells was not associated withthe N-myc expression status.

The chromogranin and NPY expression data suggested that,compared with cells grown at normoxic conditions, hypoxic NBcells acquire a less mature phenotype. Therefore, we investigatedthe expression patterns of HASH-1, dHAND, and N-myc, threetranscription factor genes involved in early sympathetic lineagespecification and development (23, 34). Hypoxia caused down-regulation of HASH-1 and dHAND, but SH-SY5Y cells wereagain an exception (Fig. 3 B and C). N-myc is expressed duringneural crest cell migration and early phases of ganglionic neuralcrest differentiation (34), and human embryonal sympatheticganglia express N-myc at least up to week 8.5 (S.P., unpublished

observation). In hypoxic NB cells with N-myc amplification,N-myc expression was distinctly down-regulated after 72 h(Fig. 3B).

Hypoxia-Induced Expression of Neural Crest Genes. Because ourresults suggested that hypoxia-treated NB cells lose their sym-pathetic ganglionic phenotype, we examined the effects ofhypoxia on expression of neural crest genes. Id2, Notch-1, HES-1,and c-kit are involved in determination of neural crest cell fate(35–38) and are expressed in sympathetic precursor cells atearlier developmental stages than, for instance, N-myc, HASH-1,and dHAND (21). We observed Id2 expression in all four testedcell lines, and this expression decreased with time in culture.Hypoxia led to an increased Id2 expression after both 4 and 72 h(Fig. 3B). Expression of c-kit was detected in only two cell linesat normoxia but occurred in all four tested NB cell lines aftergrowth at low oxygen levels (Fig. 3E). Hypoxia also increasedNotch-1 levels (Fig. 3D), whereas HES-1 protein was detected inthree cell lines under normoxic conditions, and hypoxia in-creased the levels in two of these cell lines (not shown).

In Vivo Gene Expression Patterns in Human NB Xenografts. To studythe effects of hypoxia on gene expression in solid tumors in vivo,we mimicked such conditions by growing SK-N-BE(2) cells asxenografts in nude mice (28). Viable cells surrounding necroticareas in the tumors exhibited high VEGF expression, whereas

Fig. 2. Hypoxia-induced gene expression in oxygen-deprived NB cells. (A) Accumulation of HIF-1� and HIF-2� in NB cells in response to oxygen deprivation.Human NB cells were exposed to the indicated oxygen levels for 4 h and subjected to Western blot analysis. Membranes containing identical protein lysates wereanalyzed with anti-HIF-1� or anti-HIF-2� Abs, which revealed accumulation of proteins of about 120 kDa, in cells grown at 0% or 1% oxygen. A protein unknownto us (indicated by *) appeared in the HIF-2� blots in all experiments. Hypoxia-exposed HeLa cells were used as positive controls for HIF-1� and HIF-2� proteinaccumulation, also verified with in vitro-translated (i v t) human HIF-1� and mouse HIF-2�, respectively. (B) Western blot analysis showing increased TH levelsin NB cells after a 72-h exposure to hypoxia. The non-NB cell lines, HeLa and SK-N-MC, did not express TH. TH expression was detected in SH-SY5Y cells whenthe film was exposed for a longer time (Right). (C) Northern blot analysis of total RNA from NB cell lines, showing hypoxia-induced expression of VEGF, GAPDH,and IGF-2. The 28S-RNA level was used as a loading control. Representative data of more than two experiments are shown.

Jögi et al. PNAS � May 14, 2002 � vol. 99 � no. 10 � 7023

MED

ICA

LSC

IEN

CES

Dow

nloa

ded

by g

uest

on

Apr

il 8,

202

1

most cells in other parts of the tumors were VEGF-negative (Fig.4 B–D and not shown). The VEGF-positive cells also expressedhigh levels of TH and IGF-2, but low levels of chromogranin (notshown). Furthermore, expression of the neuronal differentiationmarker genes dHAND and GAP-43 was low in cells expressingVEGF, IGF-2, and TH, as compared with the expression innonnecrotic, well vascularized areas of the tumor (Figs. 4 and 5).In summary, hypoxic SK-N-BE(2) cells seem to develop essen-tially the same immature phenotype, regardless of whether theyare grown in experimental tumors or under hypoxic conditionsin vitro. To test whether the hypoxia-induced phenotypicalchanges render SK-N-BE(2) cells other growth properties invivo, these cells were cultured in vitro under normoxic or hypoxicconditions for 3 days before injection into nude mice. Thehypoxia-treated cells tended to form palpable tumors earlierthan control cells (Fig. 6A). In addition, the tumors generatedfrom hypoxia-pretreated cells appeared to grow faster andbecome slightly larger in a shorter time than the correspondingtumors from control cells (Fig. 6 A and B).

DiscussionThe observations that NB tumors display distinct differences instages of cell maturation, and that there is a strong correlationbetween tumor cell differentiation stage and prognosis, haveprovided an important basis for the concept of tumor celldifferentiation. We have found that spontaneous neuroendo-crine differentiation can occur in tumor areas with poor oxy-genation in a subset of NBs (20, 24). Given the apparentrequirement of HIF-2� for proper SNS development and theexpression of HIF-2� in SNS cells during embryogenesis (7),these observations prompted us to determine whether hypoxiainfluences the differentiation status of NB cells. We found thatall hypoxic NB cell lines accumulated HIF-1� but not allaccumulated HIF-2�, which may reflect that these cell lines arearrested at different stages of maturation, given the temporalSNS expression of HIF-2� during normal development. Thehypoxic NB cells increased the expression of TH and otherhypoxia-inducible genes, such as VEGF, IGF-2, and GAPDH.

Concurrent with the induction of these genes, hypoxic NB cellsdown-regulated neuronal and neuroendocrine marker genes andup-regulated genes expressed during normal development of theneural crest (Fig. 6C). Furthermore, xenografted hypoxia-pretreated cells tended to form tumors earlier and grow slightlyfaster than grafted control cells. Thus, over a period of days,hypoxia induces complex changes in the gene expression patternboth in vitro and in vivo, which strongly suggest that dedifferen-tiation and acquisition of a neural crest-like phenotype in NBcells is an overall effect of growth at low oxygen levels. Theinitiating molecular event leading to dedifferentiation has notbeen explored, but the hypoxia-induced expression of Id2,Notch-1, and HES-1 will contribute to development of a non-neuronal phenotype. Id2 would act by sequestering E proteins,thereby preventing the proneuronal effect of HASH-1 anddHAND and Notch-1�HES-1 by inhibition of HASH-1 expres-sion (36, 37). Thus, these changes in expression will act in concertand counteract a neuronal phenotype.

The tested NB cell lines showed no induced neuroendocrineor neuronal differentiation at 1% oxygen. Indeed, oxygen pres-sure can be important for SNS precursor cell development, andit was recently shown that growth of primary neural crest cells at5% oxygen, which resembles physiological levels, promotessympathoadrenal differentiation (30). At this oxygen level therewas no accumulation of either HIF-1� or HIF-2� in the NB cells(Fig. 2 A), suggesting that major hypoxia-driven changes intranscription were not activated under this condition. SH-SY5Ycells, which have a normal N-myc copy number, seemed resistantto hypoxia-induced decrease in neuronal�neuroendocrinemarker gene expression. This lack of response could not beattributed to low N-myc expression as two other non-N-myc-amplified NB cell lines did down-regulate these marker genes.Our data could suggest that SH-SY5Y cells are phenotypicallycloser to differentiating NB cells found in tumors exhibiting aspontaneous neuronal-to-neuroendocrine lineage shift in hy-poxic regions (20, 24). Such tumors are usually of low clinicalstage, whereas virtually all NB cell lines are derived fromhigh-stage tumors. Consequently, cell line studies may be inap-

Fig. 3. Down-regulation of SNS marker genes and induced expression of neural crest genes in hypoxic NB cells. (A) Western blot analysis of chromogranin andTH in NB cells exposed to 21% or 1% oxygen for 72 h. (B and C) Northern blot analysis of neuronal and neural crest marker genes in normoxic and hypoxic NBcells grown for 4 (B) and 72 h (B and C). We used 28S-RNA as a loading control. (D) Western blot analysis of Notch-1 in cells exposed to normoxia or hypoxia for72 h. (E) Reverse transcription–PCR analysis of c-kit expression in hypoxic (72 h) NB cells. Beta-2-microglobulin (�2M) was a reference for input cDNA.Representative data of more than two experiments are shown.

7024 � www.pnas.org�cgi�doi�10.1073�pnas.102660199 Jögi et al.

Dow

nloa

ded

by g

uest

on

Apr

il 8,

202

1

propriate for elucidating the putative capacity of low-stagetumor cells to undergo neuroendocrine differentiation in re-sponse to hypoxia.

To our knowledge, no investigations have addressed thequestion of whether cells in hypoxic regions of aggressive solidtumors are generally less mature than oxygenated cells. How-ever, it has been reported that hypoxia leads to up-regulation oftelomerase activity (39), which is a characteristic sign of dividingimmature progenitor cells. Moreover, the level of HIF-1� pro-tein in breast carcinomas, presumably reflecting hypoxic condi-tions, is higher in poorly differentiated than in well differentiatedlesions (40). Tumor aggressiveness and tumor cell differentiationstage are clearly correlated in NBs, because, when highly ma-lignant (clinical stages 3 and 4), these tumors express low levelsof neuronal differentiation marker genes (22). In light of the datapresented here, it will be important to examine whether there isa reciprocal correlation between neuronal marker gene expres-sion and expression of neural crest genes in high-stage tumors.Clinically, NB aggressiveness depends on the extent to whichthey metastasize (41). Accordingly, the metastatic phenotype ofhigh-stage NBs might reflect that these cells have features that

mimic the high migratory capacity of neural crest-derived pro-genitor cells. Therefore, we suggest that the hypoxia-inducedshift toward a neural crest-like phenotype reported here resultsin more aggressive tumor cells with increased potential to

Fig. 4. Marker gene expression in human SK-N-BE(2) NB cells grown for 3weeks as xenograft tumors in nude mice. (A) Hematoxylin�eosin (H&E)-stained section of a tumor, indicating areas analyzed in the experimentsillustrated in this figure and Fig. and 5. Necrotic (*) and well vascularized (‚)areas are indicated. (B–D) In situ hybridization showing VEGF expressionadjacent to a necrotic (*) area. C is a magnification of the framed area in B. Dis a dark-field representation of B. (E–H) TH and IGF-2 expression in adjacentsections. G depicts a magnification of the framed area in E, and H is thedark-field representation of F. [Bars � 850 (A); 100 (B, D, and F–H); 25 (C); and400 �m (E).]

Fig. 5. Neuronal marker gene expression in hypoxic and vascularized regionsof an NB xenograft tumor. Analysis of tumor sections taken adjacent to thoseanalyzed in Fig. 4. In situ hybridization of GAP-43 (A–D) and dHAND (E–H) inhypoxic (*) and vascularized (‚) areas. (Bars � 25 �m.)

Fig. 6. (A and B) Formation of tumors in nude mice after injection of SK-N-BE(2)cells precultured for 72 h at 1% or 21% O2. The animals were killed when thetumors reached a diameter of 15 mm. (A) Mean time until tumor take (open bars)and termination (filled bars). Data are given as mean of two separate experi-ments with five mice in each group. In one experiment, one mouse in each groupdidnotdevelopatumorwithin3weeks. (B)Weightsof therecoveredtumors (P�0.068). (C) Schematic representation of changes in gene expression in N-myc-amplified NB cells exposed to hypoxia. In hypoxic cells, neuronal marker geneschromogranin, dHAND, GAP-43, HASH-1, N-myc, and NPY (solid line) are down-regulated, whereas hypoxia-driven genes GAPDH, IGF-2, TH, and VEGF (dots) andneural-crestmarkergenesc-kit,HES-1, Id2,andNotch-1 (dashedline)are induced.

Jögi et al. PNAS � May 14, 2002 � vol. 99 � no. 10 � 7025

MED

ICA

LSC

IEN

CES

Dow

nloa

ded

by g

uest

on

Apr

il 8,

202

1

metastasize (Fig. 6). This suggestion is in accordance with recentinvestigations demonstrating that cells in hypoxic tumors (otherthan NBs) show an increased tendency toward metastatic spread(3, 4). Interestingly, and consistent with our findings, a corre-lation has been found between congenital heart disease withcyanosis in infants and occurrence of NB (42).

Hypoxia-induced cell heterogeneity may also affect tumor cellselection mechanisms. The subpopulation of treatment-resistantcells in solid tumors includes hypoxic cells, which, in poorlyvascularized and hypoxic regions, can elude treatment becausethey are inaccessible to cytostatic drugs, and irradiation-inducedfree-radical production is low because of low oxygen levels.Moreover, the mutation rate has been shown to increase intumor cells in hypoxic regions (43). Many of the genes expressedin hypoxic NB cells determine stem cell features, such asself-renewal, survival, and migration, and these genes are alsoimplicated in the growth and spread of aggressive cancers. Ifmutations or rearrangements occur in such genes and lead toconstitutive active proteins, the affected cells may acquire a

growth and�or survival advantage that improves their ability towithstand selection pressure.

Together, our results show that low oxygen tension in NB cellsleads to dedifferentiation and an immature neural crest-likephenotype. We hypothesize that dedifferentiation is a generalphenomenon in solid tumors, and that a low oxygen level, whichis known to increase mutation frequency and promote metastaticspread, contributes to selection of immature, highly malignanttumor cells with stem cell characteristics. This phenomenonwould define a novel mechanism by which hypoxia contributes tothe malignant progression of solid tumors.

We thank Dr. Kristian Riesbeck for valuable help and Ms. CarolinJönsson for skillful technical assistance. This work was supported bygrants from the Swedish Cancer Society, the Children’s Cancer Foun-dation of Sweden, the Swedish Foundation for Strategic Research, HKHKronprinsessan Lovisas Förening för Barnasjukvård, Hans von Kant-zows Stiftelse, and by the research funds of Malmö and Lund UniversityHospitals.

1. Brown, J. M. (2000) Mol. Med. Today 6, 157–162.2. Semenza, G. L. (2000) Crit. Rev. Biochem. Mol. Biol. 35, 71–103.3. Brizel, D. M., Scully, S. P., Harrelson, J. M., Layfield, L. J., Bean, J. M.,

Prosnitz, L. R. & Dewhirst, M. W. (1996) Cancer Res. 56, 941–943.4. Zhong, H., De Marzo, A. M., Laughner, E., Lim, M., Hilton, D. A., Zagzag,

D., Buechler, P., Isaacs, W. B., Semenza, G. L. & Simons, J. W. (1999) CancerRes. 59, 5830–5835.

5. Wang, G. L., Jiang, B. H., Rue, E. A. & Semenza, G. L. (1995) Proc. Natl. Acad.Sci. USA 92, 5510–5514.

6. Ema, M., Taya, S., Yokotani, N., Sogawa, K., Matsuda, Y. & Fujii-Kuriyama,Y. (1997) Proc. Natl. Acad. Sci. USA 94, 4273–4278.

7. Tian, H., Hammer, R. E., Matsumoto, A. M., Russell, D. W. & McKnight, S. L.(1998) Genes Dev. 12, 3320–3324.

8. Wiesener, M. S., Turley, H., Allen, W. E., Willam, C., Eckardt, K. U., Talks,K. L., Wood, S. M., Gatter, K. C., Harris, A. L., Pugh, C. W., et al. (1998) Blood92, 2260–2268.

9. Cockman, M. E., Masson, N., Mole, D. R., Jaakkola, P., Chang, G. W., Clifford,S. C., Maher, E. R., Pugh, C. W., Ratcliffe, P. J. & Maxwell, P. H. (2000) J. Biol.Chem. 275, 25733–25741.

10. Ohh, M., Park, C. W., Ivan, M., Hoffman, M. A., Kim, T. Y., Huang, L. E.,Pavletich, N., Chau, V. & Kaelin, W. G. (2000) Nat. Cell Biol. 2, 423–427.

11. Tanimoto, K., Makino, Y., Pereira, T. & Poellinger, L. (2000) EMBO J. 19,4298–4309.

12. Maxwell, P. H., Wiesener, M. S., Chang, G. W., Clifford, S. C., Vaux, E. C.,Cockman, M. E., Wykoff, C. C., Pugh, C. W., Maher, E. R. & Ratcliffe, P. J.(1999) Nature (London) 399, 271–275.

13. Carmeliet, P., Dor, Y., Herbert, J. M., Fukumura, D., Brusselmans, K.,Dewerchin, M., Neeman, M., Bono, F., Abramovitch, R., Maxwell, P., et al.(1998) Nature (London) 394, 485–490.

14. Iyer, N. V., Kotch, L. E., Agani, F., Leung, S. W., Laughner, E., Wenger, R. H.,Gassmann, M., Gearhart, J. D., Lawler, A. M., Yu, A. Y. & Semenza, G. L.(1998) Genes Dev. 12, 149–162.

15. Ryan, H. E., Lo, J. & Johnson, R. S. (1998) EMBO J. 17, 3005–3015.16. Peng, J., Zhang, L., Drysdale, L. & Fong, G. H. (2000) Proc. Natl. Acad. Sci.

USA 97, 8386–8391.17. Le Douarin, N. M. & Smith, J. (1988) Annu. Rev. Cell Biol. 4, 375–404.18. Padbury, J. (1989) Baill. Clin. Endocrinol. Metab. 3, 689–705.19. Hoehner, J. C., Gestblom, C., Hedborg, F., Sandstedt, B., Olsen, L. & Påhlman,

S. (1996) Lab. Invest. 75, 659–675.20. Gestblom, C., Hoehner, J. C., Hedborg, F., Sandstedt, B. & Påhlman, S. (1997)

Am. J. Pathol. 150, 107–117.21. Påhlman, S. & Hedborg, F. (2000) in Neuroblastoma, eds. Brodeur, G. M.,

Sawada, T., Tsuchida, Y. & Voute, P. A. (Elsevier Science, Amsterdam), pp.9–19.

22. Hedborg, F., Bjelfman, C., Sparen, P., Sandstedt, B. & Påhlman, S. (1995) Eur.J. Cancer 4, 435–443.

23. Gestblom, C., Grynfeld, A., Øra, I., Örtoft, E., Larsson, C., Axelson, H., Sandstedt,B., Cserjesi, P., Olson, E. N. & Påhlman, S. (1999) Lab. Invest. 79, 67–79.

24. Hoehner, J. C., Gestblom, C., Olsen, L. & Påhlman, S. (1997) Br. J. Cancer. 75,1185–1194.

25. Hammerling, U., Bjelfman, C. & Påhlman, S. (1987) Oncogene 2, 73–77.26. Zelzer, E., Levy, Y., Kahana, C., Shilo, B. Z., Rubinstein, M. & Cohen, B.

(1998) EMBO J. 17, 5085–5094.27. Cohen, P. S., Chan, J. P., Lipkunskaya, M., Biedler, J. L. & Seeger, R. C. (1994)

Blood 84, 3465–3472.28. Øra, I., Bondesson, L., Jönsson, C., Ljungberg, J., Pörn-Ares, I., Garwicz, S. &

Påhlman, S. (2000) Biochem. Biophys. Res. Commun. 277, 179–185.29. Kallio, P. J., Pongratz, I., Gradin, K., McGuire, J. & Poellinger, L. (1997) Proc.

Natl. Acad. Sci. USA 94, 5667–5672.30. Morrison, S. J., Csete, M., Groves, A. K., Melega, W., Wold, B. & Anderson,

D. J. (2000) J. Neurosci. 20, 7370–7376.31. Czyzyk-Krzeska, M. F., Bayliss, D. A., Lawson, E. E. & Millhorn, D. E. (1992)

J. Neurochem. 58, 1538–1546.32. Bae, S. K., Bae, M. H., Ahn, M. Y., Son, M. J., Lee, Y. M., Bae, M. K., Lee,

O. H., Park, B. C. & Kim, K. W. (1999) Cancer Res. 59, 5989–5994.33. Zhong, H. & Simons, J. W. (1999) Biochem. Biophys. Res. Commun. 259,

523–526.34. Wakamatsu, Y., Watanabe, Y., Nakamura, H. & Kondoh, H. (1997) Develop-

ment (Cambridge, U.K.) 124, 1953–1962.35. Martinsen, B. J. & Bronner-Fraser, M. (1998) Science 281, 988–991.36. Wakamatsu, Y., Maynard, T. M. & Weston, J. A. (2000) Development (Cam-

bridge, U.K.) 127, 2811–2821.37. Ishibashi, M., Ang, S. L., Shiota, K., Nakanishi, S., Kageyama, R. & Guillemot,

F. (1995) Genes Dev. 9, 3136–3148.38. Langtimm-Sedlak, C. J., Schroeder, B., Saskowski, J. L., Carnahan, J. F. &

Sieber-Blum, M. (1996) Dev. Biol. 174, 345–359.39. Seimiya, H., Tanji, M., Oh-hara, T., Tomida, A., Naasani, I. & Tsuruo, T. (1999)

Biochem. Biophys. Res. Commun. 260, 365–370.40. Bos, R., Zhong, H., Hanrahan, C. F., Mommers, E. C., Semenza, G. L., Pinedo,

H. M., Abeloff, M. D., Simons, J. W., van Diest, P. J. & van der Wall, E. (2001)J. Natl. Cancer Inst. 93, 309–314.

41. Brodeur, G. M., Pritchard, J., Berthold, F., Carlsen, N. L., Castel, V.,Castelberry, R. P., De Bernardi, B., Evans, A. E., Favrot, M., Hedborg, F., etal. (1993) J. Clin. Oncol. 11, 1466–1477.

42. de la Monte, S. M., Hutchins, G. M. & Moore, G. W. (1985) Am. J. Pediatr.Hematol. Oncol. 7, 109–116.

43. Reynolds, T. Y., Rockwell, S. & Glazer, P. M. (1996) Cancer Res. 56,5754–5757.

7026 � www.pnas.org�cgi�doi�10.1073�pnas.102660199 Jögi et al.

Dow

nloa

ded

by g

uest

on

Apr

il 8,

202

1