Bone morphogenetic protein antagonist gremlin 1 is widely expressed by cancer-associated stromal cells and can promote tumor cell proliferation


Julie B. Sneddon*, Hanson H. Zhen, Kelli Montgomery, Matt van de Rijn, Aaron D. Tward, Robert West, Hayes Gladstone, Howard Y. Chang, Greg S. Morganroth, Anthony E. Oro, and Patrick O. Brown*,||

Departments of *Biochemistry, Dermatology, and Pathology, and Howard Hughes Medical Institute, Stanford University Medical Center, Stanford, CA 94305; and G. W. Hooper Foundation, University of California, San Francisco, CA 94143

Contributed by Patrick O. Brown, August 9, 2006


Although tissue microenvironments play critical roles in epithelialdevelopment and tumorigenesis, the factors mediating these effectsare poorly understood. In this work, we used a genomic approachto identify factors produced by cells in the microenvironmentof basal cell carcinoma (BCC) of the skin, one of the most commonhuman cancers. The global gene expression programs of stromalcell cultures derived from human BCCs showed consistent, systematicdifferences from those derived from nontumor skin. The genemost consistently expressed at a higher level in BCC tumor stromalcells compared with those from nontumor skin was GREMLIN 1,which encodes a secreted antagonist of the bone morphogeneticprotein (BMP) pathway. BMPs and their antagonists are knownto play a crucial role in stem and progenitor cell biology asregulators of the balance between expansion and differentiation.Consistent with the hypothesis that BMP antagonists might havea similar role in cancer, we found GREMLIN 1 expression in thestroma of human BCC tumors but not in normal skin in vivo. Furthermore,BMP 2 and 4 are expressed by BCC cells. Ex vivo, BMP inhibits,and Gremlin 1 promotes, proliferation of cultured BCC cells.We further found that GREMLIN 1 is expressed by stromal cellsin many carcinomas but not in the corresponding normal tissuecounterparts that we examined. Our data suggest that BMP antagonistsmay be important constituents of tumor stroma, providing a favorablemicroenvironment for cancer cell survival and expansion in manycancers.

cancer biology | stem cell regulation | tissue microenvironment | tumor stroma

PNAS | October 3, 2006 | vol. 103 | no. 40 | 14842-14847. OPEN ACCESS ARTICLE.


Tissue microenvironments play a critical role in specifyingcellular niches in both the developing embryo and adult organisms(1, 2). In development, cell fate decisions are dictated notonly by cell-autonomous signals but also by stimuli from thesurrounding tissue microenvironment (3, 4). Similarly, in adulttissues that continue to renew throughout the lifetime of theorganism, such as the skin, intestinal epithelium, and hematopoieticsystem, the self renewal and maturation of the stem cell populationare regulated by specific molecular cues derived from the correspondingmicroenvironments (57). In the skin, hair follicle morphogenesisis regulated by signals coming from the dermal papilla, a specializedmesenchymal structure that signals to matrix stem cells locatedacross the basement membrane (8, 9). Similarly, the modulationof stem cell activity in the intestine is also subject to cuesderived from underlying mesenchymal cells that surround thecrypt (10, 11). Hematopoietic stem cells are regulated in partby osteoblasts, cells that reside in the adjacent bone spicule(12, 13). In all of these cases, a crucial feature of the regulationof stem cell compartment size, location, and timing of selfrenewal is the production of critical factors by a specializedset of mesenchymal cells that create a customized microenvironment.

During carcinogenesis, an analogous system of specialized tissuemicroenvironment cells may also be important in specifying a"tumor cell niche" that supports a self-renewing populationof tumor cells. Paradoxically, although uncontrolled proliferationand survival are the cardinal characteristics of cancer cells,it can be difficult to sustain these cells away from their correspondingmicroenvironment, either in culture or as explants (14). Thereis accumulating evidence that tumor stroma influences tumordevelopment (15, 16). Genetic studies have shown that stromalcells are altered in some inherited cancer-susceptibility syndromes(17). In breast cancer, rearrangements at several loci havebeen noted exclusively in tumor-associated stromal cells (18).In vivo and in vitro experiments demonstrated that human prostaticepithelial cells showed dramatic changes both in histology andgrowth rate when grown with human fibroblast cells derived fromprostatic carcinoma, suggesting that carcinoma-derived fibroblastscan stimulate tumorigenesis (19). Others have shown that coinjectionof fibroblasts with tumor epithelial cells into mice can enhancetumor formation (20).

To identify factors produced by tumor stromal cells that contributeto the initiation or maintenance of the tumor, we used a genomicapproach with basal cell carcinoma (BCC) of the skin, one ofthe most common human neoplasms, as our model system. Previouswork with human autotransplants of BCC lesions has suggestedthat stromal cells in the tumor tissue play a crucial role insustaining the tumor (21). Mouse models of the disease haveshown that sustained activation of the Sonic Hedgehog pathway,a major genetic component of BCC, is maintained only in thecontext of the animal in vivo; when explanted in culture, tumorcells lose pathway activity (22).

We cultured stromal cells from BCC tumor and nontumor humanskin and compared those two cell populations by cDNA microarrayanalysis. Antagonists of the bone morphogenetic protein (BMP)pathway were among the genes most consistently and significantlydifferentially expressed between the two populations. Givenwhat is already known about the role of BMPs and their antagonistsin regulating stem cell compartments in normal development andphysiology, we hypothesized that a similar role could be playedby BMPs and BMP antagonists in the context of the tumor.

BMPs are important regulators of stem cell fate (23). In diversesettings, BMPs promote differentiation of stem cells, thus promotingexit from the stem cell compartment (12, 24). In the skin, conditionalgene targeting of BMPRIA in mice has demonstrated that BMPRIAis required for proper differentiation of progenitor cells inthe hair shaft (25, 26). The BMP inhibitor noggin is expressedby cells in the follicular mesenchyme, and mice lacking noggindisplay defects in hair follicle induction and morphogenesis(9, 27). High levels of GREMLIN 1 transcript have been observedin mouse embryonic fibroblast cells that are capable of maintaininghuman embryonic stem cells in culture (28). These observationsled us to investigate the hypothesis that BMP antagonists secretedby stromal cells in cancer tissues might be an important partof the specialized tumor microenvironment that allows continuedproliferation and self renewal of cancer cells.


Identification of Stromal Factors Important for BCC Tumorigenesis. To identify factors produced by tumor-associated stromal cellsthat contribute to the initiation or maintenance of BCC, wecultured stromal cells from fresh samples of human BCC or nontumorskin. The cells adhered readily to untreated plastic platesand were spindle-shaped and elongated. We used the SignificanceAnalysis of Microarrays algorithm to identify genes differentiallyexpressed between tumor- and nontumor-associated stromal cells(Fig. 1A; ref. 29). Fourteen genes were identified at an estimatedfalse discovery rate of 5%, with 13 genes expressed at higherlevels and one gene expressed at a lower level in the tumor-associatedcells (Fig. 1B and Table 1, which is published as supporting informationon the PNAS web site). Two of the 13 genes more highly expressedin BCC-derived stromal cells, GREM1 (GREMLIN 1) and FST (FOLLISTATIN),both encode antagonists of the BMP pathway.

GREMLIN 1 Expression Is Elevated in BCC, and BMPs Are Highly Expressed by BCC Tumors. We analyzed expression of GREMLIN 1 in vivo in human tissueby quantitative RT-PCR analysis of independent samples of wholetissue from eight matched BCC and adjacent nontumor skin samples.GREMLIN 1 transcripts were, indeed, expressed at higher levelsin BCC tissue than in adjacent nontumor tissue from the samepatient (Fig. 1C). We then performed in situ hybridization (ISH)in 15 paraffin-embedded BCC tissue samples and found detectableGREMLIN 1 mRNA expression in 12 of 15 samples (80%). Expressionwas localized predominantly to stromal cells in the tumor, andimmunohistochemistry (IHC) localized gremlin 1 protein to thestroma surrounding the tumor cell nests (Fig. 2B and D and Fig.6, which is published as supporting information on the PNASweb site). In contrast, no expression of GREMLIN 1 RNA or proteinwas detected in normal skin (Fig. 2 A and C). Thirty-nine sectionsof normal skin from multiple anatomical sites, including arm(dorsal, ventral, posterior, and anterior), hand (dorsal andventral), digits (posterior), palm, foot (dorsal and plantar),and leg (anterior, posterior, dorsal, and midline), were allnegative for GREMLIN 1 RNA, with only two exceptions: a fewstromal cells surrounding a neuromuscular junction in one sectionof skin from below the knee, and a small number of stromal cellsdeep in the dermis of the foot dorsum (data not shown). Theseresults indicate that GREMLIN 1 RNA expression is below levelsof detection or absent in the vast majority of normal humanskin sites.

An implicit aspect of our hypothesis is that there exists asource of BMP in BCC tumors that needs to be antagonized topromote proliferation of tumor cells. We found that BMP 2 and4 are, indeed, expressed in BCC tumor nests (Fig. 2 E and F).BMP antibody staining localized mostly to tumor cells, withmacrophages occasionally demonstrating positive staining.

To better characterize the stromal cell population that expressedGREMLIN 1 in BCC tumors, we analyzed adjacent serial sectionsof tumor by ISH for GREMLIN 1 and IHC for various cell lineagemarkers: vimentin (characteristic of mesenchymal cells), CD45(hematopoietic lineage), CD31 (endothelial cells), desmin (smoothmuscle cells), cytokeratins (epithelial cells), and glial fibrillaryacid protein (astrocytes and Schwann cells). GREMLIN 1-expressingcells were also strongly positive for vimentin, mostly or entirelynegative for CD45 and desmin, and completely negative for CD31,keratins, and glial acid fibrillary protein (GFAP) (Fig. 2 G–N).

A Functional Response to gremlin 1 in Cultured Human Skin Epithelial Cells. We reasoned that if the functional role of gremlin 1 in maintaininga tumor cell niche was analogous to its role in the normal skinprogenitor cell niche, gremlin 1 might be capable of inhibitingdifferentiation and promoting expansion of keratinocytes. Todirectly examine the effects of gremlin 1 on BCC tumor cells,cells isolated from fresh BCC tumors were cultured in the presenceof recombinant human BMP 4, recombinant mouse gremlin 1, orboth, and allowed to expand for 7 days. The resulting cell populationswere compared by using quantitative RT-PCR to characterize theirdifferentiation state (Fig. 3A). Compared with untreated controls,cells maintained in BMP 4 exhibited elevated mRNA levels ofSPRR1A, SPRR1B, SPRR3, and SPRR4, established markers of differentiatedkeratinocytes. Gremlin 1 strongly attenuated this effect. Gremlin1 protein alone, in the absence of exogenously added BMP 4,had little effect on SPRR expression. [Note that basal mediacontains no detectable BMP (data not shown).]

Gremlin 1 also antagonized BMP-mediated repression of cell proliferation.Primary BCC keratinocytes were cultured and cell growth assessedin the presence of varying concentrations of recombinant humanBMP 4 and recombinant mouse gremlin 1 (Fig. 3B). The doublingtime of these cells in culture with no added BMP or gremlin1 was 3.1 (+/– 0.1) days. Addition of gremlin 1 in theabsence of added BMP 4 did not significantly affect growth rate,even at the highest concentration of gremlin (2.105 µg/ml).In the absence of gremlin 1, doubling time increased steadilywith increasing BMP 4 concentration, reaching a maximum of 7.4(+/– 0.1) days, 2.4 times the baseline doubling rate.At the highest level of BMP 4, increasing the concentrationof gremlin 1 protein steadily lowered the doubling time backto baseline. These results indicated that BMP 4 inhibits theexpansion of BCC cell populations in culture, and that gremlin1 attenuates this inhibition.

GREMLIN 1 Is Expressed by Stromal Cells in Diverse Human Carcinomas. GREMLIN 1 is highly expressed in the fibroblasts of most BCCsand undetectable in most normal skin sites. Evidence that BMPsregulate stem cell expansion in many tissues (skin, intestine,and blood) raised the possibility that expression of gremlin1 may be an important feature of the tumor microenvironmentin other cancers (12, 24, 25). We therefore examined GREMLIN1 RNA expression in a total of 774 tumors, including melanomaand carcinomas of the liver, testis, ovary, uterus, kidney,thyroid, prostate, head and neck, bladder, breast, lung, colon,pancreas, and esophagus (n = 11–260 samples of each) byISH to tissue microarrays. GREMLIN 1 was expressed by stromalcells in at least 50% of samples in carcinoma of the bladder,breast, lung, colon, pancreas, and esophagus, and in at least25% of prostate and head and neck cancers (Fig. 4). Expressionof GREMLIN 1 was exclusively localized to the stromal cells,with the exception of some breast and prostate samples, whichshowed limited expression in the tumor cells themselves.

We also examined large sections of breast, pancreas, lung, andintestine, both tumor and nontumor. GREMLIN 1 expression wasundetectable in normal and benign breast tissue. In a seriesof 165 samples of pancreas, including normal tissue and benignand malignant lesions, we detected GREMLIN 1 RNA in only 5%(2/37) of normal samples, compared with 71.5% of pancreatictumors (68/95) (Fig. 7, which is published as supporting informationon the PNAS web site). GREMLIN 1 expression was also detectedin 45% (15/33) of benign pancreatic disease samples, includingpancreatitis, benign neuroendocrine tumors, and benign adenomas.In normal lung tissue, we observed GREMLIN 1 RNA in only a fewsmooth muscle cells. In large sections of both adenocarcinomaof the lung and adjacent normal lung tissue, there was no detectableGREMLIN 1 mRNA in the normal lung, whereas the tumor stromaand not the tumor cells themselves showed expression of GREMLIN1 mRNA (Fig. 8, which is published as supporting informationon the PNAS web site). In normal intestine, no GREMLIN 1 expressionwas observed except in the lamina propria, in what appear bymorphology to be smooth muscle cells (data not shown).


We found that the BMP antagonist GREMLIN 1 is frequently expressedby stromal cells in the microenvironment of human carcinomas,including BCC, and can enhance cell expansion and block differentiationin vitro. Carcinomas are histologically complex tissues comprisingnot only tumor cells but also fibroblasts, smooth muscle cells,endothelial cells, adipocytes, and leukocytes, as well as componentsof the extracellular matrix. Interactions with these cells andfactors in the tumor microenvironment, or tumor cell niche,may also play a critical role in the initiation and progressionof cancer (15, 16, 30). We used a genomic approach to identifyfactors differentially expressed by BCC-associated fibroblastscompared with their nontumor-associated counterparts. Globalgene expression profiling of these two cell populations revealedintrinsic, systematic differences in gene expression programs.We chose to focus on one gene in particular, the BMP antagonistGREMLIN 1. In many settings, BMPs promote differentiation ofstem cells, thus promoting exit from the stem cell compartment(23, 31). These observations led us to hypothesize that BMPantagonists may define a niche for a self-renewing populationin some cancers.

In this model, the role of BMPs and their antagonists in regulatinga self-renewing tumor cell compartment parallels their rolein regulating the normal stem cell compartment (Fig. 5). Innormal physiology, factors (including the BMP antagonists) thatsupport "stemness" of stem cells are often provided by a stemcell "niche," a molecular microenvironment defined by a localizedpopulation of cells that regulates the size of the stem cellcompartment (32, 33). Our data suggest a directly analogousmodel for the tumor context in which the tumor cells requireBMP antagonists coming from the tumor fibroblasts (another specializedstromal compartment) to maintain their expansion.

Our results represent a dramatic example of the differencesbetween stromal cells in cancer and those in the normal tissuecounterpart. Elevated expression of GREMLIN 1 has previouslybeen documented in a small subset of cells in normal skin, theputative epithelial stem cells, compared with other normal skinepithelial cells (34). In our study, GREMLIN 1 RNA was expressedin stromal cells of nearly all BCC samples examined, but undetectablein the vast majority of normal skin sites. The cells that expressGREMLIN 1 have the appearance and immunohistochemical characteristicsof fibroblasts and not cells of epithelial, lymphocytic, endothelial,smooth muscle, or glial origin.

How is this distinct, specialized tumor stromal compartmentinitially established? Does the gremlin 1-rich tumor niche developin response to signals derived from the tumor? If so, the presenceof gremlin 1-expressing fibroblasts could be the product ofeither de novo differentiation, recruitment from a distant site,or preferential expansion of an otherwise rare population inresponse to molecular signals from the tumor cells. In an alternativemodel, the chronology is reversed, that is, a specialized nichefavorable to tumor initiation and expansion may be establishedbefore the tumor can form, perhaps as a result of clonal expansionof a mutant or epigenetically modified clone of fibroblasts.Indeed, the familiar focal, patchy alterations in skin pigmentationand texture, hair morphology, and vascularization seen in aging,sun-exposed skin are consistent with preexisting local clonalfields of altered cells (35). Whatever events lead to the accumulationof gremlin 1-expressing fibroblasts in diverse carcinomas, theability of tumor-derived fibroblasts to maintain this distinctiveexpression program even after many generations of culture exvivo, away from the influence of their tumor counterpart, suggeststhis maintenance is specified by a stable genetic or epigeneticprogram.

The addition of gremlin 1 alone to basal media was not enoughto sustain long-term culture of BCC-derived cells. Thus, futurework is needed to define additional supporting factors presentin the tumor cell niche. As a preliminary step, we have usedRT-PCR to examine the expression of a number of other reportedBMP antagonists, including TSG1, FOLLISTATIN, NOGGIN, and CHORDIN,in whole tissue samples of human BCC and matched nontumor tissue.Like GREMLIN 1, both TSG1 and CHORDIN were typically expressedat higher levels in tumors compared with nontumor controls (Fig.9, which is published as supporting information on the PNASweb site). Further characterization of other factors in thetumor cell niche, combined with the identification of signalsderived from basal cell tumor cells, will help elucidate thereciprocal crosstalk that occurs between the tumor and its microenvironment.Along with GREMLIN 1, other genes that were elevated in BCCtumor-associated fibroblasts included a number of componentsof the Wnt signaling pathway, such as DICKKOPF HOMOLOG 1 (DKK1),a secreted protein inhibitor of the Wnt signaling pathway. TheWnt proteins (along with BMPs) are targets of the Sonic Hedgehogpathway (36). In one report of Wnt pathway activity in BCC,the pattern of nuclear b-catenin showed increased staining atthe periphery of tumor nests, as well as some staining in tumor-adjacentfibroblasts (37). Additional experiments will be useful in uncoveringthe connections between Wnt, Sonic Hedgehog, and BMP signalingin BCC.

We have shown that BMP inhibits expansion of BCC cells in culture,and that gremlin 1 can overcome this inhibition. The mechanismof gremlin 1/BMP action and downstream signaling events, however,remains unclear. Although we have not definitively addressedwhether the effects of gremlin are mediated exclusively throughthe BMP pathway, our data on cultured cells from BCC tumorssuggest that this is likely, because gremlin 1 had no appreciableeffect on cell expansion unless BMP was present. Although wedid observe BMP in some of the tumor cells in vivo, the tumorcells in vitro showed a response to gremlin 1 only in the presenceof exogenous BMP. Thus, the level of BMP production by the culturedtumor cells was not high enough to produce a clear effect atthe plated cell density, possibly because of the effects ofdilution by the media or by loss of normal cell–cell interactionnormally seen in vivo.

The expression of GREMLIN 1 by stromal cells in diverse humancarcinomas, in contrast to its rare expression in correspondingnormal tissues, suggests that expression by cells in the tumormicroenvironment of factors that regulate the self renewal ofthe tumor cells may be a general feature of human cancer. Inhibitingthese critical molecular signals from the tumor microenvironmentmay thus be a useful therapeutic strategy. The potential parallelsbetween stem cell–microenvironment interactions in normaldevelopment and cancer should provide fertile ground for furtherinvestigations.

Materials and Methods

Primary Human Cells. Stromal cells were isolated from discarded skin tissue fromthe Dermatology Clinic with approval from the InstitutionalReview Board (Stanford University Medical Center). Fat was removedby using a sterile scalpel and forceps; tissue was minced into ~1-mm cubes. Incubation in a six-well dish without medium at37°C for 10 min allowed for adhesion of the tissue to theplate. Fresh media containing DMEM, 10% FBS, and penicillin-streptomycinwere added, and samples were maintained at 37°C and 5% CO2.Media were replaced every 2 days. Outgrowth of spindle-shapedcells was typically apparent after 5–15 days in cultureand had a success rate of ~60%. When the cells were near confluence,they were subcultured with 0.25% trypsin-EDTA. Cultures wereexpanded until sufficient for RNA isolations (typically fourpassages).

Human BCC keratinocyte cultures were derived from fresh skintissue as described (38). A small crosssectional piece of eachsample was cut and fixed in 10% buffered formalin for histologicalconfirmation. The remaining tissue was placed overnight in 5mg/ml dispase (Gibco, Carlsbad, CA) at 4°C. The next day,epidermis was separated from dermis with dissecting forceps,minced by using sterile forceps and scalpel, and incubated in0.05% trypsin-EDTA at 37°C for 15 min, with occasional mixingto disperse cells. After neutralization with HBSS containing15% FBS, cells were spun down at 900 rpm in a Beckman AllegraGR centrifuge for 5 min, then resuspended in Keratinocyte serum-freemedia supplemented with EGF, bovine pituitary extract, and penicillin-streptomycin(Gibco). Cells were plated onto 12-well collagen I-coated plates(BD Biosciences, Franklin Lakes, NJ) and incubated at 37°Cin 5% CO2. Media were replaced every 2 days. Contamination fromfibroblasts or normal keratinocytes was avoided by subjectingthe culture to differential trypsinization and a transient increasein calcium concentration, respectively (39).

Microarray Procedures. Construction of human cDNA microarrays with ~42,000 elements,representing ~24,000 genes, and hybridizations was as described(40). Forty-eight hours before RNA harvest of stromal cultures,cells were washed three times in prewarmed PBS and then maintainedin low serum media containing DMEM and 0.1% FBS. mRNA was harvestedby using the FastTrack kit (Invitrogen, Carlsbad, CA). UniversalHuman Reference RNA (Stratagene, La Jolla, CA) was used as referencefor array experiments.

Arrays were scanned with a GenePix 4000A scanner and imagesanalyzed with GenePix 3.0 (Axon Instruments, Union City, CA).Microarray data were stored in the Stanford Microarray Database(41). All microarray data are available at the web site

Data Analysis. We considered only genes for which the cognate array elementhad a fluorescent signal at least 1.5-fold greater than thelocal background signal in both channels. Significance Analysisof Microarrays (29) was then used to identify a set of geneswhose expression levels were significantly different betweenfive tumor- and five nontumor-derived stromal cell culturesat a false discovery rate of 15% or 5%. Resulting expressionpatterns were organized by hierarchical clustering (42).

ISH. Digoxigenin-labeled sense and antisense riboprobes for GREMLIN1 were synthesized by using T7 polymerase-directed in vitrotranscription of linearized plasmid DNA (IMAGE clone 7262108)by using the DIG RNA Labeling Kit (Roche Diagnostics). ISH onparaffin sections was performed by using a biotinyl tyramideamplification procedure, essentially as described (43). Resultswere considered specific when a strong pattern of distinct punctatestaining was seen for the antisense probe, and little or nostaining was observed for the corresponding sense probe. Tissuemicroarrays of tumor samples were made as described (44).

IHC. IHC staining for Gremlin 1 was performed with Dako EnvisionPlus (Glostrup, Denmark). Anti-gremlin 1 antibody (Imgenex,San Diego, CA) was used at 1:10 dilution. IHC for BMPs was performedby using Vectastain ELITE ABC Rabbit IgG (Vector Laboratories,Burlingame, CA). Anti-BMP 2 and 4 antibodies (Santa Cruz Biotechnology,Santa Cruz, CA) were used at 1:50 dilution. IHC for cell lineagespecific markers was performed by using the Vectastain ELITEABC Mouse IgG kit with antibodies against Vimentin (1:200),CD31 (1:30), CD45 (1:100), GFAP (1:100), Desmin (1:100), andpancytokeratin (1:100; Dako). In all cases, antigen retrievalconsisted of a microwave step in 10 mM citrate buffer. Nucleiwere stained with hematoxylin.

As positive and negative controls, each antibody was also testedon a tissue microarray containing a large variety of normaland tumor human tissue samples to confirm the nominal specificity.ISH and IHC images were acquired with the BLISS Microscope System(Bacus Laboratories, Lombard, IL).

In Vitro Expansion and Differentiation Assays. To assess the effects of gremlin 1 and BMP proteins on expansionof cells in vitro, BCC-derived cells were maintained in keratinocytegrowth media containing bovine pituitary extract, human EGF,bovine insulin, hydrocortisone, gentamicin, and amphotericinB (Clonetics, San Diego, CA). Cells were incubated with recombinantmouse gremlin 1 and/or recombinant human BMP 2 or 4 (R&DSystems, Minneapolis, MN) at the concentrations indicated. Cellnumber was assessed by using triplicate counts with a hemacytometer,or RNA was collected for RT-PCR analysis.

Quantitative RT-PCR. Total RNA was isolated from whole tissue, either tumor or adjacentnontumor tissue from the same patient, by using the RNeasy FibrousTissue Mini kit (Qiagen, Chatsworth, CA) and a rotor homogenizer.Total RNA was isolated from cultured cells by using RNeasy Mini(Qiagen). First-strand DNA was generated from mRNA by usingthe SuperScript III First-Strand Synthesis System (Invitrogen).RT-PCR (TaqMan) was performed by using ABI 7300 (Applied Biosystems,Foster City, CA) with duplicate experimental samples for eachsample and each probe/primer set. GAPDH was used for normalizingPCR results.



We thank Darien Whang, Michael Knittel, and Anna Bar for helpin human tissue sample collection; Mindy Hebert for discussions;Jason Casolari, Michael Clarke, and Catriona Jamieson for helpfuldiscussions and critical reading of the manuscript; and theStanford Histology Core Facility and the Stanford FunctionalGenomics Facility (Stanford University Medical Center). Thiswork was supported by National Cancer Institute Grant CA77097(to P.O.B.), National Institutes of Health Grants ARO46786 (toA.E.O.) and K08-AR0008 (to H.Y.C.), a National Science FoundationPredoctoral Fellowship (to J.B.S.), and the Howard Hughes MedicalInstitute. H.Y.C. is a Damon Runyon Cancer Research Foundationscholar. P.O.B. is an Investigator of the Howard Hughes MedicalInstitute.


Abbreviations: BCC, basal cell carcinoma; BMP, bone morphogenetic protein; ISH, in situ hybridization; IHC, immunohistochemistry.

||To whom correspondence should be addressed. E-mail:

Freely available online through the PNAS open access option.

Author contributions: A.E.O. and P.O.B. contributed equallyto this work; J.B.S., A.E.O., and P.O.B. designed research;J.B.S., H.H.Z., K.M., and A.D.T. performed research; J.B.S.,H.H.Z., K.M., M.v.d.R., A.D.T., H.G., H.Y.C., G.S.M., and A.E.O.contributed new reagents/analytic tools; J.B.S., M.v.d.R., A.D.T.,R.W., A.E.O., and P.O.B. analyzed data; and J.B.S. and P.O.B.wrote the paper.

The authors declare no conflict of interest.

Data deposition: The array data reported in this paper havebeen deposited in the Gene Expression Omnibus (GEO) database, (accession no. GSE5502).

© 2006 by The National Academy of Sciences of the USA


  1. Xie, T & Spradling, AC. (2000) Science 290, 328–330
  2. Moore, KA & Lemischka, IR. (2006) Science 311, 1880–1885
  3. Karlsson, L, Lindahl, P, Heath, JK & Betsholtz, C. (2000) Development (Cambridge, UK) 127, 3457–3466
  4. Kaestner, KH, Silberg, DG, Traber, PG & Schutz, G. (1997) Genes Dev 11, 1583–1595
  5. Marshman, E, Booth, C & Potten, CS. (2002) BioEssays 24, 91–98
  6. Alonso, L & Fuchs, E. (2003) Proc Natl Acad Sci USA 100, Suppl 1, 11830–11835
  7. Lemischka, IR & Moore, KA. (2003) Nature 425, 778–779
  8. Millar, SE. (2002) J Invest Dermatol 118, 216–225
  9. Botchkarev, VA, Botchkareva, NV, Roth, W, Nakamura, M, Chen, LH, Herzog, W, Lindner, G, McMahon, JA, Peters, C & Lauster, R, et al. (1999) Nat Cell Biol 1, 158–164
  10. Bjerknes, M & Cheng, H. (2001) Proc Natl Acad Sci USA 98, 12497–12502
  11. Haramis, AP, Begthel, H, van den Born, M, van Es, J, Jonkheer, S, Offerhaus, GJ & Clevers, H. (2004) Science 303, 1684–1686
  12. Zhang, J, Niu, C, Ye, L, Huang, H, He, X, Tong, WG, Ross, J, Haug, J, Johnson, T, Feng, JQ, Harris, S, Wiedemann, LM, Mishina, Y & Li, L. (2003) Nature 425, 836–841
  13. Calvi, LM, Adams, GB, Weibrecht, KW, Weber, JM, Olson, DP, Knight, MC, Martin, RP, Schipani, E, Divieti, P & Bringhurst, FR, et al. (2003) Nature 425, 841–846
  14. Williams, JA, Guicherit, OM, Zaharian, BI, Xu, Y, Chai, L, Wichterle, H, Kon, C, Gatchalian, C, Porter, JA & Rubin, LL, et al. (2003) Proc Natl Acad Sci USA 100, 4616–4621
  15. Bissell, MJ & Radisky, D. (2001) Nat Rev Cancer 1, 46–54
  16. Elenbaas, B & Weinberg, RA. (2001) Exp Cell Res 264, 169–184
  17. Howe, JR, Roth, S, Ringold, JC, Summers, RW, Jarvinen, HJ, Sistonen, P, Tomlinson, IP, Houlston, RS, Bevan, S & Mitros, FA, et al. (1998) Science 280, 1086–1088
  18. Moinfar, F, Man, YG, Arnould, L, Bratthauer, GL, Ratschek, M & Tavassoli, FA. (2000) Cancer Res 60, 2562–2566
  19. Olumi, AF, Grossfeld, GD, Hayward, SW, Carroll, PR, Tlsty, TD & Cunha, GR. (1999) Cancer Res 59, 5002–5011
  20. Camps, JL, Chang, SM, Hsu, TC, Freeman, MR, Hong, SJ, Zhau, HE, von Eschenbach, AC & Chung, LW. (1990) Proc Natl Acad Sci USA 87, 75–79
  21. Van Scott, EJ & Reinertson, RP. (1961) J Invest Dermatol 36, 109–131
  22. Aszterbaum, M, Epstein, J, Oro, A, Douglas, V, LeBoit, PE, Scott, MP & Epstein, EH Jr. (1999) Nat Med 5, 1285–1291
  23. Zhang, J & Li, L. (2005) Dev Biol 284, 1–11
  24. He, XC, Zhang, J, Tong, WG, Tawfik, O, Ross, J, Scoville, DH, Tian, Q, Zeng, X, He, X, Wiedemann, LM, Mishina, Y & Li, L. (2004) Nat Genet 36, 1117–1121
  25. Kobielak, K, Pasolli, HA, Alonso, L, Polak, L & Fuchs, E. (2003) J Cell Biol 163, 609–623
  26. Andl, T, Ahn, K, Kairo, A, Chu, EY, Wine-Lee, L, Reddy, ST, Croft, NJ, Cebra-Thomas, JA, Metzger, D & Chambon, P, et al. (2004) Development (Cambridge, UK) 131, 2257–2268
  27. Jamora, C, DasGupta, R, Kocieniewski, P & Fuchs, E. (2003) Nature 422, 317–322
  28. Pera, MF, Andrade, J, Houssami, S, Reubinoff, B, Trounson, A, Stanley, EG, Ward-van Oostwaard, D & Mummery, C. (2004) J Cell Sci 117, 1269–1280
  29. Tusher, VG, Tibshirani, R & Chu, G. (2001) Proc Natl Acad Sci USA 98, 5116–5121
  30. Bhowmick, NA & Moses, HL. (2005) Curr Opin Genet Dev 15, 97–101
  31. Xu, RH, Peck, RM, Li, DS, Feng, X, Ludwig, T & Thomson, JA. (2005) Nat Methods 2, 185–190
  32. Li, L & Xie, T. (2005) Annu Rev Cell Dev Biol 21, 605–631
  33. Ohlstein, B, Kai, T, Decotto, E & Spradling, A. (2004) Curr Opin Cell Biol 16, 693–699
  34. Morris, RJ, Liu, Y, Marles, L, Yang, Z, Trempus, C, Li, S, Lin, JS, Sawicki, JA & Cotsarelis, G. (2004) Nat Biotechnol 22, 411–417
  35. Wulf, HC, Sandby-Moller, J, Kobayasi, T & Gniadecki, R. (2004) Micron 35, 185–191
  36. Cohen, MM Jr. (2003) Am J Med Genet 123, 5–28.
  37. Saldanha, G, Ghura, V, Potter, L & Fletcher, A. (2004) Br J Dermatol 151, 157–164
  38. Fan, H, Oro, AE, Scott, MP & Khavari, PA. (1997) Nat Med 3, 788–792
  39. Grando, SA, Schofield, OM, Skubitz, AP, Kist, DA, Zelickson, BD & Zachary, CB. (1996) Arch Dermatol 132, 1185–1193
  40. Perou, CM, Sorlie, T, Eisen, MB, van de Rijn, M, Jeffrey, SS, Rees, CA, Pollack, JR, Ross, DT, Johnsen, H & Akslen, LA, et al. (2000) Nature 406, 747–752
  41. Sherlock, G, Hernandez-Boussard, T, Kasarskis, A, Binkley, G, Matese, JC, Dwight, SS, Kaloper, M, Weng, S, Jin, H & Ball, CA, et al. (2001) Nucleic Acids Res 29, 152–155
  42. Eisen, MB, Spellman, PT, Brown, PO & Botstein, D. (1998) Proc Natl Acad Sci USA 95, 14863–14868
  43. Iacobuzio-Donahue, CA, Ryu, B, Hruban, RH & Kern, SE. (2002) Am J Pathol 160, 91–99
  44. Kononen, J, Bubendorf, L, Kallioniemi, A, Barlund, M, Schraml, P, Leighton, S, Torhorst, J, Mihatsch, MJ, Sauter, G & Kallioniemi, OP. (1998) Nat Med 4, 844–847


Fig. 1. GREMLIN 1 mRNA is elevated in BCC tumor-derived stromal cultures ex vivo and in BCC tumors in vivo. (A and B) Gene expression in stromal cells from BCC tumor and nontumor skin. Each row in the heat map represents a gene, and each column represents a sample; cultures derived from BCC and nontumor skin are indicated by red and black branches, respectively. (A) Hierarchical clustering of samples based on expression of 403 array elements selected for differential expression by two-class Significance Analysis of Microarrays analysis at a false discovery rate of 15%. The level of expression of each gene in each stromal cell sample is relative to the mean level of expression of that gene across all samples and is represented by using a red-green color scale. (B) Hierarchical clustering of samples according to their expression of a more stringently selected set of genes (false discovery rate of 5%). (C) RT-PCR analysis of whole-tissue samples of BCC tumor or adjacent nontumor skin from eight patients. The relative level of GREMLIN 1 RNA in each sample was normalized to GAPDH for that sample.  

Fig. 2. Expression of GREMLIN 1 and BMP 2 and 4 in BCC tumor tissues. (A and B) ISH for GREMLIN 1 RNA in normal scalp (A) and BCC tumor skin (B). GREMLIN 1 is expressed by stromal cells surrounding the tumor (indicated by arrows) but is undetectable in normal scalp. Positive signal appears as dark purple dots. (C and D) IHC for gremlin 1 protein in normal scalp (C) and BCC tumor skin (D). Positive signal appears as diffuse brown staining. (E and F) IHC with antibodies against BMP 2 (E) and 4 (F) in large sections of human BCC. Signal is represented by brown color. (G–N) serial sections of a BCC tumor showing that GREMLIN 1-expressing cells have properties of fibroblasts. (G and H) RNA ISH for GREMLIN 1 RNA in nontumor (G) and tumor (H) skin. GREMLIN 1 expression is indicated by dark purple dots. (I–N) IHC for cell lineage markers vimentin (I), CD45 (J), CD31 (K), desmin (L), pancytokeratin (M), and GFAP (N). Signal is represented by brown color.

Fig. 3. Effects of BMP and gremlin 1 on BCC cell differentiation and expansion in vitro. (A) Cells were cultured from a human BCC tumor and then treated for 7 days in culture with recombinant human BMP 4 (833 ng/ml), recombinant mouse gremlin 1 (2,105 ng/ml), or both. Populations were then compared by using quantitative RT-PCR to detect the levels of SPRR1A, SPRR1B, SPRR3, and SPRR4 transcripts. (B) Cells were cultured in vitro from human BCCs then treated with varying concentrations of gremlin 1 or BMP 4 protein for 7 days. Cells were counted by using a hemacytometer, with triplicate counts taken for each measurement; each measurement is the average of duplicate experiments.

Fig. 4. GREMLIN 1 RNA is widely expressed in cancer. (A) Representative images for ISH in other carcinomas. Tissue microarrays representing 774 human cancers of diverse tissue types were analyzed for expression of GREMLIN 1 RNA. Positive signal is denoted by dark brown staining. (B) Tabulation of results for all tumor types examined for which n was >10.

Fig. 5. A possible parallel for the role of BMPs and BMP antagonists in the tumor stem cell niche and normal stem cell niche. BMPs can induce differentiation and block expansion of normal stem and progenitor cells. (Upper) BMP antagonists can reverse this effect and favor expansion. Signals that are critical for stem cell self renewal, such as BMP antagonists, derive from the stem cell niche. BMPs may often come from epithelial or stem cells. (Lower) In tumor biology, BMPs can also inhibit expansion of tumor cells. BMP antagonists secreted by the tumor cell niche may reverse this inhibition, allowing tumor cell expansion to continue.