Experimental Hematology 2008;36:799–806

Transcriptional regulationof bone marrow thrombopoietin by platelet proteins

Bryan McIntosh and Kenneth Kaushansky

Department of Medicine, University of California, San Diego, San Diego, Calif., USA

(Received 28 November 2007; revised 19 February 2008; accepted 21 February 2008)

Platelet production is regulated primarily by the cytokine thrombopoietin (TPO). AlthoughTPO is expressed in several different tissues, only in the bone marrow has the level of expres-sion been reported to increase in response to reduced numbers of platelets. In these studies, wedemonstrate that platelet granule proteins are able to transcriptionally repress TPO mRNAexpression in a marrow stromal cell line as well as in primary bone marrow stromal cell cul-tures. Like TPO mRNA, secretion of TPO protein was also suppressed by serum treatment.Reporter gene constructs indicate that DNA elements located in an approximately 1.9-kbregion between 250-bp upstream of the transcriptional initiation site and the middle of thesecond intron are able to mediate the transcriptional repression. � 2008 ISEH - Societyfor Hematology and Stem Cells. Published by Elsevier Inc.

Hematopoiesis, the development of bone marrow progeni-tor cells into mature blood cells, is regulated by a largenumber of cytokines and growth factors, each with a limitedand defined set of target cells. Physiologically appropriatenumbers of each cell type are maintained by continual ad-justments in the levels of hematopoietic growth factors. Forplatelet the lineage, thrombopoietin (TPO) is the primaryregulator of thrombopoiesis. Disruption of the Tpo gene,or that of its receptor, c-Mpl, results in a dramatic reductionin the number of mature platelets produced by marrow pro-genitors in both mouse and man [1–3]. As the primary reg-ulator of thrombopoiesis, blood levels of TPO are inverselyrelated to platelet count [4]. One mechanism that accountsfor this phenomenon is the binding of TPO to its receptorand cellular uptake by the receptor, which is present on ma-ture platelets [5,6]. Thus, when platelet levels are high,a larger quantity of TPO is removed from the blood, leadingto a lower concentration of the hormone. Conversely, whenplatelet levels are reduced, little TPO is removed and bloodlevels of the hormone rise, resulting in increased plateletproduction. A tenet of this model of regulation is thatTPO expression in the liver, the primary site of TPO pro-duction [7] is unchanged [8]. However, several reportshave indicated that TPO mRNA expression in the bonemarrow can increase in response to pathologic reductions

in platelet levels and experimental manipulation [8,9].Moreover, a detailed analysis indicated that several platelet-specific granule proteins, such as platelet factor 4, thrombo-spondin (TSP), transforming growth factor (TGF)–b,among others, could alter TPO mRNA and protein expres-sion in the marrow stroma [10]. It is well-established thatsuch paracrine production of a cytokine such as TPO wouldhave a far greater effect on blood cell development thanwould its endocrine functioning.

In order to better understand what regulates TPO mRNAexpression in the bone marrow, and the responsible molec-ular mechanism(s), we used quantitative polymerase chainreaction (Q-PCR) to measure changes in TPO mRNA ex-pression provoked by treatment with serum and plateletproteins in model cell lines, cultured primary cells, and invivo. Additionally, we used a modified TPO-specific Q-PCR assay to test for expression of mRNAs derived fromTPO genomic DNA after its introduction into a bone mar-row stromal cell line. We found that TPO is subject to tran-scriptional regulation in marrow stromal cells in response toreleased platelet proteins.

Materials and methods

Cell cultureThe murine marrow stromal cell line OP9 [11] (kindly providedby Dr. Sanford Shattil) was cultured at 37�C in 5% CO2 in a-modification of Eagle’s medium (a-MEM), freshly preparedfrom powder (HyClone, South Logan, UT, USA) every 7 to 10

Offprint requests to: Kenneth Kaushansky, M.D., Department of Medi-

cine, University of California, San Diego, 9500 Gilman Drive, San Diego,

CA 92103-8811; E-mail: kkaushansky@ucsd.edu

0301-472X/08 $–see front matter. Copyright � 2008 ISEH - Society for Hematology and Stem Cells. Published by Elsevier Inc.

doi: 10.1016/j.exphem.2008.02.012

800 B. McIntosh and K. Kaushansky/ Experimental Hematology 2008;36:799–806

days and containing 20% preselected bovine calf serum, penicil-lin, streptomycin, and supplemental glutamine. The murine hepa-toma cell line Hepa1c1c7 [12] (American Type Culture Company,Manassas, VA, USA) was cultured at 37�C in 5% CO2 in Dulbec-co’s modified essential medium (DMEM) containing penicillin,streptomycin, and supplemental glutamine. Primary bone marrowstromal cells were derived from normal murine bone marrow thathad been flushed from femurs and placed directly into Iscove’smodified Dulbecco’s medium supplemented with 12.5% bovinecalf serum, 12.5% horse serum, penicillin, streptomycin, pyruvate,and 100 nM dexamethasone. Cells were maintained at 37�C in 5%CO2, with weekly half media changes until only an adherent layerwas present. Adherent cells were then used for up to four passagesonce confluent.

Quantitative RT-PCRCell cultures were treated as indicated and total RNA was purifiedwith Trizol (Invitrogen, Carlsbad, CA, USA) except for trans-fected cells, where the RNeasy Plus mini kit (Qiagen, Valencia,CA) was used instead. To further reduce DNA contamination,RNA samples were subjected to DNAse (Roche, Indianapolis,IN, USA) digestion and ethanol precipitation. When collectingbone marrow RNA, femurs were first flushed lightly with severalmilliliters of low serum medium applied through a 25½-g needle.Two femurs from each mouse were then immersed together in onetube in 750 mL Trizol. To complete the stromal cell solubilization,the Trizol was passed through the empty bone cavities with a 27½-gneedle, and the marrow cavity scraped with the tip of the needle. Areal-time reverse transcriptase (RT)-PCR was then conducted onan iCycler (BioRad, Hercules, CA, USA) thermal cycler usingTaqman assay chemistry. The RT reaction (20 mL) was performedwith Maloney murine leukemia virus RT (50 U) at 37�C for 1 hourwith random hexamer primers after an initial 20-minute roomtemperature preincubation. The PCR reaction (40 mL) contained15 mM Tris (pH 8.0), 50 mM KCl, 0.3 mM dNTPs, 1 U AmpliTaqGold (Applied Biosystems, Foster City, CA, USA), and 250 mM ofeach primer. Magnesium and FAM-TAMRA–labeled probes wereadded as indicated in Table 1. The PCR protocol was: 95�C for 10minutes followed by 50 cycles of 95�C for 15 seconds and 60�Cfor 1 minute.

Table 1. Quantitative reverse transcription polymerase chain reaction

conditions

Gene/acc. no. Primers/Probe MgCl2 (mM)

Thrombopoietin (TPO)

NM_009379.3

L: ccccatacagggagccacttR: gacagagttagtcttgccactgcaaPrb: actgatttgctcctggcggccatMutant prb: tggagctgactgatt

tactactagcagcaatgc

3.0

Cyclophilin (CYC)

NM_008907

R: gtctccttcgagctgtttgcL: tggcacatgaatcctggaataPrb: ttcgagctctgagcactggagaga

4.0

Bone sialoprotein (BSP)

L20232.1

L: ggcgatagttccgaagaggR: cccgagagtgtggaaagtgtPrb: aggcggaggcagagaacgcc

2.0

Firefly luciferase

(from pGL3, Promega)

F: ggttgtggatctggataccgR: cgcttcggattgtttacatPrb:aggcgaactgtgtgtgagaggtccta

3.0

18S NR_003278.1 L: cgcggttctattttgttggtR: caaatgctttcgctctggtcPrb: ggcattcgtattgcgccgct

2.0

Platelet-rich and platelet-poor plasma-derived serumTPO receptor (c-Mpl) null mice were backcrossed onto a C57/Bl6background for greater than 10 generations (kindly provided byDr. Warren Alexander, Melbourne, Australia). Whole blood wascollected into acid-citrate-dextrose by cardiac puncture ofc-Mpl–/– (mean platelet count: w327,000/mL) and C57/Bl6 (meanplatelet count: w1,050,000/mL) control mice. Samples without ev-ident clotting were pooled and centrifuged at 800g for 15 minutesto generate platelet-rich plasma. Half of each sample was thencentrifuged at 1500g for 10 minutes to generate platelet-poorplasma. Plasma-derived serum was obtained by adding CaCl2 to100 mM, HEPES buffer pH7.5 to 100 mM, and 0.5 U/mL throm-bin (Sigma-Aldrich, St. Louis, MO, USA) and then incubating at37�C for 30 minutes. Fibrin clots were disrupted by passagethrough a 27½-g needle and removed by centrifuging at 15,000gfor 5 minutes. Samples were dialyzed against phosphate-bufferedsaline (PBS) overnight, and the final protein concentrations ad-justed to 20 mg/mL with PBS.

Metabolic labelingOP9 cells were grown to confluency. The medium was then re-placed with methionine- and cysteine-free medium containing20% dialyzed bovine calf serum and incubated for 1 hour. The me-dium was then replaced with fresh methionine- and cysteine-freemedium (MP Biomedicals, Solon, OH, USA) containing 20% or0.5% bovine calf serum and incubated for an additional 1 hour,at which time 1.5 mCi of a mixture of 35S-labeled methionineand cysteine (MP Biomedicals) was added to each T-75 cultureflask (BD Bioscience, Bedford, MA, USA) containing 7 mL me-dium. After 8 hours of further incubation in a container packedwith a large volume of activated charcoal, the culture supernatantswere collected. Serum or PBS was added to standardize the me-dium composition, and trichloroacetic acid precipitable countswere measured by liquid scintillation counting. An equal quantityof cpm was combined with the primary biotinylated precipitatingreagents (anti-TPO antibodies or soluble c-Mpl receptor; R&DSystems, Minneapolis, MN, USA) and incubated at 4�C overnightwith rocking. Streptavidin-agarose beads (Cedarlane Laboratories,Burlington, Canada) were added and incubated for 2 hours. Theagarose beads were collected by centrifugation and washed severaltimes in PBS. The bound protein was eluted with 1� sodium do-decyl sulfate polyacrylamide gel electrophoresis sample bufferand separated under standard polyacrylamide gel electrophoresisconditions. Before drying, the gel was soaked in ENLIGHTNINGautoradiography enhancer (Fischer Scientific, Pittsburgh, PA,USA) for 1 hour. Finally, the dried gel was exposed to film.

Subcloning of murine TPO genomicDNA by homologous recombinationSubcloning was achieved with the RED/ET BAC Subcloning kit(Gene Bridges, Dresden, Germany). The protocol consisted of firsttransforming a bacterial artificial chromosome clone known tocontain TPO genomic region (RP23-187G15) with the pSC101-BAD-gbaA plasmid, which contained the recombination enzymes,and selecting for transformants at 30�C. The recombination en-zymes were induced with arabinose for 1 hour. Cells were thenelectroporated with the subcloning vector pColE1-Amp that hadbeen prepared previously by PCR to incorporate ends 50-bp of se-quence identical to the TPO gene (specifically, at the 50 end ofTPO: gattgctcaatgatcctcttctagtcctaccactatacatgtgacaagag, and at

801B. McIntosh and K. Kaushansky/ Experimental Hematology 2008;36:799–806

the 30 end of TPO: gaatctcagtgtcagtttctatgtgacagagaccttcatcttc-taattct). The subclones were selected on ampicillin containingplates at 37�C and screened by restriction digest.

DNA reporter constructsThe bacterial artificial chromosome subclone containing 7.7 kbTPO genomic DNA served as the basis of two series of constructs,pmutTPO and pTþL. The pmutTPO series of constructs was cre-ated by using Quikchange (Stratagene, La Jolla, CA, USA) muta-genesis to introduce five single base changes that preserve thereading frame and splicing signals. The pmutTPO-250pro con-struct was the result of restriction digest at the immediate 50 endof the cloned genomic fragment with SpeI and a XbaI site 1 kbdownstream, followed by blunting and recircularization. ThepmutTPO-30del construct was produced in the same mannerwith a XhoI at the 30 end of the TPO gene and a AgeI site incor-porated at the immediate 30 end of the subcloned fragment. The4th intron was removed, cutting with AatII in exon 4 and SbfIin exon 5, and replaced with the corresponding cDNA fragmentto produce the pmutTPO-din4 construct. The firefly luciferasecDNA followed by an SV40 polyadenylation sequence was re-moved from pGL3-basic (Promega, Madison, WI, USA) at theHindIII and BamHI sites, blunted, and inserted into the EcoRVsite in the first intron of the TPO gene. The pTþL-dBam andpTþL-dPst were subsequently produced by digestion the BamHIsite (2860 bp) or the PstI site (5100 bp) and a AgeI site at the im-mediate 30 end of the subcloned fragment, followed by bluntingand recircularization.

Results

Bone marrow TPO mRNA levelsare regulated by thrombocytopeniaIn order to confirm earlier reports of selective upregulationof bone marrow TPO mRNA levels in response to myelo-suppression, and to provide a quantitative measure of thechanges, a quantitative RT-PCR assay was developed. Wefirst tested whether murine marrow TPO mRNA levelswere altered in response to thrombocytopenia. We foundthat 9 days after mice receive 9 Gy g-radiation, a time pointcharacterized by severe thrombocytopenia, TPO-specificmRNA levels in the marrow stromal cells are greatly en-hanced compared to control animals (Fig. 1). In contrast,TPO mRNA levels in the liver were stable or decreasedslightly. These results are consistent with those reportedpreviously [8].

Platelet products regulate TPO-specificmRNA expression in marrow stromal cellsTo investigate the role that soluble products of platelet re-lease or blood coagulation might play in the regulation ofTPO expression, cultured primary murine bone marrowstromal cells and model stromal cell lines were treatedwith varying concentrations of normal serum. OP9 cellswere selected as a model of bone marrow stromal cellsbased on their ability to support the in vitro differentiation

of ES cells toward hematopoietic lineages [13]. The murinehepatoma–derived cell line, Hepa1c1c7, was selected torepresent liver cells. Interestingly, the main effect of normalmurine serum on the OP9 cells was found to be suppressionof TPO-specific transcripts (Fig. 2A), whereas in Hep-a1c1c7 cells, TPO expression was reduced by serum starva-tion (Fig. 2b). The suppressive ability of normal murineserum was likewise matched by bovine calf serum in OP9cells (Fig. 2C) and in bone marrow stromal cells (Fig. 2D).

Having demonstrated the effect of serum on TPO mRNAlevels, it was important to also demonstrate equivalent effectson TPO protein production. Unfortunately, OP9 cells do notsecrete sufficient TPO in the culture supernatant to be deter-mined by the available enzyme-linked immunosorbent assayassays, nor could the culture supernatant be concentrated suf-ficiently owing to the 20% serum already present in the cul-ture medium. Simple immunoprecipitation was alsoprecluded by the cross-reactivity of bovine TPO and IgG inthe calf serum. Metabolic labeling of proteins in OP9 cellswas undertaken to establish the concordance of RNA andprotein production. Both an anti-murine TPO antibody(Fig. 3, left) and a soluble form of the TPO receptor(Fig. 3, middle) were able to pull down a 35S-labeled proteinof the appropriate Mr. Metabolic labeling of BHK cells thatoverexpress recombinant murine TPO provided a suitablepositive control immunoprecipitation (Fig. 3, right).

Serum is a complex milieu of proteins, lipids, and othersmall molecules. The simplest explanation of the serum

Figure 1. Thrombopoietin (TPO) expression in murine bone marrow and

liver. Mice were irradiated with 9 Gy g-radiation. Nine days later, quanti-

tative polymerase chain reaction was used to determine expression of

TPO-specific mRNA in bone marrow stromal cells as well as in the liver.

TPO transcript levels were normalized to bone sialoprotein (BSP) expres-

sion in marrow cells and to cyclophilin (CYC) in the liver. Results are ex-

pressed as the mean values from a representative experiment conducted

with four mice in each group and two independent tissue samples from

each animal in each condition 6 SEM.

802 B. McIntosh and K. Kaushansky/ Experimental Hematology 2008;36:799–806

suppression would be that TPO in serum suppresses bonemarrow production through feedback inhibition. However,recombinant TPO was unable to suppress TPO mRNAlevels in OP9 or marrow stromal cells (data not shown).In addition, bone marrow TPO mRNA is elevated in set-tings where endogenous levels of TPO are known to behigh, such as after myelosuppressive therapies or in otherthrombocytopenic conditions [4]. To further explore thecomponent(s) of serum that suppresses TPO production,we tested serum from thrombocytopenic c-Mpl–/– mice,which display extraordinarily high levels of TPO [1]. Wefound that c-Mpl–/– mouse-derived serum does not suppressTPO production as well as control murine serum (Fig. 4).

Many of the important growth factors in serum that helpto maintain cells in culture are platelet-derived proteins

released during blood clotting. The critical role that TPOplays in platelet production prompted experiments aimedat quantifying the suppressive ability of platelet proteinsrelative to plasma proteins. The results illustrated inFigure 4A and 4B demonstrate that platelet proteins areable to suppress TPO expression. First, thrombin activatedhuman platelet releasate is able to inhibit TPO expression(Fig. 4B). In these experiments, thrombin did not demon-strate a direct effect to change TPO expression in stromalcells. In addition, serum produced by thrombin activationof platelet-rich plasma reduces TPO expression more thanserum produced from platelet-poor plasma (Fig. 4A), an ef-fect that is absent in samples from c-Mpl–/– mice, likely dueto their severe thrombocytopenia. These results stronglysuggest that thrombocytopenia stimulates marrow TPO

Figure 2. Thrombopoietin (TPO) expression in cultured primary cells and cell lines in response to serum. Two cell lines, OP9 (A) and Hepa1c1c7 (B) were

treated with increasing doses of murine serum for 10 hours. OP9 cells (C) and primary murine bone marrow stromal cells (D), were treated with bovine calf

serum for 10 hours. The results are expressed as the mean value of four independent experiments 6 SEM.

803B. McIntosh and K. Kaushansky/ Experimental Hematology 2008;36:799–806

expression not by increasing a stimulatory factor, but ratherby reducing a negative regulatory factor.

Thrombopoietin is regulation transcriptionallyThe molecular mechanism(s) that underlies serum-inducedchanges in TPO expression was next investigated by intro-ducing TPO DNA constructs into OP9 cells and evaluatingthe changes in transgene expression levels resulting from se-rum treatment by Q-PCR. The first set of constructs (Fig. 5A)are deletion mutants of a 7.7-kb fragment of genomic DNAthat contains all the exons of murine TPO. The 50 most 1kb of DNA has been deleted in the pmutTPO-250pro con-struct, leaving only 250-bp upstream of the transcriptionalinitiation site. The last 900-bp have been deleted from the30del construct. The fourth, and largest, intron of TPO was

deleted from the pmutTPO-din4 construct. The expressionof specific transcripts from this series of constructs was dis-tinguished from endogenous TPO by the incorporation ofa 5-bp change that preserves the reading frame and splicingsignals, but to which a corresponding Q-PCR probe anneals.Thus, a Q-PCR signal is only detected in transfected cells, butnot in nontransfected controls. As shown in Figure 5B, se-rum-induced suppression of TPO expression is maintainedin each of these constructs, indicating that none of the deletedregions contain required cis-acting sequence element(s) re-sponsible for TPO regulation.

A second set of reporter gene constructs was then gener-ated inserting the firefly luciferase cDNA and polyadenyla-tion signal into the first intron of the TPO gene,downstream of the normal TPO transcriptional initiationsite. The resulting message would be expected to terminateat the polyadenylation sequence incorporated with the lu-ciferase gene and contain no TPO RNA sequence 30 tothe insertion site in intron 1. The expression of luciferasemRNA is thus driven by TPO genomic sequences and canbe followed by Q-PCR for luciferase. Based on this con-struct, a series of successive deletions of the 30 end ofTPO were generated (Fig. 5C). As shown in Figure 5D,none of the constructs was able to completely abrogate se-rum-induced repression. Therefore, the shortest construct(pTþL-dBam; 2860 bp) contains sequence element(s) suf-ficient to cause serum suppression of TPO in bone marrowcells. Taking into account the first series of constructs(Fig. 5A) in which the pmutTPO-250pro construct wasable to be regulated normally, the most likely region to con-tain all necessary serum repressive sequence elements liesbetween 1000 bp and 2860 bp. This series of reporter

Figure 3. Metabolic labeling of proteins in OP9 cells. Cells were labeled

with a mixture of 35S-cysteine and 35S-methionine for 8 hours in the pres-

ence of 20% (þ) or 0.5% (–) bovine calf serum. Labeled proteins were pre-

cipitated from culture supernatants using either an anti-thrombopoietin

(TPO) antibody (left) or a soluble TPO receptor (middle). As a control, la-

beled TPO was precipitated from the supernatant of BHK cells overex-

pressing recombinant murine TPO (right).

Figure 4. The effect of platelet factors on the expression on thrombopoietin (TPO). (A) OP9 cells were treated for 10 hours with serum derived from throm-

bin-activated plasma that was either platelet-rich (dark bars) or platelet-poor (light bars). Results represent the mean relative TPO mRNA levels of seven

samples from two separately prepared serum batches and independently executed experiments 6 SEM. In order to minimize the effect of factors not released

from platelets, both platelet-rich and platelet-poor serums were produced from the same batch of murine blood collected into acid-citrate-dextrose. (B) OP9

cells were treated with increasing doses of thrombin-activated human platelet releasate. The platelet concentration was 400/mL, the same as normal human

blood. Results represent the mean TPO mRNA levels in three independent experiments 6 SEM.

804 B. McIntosh and K. Kaushansky/ Experimental Hematology 2008;36:799–806

gene constructs also indicates that at least some portion ofserum-mediated TPO suppression is mediated by changesin transcription, because essentially no TPO sequences onwhich mRNA stability mechanisms might act are incorpo-rated into mature luciferase messages.

DiscussionQuantitative RT-PCR demonstrates that bone marrow TPOmRNA expression is increased in response to experimen-tally induced thrombocytopenia. Treating cultured primary

stromal cells and OP9 cells with serum supports the conclu-sion that the elevation of TPO RNA results from disruptionof a normal negative feedback loop, rather than the actionof stimulatory factors generated during the thrombocytope-nia. Platelet-derived proteins seem to account for a signifi-cant portion of the suppression of TPO transcripts becauseserum derived from platelet-rich plasma suppresses morethan that from platelet-poor plasma. In addition, retentionof suppressing activity by dialyzed serum indicates thatthe responsible factors are proteins of greater Mr than 10kDa. In contrast, the same stimuli seem to cause TPO

Figure 5. Expression of transcripts derived from on thrombopoietin (TPO) genomic DNA constructs. (A) A TPO genomic construct was generated encom-

passing w7.7 kb DNA as illustrated. A quantitative polymerase chain reaction (Q-PCR) was developed in which the probe was complementary to five silent

point mutations that allowed mRNA expression derived from the DNA construct to be distinguished from that of endogenous TPO-specific mRNA (shown at

arrow). A series of deletion mutants was also generated. (B) About 40 hours after transfection of the TPO DNA constructs, the OP9 cells were treated with or

without bovine calf serum for 10 hours and cells harvested for Q-PCR. Levels of construct-specific mRNA are shown from four independent experiments 6

SEM. (C) A second series of TPO gene constructs that include a firefly luciferase reported genes and corresponding polyadenylation signal in the first intron

of TPO was generated. (D) The effect of serum on luciferase gene expression in the presence or absence of bovine calf serum was measured by Q-PCR. The

results represent the mean luciferase mRNA levels adjusted for cyclophilin levels as shown from four independent experiments 6 SEM.

805B. McIntosh and K. Kaushansky/ Experimental Hematology 2008;36:799–806

expressed in the liver or a liver-derived cell line to move inthe opposite direction, which is most likely due to the gen-eral suppression of gene expression seen in most culturedcells deprived of the growth-promoting action of serum.Our finding that liver cells cultured in serum depleted me-dium or 1.25% serum displayed reduced levels of TPO, butthat there was no evidence of a dose response relationshipwhen the cells were cultured in medium supplementedwith between 2.5% and 20% serum also suggests that lowconcentrations of serum merely maintain cell metabolism,and that once cell viability is insured, there is little addi-tional effect of serum on TPO expression. Enhanced TPOexpression was also confirmed to occur at the level of se-creted protein by metabolic labeling.

As shown in Figure 5, a 1.9-kb region between 250-bpupstream of the transcriptional initiation site and the middleof the second intron of the TPO gene appears to contain ge-netic elements sufficient to mediate serum-induced suppres-sion of TPO gene transcription. One notable transcriptionfactor for which five predicted binding sites [14,15] arefound within the region between 1000 bp and 2860 bp isd-crystallin enhancer factor-1 (dEF-1). This transcriptionfactor is widely expressed and has been shown to translo-cate to the nucleus in response to serum [16]. dEF-1 gener-ally acts to downmodulate gene expression [17–26] Ofparticular interest, previous reports indicate the importanceof dEF-1 in contributing to the suppression of osteoblasticgenes, such as collagen Ia and osteocalcin [27,28]. The as-sociation of dEF-1 with osteoblasts is interesting becauseevidence suggests that TPO expression in the bone marrowoccurs primarily in osteoblasts (Dr. Toshio Suda, personalcommunication). In fact, during the course of the currentinvestigation, a variety of methods for detection TPO inbone marrow by Q-PCR were attempted, and only afterflushing away the hematopoietic marrow components andextracting the RNA from cells lining the endosteal surfaceof the bone was TPO able to be detected (data not shown).Expression of TPO by bone marrow osteoblasts is relevant,considering the recent attention that endosteal osteoblastshave garnered as the stem cell niche [29,30] and the well-known role of TPO in supporting stem cells [3,31,32].

When considering osteoblasts as the location of TPO ex-pression in the bone marrow, a body of literature concerningthe interconnection of osteoblast, osteoclast, and megakaryo-cyte development may be enlightening. Osteoblasts and oste-oclasts participate in a well-developed interrelated regulatorynetwork of cytokine and growth factors, each inhibiting thedevelopment of the other and stimulating their own (for a re-view see [33]). Interestingly, megakaryocytes have been rec-ognized as a component of this interplay (reviewed by [34]).A large number of reports indicate that megakaryocytesstimulate osteoblastogenesis and inhibit osteoclastogenesis[35–41]. Thus, one might postulate that osteoclasts, whileacting to inhibit osteoblast development, might also act indi-rectly to inhibit megakaryocyte growth by secreting factors

that suppress TPO expression and inhibit megakaryocytedevelopment.

With regard to potential platelet-derived serum factorthat may suppress TPO, one prior study demonstrated thattransforming growth factor-b1 and platelet factor-4 wereable to suppress TPO expression in human bone marrowcultures [42]. In addition, insulin and insulin-like growthfactor have been show to induce dEF-1 expression [43].In actuality, the possibilities are substantial, consideringthat platelets are known to absorb a variety of nonplateletproteins from the circulation, which they are then able torelease in abundance when activated.

In summary we found that platelet granule proteins exerta potent effect on bone marrow stromal cell production ofTPO, and that the mechanism of regulation is transcrip-tional suppression. As paracrine hormone production islikely to exert a far greater effect on platelet homeostasisthan would endocrine production from the liver or othernonmarrow organs, we believe that while platelet adsorp-tion of plasma TPO might be responsible for basal regula-tion of TPO levels, thrombocytopenia can further modulatean important source of TPO by loss of an important sup-pressor of stromal cell production of the hormone.

References1. Gurney AL, Carver-Moore K, de Sauvage FJ, Moore MW. Thrombocy-

topenia in c-mpl-deficient mice. Science. 1994;265(5177):1445–1447.

2. de Sauvage FJ, Carver-Moore K, Luoh SM, et al. Physiological regu-

lation of early and late stages of megakaryocytopoiesis by thrombo-

poietin. J Exp Med. 1996;183:651–656.

3. Ballmaier M, Germeshausen M, Schulze H, et al. c-mpl mutations are

the cause of congenital amegakaryocytic thrombocytopenia. Blood.

2001;97:139–146.

4. Kuter DJ, Rosenberg RD. The reciprocal relationship of thrombopoie-

tin (c-Mpl ligand) to changes in the platelet mass during busulfan-

induced thrombocytopenia in the rabbit. Blood. 1995;85:2720–2730.

5. Fielder PJ, Gurney AL, Stefanich E, et al. Regulation of thrombopoie-

tin levels by c-mpl-mediated binding to platelets. Blood. 1996;87:

2154–2161.

6. Stoffel R, Wiestner A, Skoda R. Thrombopoietin in thrombocytopenic

mice: evidence against regulation at the mRNA level and for a direct

regulatory role of platelets. Blood. 1996;87:567–573.

7. Lok S, Kaushansky K, Holly RD, et al. Cloning and expression of mu-

rine thrombopoietin cDNA and stimulation of platelet production in

vivo. Nature. 1994;369(6481):565–568.

8. McCarty JM, Sprugel KH, Fox NE, Sabath DE, Kaushansky K. Mu-

rine thrombopoietin mRNA levels are modulated by platelet count.

Blood. 1995;86:3668–3675.

9. Sungaran R, Markovic B, Chong BH. Localization and regulation of

thrombopoietin mRNa expression in human kidney, liver, bone mar-

row, and spleen using in situ hybridization. Blood. 1997;89:101–107.

10. Kacena MA, Nelson T, Clough ME, et al. Megakaryocyte-mediated

inhibition of osteoclast development. Bone. 2006;39:991–999.

11. Sungaran R, Chisholm OT, Markovic B, et al. The role of platelet al-

pha-granular proteins in the regulation of thrombopoietin messenger

RNA expression in human bone marrow stromal cells. Blood. 2000;

95:3094–3101.

12. Nakano T, Kodama H, Honjo T. Generation of lymphohematopoietic

cells from embryonic stem cells in culture. Science. 1994;265(5175):

1098–1101.

806 B. McIntosh and K. Kaushansky/ Experimental Hematology 2008;36:799–806

13. Bernhard HP, Darlington GJ, Ruddle FH. Expression of liver pheno-

types in cultured mouse hepatoma cells: synthesis and secretion of se-

rum albumin. Dev Biol. 1973;35:83–96.

14. Feugier P, Li N, Jo D, et al. Osteopetrotic mouse stroma with throm-

bopoietin, c-kit ligand, and flk-2 ligand supports long-term mobilized

CD34þ hematopoiesis in vitro. Stem Cells Dev. 2005;14:505–516.

15. Akiyama Y. TFSEARCH: searching transcription factor binding sites.

Available at: http://www.rwcp.or.jp/papia/. Accessed November 2006.

16. Heinemeyer T, Wingender E, Reuter I, et al. Databases on transcrip-

tional regulation: TRANSFAC, TRRD and COMPEL. Nucl Acids

Res. 1998;26:362–367.

17. Franklin AJ, Jetton TL, Shelton KD, Magnuson MA. BZP, a novel se-

rum-responsive zinc finger protein that inhibits gene transcription. Mol

Cell Biol. 1994;14:6773–6788.

18. Jethanandani P, Kramer RH. {alpha}7 Integrin expression is nega-

tively regulated by {delta}EF1 during skeletal myogenesis. J Biol

Chem. 2005;280:36037–36046.

19. Shirakihara T, Saitoh M, Miyazono K. Differential regulation of epi-

thelial and mesenchymal markers by {delta}EF1 proteins in epithelial

mesenchymal transition induced by TGF-beta. Mol Biol Cell. 2007;

18:3533–3544.

20. van Grunsven LA, Michiels C, Van de Putte T, et al. Interaction be-

tween Smad-interacting protein-1 and the corepressor C-terminal

binding protein is dispensable for transcriptional repression of

E-cadherin. J Biol Chem. 2003;278:26135–26145.

21. Postigo AA. Opposing functions of ZEB proteins in the regulation of

the TGFbeta/BMP signaling pathway. EMBO J. 2003;22:2443–2452.

22. Ponticos M, Partridge T, Black CM, Abraham DJ, Bou-Gharios G.

Regulation of collagen type I in vascular smooth muscle cells by com-

petition between Nkx2.5 and deltaEF1/ZEB1. Mol Cell Biol. 2004;24:

6151–6161.

23. Postigo AA, Depp JL, Taylor JJ, Kroll KL. Regulation of Smad signal-

ing through a differential recruitment of coactivators and corepressors

by ZEB proteins. EMBO J. 2003;22:2453–2462.

24. van Grunsven LA, Schellens A, Huylebroeck D, Verschueren K. SIP1

(Smad interacting protein 1) and deltaEF1 (delta-crystallin enhancer

binding factor) are structurally similar transcriptional repressors.

J Bone Joint Surg. 2001;83-A(Suppl 1; Pt 1):S40–S47.

25. Sekido R, Murai K, Funahashi J, et al. The delta-crystallin enhancer-

binding protein delta EF1 is a repressor of E2-box-mediated gene

activation. Mol Cell Biol. 1994;14:5692–5700.

26. Murray D, Precht P, Balakir R, Horton WE. The transcription factor

deltaEF1 is inversely expressed with type II collagen mRNA and

can repress Col2a1 promoter activity in transfected chondrocytes.

J Biol Chem. 2000;275:3610–3618.

27. Sekido R, Murai K, Kamachi Y, Kondoh H. Two mechanisms in the

action of repressor deltaEF1: binding site competition with an activa-

tor and active repression. Genes Cells. 1997;2:771–783.

28. Terraz C, Toman D, Delauche M, Ronco P, Rossert J. delta Ef1 binds

to a far upstream sequence of the mouse pro-alpha 1(I) collagen gene

and represses its expression in osteoblasts. J Biol Chem. 2001;276:

37011–37019.

29. Sooy K, Demay MB. Transcriptional repression of the rat osteocalcin

gene by {delta}EF1. Endocrinology. 2002;143:3370–3375.

30. Calvi LM, Adams GB, Weibrecht KW, et al. Osteoblastic cells regu-

late the haematopoietic stem cell niche. Nature. 2003;425(6960):

841–846.

31. Zhang J, Niu C, Ye L, et al. Identification of the haematopoietic stem

cell niche and control of the niche size. Nature. 2003;425(6960):836–

841.

32. Solar GP, Kerr WG, Zeigler FC, et al. Role of c-mpl in early hemato-

poiesis. Blood. 1998;92:4–10.

33. Sitnicka E, Lin N, Priestley GV, et al. The effect of thrombopoietin on

the proliferation and differentiation of murine hematopoietic stem

cells. Blood. 1996;87:4998–5005.

34. Boyce BF, Xing L. Biology of RANK, RANKL, and osteoprotegerin.

Arthritis Res Ther. 2007;9(Suppl 1):S1.

35. Kacena MA, Gundberg CM, Horowitz MC. A reciprocal regulatory in-

teraction between megakaryocytes, bone cells, and hematopoietic

stem cells. Bone. 2006;39:978–984.

36. Chagraoui H, Sabri S, Capron C, et al. Expression of osteoprotegerin

mRNA and protein in murine megakaryocytes. Exp Hematol. 2003;31:

1081–1088.

37. Bord S, Vedi S, Beavan SR, Horner A, Compston JE. Megakaryocyte

population in human bone marrow increases with estrogen treatment:

a role in bone remodeling? Bone. 2000;27:397–401.

38. Bord S, Frith E, Ireland DC, et al. Synthesis of osteoprotegerin and

RANKL by megakaryocytes is modulated by oestrogen. Br J Haema-

tol. 2004;126:244–251.

39. Bord S, Frith E, Ireland DC, et al. Megakaryocytes modulate osteo-

blast synthesis of type-l collagen, osteoprotegerin, and RANKL.

Bone. 2005;36:812–819.

40. Miao D, Murant S, Scutt N, Genever P, Scutt A. Megakaryocyte-bone

marrow stromal cell aggregates demonstrate increased colony forma-

tion and alkaline phosphatase expression in vitro. Tissue Eng. 2004;

10:807–817.

41. Kacena MA, Shivdasani RA, Wilson K, et al. Megakaryocyte-

osteoblast interaction revealed in mice deficient in transcription

factors GATA-1 and NF-E2. J Bone Miner Res. 2004;19:652–660.

42. Thiede M, Smock S, Petersen D, et al. Presence of messenger ribonu-

cleic acid encoding osteocalcin, a marker of bone turnover, in bone

marrow megakaryocytes and peripheral blood platelets. Endocrinol-

ogy. 1994;135:929–937.

43. Alemany J, Zelenka P, Serrano J, de Pablo F. Insulin-like growth factor

I and insulin regulate delta-crystallin gene expression in developing

lens. J Biol Chem. 1989;264:17559–17563.