Essential Role for Cyclin D3 in Granulocyte Colony-Stimulating Factor-Driven Expansion of Neutrophil Granulocytes

The proliferation of neutrophil granulocyte lineage is drivenlargely by granulocyte colony-stimulating factor (G-CSF) actingvia the G-CSF receptors. In this study, we show that mice lackingcyclin D3, a component of the core cell cycle machinery, arerefractory to stimulation by the G-CSF. Consequently, cyclinD3-null mice display deficient maturation of granulocytes inthe bone marrow and have reduced levels of neutrophil granulocytesin their peripheral blood. The mutant mice are unable to mounta normal response to bacterial challenge and succumb to microbialinfections. In contrast, the expansion of hematopoietic stemcells and lineage-committed myeloid progenitors proceeds relativelynormally in mice lacking cyclin D3, revealing that the requirementfor cyclin D3 function operates at later stages of neutrophildevelopment. Importantly, we verified that this requirementis specific to cyclin D3, as mice lacking other G₁ cyclins (D1,D2, E1, or E2) display normal granulocyte counts. Our analysesrevealed that in the bone marrow cells of wild-type mice, activationof the G-CSF receptor leads to upregulation of cyclin D3. Collectively,these results demonstrate that cyclin D3 is an essential cellcycle recipient of G-CSF signaling, and they provide a molecularlink of how G-CSF-dependent signaling triggers cell proliferation.

INTRODUCTION

Top

Introduction

The generation of hematopoietic cells from hematopoietic stemcells is a tightly regulated process. Within the myeloid lineage,approximately 120 billion granulocytes are generated every dayin the human body from a limited number of stem cells (22).The production of granulocytes is controlled by granulocytecolony-stimulating factor (G-CSF), a cytokine acting via theG-CSF receptor present on granulocyte progenitors and on maturegranulocytes (8). Granulocytes provide a defense mechanism againstmicrobial infections. These cells have a very short half-lifeof approximately 8 h (29), and their number can be rapidly increasedup to 10-fold in response to an infection (22). Abnormal, excessiveproduction of cells of the granulocytic lineage is seen in severalhematopoietic malignancies, such as acute myelogenous leukemiaand chronic myelogenous leukemia (24). Conversely, insufficientproduction of granulocytes leads to severe congenital neutropenia(SCN), also known as Kostmann syndrome (18, 19).

Much information has been gathered about the signal transductionpathways operating in granulocyte progenitors in response toG-CSF. It is less clear how these pathways activate the corecell cycle machinery, which is required to trigger cell proliferation.The key components of the core cell cycle machinery are proteinscalled cyclins (30). Cyclins represent regulatory subunits thatbind, activate, and provide substrate specificity for theirassociated cyclin-dependent kinases (CDKs) (30). In particular,cyclins operating in the G₁ phase of the cell cycle from theD-type (D1, D2, and D3) and E-type (E1 and E2) families representthe recipients of many mitogenic signals (15, 31).

In order to dissect the role of individual cyclins in tissue-specificsignaling pathways, we generated mice lacking cyclin D1, D2,D3, E1, or E2 by using gene targeting (10, 32-34). Cyclin D3-nullmice are viable, fertile, and show defects in the expansionof immature lymphocytes. Despite this, mutant mice display relativelynormal levels of mature lymphocytes in peripheral lymphoid organsand normal expansion of mature lymphocytes in response to T-cellreceptor and B-cell receptor stimulation (4, 32). In this study,we report that mice lacking cyclin D3 also display severe neutropenia.We found that cyclin D3-null animals are resistant to stimulationby the G-CSF. We also provide evidence that the mitogenic pathwaysemanating from the G-CSF receptor activate the cell cycle machineryby inducing cyclin D3.

MATERIALS AND METHODS

Top

Materials and Methods

Mice. Cyclin D1, D2, D3, E1, and E2 knockout mice were generated andPCR genotyped as described before (10, 32-34). For all experiments,we used 2- to 4-month-old cyclin-deficient and wild-type littermates.Peripheral blood was collected from the retro-orbital venousplexus into tubes containing EDTA (Vacutainer; BD). Completeblood counts were determined by using a hematology analyzer(Abbott), while the fraction of neutrophils was determined manuallyby analyzing blood smears stained with Wright-Giemsa. Bone marrowwas harvested by flushing both femoral bones with Dulbecco modifiedEagle medium plus 2% fetal bovine serum. Bone marrow cytospinswere then stained with Wright-Giemsa. All animal experimentscomplied with the relevant federal guidelines.

Administration of G-CSF. Recombinant G-CSF (Amgen) was administered to wild-type (n =6) or cyclin D3^–/– (n = 6) mice by twice-daily subcutaneousinjection at a dose of 250 µg/kg of body weight for 5days. Peripheral blood was collected and analyzed as describedabove before the first G-CSF administration and 2 h after thelast dose. The data presented in Fig. 4A are pooled from twoseparate experiments. To detect induction of cyclin D3 by G-CSF,wild-type mice (n = 3) were injected with a single dose of G-CSF(250 µg/kg). Bone marrow cells were collected after 6h, pooled, and used for Western blotting. Bone marrow cellsfrom noninjected mice (n = 3) were also pooled and served asa control.

View larger version (11K):
[in this window]
[in a new window]
FIG. 4. G-CSF signaling impacts the core cell cycle machinery through cyclin D3. (A) Resistance of cyclin D3-null mice to G-CSF. Wild-type (WT) (n = 6) and cyclin D3-null (n = 6) mice received twice daily injections of G-CSF for 5 days, and the number of granulocytes in the peripheral blood was determined at the beginning and at the end of the experiment. Mean values are shown. Error bars denote standard deviations (error bars for the values observed for cyclin D3^–/– mice are too small to be seen on the graph). (B) Wild-type mice were injected with G-CSF. Bone marrow cells were collected after 6 h and processed for Western blotting (+). Bone marrow cells from noninjected wild-type mice served as a control (–). Immunoblots were probed with an antibody against cyclin D3 or against actin (loading control). (C) 32Dcl3 cells expressing G-CSF receptor were mitogen deprived and then stimulated with G-CSF for 2 h. Immunoblots were probed with an antibody against cyclin D3 or against actin (loading control). (D) BaF3 cells engineered to express G-CSF receptor were mitogen deprived and then stimulated with G-CSF for the indicated periods. Immunoblots were probed with an antibody against cyclin D3 or against actin (loading control). (E) Northern blot analysis of RNA prepared from BaF3 cells treated as described for panel D. Blots were probed with a cyclin D3-specific probe. An ethidium bromide-stained gel is shown below (loading control).

Methylcellulose cultures. Bone marrow cells collected from wild-type (n = 5) or cyclinD3^–/– (n = 5) mice were plated in methylcellulose(no. M3434; Stem Cell Technologies), 2 x 10⁴ per dish, plus10 ng/ml G-CSF (Amgen), 10 ng/ml granulocyte-macrophage colony-stimulatingfactor (GM-CSF; R&D Systems), or 10 ng/ml interleukin-3(IL-3; R&D Systems). Colonies were counted on day 7.

Analyses of G-CSF receptor. Bone marrow cells were collected from wild-type (n = 3) or cyclinD3^–/– (n = 3) mice, stained with phycoerythrin-conjugatedanti-G-CSF receptor antibody (Abcam, Inc.), and analyzed byfluorescence-activated cell sorting (FACS). To detect G-CSF-dependentphosphorylation of STAT3, bone marrow cells were collected fromwild-type (n = 3) or cyclin D3^–/– (n = 3) mice.The cells from the same genotype were pooled, incubated withDulbecco modified Eagle medium plus 10 ng/ml G-CSF (Amgen) for15 min at 37°C, and processed for Western blotting.

Experimental infection with Listeria monocytogenes. Listeria monocytogenes (strain EGD, ATCC BAA-679) was cultured,and the titer was determined as described previously (5). Adose of 5 x 10³ CFU/mouse was inoculated intravenously via thetail vein on day 0. On days 1 and 5, groups of mice were sacrificedand livers and spleens were removed aseptically and homogenizedin saline with a Precision disposable tissue grinder system(Tyco Healthcare). To determine organ loads of bacteria, 10-foldserial dilutions of organ homogenates were plated onto bloodhorse agar and bacterial colonies were counted after incubationat 37°C for 24 h. To determine the hematological responseto infection, peripheral blood was collected and analyzed atbaseline (day 0) and on days 1 and 5, as described above. Thedata presented are the pooled results of three separate experiments,each of which used three to four mice of each genotype per timepoint.

Measurements of cytokine levels. Serum was collected on day 5 after Listeria monocytogenes challenge.The concentration of tumor necrosis factor alpha (TNF- ${alpha}$ ), IL-1ß,and IL-6 in serum was measured by an enzyme-linked immunosorbentassay using mouse TNF- ${alpha}$ , IL-1ß/IL-1F2, and IL-6 enzyme-linkedimmunosorbent assay kits (R&D Systems, Minneapolis, MN).The absorbance was measured at 450 nm using a microplate spectrophotometer(Benchmark Plus; Bio-Rad).

Bone marrow reconstitution assays. Bone marrow cells were collected from six cyclin D3^–/–mice (Ly 5.2⁺) or from six C57BL/6 mice (B6.SJL-Ptprs^a Pep3^b/BoyJ,Ly 5.1⁺; purchased from the Jackson Laboratory), and the cellswere mixed at a 1:1 or 9:1 ratio. A total of 5 x 10⁶ bone marrowcells were injected into tail veins of ¹³⁷Cs-irradiated (950cGy, split dose) Ly 5.1⁺ C57BL/6 recipients (n = 5 for eachratio). The animals were sacrificed after 3 months, and bonemarrow cells were stained for Ly 5.2 and Gr-1 or for hematopoieticstem cells (HSC) and myeloid progenitors (see below) and analyzedby FACS. Peripheral blood was stained for Ly 5.2 and Gr-1.

Stem cell and progenitor analyses. Bone marrow cells were collected from five cyclin D3^–/–and five wild-type mice. Hematopoietic stem cells were stainedas IL7R ${alpha}$ ^–Lin^–Sca-1^hic-Kit^hi, common myeloid progenitors(CMP) as IL7R ${alpha}$ ^–Lin^–Sca-1^–c-Kit⁺CD34⁺Fc ${gamma}$ RII/III^lo,and granulocyte-macrophage progenitors (GMP) as IL7R ${alpha}$ ^–Lin^–Sca-1^–cKit⁺CD34⁺ Fc ${gamma}$ RII/III^hi, essentially as described previously (2).

Assays of in vitro-cultured cells. A subline of the IL-3-dependent murine myeloid cell line 32D,called 32Dcl3, engineered to express the wild-type G-CSF receptorwas described before (7). 32Dcl3 cells were starved for 4 hand then stimulated with 10 ng/ml G-CSF (Amgen) for 2 h. BaF3cells engineered to express G-CSF receptor were described before(27). Cells were cultured in RPMI 1640 medium with 10% fetalbovine serum and 10 ng/ml recombinant murine IL-3 (PeproTech).Prior to stimulation with G-CSF, cells were starved for 6 hin complete serum-free medium (Cellgro) and then stimulatedwith 10 ng/ml G-CSF (Amgen). Cells were lysed and processedfor Western blotting.

Western and Northern blotting. Protein lysates (40 µg) were separated on 12% sodium dodecylsulfate-polyacrylamide gel electrophoresis gels and transferredto Immobilon-P membranes (Millipore). Immunoblots were probedwith antibodies against cyclin D3 (C-16; Santa Cruz), phospho-STAT3(Tyr 705; Cell Signaling), STAT3 (C-20; Santa Cruz), or actin(MAB1501; Chemicon International). Northern blot analyses wereperformed as described previously (32, 39).

RESULTS

Top

Results

Severe neutropenia in cyclin D3-null mice. Generation of cyclin D3-deficient mice has been described previously(32). Analyses of the peripheral blood in cyclin D3-null micerevealed that the numbers of neutrophil granulocytes were reducedmore than fourfold in the mutant animals (Fig. 1A and B); cyclinD3^+/– heterozygotes displayed normal neutrophil levels(data not shown).

View larger version (20K):
[in this window]
[in a new window]
FIG. 1. Neutropenia in cyclin D3-null mice. (A) Mean values of white blood cell counts, percentages of neutrophils, and total numbers of neutrophils in peripheral blood of wild-type (WT) or knockout mice lacking D-type (D1, D2, and D3) or E-type (E1 and E2) cyclins. Standard deviations are shown in parentheses. (B) Mean neutrophil counts in peripheral blood of wild-type and cyclin D3-null mice, presented as a graph. Error bars denote standard deviations. (C) Analyses of bone marrow composition in wild-type and cyclin D3-null mice. Immature myeloid cells are comprised of myeloblasts, promyelocytes, myelocytes, and metamyelocytes. Mature cells are comprised of bands and polymorphonuclear neutrophil granulocytes. Mean values (±standard deviations) are shown.

We verified that this phenotype was specific to mice lackingcyclin D3, as knockout animals deficient in cyclin D1, D2, E1,or E2 displayed normal neutrophil counts (Fig. 1A). The neutropeniaseen in cyclin D3-null mice resembled the phenotype of knockoutmice lacking G-CSF (21) or the G-CSF receptor (22).

The development of neutrophil granulocytes from hematopoieticstem cells takes place within the bone marrow. Therefore, weanalyzed the composition of bone marrow in cyclin D3-null animals.We observed that the total number of bone marrow cells was reducedmore than twofold in the mutant animals, due largely to a fourfoldreduction in the number of bone marrow granulocytes (Fig. 1C).Importantly, the fraction of immature granulocyte progenitors(myeloblasts, promyelocytes, myelocytes, and beta-myelocytes)was increased in the mutant animals, suggesting a partial blockin granulocyte maturation (Fig. 1C). Again, a similar phenotypewas reported in mice lacking G-CSF receptors (22). Collectively,these results suggested that cyclin D3 might represent a cellcycle recipient of the G-CSF receptor-dependent pathways.

During granulocyte development within the bone marrow, HSC giverise to lineage-committed CMP. CMP in turn give rise to GMP,which, through a series of divisions, generate mature granulocytes(Fig. 2A) (2). In order to gauge the requirement for cyclinD3 function at these various stages of myeloid development,we determined the numbers of HSC, CMP, and GMP in bone marrowof cyclin D3-null mice. We observed that the mutant bone marrowcontained normal numbers of HSC, CMP, and GMP (Fig. 2B and C),whereas the numbers of mature granulocytes were greatly decreased(Fig. 1). These analyses indicate that cyclin D3 is largelydispensable for the expansion of hematopoietic stem cells andmyeloid progenitors (CMP and GMP) but that it is required inmyeloid development downstream of GMP.

View larger version (20K):
[in this window]
[in a new window]
FIG. 2. Analyses of cyclin D3-null bone marrow cells. (A) Mean contribution of cyclin D3^–/– bone marrow cells to HSC, CMP, GMP, and Gr1⁺ cells in the bone marrow and Gr1⁺ granulocytes in the peripheral blood in mice reconstituted with a 1:1 or 9:1 ratio of cyclin D3^–/– and wild-type bone marrow cells. (B) Staining of bone marrow cells isolated from wild-type (WT) or cyclin D3-null mice for the presence of HSC, CMP, and GMP. The fractions of particular populations among all bone marrow cells are indicated. (C) Mean absolute numbers of HSC, CMP, and GMP, calculated per femur, in wild-type and in cyclin D3-deficient mice. Error bars denote standard deviations. Note that cyclin D3-null mice have a reduced number of total bone marrow cells; hence, they show increased fractions (see panel B) but normal absolute numbers of HSC and myeloid progenitors.

Bone marrow reconstitution analyses. As mentioned above, cyclin D3-null mice display a defect inthe expansion of immature lymphocytes (4, 32). Although cyclinD3-deficient animals have nearly normal levels of peripheralmature T and B lymphocytes, it was nevertheless possible thatthe defect in granulocytic lineage within the bone marrow wassecondary to the abnormalities in other compartments (such asthe lymphoid compartment). To determine whether the neutrophilabnormalities described here were intrinsic to the myeloid lineage,we employed competitive bone marrow reconstitution assays. Specifically,we mixed cyclin D3-null and wild-type bone marrow cells at a9:1 or 1:1 ratio, and we injected the cells into lethally irradiatedwild-type hosts. We assayed the recipient mice 3 months afterthe reconstitution. At this time, all transplanted granulocytesor lineage-committed progenitors had died off and all hematopoieticcells of the recipient mice arose from the transplanted hematopoieticstem cells. This approach allowed us to study the neutrophildevelopment from cyclin D3-null hematopoietic stem cells inthe context of a cyclin D3 wild-type animal. Importantly, thetransplanted cyclin D3-null and wild-type hematopoietic stemcells express different haplotypes of the surface Ly 5 antigen,thereby allowing unequivocal distinction of mutant-derived (Ly5.2-positive) and wild-type (Ly 5.1-positive) cells. If no competitivedisadvantage for cyclin D3-null cells exists at a particulardevelopmental stage, one expects that the Ly 5.2-positive cellswould contribute to approximately 50% (1:1 reconstitution) or90% (9:1 reconstitution) of hematopoietic cells. Conversely,lower-than-expected levels of reconstitution by the mutant cellsindicate a competitive disadvantage for cyclin D3-null hematopoieticcells.

We collected bone marrow cells from the reconstituted chimericmice and determined the contribution of Ly 5.2- and Ly 5.1-positivecells to HSC, CMP, GMP, and Gr-1⁺ from a neutrophil lineage.Our analyses revealed a relatively robust contribution of cyclinD3-null cells to hematopoietic stem cells as well as to CMPand GMP compartments (Fig. 2A), consistent with our findingsthat cyclin D3 is not required at this stage of development.In contrast, the contribution of transplanted cyclin D3-nullHSC to Gr-1⁺ neutrophils within the bone marrow and in the peripheralblood was drastically reduced (Fig. 2A). These results indicatethat the defect seen in cyclin D3-null mice is intrinsic tothe myeloid lineage.

Susceptibility of cyclin D3-null mice to infections. Mice lacking G-CSF were shown to be unable to activate "emergency"granulopoiesis that normally takes place upon exposure of miceto bacteria (21). Consequently, G-CSF-null mice are highly susceptibleto microbial infections (21). Given the similarities betweenthe phenotypes of mice lacking G-CSF and cyclin D3, we wishedto gauge the response of cyclin D3-null mice to bacterial challenge.Of note, although our cyclin D3-null mice were kept in a pathogen-freeenvironment, these mutants occasionally succumbed to infections.Upon necropsy, we observed the formation of large abscessesin the affected animals (Fig. 3A). In contrast, we never observedsuch infections among wild-type mice housed under identicalconditions. In order to rigorously gauge the susceptibilityof cyclin D3-null mice to bacterial infections, we challengedthe mutant and control wild-type animals with a single inoculumof gram-positive bacteria, Listeria monocytogenes. Wild-typemice responded to the infection by increasing the neutrophilcounts in the peripheral blood. In contrast, in cyclin D3-nullmice, the neutrophil levels never reached the values seen forinfected wild-type controls (Fig. 3B). As a consequence, thebacterial titers in the livers and spleens of the infected cyclinD3^–/– animals were elevated more than 100-fold comparedto those of their infected wild-type counterparts (Fig. 3C).At the end of the experiments, the livers and spleens of themutant mice were overwhelmed with infection, in contrast tothe normal, healthy appearance of these organs in control animals(Fig. 3D); cyclin D3^+/– heterozygotes behaved like wild-typecontrols (data not shown). Hence, like G-CSF-deficient mice,mice lacking cyclin D3 are unable to mount an "emergency" granulopoiesisresponse, and consequently, they show greatly increased susceptibilityto bacterial infections. The normal life spans of these animalsare most likely possible due to the fact that the mice are keptin an artificial, pathogen-free environment.

View larger version (38K):
[in this window]
[in a new window]
FIG. 3. Increased susceptibility of cyclin D3^–/– mice to bacterial infections. (A) Necropsy of a cyclin D3^–/– animal that developed spontaneous infection despite being kept in a pathogen-free environment. The arrow points to a large abscess of the right kidney. (B to D) Wild-type (WT) and cyclin D3-null mice were inoculated with Listeria monocytogenes. Animals were sacrificed 1 day or 5 days after bacterial challenge, and the numbers of neutrophils in peripheral blood (B) and bacterial titers in livers and spleens (C) were determined. Mean values are shown, and error bars represent standard errors. In panel B, the relative values of neutrophil counts were as follows: wild type, day 0, 100%; day 1, 209%; day 5, 138%; and cyclin D3^–/–, day 0, 100%; day 1, 187%; day 5, 109%. (D) Appearance of livers and spleens 5 days after bacterial challenge. Arrowheads point to microabscesses present in cyclin D3-null mice. Bars, 5 mm.

It was previously shown that IL-1, IL-6, and TNF- ${alpha}$ are criticallyrequired to trigger a normal response to Listeria monocytogeneschallenge (6, 11, 13, 26, 28). The possibility that the phenotypeof cyclin D3-null mice was caused by insufficient cytokine levelswas ruled out by the observations that the levels of these cytokineswere elevated in Listeria-infected cyclin D3-null mice comparedto those of their infected wild-type counterparts (IL-1ß,40.39 ± 18.56 pg/ml versus 31.53 ± 12.01 pg/ml;IL-6, 579.25 ± 60.08 pg/ml versus 62.21 ± 2.41pg/ml; TNF- ${alpha}$ , 372.28 ± 25.91 pg/ml versus 86.54 ±35.21 pg/ml; values represent means ± standard deviations).Hence, the inability of cyclin D3^–/– mice to mountthe response to Listeria infection cannot be attributed to insufficientcytokine levels. Instead, we propose that this phenotype reflectsan intrinsic proliferative deficiency of cyclin D3-null granulocytes.

Resistance of cyclin D3-deficient mice to G-CSF treatment. Given the similarities between the phenotypes of cyclin D3-nullmice and animals lacking the G-CSF or G-CSF receptor, we decidedto directly test the response of cyclin D3-null mice to G-CSF.To this end, we challenged cyclin D3^–/– and controlwild-type animals with five consecutive daily doses of G-CSFand we monitored the response. As expected, wild-type animalsresponded to G-CSF treatment with a more-than-ninefold increasein their neutrophil counts. In contrast, cyclin D3-null micewere completely refractory to G-CSF treatment (Fig. 4A).

Importantly, we confirmed that cyclin D3-null bone marrow cellsexpressed the G-CSF receptor (Fig. 5A). We also verified that,as with wild-type animals, stimulation of cyclin D3-null micewith G-CSF led to rapid phosphorylation of STAT3 within thebone marrow cells (Fig. 5B), revealing preserved signaling bythe activated G-CSF receptor (note that the levels of phosphorylatedSTAT3 were unperturbed in cyclin D3-null bone marrow, despitethe decreased overall STAT3 levels). Hence, the resistance ofcyclin D3-null mice to G-CSF treatment is unlikely to be causedby the deficient G-CSF receptors. Instead, we propose that thesignaling pathways operating downstream of the G-CSF receptorsfeed to the core cell cycle machinery through cyclin D3. Consequently,in the absence of cyclin D3, G-CSF fails to elicit a normalmitogenic response in granulocytic progenitors.

View larger version (12K):
[in this window]
[in a new window]
FIG. 5. Analyses of G-CSF receptor signaling in cyclin D3-null mice. (A) Bone marrow cells collected from wild-type (WT) (n = 3) or cyclin D3^–/– (n = 3) mice were stained with an antibody against the G-CSF receptor and analyzed by FACS. The mean percentage (± standard deviation) of G-CSF receptor-positive cells is indicated. The negative control (unstained cells) is also shown. The horizontal axis represents the intensity of G-CSF receptor staining; the vertical axis represents the number of events. (B) Bone marrow cells isolated from wild-type or cyclin D3-null mice were stimulated with G-CSF for 15 min (+G-CSF) and processed for Western blotting along with bone marrow cells from nonstimulated animals (–G-CSF). Immunoblots were probed with an antibody against the phosphorylated form of STAT3 (pSTAT3) (phosphorylation of STAT3 represents one of the early events in G-CSF receptor signaling, in response to the receptor's activation by G-CSF), with an antibody recognizing both unphosphorylated and phosphorylated forms of STAT3 (Total STAT3) or with an antiactin antibody (loading control).

To further test the connection between G-CSF signaling and cyclinD3, we injected wild-type mice with G-CSF, collected bone marrow,and analyzed it by Western blotting. We observed upregulationof cyclin D3 by G-CSF (Fig. 4B). Note that for our analyses,we used total, unfractionated bone marrow, in which G-CSF-responsivecells represent only a minor fraction of cells (20). Hence,the magnitude of cyclin D3 induction is likely much higher amongthe responsive cells. We extended these findings using murinemyeloid cell line 32Dcl3 engineered by us to express the G-CSFreceptor (36). This is one of the very few cell lines that canterminally differentiate into neutrophils upon G-CSF challenge(12). Again, we observed that treatment of 32Dcl3 cells withG-CSF led to upregulation of cyclin D3 (Fig. 4C). We obtainedsimilar results using another cell line, BAF3, engineered toexpress the G-CSF receptor (27). Stimulation of these cellswith G-CSF led to a rapid induction of cyclin D3 protein (Fig.4D) and transcripts (Fig. 4E). Importantly, G-CSF stimulationupregulated specifically cyclin D3 but not related cyclins D1and D2 (data not shown). These results further strengthen ourconclusion that cyclin D3 represents the essential cell cyclerecipient of G-CSF receptor signaling.

We wished to determine whether the requirement for cyclin D3function in the expansion of the neutrophil lineage is specificto G-CSF. Therefore, we tested the ability of cyclin D3-nullhematopoietic progenitors to respond to other cytokines thatstimulate granulopoiesis. We plated cyclin D3^–/–and wild-type bone marrow cells in methylcellulose in the presenceof GM-CSF, IL-3, or G-CSF, and we scored the formation of granulocyticcolonies. Consistent with the analyses described above, we foundthat the number of colonies growing in G-CSF-containing mediumwas reduced in the mutant bone marrow. Strikingly, cyclin D3-nullbone marrow cells yielded normal numbers of neutrophil coloniesin GM-CSF- or in IL-3-containing medium (Fig. 6). These resultssuggest that the defect in granulocyte development seen in cyclinD3-null animals is specific to G-CSF signaling.

View larger version (8K):
[in this window]
[in a new window]
FIG. 6. In vitro cultures of cyclin D3-null bone marrow cells. Equal numbers of bone marrow cells (2 x 10⁴) collected from wild-type (WT) or cyclin D3^–/– mice were plated in methylcellulose, and the formation of colonies in response to the indicated cytokines was scored. Mean values are shown. Error bars denote standard deviations.

DISCUSSION

Top

Discussion

D-type cyclins (cyclin D1, D2, and D3) represent the componentsof the core cell cycle machinery in mammalian cells. The expressionof the D cyclins is controlled primarily by the extracellularenvironment. Once induced, D cyclins bind and activate theirassociated cyclin-dependent kinases, CDK4 and CDK6 (30).

In the past, we and others generated mice lacking D-type cyclinsby gene targeting. We observed that cyclin D-deficient micewere viable and displayed very circumscribed phenotypes (9,32-34). These cyclin D-null mice provided us with tools to studyhow various tissue-specific mitogenic and oncogenic pathwaysimpact the core cell cycle machinery. Analyses of cyclin D1-deficientmice revealed that the ErbB-2 $->$ Ras pathway signals to the cellcycle machinery in mammary epithelial cells through cyclin D1.We observed that mice lacking cyclin D1 were completely resistantto mammary carcinomas triggered by the ErbB-2 and Ras oncogenes(39). Likewise, we reported that the follicle-stimulating hormone $->$ cyclicAMP pathway induces cyclin D2 and that cyclin D2-null mice areresistant to the stimulation by the follicle-stimulating hormone(33). Lastly, we observed that the pre-T-cell receptor $->$ p56^LCKpathway signals through cyclin D3 (32). In this study, we extendthese observations by demonstrating that during myeloid development,the mitogenic pathways emanating from the G-CSF receptor impactthe core cell cycle machinery through cyclin D3. Importantly,we show that cyclin D3 represents an essential cell cycle recipientof this pathway, as evidenced by our observations that micelacking cyclin D3 are refractory to the stimulation by G-CSF.These results provide a molecular link of how G-CSF-dependentsignaling triggers cell proliferation. Given that abnormal proliferationof hematopoietic cells is an underlying cause of several humandiseases, such as leukemias and aplastic syndromes (16, 24,42), it is important to elucidate the mechanisms that governnormal proliferation of these lineages.

It remains to be determined whether the critical function ofcyclin D3 in the granulocyte lineage is dedicated solely topromote cell cycle progression. Ectopic expression of cyclinD3 or D2 (but not D1) in 32Dcl3 cells was shown to prevent granulocytedifferentiation (17). Conversely, inhibition of cyclin D-CDK4/6function (by ectopic expression of p19^INK4d) induced premature,accelerated granulocytic differentiation of 32Dcl3 cells inresponse to G-CSF (1). These findings raise the possibilitythat cyclin D3 activity may negatively regulate the differentiation-specificprograms. Consistent with this notion, cyclin D3 was shown tophysically interact with AML1, a transcription factor involvedin myeloid differentiation (25).

The impaired neutrophil development seen in cyclin D3-null miceresembles the appearance of patients with SCN, also known asKostmann syndrome. SCN is a rare, genetically heterogeneousautosomal recessive disorder (18, 19). The genetic basis ofKostmann syndrome is currently unknown. The affected individualsdisplay greatly reduced neutrophil counts in the peripheralblood (<500/mm³), show a block in the early stages of neutrophildevelopment in the bone marrow, and are highly prone to life-threateningbacterial infections (3, 23, 37, 41, 42). A subset of patientsdisplays a G-CSF-resistant form of the disease (14, 38, 40,41). These similarities prompted us to test whether mutationswithin the cyclin D3 gene might underlie human severe congenitalneutropenia cases. We collected peripheral blood from 16 G-CSF-resistantand 23 G-CSF-responsive patients, along with 17 normal controls,and we sequenced the entire coding sequence of the cyclin D3gene. We did not detect any abnormalities except for a polymorphism(T940G), seen in all three groups at similar frequencies, whichchanges an alanine into a serine at amino acid position 259(data not shown). These results essentially rule out mutationswithin cyclin D3 as a genetic cause of SCN. It remains possible,however, that the expression of cyclin D3 is compromised inKostmann syndrome patients through other mechanisms, such asmethylation of cyclin D3's promoter and lesions in the pathwaythat control cyclin D3 expression, including transcription factors.The promoter of the cyclin D3 gene was shown to contain potentialbinding sites for several transcription factors, including GATA,C/EBP, NF- ${kappa}$ B, Myb, and ATF (35). However, the repertoire of transcriptionfactors that control cyclin D3 expression in the granulocyticlineage remains to be determined. Regardless of the exact molecularlesion in human Kostmann syndrome patients, cyclin D3-null miceprovide a model for testing various therapeutic strategies forindividuals affected with G-CSF-resistant severe congenitalneutropenia.

ACKNOWLEDGMENTS

We thank the Sicinski lab for help and advice.

This work was supported by R01 grant CA108420 and P01 grantCA109901 from the NIH to P.S. P.S. is a scholar of the Leukemiaand Lymphoma Society. J.G. and I.P.T. are supported by the DutchCancer Society KWF Kankerbestrijding. Y.-M.L. was partiallysupported by a fellowship from the Korea Science and ResearchFoundation (KOSEF).

FOOTNOTES

* Corresponding author. Mailing address: Dana-Farber Cancer Institute, Smith Research Building, 44 Binney Street, Boston, MA 02115. Phone: (617) 632-5005. Fax: (617) 632-5006. E-mail: peter_sicinski{at}dfci.harvard.edu.

Published ahead of print on 5 September 2006.

These authors contributed equally to this work.

REFERENCES

Top