The biological consequences of excess GM-CSF levels in transgenic mice also lacking high-affinity receptors for GM-CSF

Leukemia (1998) 12, 353–362 1998 Stockton Press All rights reserved 0887-6924/98 $12.00

The biological consequences of excess GM-CSF levels in transgenic mice also lackinghigh-affinity receptors for GM-CSFD Metcalf, S Mifsud, L Di Rago, L Robb, NA Nicola and W Alexander

The Walter and Eliza Hall Institute of Medical Research, PO Royal Melbourne Hospital, 3050 Victoria, Australia

GM-CSF transgenic mice were crossed with mice with homo-zygous inactivation of the gene encoding the common b chain(bc) of the GM-CSF receptor to produce mice with constitut-ively elevated GM-CSF levels but no high-affinity GM-CSFreceptors. GM-CSF transgenic bc 2/2 mice had exceptionallyelevated serum GM-CSF levels but failed to develop the abnor-mal peritoneal cell population, eye destruction or tissue lesionscharacteristic of GM-CSF transgenic bc 1/1 mice. The alveolarproteinosis of bc 2/2 mice was not altered in GM-CSF trans-genic bc 2/2 mice. Levels of GM-CSF mRNA in transgenic GM-CSF bc 2/2 were elevated but lower than in transgenic b 1/1mice and the higher serum GM-CSF levels were traced in partto the longer serum half-life of GM-CSF in bc 2/2 than in bc1/1 mice although urinary loss of GM-CSF was higher in bc2/2 than in 1/1 mice. The data indicate that the transgenicphenotype was due to stimulation by GM-CSF and not an inser-tional effect, that low-affinity receptors are not capable of initi-ating tissue pathology even in the presence of excess GM-CSFlevels and that autocrine production of GM-CSF by GM-CSF-responsive cells also fails to induce changes in these cells. Theresults support current dogma that the action of polypeptideregulators is mediated exclusively by activation of high-affinitymembrane receptors.Keywords: granulocyte–macrophage colony-stimulating factor;membrane receptors; transgenic; homozygous inactivation; tissuepathology

Introduction

Commencing two decades ago, there has been a progressiverecognition that specific membrane receptors exist for poly-peptide regulators and that these mediate interactionsbetween the regulators and their responding target cells. Indi-vidual specific receptor chains are transmembrane proteins,of low affinity for their respective ligands.

Although polypeptide regulators themselves have a highlycomplex three-dimensional structure, current views havedowngraded the role of regulators to that of merely acting ascross-linking agents bringing together two or more receptorchains, with all signaling events originating from subsequentcross-activation of the receptor chains. Receptor–ligand com-plexes are rapidly internalized, the complex is then dis-sociated in intracellular endosomes, usually after fusion withlysosomes, and the ligand (regulator) degraded by proteo-lysis.1 No significant functional role is ascribed either to low-affinity complexes of the regulator with its specific a receptorchain or to the regulator molecules themselves after entry intothe cells.

In the context of leukemic cells, where transformation ofteninvolves an acquired capacity by the cells to produce theirown relevant polypeptide growth factors,2–4 these currentviews warrant reanalysis. Is it certain that intracellular growthfactors have no action on responsive cells in the absence ofhigh-affinity receptors? Can the regulator molecules them-

Correspondence: D Metcalf; Fax: 61-3-9347-0852Received 29 September 1997; accepted 14 November 1997

selves elicit some cellular responses, acting alone or whencomplexed merely with low-affinity a-chain receptors? Whatis the situation when excessive amounts of regulators arebeing produced inside responsive cells?

The present study involved granulocyte–macrophagecolony-stimulating factor (GM-CSF), a glycoprotein hemato-poietic regulator with major proliferation actions on neutro-philic granulocytes, monocyte–macrophages and eosinophilsand an ability to functionally activate macrophages.5,6 TheGM-CSF receptor is a heterodimer composed of a low-affinityspecific a-chain and a signaling b-chain (bc) which is sharedin common by the specific a-receptor chains for IL-5 and IL-3.5 Most evidence indicates that the b-chain initiates signalingevents5 but there has been one report of membrane transportresponses initiated by the a-chain.7

The specific questions concerning the relative roles playedby GM-CSF and its receptor in eliciting cellular responseswere addressed by developing a transgenic mouse producingexcessive levels of GM-CSF but possessing no signaling GM-CSF b-receptor chains, and thus no high-affinity GM-CSFreceptors. GM-CSF transgenic, bc −/− mice were analyzed forthe development of the macrophage population changes andorgan damage characteristic of GM-CSF transgenic mice.8–10

The opportunity was also taken with these mice to study thepattern of tissues producing GM-CSF in excessive amountsand, in non-transgenic bc −/− mice, to study the plasma half-life of GM-CSF and in particular the renal clearance of GM-CSF in animals lacking high-affinity GM-CSF receptors.

Materials and methods

Mice

The production of heterozygous GM-CSF transgenic(C57BL × SJL)F2 mice (GM-CSF transgenic mice) has beendescribed previously.9 The transgenic state is identifiable clini-cally by the small size and opaque appearance of the eye.For the present studies, mice were also identified by Southernanalysis of tail DNA after enzyme digestion to identify theadditional band originating from the insertion of two trans-gene copies of the GM-CSF gene.9

The production of mice with homozygous inactivation ofthe gene encoding the common b-chain of the GM-CSF recep-tor has been described previously.11 These mice were ident-ified by a PCR-based analysis of tail DNA. Tail DNA was pre-pared and 1 ml was amplified in a PCR reaction mixturecontaining bc primers 5′ GTG TAG ACA CTG GCC CCC G-3′and 5′ GAA CCT TCA ATG CTT CTT TGA TGG GAT-3′ andneo primer 5′-ATA TTG CTG AAG AGC TTG GCG GC-3′.Reaction conditions were 96°C 30 s, 60°C 50 s, 72°C 3 minfor 35 cycles in a 50 ml reaction mixture with 200 ng eachprimer, 0.2 m mol/l each dNTP, 1 × buffer (Boehringer,Mannheim, Germany) and 2.5 U Taq (Boehringer).

Production of GM-CSF transgenic mice with homozygous

Transgenic GM-CSF miceD Metcalf et al

354inactivation of the bc GM-CSF receptor was achieved by firstcrossing heterozygous GM transgenic mice with homozygousbc −/− mice then mating offspring that were bc +/− and GM-CSF transgenic +/−. Grouping together the mice that werehomozygous or heterozygous for the GM-CSF transgene, thiscross yielded mice of six genotypes as shown in Table 1.

All analyses were made on mice aged between 2 and 3months that were raised in specific pathogen-free quarters.

Analyses

White cell counts were performed under penthrane anesthesiaon orbital plexus blood then the mice were killed by col-lecting the blood from the axillary vessels. At this time, urinesamples were also collected and subsequently passed througha NAP-5 column (Pharmacia, Uppsala, Sweden) to removetoxic low molecular weight inhibitors. Sera were collected,diluted 1:4 with saline then millipore filtered. Peritoneal cellswere collected by injecting 2 ml of 5% fetal calf serum in0.9% saline intraperitoneally into the intact abdomen, thengently massaging the abdomen and harvesting the cell suspen-sion using a soft plastic Pasteur pipette. Marrow cells werecollected from one femur shaft for counting and analysis ofcytocentrifuge suspensions after staining with May–Grune-wald Giemsa.

The organs were then removed and fixed in 10% formalinin 0.9% saline. Histological analysis was performed on botheyes, the salivary gland, thymus, heart, lung, liver, diaphragm,spleen, mesenteric lymph node, kidney, bladder, small andlarge intestine, pancreas, uterus and ovaries or testis and semi-nal vesicles, brain, skin, skeletal muscle, sternum and onefemur plus tibia.

GM-CSF assays

Assays for GM-CSF in serum, urine and peritoneal cell-conditioned media were performed using 60 well microtitertrays, each well containing 200 FDC-P1 cells in 10 ml vol-umes of Dulbecco’s modified Eagle’s medium with 10% new-born calf serum and duplicate serial two-fold dilutions of thetest material in 5 ml volumes.12 Cultures were scored after48 h of incubation at 37°C in a fully humidified atmosphereof 10% CO2 and the GM-CSF content of test specimens calcu-lated as ng/ml, using for calibration parallel titrations of a pur-ified recombinant murine GM-CSF standard containing1 ng/ml. The lower detection level of the assay was 100 pg/mland because serum samples were initially diluted 1:4, thelower detection limit for serum GM-CSF was 400 pg/ml.

Table 1 Frequency of observed genotypes in the progeny ofintermated heterozygous GM-CSF transgenic bc +/− mice

bc status GM-CSF Expected Observedtransgene frequency ratios numbers

status

bc −/− + 3 15− 1 6

bc +/− + 6 44− 2 15

bc +/+ + 3 19− 1 7

Because of the use of serial two-fold dilutions in the assays,the mean data from groups of mice were expressed as geo-metric means and in the Tables the actual ranges of valuesobserved are also indicated.

Serum GM-CSF half-life and renal clearance studies

bc −/− and bc +/+ mice were injected intravenously with0.2 ml of 5% fetal calf serum in 0.9% saline containing either400 ng or 2000 ng purified recombinant murine GM-CSF,either non-glycosylated as produced by E. coli or glycosylatedas produced by murine erythroleukemia cells. Renal clearancestudies were performed by holding individual mice in metab-olism cages for 5 h following the intravenous injection of GM-CSF in a volume of 0.4 ml to promote unine production.

Production of GM-CSF by peritoneal cells

Suspension cultures were prepared in 1 ml volumes of Dul-becco’s modified Eagle’s medium containing 10% newborncalf serum in 35-mm Petri dishes using 1 × 106 peritonealcells. Medium was collected from replicate cultures after 3and 6 h of incubation at 37°C in a fully humidified atmos-phere of 10% CO2 in air. Production of GM-CSF by adherentperitoneal cells (.95% macrophages) was performed by incu-bating 1 × 106 peritoneal cells in the culture dishes for 3 hthen washing non-adherent cells free by three cycles of wash-ing. Adherent cells were then incubated in fresh medium fora second 3-h period before collection of the medium for GM-CSF assays.

RNA extraction and RT-PCR

Total RNA was extracted from the organs of adult male miceas previously described.13 Oligo-dT-primed first strand cDNAwas synthesized from 5 mg of total RNA using ‘Ready-To-Go’reaction mixes (Pharmacia). RT-PCR was performed using pre-viously described oligonucleotide primers and methods12 withthe following modifications. Template cDNA from each organwas initially tested for b-actin amplification and then appro-priate amounts used for specific amplification of GM-CSFsequences. 20 ml reactions containing template cDNA,200 mM dNTPs, 1 mM primers, 0.5 Units Taq DNA polymerase(Boehringer) and the buffer supplied by the manufacturer,were cycled 25 times through a 94°C/30 s, 60°C/30 s,72°C/30 s amplification protocol. Amplified products wereseparated by 1% agarose gel electrophoresis, transfered toGeneScreen Plus nylon membrane (NEN Life Sciences, Bos-ton, MA, USA) and hybridized with oligonucleotides locatedinternally to the primers used for amplification. Specific RT-PCR products were visualized by autoradiography.

Autoradiography of tissues

Non-glycosylated purified recombinant GM-CSF was labeledwith 125I using methods described previously.14 125I-GM-CSFin Dulbecco’s modified Eagle’s medium containing 10% fetalcalf serum was injected into 2-month-old bc +/+ and bc −/−mice using 3 × 106 c.p.m. in 0.2 ml. Four hours later, the micewere killed and sections of the organs subjected to autoradio-graphic analysis as described previously.15


355Results

Table 1 shows the frequency of the six genotypes resultingfrom the intermating of heterozygous GM-CSF transgenic bc+/− mice. Although some GM-CSF +/+ mice resulted from thisintercross, such mice were grouped with the GM-CSF +/−mice because previous studies had indicated no major differ-ences between mice of these two types. Analysis of the first106 mice from this mating revealed a distribution of genotypesthat was very close to the expected ratio. This indicated thatnone of the genotypes was associated with fetal or neonataldeath. Unexpectedly, heterozygous GM-CSF transgenic bc+/− mice died somewhat earlier than standard GM-CSF trans-genic bc +/+ mice and, with the progeny of the mating, allobservations were performed on mice under the age of 3months to avoid complications that could have arisen from theearly death of mice of several of the genotypes under analysis.

Analysis of the mice

All observations were performed without prior knowledge ofthe genotype of the mouse, with the inevitable exception thatsome mice were readily identifiable as GM-CSF transgenicbecause of the characteristic opacity of their eyes.

Assays for GM-CSF levels in the serum and urine revealedthat all mice typed as GM-CSF transgenic had elevated GM-CSF concentrations in both the serum and urine (Table 2). Inagreement with earlier studies, male transgenic mice hadhigher GM-CSF concentrations in the urine than did femalemice but the present mice also showed a similar sex differencein serum GM-CSF concentrations. GM-CSF transgenic micethat were also bc −/− exhibited two- to five-fold higher meanGM-CSF concentrations in both the serum and urine thanwere present in GM-CSF transgenic bc +/− or bc +/+ mice. Itis of interest that serum GM-CSF levels varied over a 16-foldrange in bc −/− mice rather than the usual four-fold variationin bc +/− or bc +/+ mice. No GM-CSF was detectable in theserum of non-transgenic bc +/+ or bc +/− mice and most bc−/− mice but low levels of GM-CSF were consistently detectedin the urine of non-transgenic bc −/− mice.

These assays indicated that the tissues of GM-CSF trans-genic bc −/− mice were being exposed to high GM-CSF con-centrations, indeed to concentrations that were significantlyhigher than in standard GM-CSF transgenic mice.

Table 2 GM-CSF concentrations in serum and urine

bc status GM-CSF status Sex Number of mice Serum GM-CSF ng/ml Urine GM-CSF ng/ml

bc −/− + M 8 31.6 (8–128) 61.7 (11–160)F 7 14.5 (4–64) 13.5 (4–20)

− M 1 0 2F 3 0.3 (0–1) 2.1 (0.3–4.8)

bc +/− + M 13 8.9 (4–16) 12.3 (2–57)F 24 6.9 (4–16) 5.1 (2–29)

− M 7 0 (0–0) 0 (0–0)F 6 0 (0–0) 0 (0–0)

bc +/+ + M 6 10.0 (4–16) 19.5 (4–74)F 7 5.4 (4–16) 3.8 (2–17)

− M 3 0 (0–0) 0 (0–0)F 1 0 0

Data shown are the geometric means of the observed values and the figures in brackets are the range of values observed.

Changes in peritoneal cell populations

The results of analyzing the peritoneal populations in thepresent mice are shown in Table 3. No differences wereobserved between the peritoneal cell population of non-transgenic mice of any bc status. GM-CSF transgenic bc +/+and bc +/− mice both showed the changes characteristic ofthe transgenic state – a massive increase in peritoneal macro-phages that are enlarged, basophilic and often binucleate ormultinucleate, a rise in eosinophil numbers and an absence ofmast cells.8,9 Not shown in the Table is the fact that transgenicmacrophages also had characteristic uniformly roundednuclei in contrast to pleiomorphic morphology of normal per-itoneal macrophage nuclei (Figure 1).

In sharp contrast, the peritoneal cells from GM-CSF trans-genic bc −/− mice exhibited normal total cell numbers, plei-omorphic macrophage nuclear morphology, normal mast cellnumbers, no tendency for excessive binucleate or multi-nucleate macrophage formation, and no eosinophil accumu-lation (Table 3). These populations could not be distinguishedby any parameter from populations in non-GM-CSF trans-genic mice.

Changes in other organs

GM-CSF transgenic mice are characterized by a massivedestruction of the tissues of the eye. There is destruction ofthe photoreceptor layer of the retina, commonly destructionand infiltration of the lens and similar damage in the corneawith the presence in both chambers of infiltrating macro-phages and usually an overall size reduction and collapseinwards of the eyeball (Figure 2). Transgenic GM-CSF mice ofboth bc +/+ and +/− genotype showed an equivalent degreeof ocular damage. In sharp contrast, the eyes of GM-CSF trans-genic bc −/− mice had a normal morphology and showednone of the features characteristic of the GM-CSF transgenicstate (Figure 2).

In mice aged only 2 to 3 months, other tissue lesions inGM-CSF transgenic mice are less consistent but some micecan exhibit focal macrophage inflammatory disease in skeletalmuscles, nodule formation in the peritoneal or pleural cavityand granuloma formation in the walls of pelvic organs, parti-cularly the bladder.8,9 Such lesions were present in some ofthe GM-CSF transgenic bc +/+ and +/− mice surveyed but no


356 Table 3 Peritoneal cell populations

bc GM-CSF Number Total cells Total cells Frequency per 103

status status of mice × 10−6 × 10−6 macrophages

Neutrophils Lymphocytes Macrophages Eosinophils Mast cells Binucleate Multinucleate

−/− + 14 5.9 ± 1.3 0 ± 0 1.6 ± 0.8 4.0 ± 0.9 0 ± 0 0.2 ± 0.2 8.6 ± 7.6 1.1 ± 3.0− 5 6.4 ± 2.4 0 ± 0 2.0 ± 1.8 4.2 ± 0.9 0 ± 0 0.2 ± 0.2 6.3 ± 5.8 1.2 ± 2.6

+/− + 31 299.2 ± 158.2 2.1 ± 4.5 10.6 ± 9.1 278.1 ± 148.5 8.4 ± 9.3 0 ± 0 73.1 ± 47.1 17.2 ± 21.4− 10 8.1 ± 2.7 0.2 ± 0.5 2.1 ± 1.6 5.7 ± 1.7 0 ± 0 0.2 ± 0.1 12.1 ± 7.6 0 ± 0

+/+ + 12 300.0 ± 152.2 3.0 ± 5.0 7.0 ± 8.0 280.5 ± 139.0 9.5 ± 11.4 0 ± 0 115.7 ± 51.9 30.0 ± 31.3− 5 6.9 ± 3.7 0 ± 0 1.0 ± 0.7 5.5 ± 2.8 0.2 ± 0.2 0.2 ± 0.2 12.3 ± 7.8 0 ± 0

Mean values ± standard deviations.

lesions of this nature were seen in any of the GM-CSF trans-genic bc −/− mice.

bc −/− mice, like GM-CSF −/− mice, develop a character-istic lung pathology of accumulation of surfactant and protein-aceous material in the alveoli – the disease state, alveolar pro-teinosis.11,16,17 This is based, in part at least, on defectivefunction of the alveolar macrophages. Associated with this areprominent focal accumulations of T- and B-lymphocytes inthe peribronchial regions. GM-CSF transgenic bc −/− miceshowed lung pathology that was identical in nature and sever-ity to that seen in non-transgenic bc −/− mice (Figure 3), thehigh circulating GM-CSF levels having no impact on thisdisease.

Production of GM-CSF by peritoneal macrophages

Previous mRNA studies indicated that macrophages werelikely to be major cell populations transcribing GM-CSF inGM-CSF transgenic mice,9 and serum GM-CSF levels werefound to parallel total numbers of peritoneal and pleuralmacrophages.18 To document that GM-CSF was being pro-duced by the same peritoneal cells showing the strikingphenotypic changes in GM-CSF transgenic mice, 1 ml suspen-sion cultures of 1 × 106 peritoneal cells were prepared. Initialexperiments using peritoneal cells had shown that concen-trations of GM-CSF in the medium of such cultures were oftenlower after 24 and 48 h of incubation than at earlier time-points of 3 and 6 h of incubation. Furthermore, most tissuesexhibit extensive induction of transcription of CSFs afterincubation in vitro for more than 5 h,12 making estimates ofGM-CSF in conditioned medium progressively less likely withtime to accurately reflect levels of production that might beoccurring in vivo. Because of these problems, incubation per-iods of only 3 and 6 h were used. In a parallel set of cultures,non-adherent cells were removed after 3 h of incubation, thenthe residual purified adherent macrophages were incubatedfor a further 3-h period to establish whether macrophageswere producing GM-CSF.

The results obtained using cells of the six genotypes areshown in Table 4. GM-CSF production was detected in thecultures but the levels observed were highly variable. Despitethis, GM-CSF transgenic bc +/+ and b +/− cells clearly pro-duced more GM-CSF than corresponding non-transgeniccells. Production of GM-CSF by GM-CSF transgenic bc −/−cells was detectable but the levels were lower than with bc+/+ or +/− cells and, arguably, no higher than with non-trans-genic cells.

Production of GM-CSF by adherent macrophages was

detectable but the levels were considerably lower than thoseobserved using unfractionated populations, in part due to thelower numbers of cells being incubated following the removalof non-adherent cells.

The data were consistent with the initial description ofmajor GM-CSF transcription in the peritoneal cells of GM-CSFtransgenic mice.9 However, the low production levels by GM-CSF transgenic bc −/− cells indicated that the model used hadnot achieved a notably high level of autocrine GM-CSF pro-duction by the cell type most revealing of GM-CSF-inducedchanges, the peritoneal macrophages.

mRNA levels in various organs

Analysis by PCR of levels of expression of mRNA for GM-CSFin non-transgenic bc +/+ mice indicated that very low levelswere detectable in a variety of organs but significant levels inthe lung, thymus and eye (Figure 4). GM-CSF transgenic bc+/+ organs shared a similar pattern of expression but at elev-ated levels. In transgenic organs, expression of GM-CSFmRNA was grossly elevated in peritoneal cells, lung and eye.This confirmed and extended an earlier, less complete, analy-sis9 and suggested that the peritoneal (and pleural) transgenicmacrophages were a major source of the GM-CSF in theseanimals, the high content in lung possibly being ascribable tothe frequent presence of aggregates of pleural macrophageson the lung surface, and the high levels in the eye to the highmacrophage content of the transgenic eye. There was someelevation of mRNA in other tissues, which might again havebeen ascribable to transgenic macrophages in these tissues.

The results from a parallel analysis of tissues from GM-CSFtransgenic bc −/− mice were somewhat surprising. GM-CSFmRNA was detectable in peritoneal cells but at a far lowerlevel than in transgenic bc +/+ mice, in agreement with thebioassay results on peritoneal cell-conditioned media. LungmRNA was possibly elevated compared with non-transgenicmice but the most surprising result was the very high level ofGM-CSF mRNA in the eye of transgenic bc −/− mice, sincethis organ was histologically normal. This casts doubt on theoriginal explanation for the high GM-CSF mRNA in transgeniceyes and suggests the occurrence of a selective tissue-directedexpression of the transgene in this particular organ.

The data supported the conclusion that GM-CSF transgenicbc −/− peritoneal macrophages were not producing parti-cularly high levels of GM-CSF, making the model an imperfecttest for the action of an internally produced regulator on aresponsive cell type. The other conclusion emerging was thatoverall transcription of GM-CSF in transgenic bc −/− mice was


357

Figure 1 The morphology of peritoneal cell populations from (A)a normal mouse, (B) a GM-CSF transgenic bc +/+ mouse, and (C)a GM-CSF transgenic bc −/− mouse. Note the normal pleiomorphicappearance of the macrophages and the presence of lymphocytes andmast cells in the population from the GM-CSF transgenic bc −/−mouse.

much lower than in transgenic bc +/+ mice. This indicatedthat some other mechanism must be responsible for the factthat GM-CSF levels were five times higher in transgenic bc−/− than in transgenic bc +/+ mice, the most likely being adifference in GM-CSF clearance and/or excretion.

Clearance of plasma GM-CSF in bc −/− mice

If, as is commonly held, receptor-mediated endocytosis is amajor fate of circulating regulators,1,19–21 circulating levels ofthe regulator should be more sustained in animals lackinghigh-affinity receptors for the ligand. Previous studies using

Figure 2 The morphology of the eye from (A) a normal mouse, (B)a GM-CSF transgenic bc +/+ mouse, and (C) a GM-CSF transgenic bc−/− mouse. Note the absence of macrophage infiltrate and no destruc-tion of the retina or lens in the GM-CSF transgenic bc −/− mouse.

intravenously injected 125I-labeled non-glycosylated GM-CSFfailed to reveal a difference in the plasma half-life of theinjected GM-CSF between bc −/− and bc +/+ mice.11 How-ever, when these experiments were repeated using bioassaysto detect bioactive GM-CSF, the half-life of both non-glycosylated and glycosylated GM-CSF was significantly pro-longed in non-transgenic bc −/− compared with bc +/+ mice(Figure 5). This approach could not be used with the GM-CSFtransgenic mice because of the high pre-existing levels of GM-CSF but the slower clearance noted in non-transgenic bc −/−mice could have contributed to the higher serum GM-CSF lev-els in the GM-CSF transgenic bc −/− mice than observed inother genotypes.


358

Figure 3 The morphology of the lung from (A) a normal mouse,(B) a non-transgenic bc −/− mouse, and (C) a GM-CSF transgenic bc−/− mouse. Note that the GM-CSF transgenic state did not alter thedevelopment of alveolar proteinosis or peribronchial lymphoidaggregates.

Urine clearance of GM-CSF

When GM-CSF was injected intravenously into non-transgenic bc −/− mice, the amount of GM-CSF appearing inthe urine in the following 5 h was consistently higher thanobserved using bc +/+ mice (Table 5). This suggested that GM-CSF receptors on kidney cells might be responsible for remov-ing transfiltered GM-CSF from the urine, either for subsequentdegradation or return to the plasma. To determine whetherthis removal system could be overloaded, the experimentswere repeated using five times the amount of injected GM-CSF. However, as shown in Table 5, this resulted in only atwo- to four-fold rise in urine GM-CSF and no change in thedifference observed between bc −/− and bc +/+ mice. It

should be noted from Table 5 that although GM-CSF wasundoubtedly cleared in the urine, the amount cleared wasonly a minute fraction (at most, 0.13%) of the amount of GM-CSF injected and could not have contributed significantly tothe observed fall in plasma GM-CSF levels in bc −/− miceduring this period.

In line with the deduction that the kidney has functionallyactive GM-CSF receptors, autoradiographs of kidneys 4 h afterthe injection of 125I-labeled GM-CSF indicated selective labe-ling of proximal tubule cells (Figure 6) but no labeling ofglomerular cells or distal tubule cells. This pattern is distinctfrom that observed previously with IL-315 and G-CSF. How-ever the implications of this labeling are obscure because par-allel studies using bc −/− mice observed selective tubularlocation of labeling of similar intensity after the injectionof 125I-GM-CSF (Figure 6). In both cases, the GM-CSF usedwas non-glycosylated, eliminating the operation of somecarbohydrate-clearing cellular receptor mechanism.

Discussion

The present experiments were undertaken in part as a criticaltest of current dogma on the respective roles played by theglycoprotein regulators and their membrane receptors in initi-ating cellular responses.1 It is believed at present that regu-lators serve only to cross-link membrane receptor chains andallow their subsequent functional activation to initiate sig-naling. At no stage is the regulator itself held to interact withsignaling molecules. Furthermore, the low-affinity regulator-areceptor chain complexes are not only transient but do notlead to activation or internalization.22

While these views are supported by a large body of experi-mental evidence, they do not address one situation of knownimportance in leukemogenesis – the autocrine production ofa growth factor within a cell responsive to that growth factor.2–4

In this situation, the assumption is made that the growth factorremains cell-associated and complexes with receptors toinitiate conventional signaling responses. The acutal situationis somewhat obscure because, when antibodies to the growthfactor are added to the medium, these do not necessarilyblock the subsequent autonomous proliferation in vitro ofsome autocrine growth factor-producing cells.23

Given that growth factors have a highly specific configur-ation, are of a size approximating receptors and that regulatorsand receptor chains enter the cells in equal numbers, it seemsslightly improbable that the growth factor has no role whatso-ever other than to serve as a cross-linker of receptor chains,particularly if, in autocrine situations, the growth factor isalready intracellularly located.

The model used involved the generation of mice with over-expression of the transgenic GM-CSF in cells usually able toexhibit changes induced by the GM-CSF but now onlyexpressing low-affinity receptor a-chains and no b chains thatare needed to complete the high-affinity receptor complex toinitiate cellular signaling. Any cellular changes noted wouldthen indicate either an action of low-affinity GM-CSF–a-chainreceptor complexes or a direct intracellular action of GM-CSFitself. Essentially all information on GM-CSF receptors hasbeen derived from studies on hematopoietic cells and, in prin-ciple, novel specific GM-CSF receptors could exist on othercell types and allow responsiveness of these tissues in bc −/−mice. However, the cDNA encoding the a-GM-CSF receptorwas originally cloned from a placental cDNA library24 andcould possibly have come from a non-hematopoietic cell.


359Table 4 Peritoneal cell production of GM-CSF

bc status GM-CSF transgene Number of media GM-CSF ng/mlstatus tested

3 h 6 h 3 h adherent cells

bc −/− + 10 0.5 (0.3–1.0) 0.9 (0.3–2.0) 0.1 (0–0.5)− 4 0.2 (0–0.5) 0.3 (0.1–2.0) 0.03 (0–0.3)

bc +/− + 42 1.7 (0.5–16.0) 3.7 (0.3–64.0) 0.4 (0.1–4.0)− 9 0.2 (0–2.0) 0.2 (0–1.0) 0.2 (0–2.0)

bc +/+ + 22 1.2 (0.1–8.0) 2.5 (0.1–32.0) 0.5 (0.2–0)− 5 0.1 (0–1.0) 1.1 (0.3–4.0) 0.3 (0.1–1.0)

Cultures were initiated containing 1 × 106 cells per ml of medium. Data from GM-CSF assays on media are shown as the geometric meansof the observed values and the figures in brackets are the range of values observed.

Figure 4 Levels of tissue GM-CSF mRNA detected by RT-PCR in organs from a GM-CSF transgenic bc +/+, a GM-CSF transgenic bc −/− anda normal mouse. Note the increased levels of GM-CSF mRNA in the GM-CSF transgenic bc +/+ mouse and the lesser elevations in the GM-CSF transgenic bc −/− mouse with the notable exception of the eye.


360

Figure 5 The slower fall in serum levels of GM-CSF detected bybioassays in bc −/− compared with bc +/+ mice following the intra-venous injection of glycosylated GM-CSF (upper panel) or non-glyco-sylated GM-CSF (lower panel). Data from three mice per timepoint,showing mean values ± standard deviations.

Table 5 Clearance of i.v. injected GM-CSF to urine in a 5-h period

Amount bc Number GM-CSF Total GM- % of totalinjected status of mice ng/ml CSF ng injected

400 ng +/+ 12 0.7 (0–3.6) 0.2 (0–1.4) 0.05−/− 12 2.0 (0.3–7.7) 0.5 (0.1–2.4) 0.13

2 mg +/+ 4 1.1 (0.3–6.7) 0.5 (0.1–2.8) 0.03−/− 4 5.9 (5.4–6.7) 2.0 (0.7–2.8) 0.10

Saline +/+ 12 0 (0–0) 0 (0–0) —−/− 12 0.4 (0–2.1) 0.1 (0–0.7) —

Data shown are the geometric means of the observed values andthe figures in brackets are the range of values observed.

Figure 6 Autoradiography of the kidney from a bc +/+ mouse anda bc −/− mouse injected intravenously 4 h previously with 125I-labelednon-glycosylated GM-CSF. Note the selective labeling of proximalrenal tubules of similar intensity in both types of mouse.

Previous studies on the hematopoietic cells of bc −/− miceindicated that these cells are incapable of proliferative orsurvival responses to GM-CSF.11 The mice develop acharacteristic lung disease, alveolar proteinosis, whose basisis ascribable at least in part to failure of function of bc −/−alveolar macrophages.11,16,17

The model used did achieve a mouse with spectacularlyelevated serum GM-CSF levels although it did not achieve aperitoneal macrophage population producing particularlyhigh levels of GM-CSF. Because these cells were the majorpopulation for testing the actions of GM-CSF in the absenceof high-affinity receptors, the test system fell short of being anoptimal one that could unequivocally eliminate the possibilityof non-receptor-mediated actions. The failure of these bc −/−macrophages to produce high levels of GM-CSF strongly sug-


361gests that, in the usual GM-CSF transgenic mouse, an auto-catalytic element is operating in which macrophages produceGM-CSF or IL-1 that, in turn, stimulates the producing cellsto enlarge and become functionally activated to produce yetmore GM-CSF.

The major pathological consequences of the GM-CSF trans-genic state are destruction of tissues of the eye and accumu-lation of massive numbers of macrophages in the peritonealand pleural cavities, associated with changes in the nuclearmorphology of these macrophages, cytoplasmic enlargementand basophilia and a tendency with time to form multi-nucleated macrophages by cell fusion.8,9 Associated changesin peritoneal populations are a consistent rise in eosinophilnumbers, a depletion of mast cells and a variable rise in neu-trophils in the peritoneal cavity. More variable are the devel-opment of inflammatory macrophage-containing nodules inthe peritoneal and pleural cavity, in skeletal muscle and blad-der and a generalized wasting disease with hind limb weak-ness or paralysis.9,10

Although the transgenic bc −/− mice did not achieve opti-mal levels of autocrine GM-CSF production by macrophages,the mice unequivocally demonstrated none of this multiplicityof tissue changes. Eye structure was normal, peritoneal popu-lations were normal in total numbers, and the more subtleparameters of cellular composition and morphology and notissue lesions were observed in any organ. The completeabsence of any tissue changes is proof that the various lesionsof GM-CSF transgenic mice are indeed initiated by GM-CSFand not due to some insertional artifact dependent on theinserted transgene accidentally disrupting the function ofadjacent genes.10 The data also indicate that a-chain bindingof GM-CSF is without observable impact on subtle parametersnot previously considered such as nuclear morphology, cellfusion, cytoplasmic basophilia or size.

It has been proposed that the tissue lesions in GM-CSFtransgenic mice are macrophage-mediated with the ultimatedeath of the mice from wasting disease and paralysis, theconsequence of macrophage production of agents such asTNFa, IL-1 and g-interferon.9,18,25 The failure of bc −/− miceto develop activated macrophage populations and to developtissue lesions supports this general hypothesis but also elimin-ates a less likely possibility that other types of GM-CSF recep-tor might be present in these tissues and be able to mediatedirect toxic effects of GM-CSF on these tissues.

The lung disease of bc −/− (or GM-CSF −/−) mice is anopposite situation where the disease state has been ascribedto macrophage hypofunction rather than hyperfunction.11,16,17

The lesions in GM-CSF −/− mice can in fact be corrected bylocal lung epithelial expression of transgenic GM-CSF.26 Thepresence of excess levels of GM-CSF in GM-CSF transgenicbc −/− mice had no impact on the occurrence or severity ofthe lung pathology.

An intriguing aspect of the present transgenic bc −/− micewas the extremely high serum and urine levels of GM-CSFachieved in the absence of any comparable elevation of GM-CSF production. This argues strongly that the production ratesof GM-CSF are not the dominant parameter in determiningcirculating levels and that utilization or clearance can be fac-tors of equal or greater importance. What remains unclear isthe nature of these utilization/clearance systems. The half-lifeof plasma GM-CSF was clearly extended in bc −/− mice,documenting the role of receptor-mediated degradation.However GM-CSF levels did fall in bc −/− mice, indicatingthe extence of other types of clearance mechanisms. Becausethe same was true when non-glycosylated GM-CSF was used,

carbohydrate clearance systems cannot provide a completeexplanation.

Urine clearance of GM-CSF was documented, as was a rolefor GM-CSF receptors in decreasing this process, but againuncertainty exists because bc −/− tubule cells bound 125I-labeled GM-CSF in a similar manner to bc +/+ cells, suggest-ing the existence again of some alternative membrane recog-nition system for removing GM-CSF. Although urine GM-CSFlevels were consistently elevated in GM-CSF transgenic bc −/−mice, the data on clearance of injected GM-CSF indicated thatelimination of GM-CSF in the urine is only a minor fate ofcirculating GM-CSF and can therefore only be a minor factorin determining the fall with time in plasma GM-CSF levels.

The analysis of the GM-CSF transgenic bc −/− mice has pro-vided no evidence to contest the current view that the actionof GM-CSF or comparable glycoprotein regulators, is entirelydependent on activation of specific high-affinity membranereceptors. Interaction with specific low-affinity receptorsappeared to be unable to induce or correct any hematopoieticor tissue changes. The failure to achieve peritoneal macro-phages in these mice that produced very high levels of GM-CSF does leave the question of the consequences of autocrineproduction less satisfactorily resolved, although the levels ofproduction achieved were comparable with those often seenin autocrine myeloid leukemic cells.3

Acknowledgements

This work was supported by the Carden Fellowship Fund ofthe Anti-Cancer Council of Victoria, the National Health andMedical Research Council, Canberra, and the National Insti-tutes of Health, Bethesda, Grant No. CA-22556. The authorsare indebted to the dedicated technical assistance of Ms BettePapaevangelou of the Bone Marrow Research Laboratories ofthe Royal Melbourne Hospital for performing the tail analysesidentifying the GM-CSF transgenic and bc status of the miceand to Mrs Bronwyn Roberts for performing the chromato-graphic fractionation of the urine samples.

References

1 Gammeltoft S, Kahn CR. Hormone signaling via membrane recep-tors. In: De Groot LJ (ed). Endocrinology, 3rd edn. Saunders: Phila-delphia, 1995, pp 17–65.

2 Delwell R, Van Buitenen C, Salem M, Bot F, Gillis S, KaushanskyK, Altrock B, Lowenberg R. Interleukin-1 stimulates proliferationof acute myeloblastic leukemia cells by induction of granulocyte–macrophage colony-stimulating factor release. Blood 1989; 74:586–593.

3 Duhrsen U, Stahl J, Gough NM. In vivo transformation of factor-dependent hemopoietic cells: role of intracisternal A-particletransposition for growth factor gene activation. EMBO J 1990; 9:1087–1096.

4 Metcalf D, Rasko JEJ. Leukemic transformation of immortalizedFDC-P1 cells engrafted in GM-CSF transgenic mice. Leukemia1993; 7: 878–886.

5 Metcalf D, Nicola NA. The Hemopoietic Colony-stimulating Fac-tors: from Biology to Clinical Applications. Cambridge UniversityPress: Cambridge, 1995.

6 Metcalf D, Burgess AW, Johnson GR, Nicola NA, Nice EC, DeLa-marter J, Thatcher DR, Mermod J-J. In vitro actions on hemopoieticcells of recombinant murine GM-CSF purifed after production inEschericha coli: comparison with purified native GM-CSF. J CellPhysiol 1986; 128: 421–431.

7 Golde DW, Ding DX, Rivas CI, Vera JC. The colony-stimulatingfactors and molecular transport. Stem Cells 1994; 12 (Suppl.):61–66.


362 8 Metcalf D, Elliott MJ, Nicola NA. The excess numbers of peritonealmacrophages in granulocyte–macrophage colony-stimulating fac-tor transgenic mice are generated by local proliferation. J Exp Med1992; 175: 877–884.

9 Lang RA, Metcalf D, Cuthbertson RA, Lyons I, Stanley E, Kelso A,Kannourakis G, Williamson DJ, Klintworth GK, Gonda TJ, DunnAR. Transgenic mice expressing a hemopoietic growth factor gene(GM-CSF) develop accumulations of macrophages, blindness anda fatal syndrome of tissue damage. Cell 1987; 51: 675–686.

10 Metcalf D, Moore JG. Divergent disease patterns in GM-CSF trans-genic mice associated with differing transgene insertion sites. ProcNatl Acad Sci USA 1988; 85: 7767–7771.

11 Robb L, Drinkwater CC, Metcalf D, Li R, Kontgen F, Nicola N,Begley CG. Hematopoietic and lung abnormalities in mice witha null mutation of the common b subunit of the receptors for gra-nulocyte–macrophage colony-stimulating factor and interleukins3 and 5. Proc Natl Acad Sci USA 1995; 92: 9565–9569.

12 Metcalf D, Willson TA, Hilton DJ, Di Rago L, Mifsud S. Productionof hematopoietic regulatory factors in cultures of adult and fetalmouse organs: measurement by specific bioassays. Leukemia1995; 9: 1556–1564.

13 Alexander WS, Schrader JW, Adams JM. Expression of the c-myconcogene under control of an immunoglobulin enhancer in Em-myc transgenic mice. Mol Cell Biol 1987; 7: 1436–1444.

14 Nicola NA, Metcalf D. Binding of iodinated multipotential colony-stimulating factor (interleukin-3) to murine bone marrow cells. JCell Physiol 1986; 128: 180–188.

15 Metcalf D, Nicola NA. Tissue localization and fate in mice ofinjected multipotential colony-stimulating factor. Proc Natl AcadSci USA 1988; 85: 3160–3164.

16 Dranoff G, Crawford AD, Sadelain M, Ream B, Rashid A, BronsonRT, Dickersin GR, Bachurski CJ, Mark EL, Whitsett JA, MulliganRC. Involvement of granulocyte–macrophage colony-stimulatingfactor in pulmonary homeostasis. Science 1994; 264: 713–716.

17 Stanley E, Lieschke GJ, Grail D, Metcalf D, Hodgson G, Gall JAM,Maher DW, Cebon J, Sinickas V, Dunn AR. Granulocyte/macrophage colony-stimulating factor-deficient mice show no

major perturbation of hematopoiesis but develop a characteristicpulmonary pathology. Proc Natl Acad Sci USA 1994; 91: 5592–5596.

18 Gearing A, Metcalf D, Moore JC, Nicola NA. Elevated levels ofGM-CSF and IL-1 in the serum, peritoneal and pleural cavities ofGM-CSF transgenic mice. Immunol 1989; 67: 216–220.

19 Tushinski RJ, Oliver IT, Guilbert LJ, Tynan PW, Warner JN, StanleyER. Survival of mononuclear phagocytes depends on a lineage-specific growth factor that the differentiated cells selectivelydestroy. Cell 1982; 28: 71–81.

20 Layton JE, Hockman H, Sheridan WB, Morstyn G. Evidence for anovel in vivo control mechanism of granulopoiesis: mature cell-related control of a regulatory growth factor. Blood 1989; 741303–1307.

21 Kuter DJ, Rosenberg RD. The reciprocal relationship of throm-bopoietin (c-mpl ligand) to changes in the platelet mass duringbusulfan-induced thrombocytopenia in rabbits. Blood 1995; 85:2720–2730.

22 Nicola NA, Smith A, Robb L, Metcalf D, Begley CG. The structuralbasis of the biological actions of the GM-CSF receptor. In: BockGR, Goode JA (eds). The Molecular Basis of Cellular DefenceMechanisms. Wiley: Chichester (Ciba Foundation Symposium204) 1997, pp 19–32.

23 Lang RA, Metcalf D, Gough NM, Dunn AR, Gonda TJ. Expressionof a hemopoietic growth factor cDNA in a factor-dependent cellline results in autonomous growth and tumorigenicity. Cell 1985;43: 531–542.

24 Gearing DP, King JA, Gough NM, Nicola NA. Expression cloningof a receptor for human granulocyte–macrophage colony-stimulat-ing factor. EMBO J 1989; 8: 3667–3676.

25 Lang RA, Cuthbertson RA, Dunn AR. TNFa, IL-1a and bFGF areimplicated in the complex disease of GM-CSF transgenic mice.Growth Factors 1992; 6: 131–138.

26 Huffman JA, Hull WM, Dranoff G, Mulligan RC, Whitsett JA. Pul-monary epithelial cell expression of GM-CSF corrects the alveolarproteinosis in GM-CSF-deficient mice. J Cell Invest 1996; 97:649–655.

Documents

The biological consequences of excess GM-CSF levels in transgenic mice also lacking high-affinity receptors for GM-CSF