Progenitor cell mobilization by granulocyte colony-stimulating factor controlled by loci on chromosomes 2 and 11

B6D2F1 (F1) mice were bred by mating female C57BL/6J (B6) and male DBA/2J/WEHI (D2) mice (Walter Eliza Hall Institute of Medical Research, Victoria, Australia). Two independent backcross experiments were performed comprising 260 mice (165 male and 95 female) bred from D2 female and F1 male mice and 245 mice (122 male and 123 female) bred from F1 and D2 mice. We used 26 male mice from the BXD/Ty recombinant inbred (RI) strains (The Jackson Laboratory, Bar Harbor, ME). All mice were maintained and bred under specific pathogen-free conditions and used in all experiments at 7-11 weeks of age.

We diluted rhG-CSF (filgrastim; Amgen, Thousand Oaks, CA) in sterile normal saline containing 5% bovine newborn calf serum (Hyclone, Logan, UT). Twice daily for 5 days, mice were injected intraperitoneally with 40 μg/kg rhG-CSF. Blood was harvested on the sixth day and cultured in semisolid 1-mL agar cultures as previously described.4 After 7 days each culture was fixed and stained as described.4 Triplicate plates were counted, and the average value was represented as colonies per μL of peripheral blood colony-forming cells (PB-CFCs).

Mice were genotyped at an average 10-centimorgan (cM) resolution with fluorescent markers as described (Burt et al, unpublished data, January 1999). The genotypes of 26 BXD RI strains of mice were downloaded (Mouse Genome Informatics; The Jackson Laboratory,), supplemented with our own data, and analyzed (MapManager QTb19ppc; http://mcbio.med.buffalo.edu/mapmgr.html).

G-CSF mobilized the progenitor cells into the peripheral blood more effectively in D2 mice than in B6 mice. After 5 days of injections, the average PB-CFC counts were 5-6 colonies/μL in male D2 mice and 0-1 colonies/μL in male B6 mice.4 Figure1A shows the PB-CFC counts in the backcross mice. Female PB-CFC counts were almost half the value of the male counts; no differences were seen between the reciprocal female crosses; and males bred with F1 and D2 mice showed slightly higher PB-CFC counts than females bred with D2 and F1 mice.

The top 20% of responders or the high responders from each group were analyzed for the frequencies of homozygous D2 alleles. Linkage was determined when a deviation in segregation from the normal mendelian ratios in the high responders was observed (Figure 1B). One statistically significant linkage was found on chromosome 2, which was seen in each of the backcrosses (mice bred with F1 and D2 and mice bred with D2 and F1) (Table1). Two suggestive linkages were seen on chromosomes 4 and 14 (Figure 1B).

We calculated associations between the G-CSF–induced PB-CFC values of the 26 BXD RI strains (Figure 1C) and their genotypes at each locus across the genome (MapManager QTb19ppc, ). RI strains the inbred products of an F1 intercross and hence are homozygous for either parental chromosome in approximately equal doses. Significant linkage was seen on chromosome 11; the highest likelihood ratio statistic of 15.1 was observed at D11Ncvs42, and 16.0 was observed at the peak position (3 cM) after the interval mapping (Figure 1D). Since this linkage was not seen in the backcross experiments, the locus is likely to be D2 dominant.

Genetics is a powerful tool in studying phenotypes that only manifest at a whole animal level. Loci which contain genes that control the phenotype can be identified, as demonstrated in this study, with 2 loci being implicated in the control of progenitor release from the marrow. In addition, relationships to other phenotypes can be identified. A strong inverse correlation has been reported between the lifespan of different strains and the number of autonomously cycling progenitors (day-7 Cobblestone Area Forming Units [CAFC]) in the femur.9 A putative locus causing the variability of lifespan (small variance in D2) was mapped near the centromere on chromosome 11,10 a region which is coincident with our result using the same BXD RI mapping strains for mapping PB-CFC mobilization. It is therefore formally possible that these 2 phenotypes are controlled by the same gene.

Possible candidate genes for PB-CFC mobilization include histidine decarboxylase (chromosome 2), a gene known to be induced by G-CSF in bone marrow11 and found to have possibly relevant sequence polymorphisms between D2 and B6 mice (M. Hasegawa, unpublished data, September 1998; Genbank accession number AF109 137).11-13 The integrin subunit α4 (Genbank accession number AF109 136), another possible candidate on chromosome 4, and Transcobalamin II (Genbank accession number AF090 686),14 15 a chromosome 11 candidate (M. Hasegawa, unpublished data, September 1998), were found to have identical coding regions between the D2 and B6 mice.

In summary, 2 loci on chromosomes 2 and 11 control the progenitor mobilization response to G-CSF, with possible contributions from loci on chromosomes 4 and 14. The complexity of this multigene trait will be biologically dissected by the breeding of mice congenic for each locus.

We thank Sandra Mifsud and the supporting staff at Animal Services, the DNA Sequencing Lab, and Australian Genome Research Facility, Melbourne, Australia.