Primitive hematopoietic stem cell function in vivo is uniquely high in the CXB-12 mouse strain

Hematopoietic stem cells (HSCs) function long term and proliferate to self-renew and differentiate to produce progenitors of all hematopoietic lineages.1-5 Cells obtained from bone marrow, cord blood, or mobilized peripheral blood of healthy donors are clinically useful as sources of transplantable HSCs. After transplantation, donor HSCs compete for engraftment with residual host HSCs.6 High levels of function from donor HSCs are extremely valuable for competing effectively with host HSCs in many clinical conditions.7-10 Unfortunately, donor HSCs are often damaged by clinical manipulations so that they repopulate less well than residual host HSCs.11-13 Analysis of the genetically superior HSCs reported here may suggest how functional abilities of donor HSCs can be improved.14-17 

A broad range of genetic backgrounds and defined genetic combinations are available in mouse models,18,19 including recombinant inbred (RI) lines.20-22 The set of CXB RI lines used in the current study were derived from inbred BALB/cBy (BALB) and C57BL/6By (B6) strains by intercrossing individuals from the F2 generation and then inbreeding for at least 20 generations of brother–sister mating. All mice of a particular RI line are genetically identical and homozygous at all loci, with approximately half the loci derived from B6 and half from BALB.20 

The competitive repopulation assay directly measures the ability of a population of donor HSCs to engraft recipients relative to the repopulating ability of a population of standard competitor HSCs.23-25 When genetically distinguishable donor and standard competitor cells are mixed in a specific ratio and injected into lethally irradiated recipients, the fraction of donor-type blood cells is proportional to the fraction of donor cells injected if neither type of HSC has a repopulating advantage. If a mixture of such donor–competitor cells were engrafted in a 2:1 ratio, two thirds of the blood cells would be donor type. If the observed proportion of donor-type blood cells is significantly different from the proportion of donor cells transplanted, a repopulating advantage or disadvantage in the donor HSCs is detected. The percentage of donor type blood cells in the recipient circulation is used to calculate total repopulating units (RUs) in a donor cell population relative to the standard competitor,25 where 1 donor RU has the same repopulating ability as 10⁵ standard competitor bone marrow cells (BMCs).

We compared the relative long-term repopulating abilities of 12 CXB RI lines in vivo in the current study, using the CByB6F1 hybrid of the BALB and B6 parent strains as recipients and standard competitors because they carry all B6 and BALB alleles. Therefore, they provide the best system available in which long-term functional ability of HSCs from multiple CXB RI lines can be studied in a constant environment, though F1 hybrid resistance in the recipients must be taken into consideration. The advantage of this technique is the direct measurement of long-term function in vivo relative to a CByB6F1 standard. BMCs from CXB-12 donors had the highest long-term repopulating advantage ever reported. The new competitive dilution assay shows that CXB-12 BMCs have both an increased concentration of HSCs and an increased functional ability per HSC.

Normal inbred B6 (Gpi1^b/Gpi1^b), BALB (Gpi1^a/Gpi1^a), congenic B6, and BALB strains differing at Gpi1, hybrid CByB6F1 (Gpi1^a/Gpi1^b), and recombinant inbred CXB lines 1-12 were raised at The Jackson Laboratory in a specific pathogen-free facility. CByB6F1 competitors of Gpi1^b/Gpi1^b genotype were produced from B6 × congenic BALB (Gpi1^b/Gpi1^b) matings, whereas CByB6F1 competitors of Gpi1^a/Gpi1^a genotype were produced from BALB×congenic B6 (Gpi1^a/Gpi1^a) matings. All mice in the current study were used between 2 and 4 months of age, except for old CXB-12 donors that were 24 months old. Mice were fed NIH-31 (4% fat) diet ad libitum with standard animal husbandry management as published.18 

BMCs were flushed from 2 femurs and 2 tibias of each donor or competitor mouse through a 23-gauge needle with 2.0 mL Iscove modified Dulbecco medium and filtered through a 100-μm mesh nylon cloth to remove debris. Nucleated cells were counted using a model ZBI Coulter Counter (Coulter Electronics, Hialeah, FL). Donor BMCs were mixed with aliquots of competitor BMCs in specific ratios (see figure legends) and injected intravenously into recipients. Except where noted differently, CByB6F1 recipients were sublethally irradiated with 5 Gy (500 rads) 14 days before transplantation and intraperitoneally injected with 100 μg/mouse of a natural killer (NK) cell antibody (anti-NK1.1, clone PK 136) 3 days before transplantation to deplete NK cells (NK cell–depleted recipients). This alleviates hybrid resistance against B6, BALB, and CXB RI donor cells.26,27 In all cases, recipients were given 11 Gy (1100 rads) lethal irradiation 4 to 6 hours before marrow injection. Irradiation was from a cesium Cs 137 gamma source (Shepard Mark 1; Shepard & Associates, Glendale, CA) at 1.6 Gy (160 rads)/minute.

CXB RI lines 2, 4, 5, 6, 10, and 12 are Gpi1^a/Gpi1^a type, whereas CXB RI lines 1, 3, 7, 8, 9, and 11 are Gpi1^b/Gpi1^b type; each type of donor BMCs was mixed with BMCs from CByB6F1 competitors specially bred to have a different Gpi1 allele. Because recipients were normal CByB6F1 mice, their cells would have been detected easily by their heterozygous Gpi1^a/Gpi1^belectrophoretic bands; however, this was not found. From blood collected 1, 3, and 6 months after reconstitution by the mixtures of marrow, percentages of donor-derived erythrocytes and lymphocytes were measured after separating Gpi1 types by cellulose acetate gel electrophoresis, as reported earlier.25 

Relative repopulating abilities are most accurately compared when cells from each of the different donors are mixed with cells from the same standard competitor pool in the same experiment. Each RU is the relative repopulating ability of 10⁵ fresh marrow cells from the same standard competitor pool. So that independent experiments could be compared, fresh marrow from standard, young congenic CByB6F1 hybrid competitors was used with a variety of donors in each experiment. Numbers of RUs from each donor were calculated from the observed percentage donor cells, where the number of fresh competitor marrow cells used per 10⁵ equalled C. Then by definition, the percentage is 100 (RU/RU + C). By algebraic rearrangement, this gives RU = % (C)/(100 − %). Thus, if a donor population contains 20 RUs, it repopulates as well as 20 × 10⁵ standard competitor marrow cells, and mixtures of 20 donor RUs plus 20 × 10⁵ competitor cells produce an average of 50% donor-type erythrocytes and lymphocytes in the recipients. Total RUs per donor were estimated, assuming that the BMCs in both femurs and tibias were 25% of the total marrow cells in an adult mouse.

Blood samples were taken from young male and female B6, BALB, CByB6F1, and CXB-12 mice through the orbital sinus. Concentrations of white blood cells (WBCs) and red blood cells (RBCs) were counted using the ZBI Coulter Counter described earlier. Portions of peripheral blood samples were incubated in Gey solution to lyse RBCs and then were stained with B220, CD4, and CD8 antibodies and analyzed through fluorescence activated cell staining (FACS) using a FACScan II (Becton Dickinson, Mountain View, CA) flow cytometer to determine percentages of B220, CD4, and CD8 lymphocytes. BMCs from young B6, BALB, CByB6F1, and CXB-12 mice were also tested for lymphocyte composition through FACS analysis.

To measure colony-forming unit spleen (CFU-S), 10⁵ BMCs were injected into each of 4 strain-matched, lethally irradiated recipients. All recipients were killed at day 12. The spleen of each recipient was removed, fixed in Bouin fixative, and examined for the numbers of macroscopic colonies. Short-term production of CFU-S was tested as above, except in lethally irradiated CByB6F1 recipients, using 4 × 10⁵ BMCs from lethally irradiated CByB6F1 carriers given 10⁶ BMCs 7 days earlier or using 2 × 10⁵ BMCs from lethally irradiated CByB6F1 carriers given 10⁶ BMCs 14 days earlier.

Portions containing 5 × 10⁴ CXB-12 donor BMCs (Gpi1^a/Gpi1^a) and 5 × 10⁵ CByB6F1 standard competitor BMCs (Gpi1^b/Gpi1^b) were injected into each of 38 NK cell–depleted CByB6F1 recipients. The Poisson function, P_i = e^−N × (Nⁱ/i!), gives the probability (P_i) that a recipient gets a certain number (i) of HSCs, where N is the average number of HSCs injected into each recipient.27,28 The proportion of recipients with 0% CXB-12–type cells 6 months after reconstitution was used to compute the number of HSCs (N) in CXB-12 BMCs using the i = 0 term of the Poisson function: P₀ = e^−N or n = −lnP₀. This gave n = 1.2 for 5 × 10⁴ CXB-12 donor BMCs. Then the probabilities (P_i) of having 0, 1, 2, 3, or 4 donor HSCs in a recipient were, according to the Poisson function: 0.3012, 0.3614, 0.2169, 0.0867, or 0.0260, respectively. Because 5 × 10⁵competitor cells were used, n = 5.0 for the competitor, giving probabilities of having 0, 1, 2, 3, 4, 5, 6, 7, 8, or 9 competitor HSCs in a recipient as: 0.0067, 0.0337, 0.0842, 0.1404, 0.1755, 0.1755, 0.1462, 0.0653, 0.0363, or 0.0181, respectively. In practice, we calculated 2 arrays of probabilities, one for the donor, and one for the competitor, with i = 0 to 15 for 16 probabilities in each case. These form a 16 × 16 matrix of 256 combined probabilities, and each is associated with a donor:competitor HSC ratio that can be used to predict a value representing donor contribution. In the current study we used the highest 37 probability values because there were 37 recipients available 6 months after transplantation.

We used the 37 corresponding donor:competitor ratios in the Poisson modeling to estimate repopulating ability per cell for CXB-12 HSCs by comparing 3 hypothesized levels of repopulating abilities per cell for CXB-12 HSCs: F = 1, F = 1.4, and F = 2, each relative to a repopulating ability per CByB6F1 standard HSC of F = 1. From the value of F, a predicted percentage donor value was calculated for each of the 37 donor:competitor ratios associated with the 37 highest probabilities. This produced 3 sets of predicted donor percentage values, each ranked from low to high and compared to the 37 ranked observed data in a paired t test. The hypothesized F values that generated predictions significantly different from observed data were rejected. The F value whose predictions were not significantly different from the observed data were accepted to represent repopulating ability per cell for CXB-12 donor HSCs relative to those of the CByB6F1 standard.

The long-term colony-initiating cell (LTC-IC) assay was performed as described by Müller-Sieburg and Riblet.14 Briefly, bone marrow cells were seeded onto confluent layers of the stromal cell line S17 in 96-well plates. At least 48 wells were seeded per cell dilution, and the dilations covered the range of 1 × 10⁴ to 625 cells/well. Wells that contained colonies of small granulocytic cells were enumerated at the indicated time points, and LTC-IC frequencies were calculated from maximum likelihood statistics. The B6-Ly5 congenic mouse was used because, as described previously, it has the same frequency of LTC-IC as the parental C57BL/6 mouse.

Procedures for cobblestone area–forming cell assay (CAFC) were adopted as described by Ploemacher et al,29 with a feeder layer from an S17 stroma cell line in 96-well microplates.14,27,30 At confluence, the S17 feeder cells were irradiated with 50 Gy (5000 rads) and overlaid with BMCs from individual donors. Each well contained 0.2 mL cobblestone media (10% fetal calf serum, 10% horse serum, 80% α-MEM, 50 IU/mL penicillin, 50 μg/mL streptomycin, 50 μmol/L β-mercaptoethanol, 10 μmol/L hydrocortisone sodium succinate, and 2 mmol/L L-glutamine). On each plate, 20 wells were used for each of 4 cell doses (1644, 3780, 8695, or 20 000 donor BMCs), and 16 wells with no cells were controls.31 All plates were cultured at 33°C with 5% CO₂ and fed weekly by changing half the media in each well. At 3 and 5 weeks of incubation, wells were scored positive if they had one or more colonies of at least 6 cells within the stromal layer with a flat cobblestone shape under a phase-contrast microscope. CAFC frequencies (N) in the cell dose were calculated from the fractions of negative wells (P₀) where n = −lnP₀.

During Poisson modeling, differences between ranked predicted values and ranked observed data were analyzed through pairedt test. Strain differences were tested through variance analyses using the JMP statistical discovery software (SAS Institute, Cary, NC) on “Fit Y by X” and “Fit Model” platforms, respectively.32 

HSC function was directly compared in 12 different CXB RI lines (Figure 1) by mixing BMCs from 2 donors of each line with a portion from 1 of 2 pools of standard competitor CByB6F1 marrow. Recipients were NK cell–depleted and lethally irradiated, and each was given 4 × 10⁶ donor BMCs mixed with 2 × 10⁶ CByB6F1 competitor BMCs of the opposite Gpi1 type. After 1, 3, and 6 months, numbers of RUs were calculated as detailed in “Materials and methods.” The Gpi1 marker used to distinguish donor and competitor standard cells in competitive repopulation does not affect repopulating abilities25,28,33and thus could not cause the CXB-12 advantage.

Figure 1 shows an enormous (P < .0001) strain effect as CXB-12 donors repopulated far better than donors of the other 11 CXB RI lines. CXB-12 donors had 14 times more RUs at 6 months, representing the most primitive HSCs, than did CXB-4 donors, the next highest, and 36 times more than the average (764 RUs per donor) for the other 11 RI lines. The repopulating advantage of CXB-12 BMCs increased greatly between 1 month and 3 months. This increase with time did not occur with the other RI lines. The major histocompatibility locus (H2) does not explain why CXB-12 mice have such a high repopulating ability because the CXB-1, 4, 9, and 12 lines are all H2^d/H2^d, but CXB-1 and CXB-4 repopulate only slightly better than the other CXB RI lines (all H2^b/H2^b homozygotes), whereas CXB-9 does not repopulate any better than the others.

Repopulating abilities of CXB-12 BMCs were tested in a separate study, using 10 × 10⁵ or 30 × 10⁵ CXB-12 BMCs mixed with 100 × 10⁵ standard competitor BMCs from CByB6F1. This study is displayed with 2 independent studies in Table1. Although some CXB-12 repopulating advantage was present in the short-term HSCs responsible for repopulation after 1 month, by far the strongest advantage occurred in the more primitive HSCs, which repopulate after 3 and 6 months (Table1). In all cases, the repopulating advantage of CXB-12 BMCs increased greatly between 1 month and 3-6 months after transplantation.

Percentages of CXB-12 type erythrocytes and lymphocytes in recipients after 3 and 6 months were far higher than expected if CXB-12 donor and CByB6 F1 competitor cells repopulated equally (Table 1). However, the advantage of CXB-12 BMCs increased with total numbers of donor cells. It was only 2.4- to 2.8-fold using the lowest cell numbers, 5 × 10⁴ donor cells plus 5 × 10⁵ standard F1 cells. When cell numbers were increased, using 10⁶ donor cells plus 10⁷ standard F1 cells, the CXB-12 advantage increased to 6.4- to 6.6-fold. It further increased to a 12- to 16-fold advantage using 3 × 10⁶ donor plus 10⁷standard F1 cells (Table 1), probably because some hybrid resistance against CXB-12 cells remained in the recipients and was saturated when higher numbers of CXB-12 BMCs were used. The 11- to 31-fold advantage with 4 × 10⁶ donor plus 2 × 10⁶ standard F1 cells showed the same effect; the wide range resulted from inaccuracies when donor percentages were greater than 90%, so that a difference of a few percentage points greatly affected RU calculations.

CXB-12 mice have similar peripheral blood WBC and RBC concentrations in comparison to B6, BALB, and CByB6F1 mice (Table2). Although there were significant strain differences in peripheral blood B220 and CD4 lymphocyte percentages, total BMC numbers, and marrow percentages of B220 or CD8 cells, in all cases values for CXB-12 mice were similar to those of at least one of the controls (Table 2). Furthermore, day 12 CFU-S from fresh marrow or pre–CFU-S from carriers after 7 or 14 days was not increased in CXB-12 mice (Table 2). Despite high primitive HSC function, regulatory mechanisms in CXB-12 mice maintain blood, marrow, and short-term precursor cell concentrations at normal levels.

We used competitive dilution to test whether the CXB-12 advantage is caused by increased HSC numbers, functional ability per HSC, or both. Each of 38 CByB6F1 recipients was given 5 × 10⁴CXB-12 donor BMCs mixed with 5 × 10⁵ CByB6F1 competitor BMCs. Six months after reconstitution, 11 of 37 recipients (1 died after 2 months) had 0% CXB-12 type (Gpi1^a/Gpi1^a) blood cells, giving a Poisson (n = 1.2) for 5 × 10⁴ CXB-12 donor BMCs or 2.4 HSCs in 10⁵ CXB-12 BMCs (Table3).

In Figure 2, F values (repopulating function relative to competitor HSCs) of 1.0, 1.4, and 2.0 in donor HSCs were compared. Of the 3 sets of predictions, the F = 1.4 model best fit observed data, whereas models using F = 1 or F = 2 generated predictions significantly below or above the observed data. Thus, each CXB-12 HSC repopulated 1.4 times as well as each CByB6F1 HSC (Table 3). Because the dose of CXB-12 BMCs used in this study was very low, the hybrid resistance that remained in the CByB6F1 recipients probably reduced the CXB-12 HSC functional advantage. Nevertheless, the concentration of HSCs was 2.4 times that of the F1 hybrid standard, and the functional ability was 1.4 times the standard.

Hybrid resistance reduced but did not remove the CXB-12 advantage in recipients that were not treated to remove NK cells. Figure3 shows percentages, and RU values are given in the legend. Compared to the CByB6F1 standard competitor, CXB-12 BMCs had a 2-fold repopulating advantage after 1 month, despite the presence of recipient NK cells. After 3 and 6 months, the advantage averaged 3.9- and 5.4-fold. Although the increase with time was still obvious, the increased repopulation with donor cell numbers was only proportional to their 16-fold increase in donor BMC numbers, from 0.25 to 4.0 million BMCs per recipient, whereas the amount of standard F1 cells was constant at 2.0 million BMCs. There was no longer an increase in donor cell efficiency that would have been shown by an increase in RU concentration. Because the CByB6F1 recipients were not treated to remove hybrid resistance, its effect may have been constant, not saturated by increasing doses of donor cells, as in Table 1.

In separate studies, repopulating abilities of 2 million BALB or B6 donor BMCs in CByB6F1 recipients with intact hybrid resistance were tested with 2 million CByB6F1 competitor standard BMCs. BALB cells contained between 39% and 48% as many RU as found in CByB6F1 cells, approximately 10-fold less than the relative RU concentrations of CXB-12 donors in Figure 3. BMCs from B6 donors contained only 10% to 15% as many RU as found in CByB6F1 cells because of the strong resistance in H-2^d/H-2^bF1 hybrids against H-2^b/H-2^b donors.

In the long-term LTC-IC assay, the LTC-IC is detected by its ability to repopulate a stromal layer in limiting-dilution cultures by forming colonies of myeloid cells. Concentrations of LTC-ICs in vitro from BMCs of B6, BALB, CXB-12, and CXB-4 BMCs were compared at 4, 5, and 6 weeks (Table 4). As previously reported, B6 values were approximately 10% of those found in BALB mice.14,27,30 The CXB-12 strain had values similar to those of BALB, and the CXB-4 values were intermediate. However, in competitive repopulation, BALB cells repopulate approximately 10 times less well than CXB-12 cells, as noted above. Thus, the CXB-12 advantage is not detected by the LTC-IC assay.

The ability of B6, BALB, CXB-12, and CByB6F1 BMCs to form long-term CAFC colonies in vitro were compared at 5 weeks using S17 feeder cells in limiting-dilution analysis. Results, given as mean ± SD per 10⁵ BMCs for the following strains, were: B6 = 1.6 ± 0.7; BALB = 1.8 ± 0.8; CXB-12 = 5.3 ± 1.8, and CByB6F1 = 3.6 ± 1.2. CXB-12 donors had significantly higher CAFC frequencies than B6 and BALB donors showing nonoverlapping 95% confidence limits. However, though CXB-12 BMCs had higher CFAC concentrations than CByB6F1 BMCs, the differences were not significant. Thus, the CXB-12 advantage was not detected by the CAFC assay.

There was a substantial decline with age in HSC function in CXB-12 mice, with old/young RU ratios of 0.21 after 3 months and 0.17 after 6 months (comparing Figures 1 and 4). Still, old CXB-12 BMCs had uniquely high functional ability. Figure 4shows 24-month-old CXB-12 donors tested in the same experiment presented in Figure 1. When comparing levels of cell production by old CXB-12 donors with average values for young donors in the other 11 CXB RI lines, the old CXB-12 donors still repopulate at least 6 times better. Repopulating abilities of young donors from CXB RI lines 1 to 11 declined relative to the CByB6F1 competitor as the time after transplantation increased from 1 to 6 months, probably because of a repopulating advantage in the F1 hybrid competitor cells. This also occurred for old CXB-12 donors. Nevertheless, even after 6 months, old CXB-12 BMCs still contained approximately twice the repopulating ability of the young CByB6F1 competitors (Figure 4).

Competitive repopulation directly measures how well donor HSCs engraft recipients relative to standard competitor HSCs.23Among all the assays for HSCs, it best models reconstitution during a therapeutic transplantation in which donor HSCs are delivered into the recipient bloodstream, home to functional sites, and compete with residual host HSCs.6 A population of donor HSCs, with the large repopulating advantage seen in CXB-12 mice (Figure 1), would dramatically enhance the effectiveness of HSC transplantation as a therapeutic procedure.

F1 HSCs probably benefit from heterosis, and even after treatments to remove NK cells, recipients probably still have residual hybrid resistance. These factors explain why BMCs from CXB RI lines 1 to 11 all contain less long-term repopulating ability than BMCs from the CByB6F1 competitor (Figure 1). Hybrid resistance explains why the advantage of CXB-12 BMCs increases with higher doses of cells, even though the competitor cell numbers are increased proportionally (Table1). The residual resistance is likely saturated by using donor cell doses of 3 to 4 million.26 When NK cells were not removed, hybrid resistance reduced the CXB-12 advantage to approximately 3-5 fold; there was no effect of cell number because the resistance in untreated recipients was not saturated by the donor cell doses used (Figure 3).

The CXB-12 advantage was due to both a 2.4-fold increased HSC concentration and a 1.4-fold proliferative advantage (Figure 2, Table3). Either increased numbers of HSCs in CXB-12 marrow or increased ability to home to sites supporting long-term HSC function, or both, can explain the increase in HSC concentration. The CXB-12 advantage in this competitive dilution experiment was greatly reduced by residual hybrid resistance because the dose of donor cells was only 50 000 BMCs, and the total repopulating advantage at this dose was approximately 5 times less than at higher doses (Table 1). Hybrid resistance in CByB6F1 recipients reduced BALB HSC concentrations from 10 to 4 per million BMCs27; if hybrid resistance had the same type of effect on CXB-12 HSC, their concentration of 24 per million is several times too low.

The extremely good fit between the F = 1.4 model predictions and the observed data in Figure 2 suggests that there is no killing of standard competitor CByB6F1 HSCs by CXB-12 marrow. The substantial degree of killing required to produce the CXB-12 repopulating advantage would have reduced the observed concentration of competitor HSCs and increased the number of recipients that had no competitor HSC engraftment. This did not occur. In fact, the HSC concentration of 1 per 10⁵ competitor BMCs gave an excellent fit (Figure 2). Furthermore, marrow cells in adult mice from the other 11 CXB RI lines did not repopulate as well as the F1 hybrid competitor (Figure 1). Each of the 12 CXB RI lines tested in the current study is homozygous for either the b or the d allele at the major histocompatibility locus, so each could recognize the alternate allele carried by the CByB6F1 hybrid as foreign. If the repopulating advantage resulted from killing antigenically disparate competitor cells, all the CXB RI lines would have shown this advantage, not just the CXB-12. Finally, mouse marrow has only weak graft-versus-host activity because of low concentrations of T cells; percentages of CD4- and CD8-bearing marrow cells were low in CXB-12 marrow (Table 2), so it would not react significantly against the F1 hybrid cells.

A particularly interesting aspect of CXB-12 marrow is that its long-term repopulating advantage does not cause higher numbers of 12 day CFU-S (Table 2) than found in control inbred or F1 hybrid mice. Apparently, precursor differentiation is regulated to produce normal numbers of CFU-S and normal numbers of myeloid and lymphoid cells (Table 2).

Neither CAFC or LTC-IC (Table 4) frequencies are significantly higher in CXB-12 mice than in controls. LTC-IC populations contain precursors capable of long term repopulation in vivo30; perhaps those in CXB-12 mice have an inferior homing ability in vitro, or a superior homing ability in vivo, relative to those in controls. This is not the first case in which LTC-IC and competitive repopulation assays differ. Both here and in prior work,14 LTC-IC concentrations in BALB BMCs were approximately 10-fold higher than those in B6 BMCs. Yet competitive dilution in congenic recipients with congenic competitors gave similar HSC concentrations of approximately 10 per million BMCs in BALB and B6.34,35 

Total RUs per mouse were reduced approximately 5-fold with age in CXB-12 donors (Figures 1, 4). One possibility for this reduction is that increasing age alters HSC homing during transplantation, as in B6 mice.28 Another possibility is that accelerated HSC proliferation in young CXB-12 mice leads to early exhaustion. In the latter case, the functional ability of old CXB-12 HSCs should be lower than the average for old donors from the other 11 CXB RI lines. In fact, old CXB-12 BMCs repopulated better than those of young CByB6F1 competitors or those of young donors from CXB RI lines 1 to 11 (Figure4). We have reported previously that 9 of 11 other CXB RI lines had low old/young RU ratios.27 The loss with age in CXB-12 is similar to the reduction with age seen in these strains and in the BALB strain, but not in the other 2 CXB RI lines or in the B6 strain.27,35 The high rate of loss with age segregated in the RI lines with the BALB locus at the D12Nyu17 marker on chromosome 12.27 

Earlier reports, based on LTC-IC and CAFC frequency measurements in vitro, pointed to a few chromosome regions that may contain genes regulating HSC concentration in the mouse.14-16 It will be interesting to test whether the CXB-12 HSC hyperfunction phenotype is related to these chromosome regions. However, it is important to distinguish the phenotypes tested in prior studies from the phenotype in the current report. CXB-12 BMCs show far more effective repopulation and function than young F1 BMCs in F1 recipients for 6 months, approximately a quarter of the murine life span. It is the unprecedented degree of engraftment advantage relative to high doses of healthy competitor BMC that makes the CXB-12 hyper-HSC phenotype novel.

The phenotypes in Figure 1 were tested in a genome-wide QTL mapping analyses using the Map Manager QTX-06 program with the CXB RI mice genotype data, both available online (http://www.jax.org). This analysis did not produce any linkage that is considered significant (lod score, greater than 3.0). Only insignificant linkages to D6Mit15 (lod score, 0.8) and D17Mit22 (lod score, 1.0) were found, probably by random chance, because more than 500 markers were tested. The CXB-12 hyper-HSC phenotype is not controlled by a single Mendelian locus differing between the BALB or the B6 parent strains because the hyper phenotype does not occur in either. It is found in only 1 of the 12 RI lines, so the simplest genetic explanations are either a new mutation or a unique combination of BALB and B6 genes. Testing HSC functional phenotypes in hybrid and backcross mice between CXB-12 and its parental strains will distinguish these 2 possibilities. Based on preliminary data, it could be a codominant mutation or a combination in which a single locus has a significant effect in approximately 40% of the backcross mice, though the effect is reduced from that in the CXB-12. Future studies using B6-H-2d × BALB F1 hybrid recipients and competitors to remove the strong hybrid resistance associated with a difference at the major histocompatibility locus may give clearer results.

If high stem cell function results from a unique combination of 2 alleles, at least 1 must be recessive—CByB6F1 mice have both BALB and B6 alleles at all loci, but they do not express the phenotype. One allele must be required from each parent because neither expresses the phenotype, and the combination must occur in CXB-12 but not in any of the other 11 RI lines. In theory, the chances that this occurs are [2 (¼) (¾)¹¹] = 0.0211, or approximately 1 in 50 combinations of any 2 alleles. We tested this using a list of 253 loci defined in CXB RI lines 1 to 12 (from The Jackson Laboratory Informatics web site:http://www.informatics.jax.org/searches/riset_form.shtml). We compared alleles at pairs of loci for CXB-12 with those of the other 11 lines using an Excel function. In the first 2485 tested, there were 99 pairs unique to CXB-12, approximately 1 in 25. However, half of those pairs were both from the same parental strain, so, in fact, as expected 1 in 50 were pairs of alleles unique to CXB-12, with 1 allele from each strain. There are 253 × 252 = 63 756 possible allele pair permutations of 253 loci. Even considering that this is reduced by linkage, there are too many possibilities for our results using only 12 RI lines to have the statistical power to give significant, or even suggestive, map locations.

Regulation of the hyper-HSC phenotype appears to be focused on more primitive HSCs because CXB-12 mice have normal numbers of myeloid cells, lymphoid cells, and less primitive precursors (Tables2, 4). In competitive repopulation, the phenotype is expressed most strongly after 3 and 6 months in the recipients (Table 1; Figures 1,3), which is the time required for primitive HSCs to repopulate. The high number and function of CXB-12 primitive HSCs may be maintained by alterations in a receptor that regulates primitive HSCs or in its ligand. Increases in numbers of HSCs could result from changes in either ligand or receptor; however, the increased proliferative capacity of CXB-12 HSCs after transplantation in normal recipients suggests that the change is intrinsic to the HSCs, perhaps an altered receptor. Understanding the CXB-12 phenotype may eventually lead to improved clinical transplantation procedures to enhance the effectiveness of primitive HSCs.

We thank Avis Silva, Bee Stork, and Karen Davis for their excellent technical assistance. We thank Dr Ben Taylor for helpful discussions about RI line analyses and Dean Campagna, an academic year student, and Benjamin H. Schmidt, a summer student, for performing the CAFC assays.