SUMMARY
In mammals, the production of red blood cells is tightly regulated by the growth factor erythropoietin (EPO). Mice lacking a functional Epo gene are embryonic lethal, and studying erythropoiesis in EPO-deficient adult animals has therefore been limited. In order to obtain a preclinical model for an EPO-deficient anemia, we developed a mouse in which Epo can be silenced by Cre recombinase. After induction of Cre activity, EpoKO/flox mice experience a significant reduction of serum EPO levels and consequently develop a chronic, normocytic and normochromic anemia. Furthermore, compared with wild-type mice, Epo expression in EpoKO/flox mice is dramatically reduced in the kidney, and expression of a well-known target gene of EPO signaling, Bcl2l1, is reduced in the bone marrow. These observations are similar to the clinical display of anemia in patients with chronic kidney disease. In addition, during stress-induced erythropoiesis these mice display the same recovery rate as their heterozygous counterparts. Taken together, these results demonstrate that this model can serve as a valuable preclinical model for the anemia of EPO deficiency, as well as a tool for the study of stress-induced erythropoiesis during limiting conditions of EPO.
INTRODUCTION
The constant renewal of red blood cells (RBCs) via erythropoiesis is crucial to ensure proper tissue oxygenation (Jelkmann, 2007). In this respect, erythropoietin (EPO) has long been identified as a key factor in the regulation of erythropoietic output (Fang et al., 2008; Rankin et al., 2007; Sathyanarayana et al., 2008). During homeostasis, EPO is produced at low levels by the peritubular capillary endothelial cells in the kidney and is then released into the blood stream (Jelkmann, 2007; Koury et al., 1988). Following activation of the EPO-receptor (EPOR) complex, the expression of anti-apoptotic genes, e.g. Bcl2l1, is upregulated within erythroid progenitors, which in turn leads to their survival (Silva et al., 1996). Consequently, these cells are able to differentiate into mature reticulocytes. In times of low RBC counts, Epo expression is sharply upregulated and, in turn, erythropoietic output is dramatically increased to compensate for the loss of RBCs. This feedback system is tightly controlled by oxygen levels via hypoxia-inducible transcription factors (HIFs) (Rankin et al., 2007; Wang and Semenza, 1993).
Chronic kidney disease (CKD) is characterized by the loss of kidney function and is initiated by diabetic nephropathy, hypertension and glomerulonephritis (Go et al., 2004; Shlipak et al., 2005). The leading cause of death in patients with CKD is cardiovascular disease, regardless of disease progression (Frank et al., 2004). Late-stage CKD patients exhibit a moderate-to-severe anemia from a reduction in renal EPO production (Means, 2003). Thus, patients with CKD display a chronic anemia in the presence of low serum EPO (sEPO) levels.
Currently, recombinant human EPO (rHuEPO) is the approved treatment for patients with anemia due to CKD. Despite the success of rHuEPO as a therapy for anemia in CKD, various adverse effects have been reported (Frank et al., 2004; Goldberg et al., 1992). For example, patients can become EPO resistant and also hyporesponsive to the biologic (van der Putten et al., 2008). Furthermore, observations in a clinical trial with patients that suffer from anemia due to chemotherapy indicated that rHuEPO increases lethality in these patients (Bohlius et al., 2009). Results of another large clinical trial on patients with anemia and either diabetes or CKD determined that rHuEPO therapy did not reduce the risk of death due to either cardiovascular or renal complications in these patients (Pfeffer et al., 2009). The trial also indicated that rHuEPO therapy could lead to an increased risk of stroke in these patients. However, because rHuEPO rapidly raises the patient’s hemoglobin level, it is possible that the observed detrimental effects of rHuEPO therapy are attributable to the sharply increased levels of hemoglobin rather than being an effect of the drug itself (e.g. cross-reactivity to other targets). Nevertheless, on the basis of these observations it is now recommended that the patient’s hemoglobin level should not exceed 12 g/dl. Taken together, the development of erythropoiesis-stimulating agents (ESAs) that function without activating the EPOR complex would offer valuable alternative treatment options (Bunn, 2007; Wrighton et al., 1996).
At present, few in vivo models exist that display a phenotype similar to the anemia of CKD: specifically, low RBC production in conjunction with depleted sEPO levels. Models involving chemical injections are used to directly destroy RBCs and lower RBC indices. For instance, the injection of phenylhydrazine (PHZ) results in the destruction of RBCs by binding to hemoglobin proteins within these cells (Augusto et al., 1982). The massive loss of RBCs causes the animal to undergo a stress-induced erythropoietic response but can also bring about deleterious secondary affects, e.g. liver damage (Jonen et al., 1982). Alternatively, oxygen levels can be reduced in hyperbaric chambers, resulting in low tissue oxygenation in the animal. However, regardless of the methods employed, these current techniques lead to a rapid increase in sEPO levels and are not representative of an EPO-deficiency anemia. To overcome this caveat, nephrectomy has been used to create an EPO-limiting environment in vivo. However, this model involves an invasive procedure and is therefore not appropriate for large-scale experiments such as screening of preclinical compounds. In addition, these animals experience a shortened lifespan, which prevents the study of long-term effects of low EPO production. Finally, nephrectomized animals are still able to produce EPO in secondary tissues such as the liver (Tan et al., 1991).
The introduction of transgenic mice technology has also yielded some innovative mouse models for the study of EPO function; however, these are not without caveats. For example, Epo-knockout mice are embryonic lethal, making it an ineffective model to study anemia and erythropoiesis in the adult system (Wu et al., 1995). The authors determined that EPO was not necessary for erythroid-lineage commitment; rather, it facilitated the survival of erythroid progenitors (CFU-E) that could then differentiate into RBCs. In another model, Maxwell and colleagues established a transgenic mouse in which the SV40-TAg sequence was inserted into the 5′ sequence of the Epo gene. This gene modification leads to a dramatic decrease in Epo expression and consequently the animals display a severe chronic anemia (Maxwell et al., 1993). However, the EPO-TAg mice also develop an immune response to EPO that is probably due to a sensitizing effect of the expressed SV40 T antigen (Rinsch et al., 2002).
To facilitate the creation of a new model for the anemia of EPO deficiency, we developed a conditional Epo-knockout mouse by inserting loxP sites into the Epo gene (Claxton et al., 2008; Li et al., 2006). By crossing this floxed Epo allele to an inducible Cre transgene (Rosa26-CreERT2) Epo was postnatally ablated (Seibler et al., 2003). After induction of Cre activity, Epo expression and sEPO levels were substantially reduced. Accordingly, these conditional-knockout mice experienced a significant reduction in RBCs, although mean corpuscular hemoglobin (MCH) and mean corpuscular volume (MCV) remained normal. In addition, an analysis of the bone marrow progenitors of Epo-knockout animals showed no effect on the erythroid progenitors BFU-E and CFU-E. Therefore, these mice develop a chronic, normocytic and normochromic anemia in conjunction with low sEPO levels that correlate to the clinical display of patients with anemia due to CKD. However, in contrast to patients with CKD, these animals do not develop inflammation or uremia and therefore will allow the study of erythropoiesis in an EPO-limiting environment without the bias of possible secondary complications. Taken together, these results demonstrate that this mouse model can serve as a preclinical tool to study chronic anemia due to EPO deficiency.
RESULTS
Conditional deletion of Epo leads to a reduction in Epo mRNA and sEPO levels
We analyzed the impact of Epo deletion in adult mice by employing a conditional-knockout strategy (Claxton et al., 2008; Seibler et al., 2003). To induce Epo deletion, mice in which exon 2 through 4 was flanked by loxP sites were generated (Fig. 1A,B). These exons encode for the amino acid sequences necessary for EPOR binding. After obtaining germline transmission, EpoWT/flox mice were crossed with each other to obtain homozygous Epoflox/flox mice. These mice seemed phenotypically normal to their wild-type counterparts (data not shown). Epoflox/flox animals were bred to the EIIa-Cre mice, which ubiquitously express Cre, to generate EpoΔ/Δ embryos. These embryos died at 12.5 days post-coitum (dpc) with severe defects in erythropoiesis (Fig. 1C). An absence of definitive RBCs in the EpoΔ/Δ embryos was observed in the fetal liver; however, erythroid precursors were present in the yolk sac of these animals, as verified by CD71+/Ter119+ staining (data not shown). Thus, the phenotype of EpoΔ/Δ embryos is similar to previously described Epo-knockout mice (Suzuki et al., 2002; Wu et al., 1995).
To investigate the effects of Epo deletion in the adult, Epoflox/flox and EpoWT/flox mice were bred to mice that contained a tamoxifen-inducible Cre allele, Rosa26WT/CreERT2. The resulting animals (Epoflox/flox) displayed normal hematocrit (HCT) levels. When transgenic mice reached 8 weeks of age, tamoxifen was administered through subcutaneous pellet implantation, activating the Cre. Tamoxifen was released at 1 mg/day for 25 days, ensuring proper Cre activation and resulting in the Epoflox allele becoming EpoΔ. At 30 days after pellet implantation, Epo deletion was verified by quantitative reverse transcriptase PCR (qRT-PCR) (Fig. 1D). Under steady-state conditions, we observed a reduction of Epo kidney expression in EpoΔ/Δ mice compared with EpoWT/Δ littermates. However, owing to low baseline expression levels of Epo, an estimation of knockout efficiency by qRT-PCR was ambiguous. To overcome this limitation, we analyzed Epo expression levels during an acute hypoxic stimulus, induced by PHZ administration. At 2 days after PHZ treatment, Epo expression levels in the kidney of EpoΔ/Δ animals were approximately 95% reduced compared with EpoWT/Δ animals. Also, sEPO levels were monitored during the acute hypoxic event. In this respect, it is worth noting that homeostatic levels of sEPO are undetectable with current technologies owing to the low protein concentration in normal mouse serum. Therefore, sEPO levels were also monitored in mice undergoing a stress-induced erythropoietic response due to PHZ treatment. In agreement with the Epo mRNA data, sEPO levels in EpoΔ/Δ mice were approximately 10% of those of EpoWT/Δ mice after PHZ treatment (data not shown and see below). These results indicate that induction of Cre activity leads to a severe ablation of Epo mRNA as well as a reduction in sEPO levels. qRT-PCR analysis was next performed on a gene known to be directly regulated by EPO signaling, Bcl2l1. In agreement with the Epo mRNA and sEPO data, we observed a significant reduction of Bcl2l1 expression in the bone marrow (Fig. 1E). Taken together, these data demonstrate that the induced deletion of the Epo gene results in a decrease in sEPO levels, which in turn leads to decreased EPO signaling in the bone marrow.
rHuEPO rescues the EpoWT/Δ and EpoΔ/Δ mice phenotype
We next investigated whether the erythroid precursors of EpoΔ/Δ mice are hypersensitive to EPO and whether a smaller concentration of EPO would be able to illicit the same effect as basal levels of the protein in transgenic mice. To evaluate this, rHuEPO was injected into EpoΔ/Δ and EpoWT/flox animals at two different concentrations. A low dose (120 U/kg), representing the approximate levels observed in the EpoΔ/Δ mice, and a high concentration (1200 U/kg) were selected to evaluate the erythropoietic responses. At 4 days after rHuEPO administration, whole blood was analyzed and reticulocytes measured (Table 1 and Fig. 2A). In the placebo as well as the 120 U/kg rHuEPO group, EpoΔ/Δ mice displayed a comparable low percentage of reticulocytes. This suggests that the EpoΔ/Δ mice are not hypo-responsive to EPO. By contrast, all transgenic animals in the 1200 U/kg rHuEPO group showed a significant increase in reticulocyte and erythroid indices in the blood. These results suggest that erythroid precursors of EpoΔ/Δ were not hyper-responsive to EPO because, under high EPO concentrations, reticulocyte counts in the EpoΔ/Δ mice were the same as in EpoWT/flox animals.
Deletion of Epo in the adult causes a chronic anemia as seen in CKD patients
Induced EpoWT/Δ and EpoΔ/Δ mice were monitored for more than 300 days in order to evaluate the long-term effects of EPO deficiency in the adult. Over the initial induction period, a 40% decrease in HCT levels was observed in EpoΔ/Δ animals (Fig. 2B). Complete blood count (CBC) indices partially rebounded and stabilized after the first 30 days, although their HCT levels were still 25% below pre-induction levels. This observation suggested that tamoxifen might have a deleterious effect on erythropoiesis. To evaluate whether tamoxifen caused adverse effects, tamoxifen pellets were implanted into EpoWT/flox and Epoflox/flox animals lacking the Rosa26CreERT2 allele. Therefore, these animals experience tamoxifen without the activation of Cre recombinase. After tamoxifen administration, both Epoflox/flox Cre-negative (CreNeg) and EpoWT/floxCreNeg animals showed no reduction in the levels of HCT, RBCs and hemoglobin (Hb), indicating that it did not have a deleterious effect on erythropoiesis (Fig. 2B and Table 2). However, another explanation for the observed mild rebound could be that the activated CRE-ERT2 protein produces some toxic effects. It is known that the CRE protein is not well tolerated by cells and high levels of CRE can lead to cell death. In this respect it is curious to note that the HCT values slightly rebounded after removal of the inducer. Taken together, it seems that the induction of CRE-ERT2 by tamoxifen affects erythropoiesis. For this reason, all subsequent studies were performed with animals that were at least 50 days post-induction.
Between days 40 and 250 of the observation period, EpoΔ/Δ mice showed a 25% reduction in HCT values compared with EpoWT/floxCreNeg animals. At 50 days after pellet implant, a reduction in RBC, Hb and HCT levels was observed in EpoΔ/Δ mice compared with EpoWT/Δ mice (Table 2). The MCH, MCV and MHCV [the amount of hemoglobin relative to the size of the cell (hemoglobin concentration) per RBC] values remained normal in EpoΔ/Δ animals and no changes in other hematopoietic lineages were observed, such as in the number of megakaryocytic progenitors (data not shown). These findings are in corroboration with previously described observations in Epo-knockout mice (Jelkmann, 2007; Wang and Semenza, 1993). Taken together, these data show that the Epo conditional deletion led to the development of a chronic, normocytic and normochromic anemia.
Loss of Epo leads to a decrease in mature erythroid cells in the bone marrow
EpoΔ/Δ, EpoWT/Δ and EpoWT/WT bone marrow and spleen cells were tested for their potential to differentiate into committed erythroid progenitors. The number of progenitor colonies in bone marrow and spleen were statistically similar in EpoΔ/Δ, EpoWT/Δ and EpoWT/WT mice (Fig. 3). This observation is comparable to the previously described Epo-knockout mice (Wu et al., 1995). Also, no differences in either size nor morphology were detected within the BFU-E and CFU-E colonies of EpoΔ/Δ, EpoWT/Δ or EpoWT/WT cells. To further evaluate the erythron, bone marrow and spleen erythroid precursors were analyzed using CD71 and Ter119 markers (Zhang et al., 2003). EpoΔ/Δ mice experienced no reduction in the CD71+/Ter119Neg cell population, which represents an early erythroid stage (Fig. 4). In addition, no significant increase in bone marrow or spleen CD71+/Ter119+ erythroblast populations was observed. By contrast, EpoΔ/Δ mice displayed a decrease in the CD71Neg/Ter119+ population, which represent a more mature erythroid stage in the bone marrow. EpoΔ/Δ mice displayed a 2.5-fold decrease in the CD71Neg/Ter119+ population compared with EpoWT/WT mice. Taken together, these results demonstrate that, whereas EPO reduction in EpoΔ/Δ mice did not have an effect on the percentages of early-stage progenitors, these mice had lower numbers of intermediate progenitors in their hematopoietic compartments.
EpoΔ/Δ mice recover normally from an acute hypoxic event
Next, we investigated whether EpoΔ/Δ mice could respond to an acute erythropoietic crisis. Although HCT levels were significantly lower in the EpoΔ/Δ animals under steady conditions, sEPO levels were similar (68 pg/ml vs 76 pg/ml, respectively; Fig. 5A). Mice were treated with PHZ to induce an acute hypoxic response and were monitored for 14 days. EpoWT/Δ and EpoΔ/Δ animals showed no differences in the rate of recovery from the acute anemia, as monitored by their rise in HCT levels (Fig. 5A). As expected, EpoWT/Δ mice showed a substantial increase in sEPO levels compared with EpoΔ/Δ mice. At 2 days after PHZ injection, sEPO levels were more than tenfold higher in EPOWT/Δ mice compared with the EpoΔ/Δ animals (1000–10,000 pg/ml vs 100–800 pg/ml, respectively). Furthermore, no difference in spleen size was observed between EpoWT/Δ and EpoΔ/Δ mice, indicating that extramedullary erythropoiesis was normal in these animals. FACS analysis revealed no significant differences in erythroid progenitor populations in either bone marrow or spleen during recovery (Fig. 5B). These data suggest that, although EPO production is severely diminished in EpoΔ/Δ mice, the erythropoietic system is fully capable to respond to an acute hypoxic event.
EpoΔ/Δ mice display low EPO levels in correlation with low HCT levels, similar to CKD patients
Finally, HCT values were compared with sEPO levels in EpoΔ/Δ, EpoWT/Δ and EpoWT/WT transgenic mice (Fig. 6). The inverse relationship between sEPO levels and HCT values has been observed and is characterized by low HCT values in correlation with significantly increased sEPO levels (Artunc and Risler, 2007). As shown in Fig. 6, this relationship was observed in EpoWT/WT mice, whereas heterozygous animals (EpoWT/Δ) displayed lower sEPO level compared with HCT levels. In addition, EpoΔ/Δ mice showed a dramatic reduction of sEPO in relation to low HCT levels. For example, EpoΔ/Δ mice with an HCT of 20% displayed an sEPO level of 100–300 pg/ml. In sharp contrast to this, EpoWT/WT animals with an HCT of 30% were able to produce sEPO levels well above 10,000 pg/ml.
DISCUSSION
In this study, we describe the generation and characterization of a mouse model for the anemia of CKD. To create this model, an inducible Cre-loxP system was used that allowed for the deletion of Epo in the adult animal. After induction of Cre, the level of sEPO significantly decreased and the animals developed a moderate anemia with normal MCV and MCH indices. However, the reduction in sEPO levels did not affect early erythroid progenitors in the bone marrow, as demonstrated by CFU-C assays and FACS analysis. These results are in agreement with previously published data of EPO-deficient animals (Wu et al., 1995).
Some variations in the model characterized here versus previously generated Epo-knockout mice were observed. For instance, animals heterozygous for the Epo gene (EpoWT/Δ and EpoΔ/flox) consistently displayed a reduced HCT level compared with EpoWT/WT animals. Furthermore, the amount of late-stage erythroid precursors in the bone marrow (CD71Neg/Ter119+ cells) was reduced in a dose-dependant manner. This decrease could be the result of an early egress of reticulocytes. It has been shown that under hypoxic conditions sinusoid pores in the bone marrow widen and release reticulocytes (Stohlman et al., 1954). In agreement with this, a minor increase in peripheral blood reticulocytes of EpoΔ/Δ animals was observed. These differences could be the result of the EpoΔ locus design. The model described here was created by inserting loxP sites into the Epo allele. Although this insertion should not affect the expression of the Epo gene, it is possible that the integration influenced expression levels. We did not, however, find any known enhancer sequences within the region of loxP integration. Nevertheless, the significantly lower HCT in the EpoΔ/floxCreNeg mice compared with EpoWT/Δ animals indicates that the insertion of the loxP sites decreased Epo expression levels. Finally, a potential deleterious effect of the Cre-recombinase protein on erythropoiesis can be ruled out because EpoWT/flox mice that carry no Cre allele show the same decrease in HCT as the EpoWT/flox mice after induction of Cre activity, thus becoming EpoWT/Δ.
An interesting finding in this study is that the severe reduction of sEPO in the adult leads to a moderate rather than a severe anemia. In this respect it is important to recognize that some residual Epo mRNA and sEPO remained in the EpoΔ/Δ mice even after 20 days of continuous induction. Cre expression in these mice was driven by the ubiquitously expressed Rosa26 locus; however, expression of this locus is variable between tissues (Zambrowicz et al., 1997). Therefore, it is possible that a number of peritubular cells within the kidney escape Cre activity, leading to the residual EPO production. Alternatively, EPO production in secondary tissues such as liver or lung could contribute to the observed residual EPO. However, several previous observations argue against this view. First, EPO from secondary tissues only accounts for a small percentage of the total circulating EPO (Fried et al., 1969). Second, liver and lung tissues have been shown to express the highest levels of the CRE-ERT2 transgene and therefore excision of the Epo allele in these tissues is likely to be of a high degree (Jullien et al., 2008). Finally, in preliminary experiments we attempted to quantify hepatic Epo mRNA levels in EpoΔ/Δ mice as well as control animals but were not able to detect significant levels (data not shown). Because the renal Epo mRNA expression and sEPO levels measured in the whole blood serum are in tight agreement, it seems unlikely that EPO derived from different Cre-negative tissues is responsible for the prevention of a severe anemia. Furthermore, EpoΔ/Δ mice have a tenfold reduction of Bcl2l, a well-characterized downstream mediator of EPO activity, in the bone marrow (Dolznig et al., 2002). Taken together, these data show that the induction of Cre activity leads to a significant but incomplete loss of Epo expression throughout the adult animal in this model.
The commonly used preclinical anemia models are not suited to study the anemia of EPO deficiency, owing to the normal state of the Epo response in these animals. However, several transgenic approaches have been undertaken to overcome this hurdle. In an attempt to characterize Epo-expressing cells in vivo, Maxwell and colleagues established a transgenic mouse model in which the SV40-TAg sequence was inserted into the 5′ end of the Epo gene (EPO-TAg) (Maxwell et al., 1993). Mice homozygous for this allele develop a severe chronic anemia with sEPO levels below the detection limit. By contrast, the EpoΔ/Δ mice described here displayed a mild-to-moderate chronic anemia with measurable sEPO levels, albeit much lower than expected for the HCT observed. In this respect it is important to note that a major difference between the EPO-TAg mice and the animals described here is the nature of the Epo gene modification. Whereas the EpoΔ/Δ mice are obtained via induced Cre-mediated excision, the EPO-TAg mice are derived from breeding carriers of the modified allele so that all cells in the animal carry the modified Epo allele. By contrast, the less severe loss of sEPO levels in the EpoΔ/Δ mice is probably the result of incomplete deletion of the Epo allele. Thus, the EpoΔ/Δ mice exhibit a phenotype that is more closely related to the clinical display of anemia in CKD. Furthermore, EPO-TAg mice develop an immune response to EPO, probably owing to a sensitizing effect of the expressed SV40 T antigen (Rinsch et al., 2002). Thus, the EPO-TAg mice represent a critical EPO-deficient anemia, whereas the anemia observed in EpoΔ/Δ mice seems to be more closely related to the clinical presentation of anemia in CKD patients. In another model, Gruber and colleagues reported that the inducible deletion of the Hif2α transcription factor in the adult animal results in an anemia that is similar to the one described here for the EpoΔ/Δ model (Gruber et al., 2007). However, the observed sEPO levels in these animals are significantly higher than those observed in EpoΔ/Δ mice and therefore are not representative of an EPO-deficiency anemia. Because HIF transcription factors are known to regulate several genes, it can be suspected that the anemia observed in these animals is multifactorial and not exclusively based on EPO deficiency (Mastrogiannaki et al., 2009; Rankin et al., 2007; Sathyanarayana et al., 2008).
A surprising discovery of this study is that EpoΔ/Δ mice recover at a normal rate from an induced acute hypoxic stimulus. Whereas, overall, EpoΔ/Δ mice display lower HCT and sEPO levels, the recovery rate of these animals is similar to their wild-type counterparts. The low levels of sEPO measured in EpoΔ/Δ mice cannot account for the normal recovery to the hypoxic stress because injections of a low EPO dose had no effect on erythropoietic output. Further studies are needed to determine whether, under severe hypoxic conditions, erythroid precursors dislodge from the bone marrow in the EpoΔ/Δ mice and populate the spleen or whether expansion of splenic precursors alone is sufficient to account for the observed normal response. Another possible explanation for the observed normal recovery rate is that erythroblasts in the EpoΔ/Δ mice become hypersensitive to EPO (Perry et al., 2007). However, injecting low concentrations of rHuEPO, comparable to the sEPO levels observed in EpoΔ/Δ mice under anemic conditions, are in contrast with this hypothesis because we did not observe an increase in erythropoiesis.
In this study we demonstrate that the anemia of EpoΔ/Δ mice is similar to the clinical presentation of anemia in CKD patients (Artunc and Risler, 2007). These patients suffer from a loss of kidney function and develop a severe anemia from the lack of EPO production (van der Putten et al., 2008). Thus, these patients display a low HCT in conjunction with low sEPO levels, similar to the observation in EpoΔ/Δ mice (Artunc and Risler, 2007). In addition, CKD patients are able to maintain low sEPO levels and are therefore not completely deprived of EPO, similar to the model described here. It should be noted that patients with CKD display normal hepatic EPO production, in contrast to the model described here. In EpoΔ/Δ mice, the Epo gene is excised ubiquitously and therefore hepatic EPO is unlikely to be a major source of circulating EPO. Also, MCH and MCV values are unaffected in both EpoΔ/Δ mice and CKD patients. However, EpoΔ/Δ mice do not show signs of abnormal kidney function, inflammation or uremia, in contrast to patients with CKD. Thus, these animals will allow the study of erythropoiesis under EPO-limiting conditions free of secondary complications. The control of erythropoiesis by EPO signaling seems to be comparable between CKD patients and the model described here. However, patients with CKD develop a more severe anemia in late stages of the disease despite low levels of sEPO, in contrast to the EpoΔ/Δ mice. This suggests that other factors are contributing to the anemia in these patients. Because CKD patients experience a significant amount of inflammation, it is possible that factors that oppose erythropoiesis are increased in these patients. In this respect it is of interest that it has been shown that the iron regulator hepcidin can suppress CFU-E development and levels of hepcidin are increased in CKD patients (Ashby et al., 2009; Tomosugi et al., 2006). Therefore, the anemia of CKD seems to be multifactorial rather than an anemia of EPO deficiency alone. The EpoΔ/Δ mice described here will offer a preclinical model to study the biology of the anemia of CKD. For example, treating these animals with hepcidin might help to dissect the role of this regulator in the anemia of CKD.
In summary, our data show that the conditional deletion of Epo in the adult animal results in an anemia of EPO deficiency. We did not observe any changes in the behavior, body weight or activity level of the animal during the time observed (>50 weeks post induction), and therefore the development of a chronic moderate anemia seems to be well tolerated by the animals. Also, these transgenic animals have the ability to survive for long term studies, unlike previously used models such as those using nephrectomy. Thus, EpoΔ/Δ mice will be a valuable tool for the development of new therapeutic approaches for the treatment of the anemia of CKD. In addition, this model will allow the study of erythropoiesis under EPO-limiting conditions as well as studies analyzing the contribution of extra-renal EPO production to erythropoiesis. Furthermore, analyzing the stress-induced erythropoietic response in these animals has the potential to discover new erythropoiesis-regulating mechanisms that are masked by EPO in the currently used models. Elucidating these mechanisms has the potential to identify new target opportunities for the treatment of anemia. Finally, these animals will allow the study of the effects of EPO outside the erythropoietic system. For example, EPO has been shown to ameliorate overall cardioprotective activity and the EpoΔ/Δ mice could provide a valuable model to study these effects in detail.
METHODS
Generation of Epo conditional-knockout mice
A 10-kb targeting vector encompassing all five Epo exons and flanking genomic regions was created using recombinogenic engineering technology (Liu et al., 2003). A loxP site was inserted into intron 1 followed by a neomycin-resistance cassette and a second loxP site inserted into intron 4. The targeting vector was linearized by NotI restriction-enzyme digestion and electroporated into the Bruce4 embryonic stem cell line (Lakso et al., 1996). Clones correctly targeted at both homology arms were identified by Southern blot and injected into blastocysts to generate chimeras. The Epo allele with loxP sites flanking exons 2–4 is termed Epoflox throughout this manuscript. Mice were genotyped for Cre-excision using primers complementary to exon 4: reverse: 5′-GCCATA-GAAGTTTGGCAAGG-3′, forward: 5′-ACCCGAAGCAGTG-AAGTGAG-3′. Epoflox/flox mice (C57BL/6) were bred to EIIa-Cre mice [FVB/N-Tg(EIIa-cre)C5379Lmgd/J; Jackson Laboratories, Bar Harbor, ME], which ubiquitously express Cre early in embryonic development, to delete exons 2-4 of Epo. Deletion was confirmed by qRT-PCR and sEPO ELISA. An Epo allele with exons 2–4 deleted is described as EpoΔ throughout this paper.
Epoflox/flox mice were bred to mice carrying an inducible Cre allele, Rosa26WT/CreERT2 [B6.129S4-Gt(Rosa)26Sortm1Sor/J], from Jackson Laboratories to postnatally ablate Epo upon tamoxifen exposure (Seibler et al., 2003). Tamoxifen was administered via subcutaneous pellet (from Innovated Research of America, Sarasota, FL) implant and released at a rate of 1 mg/day for 20 days (Claxton et al., 2008). To induce an acute stress response, PHZ (Sigma-Aldrich, St Louis, MO) was intraperitonally injected at a concentration of 40 mg/kg (Agosti et al., 2009; Nakano et al., 2005). To rescue transgenic mice, rHuEPO (Amgen, Seattle, WA) was diluted in PBS and injected at 120 U/kg (for low-dose experiments) and 1200 U/kg (for high-dose experiments) in 500 μl per mouse subcutaneously (Albertengo et al., 1999). 4 days later, mouse whole blood was collected via heart stick and analyzed by CBC and flow cytometric analysis. The animal care and use program is fully accredited by the Association for Assessment and Accreditation of Laboratory Animal Care, International.
The mice used in the study were generated by Pfizer’s Genetically Modified Mice department and are available to the scientific community. To obtain these mice please contact the senior author of this manuscript.
qRT-PCR
Kidney and heart tissues were extracted then flash frozen and ground down to a powder with a mortar and pestle. Total RNA was extracted using RNeasy Mini Kit from Qiagen (Valencia, CA). First-strand cDNA was generated using reverse transcriptase SuperscriptIII and oligo (dT)20 primers. Real-time PCR was performed in duplicate on an Applied Biosystems (Carlsbad, CA) HT-7900 7900 Real-Time PCR.
The following primers were used for SYBR-Green detection: Epo forward: 5′-AGGAGGCAGAAAATGTCACG-3′, Epo reverse: 5′-GGCCTTGCCAAACTTCTATG-3′; Cyclophilin forward: 5′-AGAGAAATTTGAGGATGAGAACTTCA-3′, Cyclophilin reverse: 5′-TTGTGTTTGGTCCAGCATTTG-3. Absolute quantification of each gene was calculated by the standard curve method using ten-fold dilutions of a positive control (PHZ treated, EpoWT/WT kidney cDNA). Bcl2l1 primers were acquired from Qiagen, Unigene number, Mm.238213. Expression of individual genes was normalized to cyclophilin expression.
Cellular analysis
For flow cytometric analysis, bone marrow and spleen samples were passed through a 40-μm nylon mesh filter to create a single-cell suspension. 2×106 cells/ml were treated with 1 μL Fc blocking antibody (2.4G2) for 20 minutes at 4°C then incubated with anti-CD71-FITC (C2) and anti-Ter-119-APC (BD/Pharmingen, San Jose, CA) for 45 minutes at 4°C in the dark. Samples were washed with PBS and suspended in 200 μl FACS buffer. Isotype controls were APC-Rat IgG2bκ and FITC-Rat IgG1κ for Ter119 and CD71, respectively, from BD/Pharmingen. Cells were processed on a BD Caliber flow cytometer and the raw data analyzed using FlowJo software (Tree Star, Portland, OR).
Mouse whole blood was analyzed using a CELL-DYN 3700 (Abbott Laboratories). The remaining blood was spun and serum collected for ELISA analysis. sEPO levels were detected by mouse ELISA hypoxia kit (Meso-Scale Discovery, Gaithersburg, MD) and evaluated using a MSD Sector Imager 6000. For whole blood reticulocytes measurement, cells were stained with thiazole orange from Sigma-Aldrich for 30 minutes at room temperature in the dark and the fluorescence intensity was measured (Maltby et al., 2009).
For colony-forming assay (StemCell Technologies, Vancouver, BC) were performed as previously described (Zeigler et al., 2006). Bone marrow cells were plated at a density of 1×104 cells in 4 ml of MethoCult per 35-mm dish and spleen cells were plated 1×105 cells per dish. The number of CFU-Es was determined 2 days after culture and BFU-Es were counted on day 4.
Acknowledgements
The authors thank the Comparative Medicine group for the upkeep of animals used in this study. The authors also thank Monica Hultman and Eva Nagiec for expertise in qRT-PCR and MSD ELISA Technology. Thanks to William L. Blake, Mary K. Bauchmann and Marsha L. Roach for their assistance in electroporation, and Diane M. Nadeau for her contribution in microinjection. Finally, thanks to Don Wojchowski and his lab for helpful advice on experimental design.
AUTHOR CONTRIBUTIONS
B.Z. and K.N. performed experiments, analyzed the results, made the figures and wrote the manuscript. K.N. designed the mouse model and research project. J.V. and W.Q. created the transgenic mouse used for experiments. L.L. designed the breeding strategy and maintained the colony.
REFERENCES
COMPETING INTERESTS
At the time of this study all authors were employed and funded by Pfizer Inc.