Stage specific functional roles of integrins in erythropoiesis

Abstract

When the erythroid integrins α5β1 and α4β1 were each deleted previously at the stem cell level, they yielded distinct physiologic responses to stress by affecting erythoid expansion and terminal differentiation or only the latter, respectively. To test at what stage of differentiation the integrin effects were exerted, we created mice with α4 or α5 integrin deletion only in erythroid cells and characterized them at homeostasis and after Phenylhydrazine (PHZ)-induced hemolytic stress. The phenotype of mice with α5-erythroid deletion unlike our prior data was similar to controls, especially post-stress, and these outcomes seem to reconcile divergent prior views on its role in erythropoiesis. By contrast, α4 integrins, whether deleted early or late, have a dominant effect on bone marrow (BM) retention of erythroblasts and on terminal erythroid maturation at homeostasis and post stress.

Introduction

In erythropoiesis integrin-dependent interactions of erythroid cells with macrophages and fibronectin in their microenvironment have been extensively studied and found to be instrumental for their proper expansion/maturation [1–4]. A host of interacting macrophage/erythroblast mediators have been uncovered [5] and references therein, but several of their partners and the signaling pathways initiated have not been elucidated. Interactions of erythroid cells with fibronectin in their microenvironment have also been found to be important for proper erythroid expansion and differentiation according to in vitro studies [2, 6]. Although α4 and α5 integrins are thought to be critically involved in these interactions [2, 6], there are conflicting data as to their roles [2, 7]. Further, whether integrins are critically involved at earlier stages of erythroid differentiation was not clear until recently. By deleting β1 integrins in all hematopoietic cells we uncovered a major role for this integrin in erythroid responses and survival post stress, unlike the response seen by deleting only α4 integrins [5, 8]. As mainly α4β1 and α5β1 integrin heterodimers are present in erythroid cells, the data implied a role for α5 integrin in this process distinct from the one exerted by α4 integrin. However, it was unclear from our previous studies [5, 8] whether deletion of all β1 integrins was responsible for the differences and if so, at what stage of erythroid differentiation the integrin effects were exerted. To clarify integrin specific effects in erythropoiesis we created murine models with selective deletion of α4 or α5 integrins in erythroid cells and provide new data which demonstrate stage-dependent functional roles of these integrins in erythropoiesis.

Methods

Mice and treatments: to obtain mice with deletion of α4 or α5 integrins selectively in erythroid cells, we bred α4f/f mice [9] or α5f/f mice [10] with EpoR-Cre mice [11]. All experiments with mice were approved by the University of Washington Institutional Animal Care and Use Committee. PHZ-treatment, peripheral blood (PB) evaluation, Fluorescence Activated Cell Sorting (FACS) analysis and progenitor assays were done as described by us previously [8].

Results and Discussion

Steady state erythropoiesis in α5f/fEpo-RCre⁺(α4δEry) or α4f/fEpo-RCre⁺(α5δEry) mice

We first tested surface expression levels of α4 or α5 integrins throughout erythroid differentiation. In controls, expression of both integrins declined as erythroid cells matured; α4 integrin significantly declined only after the enucleation stage, whereas α5 declined progressively from the proerythroblast level onwards (Fig. 1A top, Suppl. Fig. 1A,B). The mean fluorescence intensity (MFI) of α4 at all levels is higher than α5 (Fig. 1B) and of interest, an α4 upregulation occurs at the basophilic erythroblast level in normal mice (Fig. 1B). This upregulation is of interest especially in the context of recently published data [12]. Specifically it was documented that the activation state of α4 integrin and its binding profile changes during erythroid maturation with the highest activation stage seen in basophilic erythroblasts. At this stage α4β1 forms stable complexes with tetraspanins CD81, CD82, and CD151, which increase the affinity and/or clustering of α4β1 and can bias erythroblast/macrophage interactions vs. erythroblast/fibronectin interactions by introducing different ligand-specific signaling [12].

Fig 1. Steady state erythropoiesis in α4δEry, α5δEry and control mice.

A. Histograms of α4 and α5 integrin expression profiles at various stages of erythroid differentiation in normal (upper row), and α4- or α5-ablated mice (bottom row). Total bone marrow (BM) cells were stained with CD71, Ter119, CD49d or CD49e and CD44 antibodies and analyzed by Fluorescence Activated Cell Sorting (FACS, see Suppl. Fig 1A for gating). Proerythroblasts (r1, red line), basophilic erythroblasts (r2, cyan line), polychromatic erythroblasts (r3, green line), orthochromatic erythroblasts (r4, orange line), reticulocytes (r5, lavender line), mature red blood cells (r6, dark green line) and isotype control (r7, black line). B. Intensity of integrin expression (mean fluorescence intensity, MFI) at various stages of erythroid maturation. Control mice (black lines), n=5, α5δEry mice (red lines), n=5, α4δEry mice (blue lines), n=4. C. Percent of Ter119-positive cells among total BM (left panel) and spleen (right panel) cells, color coding as in 1B. D. Erythroid maturation in the BM (left) and spleen (right). Percentage of cells at each maturation stage was determined by FACS as in A, color coding as in 1A. Control mice, n=6; α4δEry mice, n=5; α5δEry mice, n=4. E. Levels of circulating progenitors in control (n=10), α4δEry (n=11) and α5δEry mice (n=11). Filled bars are nonerythroid (CFU-C) and hatched bars are erythroid (BFU-e). F. Levels of circulating erythroblasts in control (n=28), α4δEry (n=5) and α5δEry mice (n=4). * indicates significant difference over control, p<0.05.

In α4δEry or α5δEry mice, both α4 and α5 integrins were virtually deleted beyond the proerythroblast (r1) level (Fig. 1A & Suppl. Fig. 1A,B). Considering the previously reported four-fold increase of Epo-R levels from late erythroid progenitor cells to the early erythroblast BFU-e progeny [13], this pattern of ablation in our mice likely reflects more efficient Cre activation and consequent integrin ablation beyond the CD71^hiTER119^lo level. These data are consistent with maintenance of functional EpoR in late (hGlycophorinA^hi) erythroblasts [14]. Earlier progenitors, i.e., BFU-e/CFU-e, present in kit⁺CD71^hi populations, were only partially ablated (Suppl. Fig. 1B). Furthermore to directly test whether and to what extent there is deletion of integrins in early progenitors, like BFU-e and CFU-e, we cultured α5⁺ vs. α5⁻ BM cells from α5δEry mice (Suppl. Table 2) in clonogenic media. A 4-fold enrichment of both CFU-e and BFU-e was seen in the α5⁺ cell fraction (Suppl. Table 2, ~51% α5⁺) and large BFU-e were only present in this fraction. Nevertheless in the α5⁻ fraction (94% α5⁻), a proportion of CFU-e/small BFU-e, was present, implying α5-deletion in a proportion of these cells. These data provide useful, updated information about stage-dependent deletion using the EpoRCre mice.

At homeostasis the erythroid populations in α5δEry mice were similar to controls (Fig. 1C,D; Suppl. Fig. 1C,D). However in α4δEry mice modest deficiencies in late erythroid cells were seen in BM, not in spleen, which had abundant erythroid cells (Fig. 1C). Peripheral blood was also notable in the latter mice by a significant increase in circulating erythroblasts, but not in BFU-e or total progenitor cells (Fig. 1E,F), which is a characteristic finding in Tie2Cre⁺α4f/f mice [15]. We attribute the modest WBC increase (Suppl. Table1) and the observed differences between BM and spleen in the α4δEry mice to the release of erythroblasts into circulation and their subsequent accumulation and further maturation in spleen. Significant erythroblast release in PB at steady state was recently reported by conditional deletion of CD169-DTR macrophages [16], but not after clodronate use [17]. Also, in the setting of HO-1 deficiency [18], decrease in α4-integrin expression was seen leading to untested speculation of premature erythroblast release in circulation. However, whether and to what extent macrophage-mediated retention [16] or other specific pathways (i.e., through Fibronectin or other α4β1ligands) are also involved has not been clear and needs to be explored further. Our data document that α4 not α5 integrin has a dominant effect on erythroblast retention but whether a combination of α4 and α5 enhances this effect, consistent with their combined contribution to erythroblast adhesion to fibronectin in vitro [2, 6], is unclear.

In PB a modest hematocrit (Hct) and red blood cell (RBC) reduction was seen with normal reticulocyte levels in α4δEry mice (Suppl. Table1). We interpret the Hct decrease as the result of suboptimal maturation of cells in BM, in which most mature forms were decreased (Fig. 1D), although a compensation by the spleen would have been anticipated. A trend towards a lower Hct with normal reticulocytes was also seen previously in Tie2Cre+α4f/f mice [5, 8].

Response of α4δEry and α5δEry mice to hemolytic stress

To test the response of our mice to erythroid stress, we treated them with PHZ. Kinetic responses to PHZ-induced erythroid stress are shown in Fig. 2A. A transient delay in Hct and reticulocyte increase was detected only in α4δEry mice. In these mice, the level of circulating erythroblasts reached new heights after PHZ-treatment, whereas only very minor increases in either control or α5δEry mice were seen (Fig. 2A). The expression profiles of α4 or α5 integrins in normal mice post stress again showed up-regulation of α4 integrin at the basophilic level (r2, Fig. 2B). In both α4δEry and α5δEry mice twice as many integrin-ablated proerythroblasts (r1) were detected compared to basal levels (Fig. 2B, red lines). This may be the result of increased EpoRCre activation (due to the early hypoxia-induced up-regulation of EpoR level) leading to a higher ablation rate and it is consistent with responses of EpoR depending on the Epo levels in the environment [19]. By contrast, unlike the picture seen at steady state, a much higher proportion of non-ablated α4 or α5 basophilic erythroblasts was seen post stress compared to proerythroblasts (Fig. 2B,C & Suppl. Fig. 2A). We speculate that these non-ablated basophilic erythroblasts are derived directly through accelerated maturation from earlier non-ablated progenitor cells with a sequestered maturation profile. These progenitors reveal themselves here through a detectable change in a surface profile. Accelerated maturation kinetics post-acute erythroid stress have been previously advocated in mice [20, 21] and also postulated in human erythropoiesis in response to acute anemia. A disordered retention of earlier markers in later cells has been the hallmark of the accelerated maturation [22, 23].

Fig 2. Response of α4δEry, α5δEry and control mice to hemolytic stress.

Mice received a single injection of Phenylhydrazine (PHZ ,100 mg/kg body weight) and were either followed for up to two weeks (A) or sacrificed three days after injection, PHZ d4 (B–D). A. Recovery from PHZ-induced stress: hematocrit (left panel), reticulocyte (middle panel) and erythroblast levels in peripheral blood (right panel) in control (n=7), α4δEry (n=4) and α5δEry (n=4) mice. B. Comparison of integrin ablation profiles in α4δEry (left upper and middle panels) and α5δEry (right upper and middle panels) and control mice (lower panels) at steady state and after PHZ treatment. Total BM cells were stained with CD71, Ter119, CD49e or CD49d and CD44 antibodies and cells were gated as shown in Fig. 1A. Proerythroblasts (r1, red line), basophilic erythroblasts (r2, cyan line), polychromatic erythroblasts (r3, green line), orthochromatic erythroblasts (r4, orange line), reticulocytes (r5, lavender line), isotype control (thin black line). C. α4 or α5 integrin expression (percent of positive cells and intensity [MFI] in integrin-positive cells) in BM erythroblasts at various stages of maturation three days after PHZ injection. Control mice, n=9; α4δEry mice, n=9 and α5δEry mice, n=5. D. Percent of Ter119-positive cells among total BM (left panel) and spleen (right panel) cells. In A, C and D: control mice, black lines or bars; α4δEry mice, blue lines or bars; α5δEry mice, red lines or bars; * indicates significant difference over control, p<0.05.

Quantitative erythropoiesis in BM or spleen post stress was also of interest. Differences in late erythroid cells were only seen in α4δEry mice, but unlike the data at homeostasis, these concerned both BM and spleen (Fig. 2D & Suppl. Fig. 2B,C). Such data, especially in view of elevated pretreatment erythroid cell levels in the spleen (Fig. 1C), strongly suggest impairment in optimal terminal erythroid maturation, similar to one noted previously with Tie2Cre+α4f/f mice [5, 8]. The splenic response to PHZ in α5δEry mice in terms of cellularity/weight was also somewhat muted (Fig. 2D & Suppl. Fig. 2C), however, it did not have an impact on the recovery kinetics (Fig. 2A). It is important to emphasize that in α4- or α5-deficient cells, upregulation of the alternative integrin (α5 in α-4 deficient, or α4 in α5-deficient) was not documented (Suppl. Fig. 3).

Taken together, our data provide insightful information about the expression levels of α4 and α5 integrins during erythropoiesis, as well as about stage-dependent and Epo-R-mediated integrin ablation both at baseline and after stress, indirectly uncovering dynamic responses of Epo-R expression at these states. An upregulation of functional α4 integrin at basophilic stages and an indirect demonstration of accelerated erythroid maturation during stress were additional findings of interest. Further, our studies in mice with deletion of α4β1 or α5β1 only in erythroid cells allow for the following secure conclusions. First, α4 integrins have a dominant effect on erythroblast retention and on terminal erythroid maturation both at steady state and post stress. Second, α5 integrin is dispensable for terminal erythroid maturation in contrast to its critical influence exerted alone or in combination with other β1 integrins deleted in all hematopoietic cells [5, 8]. Additional questions of whether a combination of α5 and α4 integrin deficiency only in erythroid cells has complementary effects and whether the signaling pathways downstream of α4 or α5 integrin triggered at each erythroid stage are distinct as we suggested [8], will require further studies.