Rats offer a superior model of human stress erythropoiesis

Abstract

Mouse models are widely used to study human erythropoiesis in vivo. One important caveat using mouse models is that mice often develop significant extramedullary erythropoiesis with anemia, which could mask important phenotypes. To overcome this drawback in mice, here we established in vitro and in vivo rat models for the studies of stress erythropoiesis. Using flow cytometry-based assays, we can monitor terminal erythropoiesis in rats during fetal and adult erythropoiesis under steady state and stress conditions. We used this system to test rat erythropoiesis under phenylhydrazine (PHZ)-induced hemolytic stress. In contrast to mice, rats did not show an increased proportion of early stage erythroid precursors during terminal differentiation in the spleen or bone marrow. This could be explained by the abundant bone marrow spaces in rats that allow sufficient erythroid proliferation under stress. Consistently, the extent of splenomegaly in rats after PHZ treatment was significantly lower than that in mice. The level of BMP4, which was significantly increased in mouse spleen after PHZ treatment, remained unchanged in rat spleen. We further demonstrated that the bone marrow c-Kit positive progenitor population underwent a phenotype shift and became more CD71 positive and erythroid-skewed with the expression of maturing erythroid markers under stress in rats and humans. In contrast, the phenotype shift to erythroid-skewed progenitor population in mice was mainly in the spleen. Our study establishes rat in vitro and in vivo erythropoiesis models that are more appropriate and superior for the study of human stress erythropoiesis than mouse models.

Introduction

Mouse models are the most commonly used laboratory animal models to study human erythropoiesis. The identification of several unique cell surface markers on the developing mouse erythroid precursors enables step-by-step cellular and molecular investigations of terminal erythropoiesis from erythroid colony forming units (CFU-E) to mature red blood cells in vitro. Among these markers, CD71 and Ter119 were first used almost two decades ago to characterize terminal erythropoiesis in mouse bone marrow and spleen [1, 2], and later in the fetal liver [3]. About a decade ago, CD44 was introduced to distinguish erythroblasts at different stages of terminal maturation [4]. Based on these two systems, several modified versions of flow cytometry-based assays were subsequently developed [5-7].

In vivo studies of erythropoiesis are also heavily relied on mice. This is not only because of the relatively rapid regeneration and short lifespan of mice, but also the ease of obtaining genetically modified animals. A paradigm for these models is the mouse with a homozygous mutation of the beta-major globin (Hbb^th-1 mice) [8]. Other mouse models with mutations in key erythroid related genes, such as EpoR [9], Gata1 [10], Eklf1 [11], Band-3 [12], Ankyrin [13], and Spectrin [14], are paramount for our understanding of the transcriptional regulation of erythropoiesis and red cell membrane biology.

Although essential tools for the study of human erythropoiesis, mouse models have their own problems. An important caveat using mouse models to study human erythropoiesis is that there is significant splenomegaly with anemia in mice. The spleen is an important organ in mice for adult erythropoiesis, especially under stress conditions [15]. Extramedullary erythropoiesis manifested by splenomegaly compensates for anemia, which often masks important phenotypes in mice. This can be revealed in that splenectomy worsens anemia in several anemia mouse models [16]. Nevertheless, stress erythropoiesis mouse models were developed to help understand the pathophysiological responses in vivo [17]. With these tools, research in the past several decades has revealed important pathways in stress erythropoiesis in mice [18, 19]. Particularly, a spleen specific bone morphogenetic protein-4 (BMP4)-dependent signaling pathway, which is not present in the bone marrow in mice, drives BFU-E stress erythroid progenitors under various stress conditions [20-23]. However, it is unclear whether the same signaling pathway is also present in humans during stress erythropoiesis.

Rats have been used to study erythropoiesis almost a century ago [24]. To a large extent, rats resemble mice in rapid regeneration, short lifespan, and convenient ex vivo manipulation. Many studies indicated that rats are more closely related to humans in which the spleen plays a far less important role in erythropoiesis under steady state or during stress[24-31]. However, the scarcity of genetically modified rat models prevented their widespread use in mechanistic studies in vivo. In addition, lack of modern technologies, especially flow cytometry-based cell sorting capacities, further limited the characterization and investigation of cellular and molecular mechanisms in rat erythropoiesis compared to that in mouse and human.

In this study, we established rat models for the study of in vitro terminal erythropoiesis and in vivo stress erythropoiesis. When compared among rats, mice, and humans, there was a distinct bone marrow c-Kit positive progenitor population that underwent a phenotype shift and became more CD71 positive and erythroid-skewed by expressing maturing erythroid markers under stress in rats and humans. In contrast, the phenotype shift of c-Kit positive progenitors to the erythroid-skewed population in mice mainly occurred in the spleen. We conclude that rat models are more suitable to study human stress erythropoiesis in vivo. Our study provides a valuable resource for the investigation of human stress erythropoiesis using rat models.

Methods

Animal experiments

C57BL/6 mice were purchased from Charles River. ACI rats were purchased from Envigo. Wild-type animals with both males and females were used. For the treatment of phenylhydrazine (PHZ, Sigma-Aldrich, MO, USA), PHZ was dissolved in PBS and injected through the peritoneal route at 40 mg/kg body weight for three days. Equal volume of PBS was injected in the control group. Peripheral blood was collected from mouse or rat tail vein every day and red blood cells (RBCs) were counted by Hemavet 950FS (Drew Scientific, FL, USA). Mice and rats were sacrificed for analyses of the bone marrows and spleens when the RBC level recovered half way to the normal level. To study the stress erythroid progenitors, 100 mg/kg PHZ was injected into rats or mice through the peritoneal route. The animals were sacrificed at 36 hour after treatment. The complete blood count was analyzed by Hemavet 950FS. All animal studies were performed in accordance with the Guidelines for Care and Use of Laboratory Animals and were approved by Institutional Animal Care and Use Committees at Northwestern University.

Flow cytometric assays

For flow cytometric assays in rats, total bone marrow cells were isolated from hind legs of ACI rats of the indicated ages in the main text. Rat fetal livers were isolated from E13.5 rat fetus. The cells were mechanically dissociated by pipetting in phosphate-buffered saline (PBS) containing 5% fetal bovine serum (Gemini, 900-108, CA, USA). Single cell suspensions were prepared by pipetting and passing the tissue through a 40 μm cell strainer (Fisher Scientific, 352340, NH, USA). The following antibodies were used for various stains in the experiments: PE-CD44H (eBioscience, 12-0444-82, CA, USA), eFlour450-CD45 (eBioscience, 48-0461-82, CA, USA), APC-CD71 (eBioscience, 17-0710-82, CA, USA), FITC-HIS49 (eBioscience, 11-9755-82, CA, USA), PE-c-Kit (Santa Cruz Biotechnology, sc-19619 PE, TX, USA) Hoechst33342 (Thermo Fisher, H3570, MA, USA). Propidium iodide was added to exclude dead cells from analysis. The cells were stained for 20 minutes at room temperature in dark and analyzed on an FACS Calibur flow cytometer (BD Biosciences, CA, USA). Post-acquisition analyses were performed using a FlowJo software V9.2.3 (Tree Star, Ashland, OR, MA, USA).

For flow cytometric assays in mice, mouse bone marrow, spleen, or fetal liver single cell suspensions were made as above. The following antibodies were used in various experiments: PE-CD71 (eBioscience, 12-0711-82, CA, USA), FITC-Ter119 (Biolegend, 116206, CA, USA), PE-CD44 (eBioscience, 12-0441-83, CA, USA), FITC-CD45 (Biolegend, 109806, CA, USA), APC-c-Kit (eBioscience, 17-1171-82, CA, USA).

For flow cytometric assays in human, single cell suspensions were prepared from thawed bone marrow aspirate samples. The following antibodies were used: FITC-c-Kit (eBioscience, 11-1178-42, CA, USA), PE-CD235a (eBioscience, 12-9987-82, CA, USA), and APC-CD71 (eBioscience, 17-0719-42, CA, USA).

Cell culture

Rat bone marrow single-cell suspensions were prepared as above. The cells were labeled with biotin-conjugated lineage specific antibodies including HIS49 for erythrocytes (BD bioscience, 550962), HIS48 for granulocytes (eBioscience, 13-0570-82, CA, USA), CD45R for B cells (eBioscience, 13-0460-82, CA, USA), and CD3 for T cells (BD bioscience, 554831, CA, USA) and incubated on ice for 20 minutes. The cells were then washed with PBS and suspended with streptavidin particles (BD Biosciences, 551728, CA, USA) followed by incubation on ice for an additional 20 minutes. Lineage negative cells were purified using EasySep column free cell isolation system according to the manufacturer’s instructions (Stem Cell Technologies, MA, USA). The purified cells were cultured in Iscove’s Modified Dulbecco’s Medium (IMDM, Thermo Fisher, 12440046, MA, USA) containing 15% fetal bovine serum (Stem Cell Technologies, 06200, MA, USA), 1% bovine serum albumin (Stem Cell Technologies, 09300, MA, USA), 200 μg/ml holo-transferrin (Sigma, T0665, MO,USA), 10 μg/ml recombinant human insulin (Sigma I9278, MO,USA), 2 mM L-glutamine (Hyclone SH30034.01, MA, USA), 10-4 M β-mercaptoethanol, and 2 U/ml recombinant human erythropoietin (Amgen, CA, USA). For the culture of fetal liver erythroblasts, total rat fetal liver cells were negatively selected using biotin-conjugated HIS49 (BD bioscience, 550962, CA, USA). The culture media is the same as that of rat bone marrow culture. Mouse bone marrow and fetal liver cell cultures were performed as described previously [32-36].

Cytospin and Benzidine-Wright-Giemsa Stain

Cytospin, benzidine-Wright-Giemsa stains were performed as described previously [7].

Thiazole Orange Stain

Rat bone marrow and peripheral blood single-cell suspensions were purified as above. The cells were stained with PE-CD44H (eBioscience, 12-0444-82), eFlour450-CD45 (eBioscience, 48-0461-82) and l00ng/ml Thiazole Orange (Sigma, 390062) for 20 min in dark. Propidium iodide was added to exclude dead cells. The cells were analyzed on a FACS Calibur flow cytometer (BD Biosciences). Post-acquisition analysis was performed using FlowJo software V9.2.3 (Tree Star, Ashland, OR).

BMP4 expression

To analyze the BMP4 expression under PHZ-induced acute anemia, mice and rats were injected with 100 mg/kg PHZ through the peritoneal route. The animals were then sacrificed at 0, 12, 24, 36, 48 hour and day 4. 3 animals were included at each time point. Total RNAs were extracted from total bone marrow cells and total spleen cells individually using TRIzol (Ambion, MA, USA), and cDNA were reverse transcribed from 1 μg total RNA using qScript cDNA Supermix (Quanta Biosciences, 95048-100, MA, USA). Quantitative RT-PCR was performed using PerfeCTa SYBR Green QPCR FastMix ROX (Quanta Bioscience, 95073-012, MA, USA) according to the manufacturer’s protocol. The primers used for the QPCR were 18s forward, GCAATTATTCCCCATGAACG; 18s reverse, GGCCTCACTAAACCATCCAA; BMP4 forward, TTCCTGGTAACCGAATGCTGA; BMP4 reverse, CCTGAATCTCGGCGACTTTTT. Fluorescence was detected in a QuantStudio™ 3 System and cycle threshold values were calculated by the QuantStudio software. BMP4 expression levels were expressed as the difference in Ct value (ΔCt) of the target gene and the housekeeping gene (18S rRNA, eukaryotic 18S ribosomal RNA) in each sample, and normalized to 18S rRNA expression.

Human Study Approval

Bone marrow aspirate samples from post-transplant patients were obtained following informed consent under Institutional Review Board–approved protocols at Northwestern University, Chicago, Illinois.

RNA sequencing

C-kit+ cells were purified from the bone marrow of hind leg of ACI rats 36 hours after peritoneal injection of 100 mg/kg PHZ or equal PBS, using a Biotin-anti-c-kit (Abcam, ab25022, MA, USA). The total RNA was then isolated using TRIzol (Ambion, MA, USA). The samples were sequenced at BGI (MA, USA) with paired-end 100bp in the BGISeq-500 platform. Each group has 3 replicates and all the sequence data has been deposited in the Gene Expression Omnibus database (accession numbers GSE122792). Pathway analysis was performed using DAVID functional annotation bioinformatic tool.

Statistical analysis

The analysis of results was performed using GraphPad Prism (GraphPad Software). All data were presented as mean ± SD except where indicated otherwise. All comparisons were carried out using the Student t test to assess the significance of the results unless otherwise specified. Statistical significance was established at P<0.05.

Results

Development of flow cytometry-based assays for terminal erythropoiesis in rats

To characterize terminal erythropoiesis in rats, we first purified the total bone marrow cells from 3-month-old rats. The cells were stained with CD71 and HIS49, a mouse IgM antibody targeting maturing rat erythroid cells [37], followed by a flow cytometric analysis. We first depleted cell debris and dead cells using forward scatter (FSC) and side scatter (SSC) combined with propidium iodide (PI) (Figure 1A). The gated live rat bone marrow cells were then divided into four populations based on their expression levels of CD71 and HIS49: Q1 (CD71⁺HIS49⁻); Q2 (CD71⁺HIS49⁺); Q3 (CD71⁻HIS49⁺); and Q4 (CD71⁻HIS49⁻). In parallel, we also stained the cells with Hoechst to reveal the enucleated reticulocytes and mature red blood cells, which showed that approximately 20% of the cells (Q7) were enucleated in rat bone marrow.

Figure 1. Development of a flow cytometry-based system to characterize rat terminal erythropoiesis in the bone marrow.

(A) Total bone marrow cells from 3-month-old ACI rats were labeled with anti-CD71, HIS49, and Hoechst. The cells were first gated by forward scatter (FSC), side scatter (SSC), and propidium iodide (PI). The gated live cells were further characterized for their differentiation based on the expression of CD71 and HIS49, and enucleation based on the percentage of Hoechst negative and HIS49 positive cells. (B) CD45 negative cells gated from live total rat bone marrow cells were separated into 6 subpopulations (I-VI) based on the levels of CD44 and cell size (FSC). (C) Populations I-VI from B were sorted and stained with benzidine and Wright-Giemsa. Representative images of I-VI are shown and represent the following stages of terminal erythropoiesis: I: proerythroblast; II: basophilic erythroblast; III: polychromatic erythroblast; VI: orthochromatic erythroblast; V: reticulocyte; VI: mature red blood cell. Data are representative of three independent experiments.

CD44, an erythroid surface glycoprotein that gradually decreases during terminal erythropoiesis, has been used in mice to differentiate erythroid precursors at various developmental stages [4]. Using the same strategy, we next stained rat bone marrow cells with antibodies against rat CD45 and CD44, followed by a flow cytometric assay. Gated CD45 negative erythroid cells were further grouped into six distinct populations (I to VI) based on their CD44 expression and cell size (Figure 1B). Benzidine and Giemsa stains of the sorted cells from these populations revealed morphologically recognizable erythroid precursors with cells in I to VI representing proerythroblasts, basophilic erythroblasts, polychromatic erythroblasts, orthochromatic erythroblasts, reticulocytes, and mature red blood cells, respectively (Figure 1C). Furthermore, we stained population V and VI with thiazole orange to determine the RNA content in these two populations, which indeed revealed significantly high level of RNA in reticulocytes (Figure 1D). Overall, these studies establish a flow cytometry-based system to monitor terminal erythropoiesis in rats.

In vitro culture of rat bone marrow and fetal liver erythroblasts

To study rat terminal erythropoiesis in vitro, we established a culture system to monitor proliferation and differentiation. We used antibodies against markers of differentiated hematopoietic cells in rats to purify lineage-negative bone marrow hematopoietic stem and progenitor cells. The cells were then cultured in erythropoietin-containing media with transferrin. In the following three days in culture, the cells underwent a significant expansion that reached the highest number at around 72 hour in culture. In comparison, mouse bone marrow lineage negative cells reached maximum proliferation at around 48 hour in culture (Figure 2A), which is consistent with old studies that mouse bone marrow erythroblasts undergo more rapid cell cycles than those in rats [28, 29, 38, 39]. In comparison to rodents, human bone marrow erythroid cells underwent a greatly extended proliferation time and greater potential in expansion (Figure 2B), possibly because the starting cell type (CD34+ cells) is at an earlier developmental stage than those in mice and rats.

Figure 2. In vitro culture of rat bone marrow lineage-negative cell.

(A) Bone marrow lineage negative cells were purified from 6 to 8-week-old wild type C57BL/6 mice (N = 3) and 9-12-week-old ACI rats (N = 3). 1 × 10⁵ purified cells from rats or mice were cultured in erythropoietin-containing media for the indicated amount of time. The cells were counted using a hemocytometer at the indicated time. Data are from three independent experiments. (B) Human bone marrow CD34 positive cells cultured in erythropoietin-containing media for the indicated amount of time. The cells were counted using a hemocytometer at the indicated time. (C) Cultured rat bone marrow lineage negative cells from A were stained with CD71, HIS49, Hoechst, CD45 and CD44 and analyzed by flow cytometry at the indicated time as in Figure 1.

To determine the differentiation and enucleation processes of the cultured lineage negative rat erythroblasts, we performed the same flow cytometric assay as in Figure 1. There was a sharp upregulation of CD71 on day 1 and a gradual gain of HIS49 during culture. CD71 level slightly decreased on day 3 when most of the cells became HIS49 positive. Enucleated cells appeared on day 2 but reached the maximum on day 3 (Figure 2C). The maximum 20-30% of the enucleation rate in cultured rat erythroid cells resembles the well-established mouse culture system [7]. As expected, CD44 gradually decreased during culture, forming six populations of different developmental stages on day 3 (Figure 2C), which is also similar to the mouse system.

We next performed the same experiments in rat fetal liver cells. Rat fetal development follows a similar timeline as that in mouse. On embryonic day 13.5, most of the fetal liver cells in rats were erythroid (CD71⁺HIS49⁺) with over 40% of the cells enucleated (Supplementary Figure S1A). We purified HIS49-negative erythroid precursors and cultured them in the erythropoietin-containing medium as the bone marrow culture system. Rat fetal liver erythroid cells proliferated and differentiated more rapidly than their bone marrow counterparts. The cells reached peak proliferation at 48 hour with approximately 15-fold increase, which is lower than mouse fetal terminal erythroid proliferation (Supplementary Figure S1B). Similar to the bone marrow-derived cells, human fetal-derived cord blood CD34 positive cells showed significantly higher proliferative potential compared to the fetal liver cells in mice and rats (Supplementary Figure S1C). We also analyzed the differentiation and enucleation profiles of in vitro cultured rat fetal liver cells. Like mouse fetal liver in vitro culture, rat fetal liver cells underwent gradual upregulation of CD71 and HIS49. Enucleation reached a maximum rate of approximately 20% at 48 hour in culture (Supplementary Figure S1D).

Bone marrow cellularity in rats is distinct from that in mice but similar to humans

Bone marrow is the primary organ for erythropoiesis in mammals. One of the important differences between mice and humans in the bone marrow histology is that mice are usually hypercellular whereas human bone marrow contains fat tissues leaving spaces to expand hematopoietic cells during stress. We next performed a comprehensive histologic analysis of the bone marrows of various locations in mice and rats. In young mice (2-3 months old), the bone marrow cellularity of most locations (Supplementary Figure S2A-C), except tail bone (Supplementary Figure S2C), was close to 100%. However, the cellularity in the bone marrow of young rats (3-4 months old, comparable life stage to mice) was variable. In the long bones, femur and humerus showed high bone marrow cellularity (~80%). But tibia and radius contained many fat tissues with cellularity close to 50% (Supplementary Figure S2A). The flat bones exhibited a high cellularity (~80%) but still contained a considerable amount of fat (Supplementary Figure S2B). Sternum and vertebrates showed similar cellularity (~50-60%). Tail bone was very hypo-cellular (<5%) (Supplementary Figure S2C). Overall, the bone marrow histology in rats resembles that in humans (Supplementary Figure S2D), which could reflect the similarities in physiology between rats and humans.

Rats and mice are distinct in response to stress erythropoiesis

The culture and flow cytometric assays establish in vitro and ex vivo systems to study rat terminal erythropoiesis. To investigate rat erythropoiesis in vivo under stress conditions, we injected phenylhydrazine (PHZ) peritoneally in rats to induce hemolysis. Mice were treated with the same weight-based dose for comparison. We injected a sub-lethal dose of PHZ (40 mg/kg) every day for three days in rats and mice. Their red cell indices were monitored daily. Rats and mice have a similar RBC count per equal volume of blood at the steady state. After sub-lethal PHZ treatment, the rats exhibited a similar extent of RBC drop compared to mice. However, rats took a longer time to recover (Figure 3A). The rats also showed a more dramatic increase in the mean corpuscular volume (MCV) compared to the mice (Figure 3A right). We sacrificed the animals when they were midway to their fully recovered RBC count, which took five days for mice and seven days for rats after their initial PHZ injection. The choice of this time point is because it is the best time to catch the differences between rats and mice when stress erythroid response is most active. We first examined their spleens given the importance of the spleen in extramedullary erythropoiesis under stress. Compared to the mice, the rats showed significantly less splenomegaly with hemolytic stress (Figure 3B). We reasoned that this could be because of the capability of rat bone marrow to expand in space during stress. Indeed, when we examined the bone marrow cellularity after PHZ treatment, mouse bone marrow remained at 100% as in the steady state whereas rat bone marrow underwent a dramatic increase from ~50% to ~90% (Figure 3C). Because of the lack of space in the bone marrow to expand erythroid population, mice exhibited a dramatic increase in spleen size with a complete replacement of the white pulp by the expanded red pulp. In contrast, although the rat spleen enlarged, the architecture remained relatively intact (Figure 3D).

Figure 3. Rats and mice are distinct during stress erythropoiesis.

(A) 3-4-month-old wild type ACI rats and 2-3-month-old wild type C57BL/6 mice were injected with phenylhydrazine (PHZ) peritoneally at 40 mg/kg daily for three days. Peripheral blood was collected for the analyses of RBC count and MCV daily at the indicated time. The animals were sacrificed for histologic and flow cytometry assays when the RBC levels were recovered to the half of normal level (day 5 for mice and day 7 for rats). N = 5 in each group. (B) The spleens of the sacrificed animals in A were harvested and their weight relative to the body weight were measured. (C-D) Histologic examination of the bone marrows (C) and spleens (D) from animals sacrificed in A. Scale bars: 100 μm. (E-F) Flow cytometric analyses of different stages of terminal erythroblasts based on CD44 and FSC as in Figure 1 in the bone marrow and spleen in mice sacrificed in A. Data are presented as means ± SD. *p <0.05, **p <0.01, ***p <0.001, NS: non-significant. (G-H) Same as E-F except rat bone marrow and spleen cells were analyzed.

To confirm the morphologic findings, we performed a flow cytometric assay on the bone marrow and spleen erythroid cells in mice and rats after PHZ treatment. In mice, the percentages of the early stage erythroid precursors, including proerythroblasts, basophilic erythroblasts, and polychromatic erythroblasts, were significantly increased after PHZ treatment in both the bone marrow and spleen (Figure 3E and 3F). In contrast, these early stage erythroid precursors were not upregulated in proportion in rat bone marrow or spleen after PHZ treatment (Figure 3G and 3H). These results indicate that mice and rats are distinct in their response to hemolytic stress.

Rats and mice contain a distinct stress responsive erythroid progenitor population in the bone marrow

Previous reports demonstrated that mice respond to acute anemia with the expansion of spleen specific stress erythroid progenitors that require bone morphogenetic protein-4 (BMP-4) signaling, which is not found in mouse bone marrow [40]. To determine whether rats also have this population under acute anemia stress, we first treated rats and mice with PHZ (100 mg/kg) and determined BMP4 transcription levels in the spleen and bone marrow at different time points. Consistent with a previous report, mouse spleen cells showed a significantly increased expression of BMP4 transcripts at 36-hour post-treatment [40]. In contrast, BMP4 levels remained unchanged in rat spleen after PHZ treatment (Supplementary Figure S3A). As reported, the increase of BMP4 was not observed in mouse bone marrow post PHZ treatment. BMP4 level remained unchanged in rat bone marrow as well (Supplementary Figure S3B). We also observed the same phenotype in later time points post-PHZ treatment (data not shown). Taken together, these data suggest that BMP4 signaling may not be required in rats under anemia stress.

As previously reported [40], the stress erythroid progenitor population in mouse spleen was characterized by the transition from c-Kit⁺CD71⁻Ter119⁻ (P1) under steady state to c-Kit⁺CD71⁻Ter119⁺ (P3) after hemolytic stress (Figures 4A and 4D). In contrast to mice, the c-Kit positive cells in rat spleen contained mostly CD71⁻ and HIS49⁻ cells (P1) under steady state. There was an increase in P2 (c-Kit⁺CD71⁺HIS49⁺) and P3 (c-Kit⁺CD71⁻HIS49⁺) after PHZ treatment, but the magnitude of changes was far less than that in mouse spleen (Figures 4B and 4D). This is also consistent with the less increase in spleen size and partial intact of the spleen histology after stress in rats (Figures 3B and 3D). In the bone marrow in mice, the major population in the c-Kit positive cells under steady state was CD71 and Ter119 double negative P1 cells, which remained unchanged during stress as reported (Figures 4E and 4H) [40]. Like mice, the major population of c-Kit positive cells in rat bone marrow under steady state was CD71 and HIS49 double negative P1 cells. However, unlike mice, there was a dramatic shift to CD71 and HIS49 double positive P2 cells in rat bone marrow under stress (Figures 4F and 4H). The c-Kit positive population indeed increased proportionally in the bone marrow and spleen in both mice and rats (Figures 4C and 4G). Overall, these results demonstrate that rat bone marrow is the major organ for stress erythropoiesis. These c-Kit⁺/erythroid-skewed stress erythroid progenitors, mostly cells in P2 population in Figure 4, are primarily in the bone marrow in rats, whereas this stress erythroid progenitor population is mainly in the spleen in mice.

Figure 4. Rat bone marrow is the major organ for stress erythropoiesis.

3-month-old wild type ACI rats (N=9) and 2-month-old wild type C57BL/6 mice (N=9) were injected with 100 mg/kg PHZ and sacrificed at 36 hour after treatment. (A-B) Flow cytometric analyses of mouse (A) and rat (B) spleen c-Kit positive cells with indicated markers under PHZ or PBS treatment. (C-D) Quantitative analyses of c-Kit positive cells (C) and the percentages of P1 (CD71⁻Ter119/HIS49⁻), P2 (CD71⁺Ter119/HIS49^med), and P3 (CD71⁻Ter119/HIS49⁺) sub-populations in c-Kit positive cells. (E-F) Same as A-B except the bone marrow cells were analyzed. (G-H) Same as C-D except the bone marrow cells were analyzed. * P <0.05; ** P <0.01; *** P <0.001; NS: non-significant. The data are representative of three independent experiments.

More important, we analyzed the stress erythroid progenitors in adult human patients within 30 days of bone marrow transplantation because of hematologic malignancies. The transplanted donor erythroid progenitors are known to undergo an active proliferation in response to the transplantation stress [41]. We analyzed the c-Kit positive cells of the frozen bone marrow aspirate from bone marrow transplant patients and compared to samples from normal adult bone marrow. Under the normal steady state situation, the main subpopulation in c-Kit positive cells were CD71 and CD235a (glycophorin A) double negative cells (P1) and CD71⁺CD235a⁻ (P2) sub-populations, which is similar to the composition of c-Kit positive cells in rat bone marrow (Figure 5A). Under bone marrow transplantation stress, the cell composition shifted with a significant increase in P2 and a decrease in P1 (Figure 5B-C), which is also similar to rat bone marrow during stress erythropoiesis.

Figure 5. Increase of c-Kit⁺/erythroid skewed population in human bone marrow during stress erythropoiesis.

(A) Flow cytometric analysis of bone marrow aspirates obtained from normal (lymphoma staging negative) and bone marrow transplanted patients. C-Kit positive cells were gated for the analysis of CD71 and CD235a expression levels. (B) Quantitative analysis of different subpopulations of c-Kit positive cells in A. N = 4 in each group. BMT: bone marrow transplantation. ** P <0.01; NS: non-significant.

Transcriptomic profile of c-kit+ cells in rat bone marrow under stress

We next performed an RNA-sequencing analysis of the c-Kit positive cells under stress to identify novel factors and pathways involved in rat stress erythropoiesis. Pairwise comparisons between PBS- and PHZ-treated c-Kit positive cells in rat bone marrow revealed high Pearson correlation coefficients among replicates, range from 0.96 to 0.99 (data not shown). The differentially expressed genes were analyzed and shown in a volcanic map (Figure 6A). The upregulated genes include those important for cell cycle, metabolism, DNA replication, and RNA transport. The downregulated genes are involved in several immune-related pathways (Figure 6B). Among these genes, Fgf23 and GPBP1L1 are highly upregulated in the c-Kit positive cells from PHZ-treated rats (Figure 6C). These highly upregulated genes could be potentially significant in bone marrow responses to stress erythropoiesis in rats, which requires future investigations.

Figure 6. Transcriptome analysis of c-kit+ cells from rats under stress erythropoiesis.

ACI rats were injected with PHZ at 100 mg/kg or PBS and sacrificed 36 hours after injection. Bone marrow c-kit+ cells from these mice were purified for RNA sequencing analyses. Each group contains three samples. (A) Volcano plot of differentially expressed genes. X axis represents log2 transformed fold change. Y axis represents −log 10 transformed significance. Red and blue points represent up- and down-regulated differentially expressed genes, respectively. Gray points represent non-differentially expressed genes. (B) Pathway functional enrichment of differentially expressed genes. Top up- (red) and down- (blue) regulated pathways are listed. (C) Top 10 up-regulated (red) and top 10 down-regulated genes (blue) in c-kit+ cell from ACI rat bone marrow after PHZ injection.

Discussion

Our study establishes rat in vitro erythroid cell culture and in vivo stress erythropoiesis models through which we can monitor the differentiation, proliferation, and enucleation of rat erythroid cells step by step in different hematopoietic organs. When compared to the cultured mouse erythroid cells, rats were found to share several similarities during steady state erythropoiesis to mice. The bone marrow and fetal liver erythroid differentiation in rodents showed rapid kinetics compared to that in humans. The enucleation rate in culture was also lower in both rats and mice when compared to that in humans. These differences between rodents and humans can be intrinsic but could also be related to the distinctions in the differentiation stages of the starting cell types in culture.

The distinctions in erythropoiesis between rats and mice are also significant, specifically for stress erythropoiesis. There are notably three major differences in rat stress erythropoiesis compared to mice at a comparable age. Importantly, these features in rats are shared in stress erythropoiesis in humans: 1) The low bone marrow cellularity in rats provides expandable spaces for the proliferation of erythroid cells during stress. 2) Under stress conditions, extramedullary erythropoiesis plays a less important role in rats. 3) More important, the bone marrow c-Kit positive stress progenitor population undergoes a phenotype shift and becomes more CD71 positive and erythroid-skewed under stress in rats and humans. In contrast, the phenotype shift to the erythroid-skewed stress progenitor population in mice occurs mainly in the spleen (Figure 7).

Figure 7. Schematic diagrams of the c-Kit positive erythroid progenitor populations during steady state and stress erythropoiesis in rat, human, and mouse.

Rat and human (left) primarily use the bone marrow for stress erythropoiesis, whereas mouse (right) uses the spleen.

We propose that the similarities between rats and humans during stress erythropoiesis are mainly due to the expandable bone marrow spaces in both species (Figure 3). This is supported by the PHZ treatment experiments in rats (Figure 3C) and well-documented bone marrow hypercellularity with erythroid hyperplasia in humans during hemolytic anemia [42]. Furthermore, studies more than two decades ago using a splenectomy approach showed that splenic erythropoiesis was negligible for the recovery from anemia in rats [43]. In contrast, splenectomy in mice in an infection-induced anemia model showed worsened anemia after the surgery [16], which is consistent with the critical roles of the spleen in stress erythropoiesis in mice. Indeed, a unique BMP4-dependent stress erythroid progenitor population in mouse spleen was reported that undergoes a transition from c-Kit⁺CD71⁻Ter119⁻ (P1) to c-Kit⁺CD71⁻Ter119⁺ (P3) under stress conditions [40]. However, this population was not identified in the rat spleen in our study. In contrast, the phenotype shift in the c-Kit positive stress erythroid progenitor population in rats, from CD71 and HIS49 double negative to double positive, occurs mainly in the bone marrow (Figures 4 and 7).

The properties of stress response in rats closely mimic those in humans. Our analyses of patients with bone marrow transplantation revealed that the c-Kit positive progenitor population showed a significant increase of the CD71 levels. Consistent with our findings, a previous report revealed that CD34 positive cells in patients with sickle cell anemia showed a strong upregulation of CD71 and CD235a [44]. In addition, it was observed that splenectomy before allogeneic hematopoietic stem cell transplantation in patients with myelofibrosis showed an improved overall survival and event-free survival [45], indicating a negligible role of the spleen during stress hematopoiesis. Together the information strongly supports that the bone marrow is the major organ in humans for stress erythropoiesis.

Our study demonstrates that rat models are more appropriate and superior to study human stress erythropoiesis than the mouse models. Given the similarities between rats and humans in response to anemia, novel pathways or genes involved in stress erythropoiesis could be shared in these two species. Our gene expression profiling study reveals genes that are upregulated in stress erythropoiesis in rat bone marrow. Future functional studies will provide important information regarding the mechanisms of these genes during stress erythropoiesis. Technology advances in CRISPR-Cas9-mediate genome editing system would also make it easy to generate genetically modified rat models for in vivo studies of these genes in rat during stress erythropoiesis.

Highlights.

• We establish in vitro and in vivo rat models to monitor erythropoiesis under steady-state and stress conditions.

• Compared to mice, rats and humans stress erythropoiesis occurs primarily in the bone marrow.