Dysregulation of erythropoiesis and altered erythroblastic NMDA receptor-mediated calcium influx in Lrfn2-deficient mice

LRFN2 encodes a synaptic adhesion-like molecule that physically interacts with N-methyl-D-aspartate (NMDA) receptor 1 and its scaffold proteins. Previous studies in humans and mice have demonstrated its genetic association with neurodevelopmental disorders such as learning deficiency and autism. In this study, we showed that Lrfn2-deficient (KO) mice exhibit abnormalities of erythropoietic systems due to altered NMDA receptor function. In mature Lrfn2 KO male mice, peripheral blood tests showed multilineage abnormalities, including normocytic erythrocythemia, and reduced platelet volume. Colony forming unit assay using bone marrow cells revealed decreases in the counts of erythrocyte progenitors (CFU-E) as well as granulocytes and monocyte progenitors (CFU-GM). Whole bone marrow cell staining showed that serum erythropoietin (EPO) level was decreased and EPO receptor-like immunoreactivity was increased. Flow cytometry analysis of bone marrow cells revealed increased early erythroblast count and increased transferrin receptor expression in late erythroblasts. Further, we found that late erythroblasts in Lrfn2 KO exhibited defective NMDA receptor-mediated calcium influx, which was inhibited by the NMDA receptor antagonist MK801. These results indicate that Lrfn2 has biphasic roles in hematopoiesis and is associated with the functional integrity of NMDA receptors in hematopoietic cells. Furthermore, taken together with previous studies that showed the involvement of NMDA receptors in hematopoiesis, the results of this study indicate that Lrfn2 may regulate erythropoiesis through its regulatory activity on NMDA receptors.


Introduction
Lrfn2 (also known as SALM [synaptic adhesion like molecule] 1) is a member of the Lrfn family proteins, which are known to be cell adhesion molecules that regulate neuronal and synaptic development in the brain [1]. Lrfn2 physically interacts with ion channels forming a glutamate receptor called N-methyl-D-aspartate receptor 1 (NMDAR1, also known as GluN1 or Grin1) [1] as well as a synaptic scaffold protein PSD95 (post synaptic density protein 95, also known as Dlg4) [1][2][3]. Recent  stored in BD microtainer tubes (Beckton, Dickinson and Company, Franklin Lakes, NJ) containing EDTA. Some mice were repeatedly bled at multiple stages. Hematological profiles were determined using a veterinary hematology analyzer (thinka CB-1010, ARKRAY, Kyoto, Japan).

Measurement of serum EPO
Sera (20-60 μL) were prepared from the peripheral blood collected from 12-14 M-old mice by cheek bleeding and stored at -80˚C. The samples were diluted six-fold, and EPO measurement was done using Quantikine ELISA Mouse Erythropoietin (R&D Systems, Minneapolis, MN), according to manufacturer's instructions. All samples were measured in duplicate.

Colony forming unit (CFU) assay
Femurs were isolated from 3-6 M-old mice (BM cells) or 11-12 M-old mice (spleen cells). Cells were plated in methyl cellulose MethoCult medium (M3434, M3234, STEMCELL Technology, Vancouver, Canada), as per the manufacturer's instructions. For BFU-E, CFU-GM, and CFU-GEMM, 1.0×10 4 BM cells or 4.0×10 5 spleen cells obtained from a mouse were cultured in two 35 mm dishes in the presence of SCF, IL-3, IL-6, and EPO (M3434) for 12 or 13 days before observation. For CFU-E, 6.8×10 5 BM or 4.0×10 5 spleen cells obtained from a mouse were cultured in three 35 mm dishes containing MethoCult medium (M3234) supplemented with 10 U/mL rhEPO for 2 days before observation. BM cells were not treated to lyse red blood cells. Images of the entire 35-mm dish field were obtained using the image-stitching function of the BZ-X800 microscope (Keyence, Osaka, Japan). Colonies were counted by an observer who was blinded to the genotypes.

Intracellular calcium concentration measurement
Calcium concentration of BM cells was measured using flow cytometry, according to a method based on the one described by Hanggi et al. [15]. BM cells were isolated from femurs of 6-12 M-old or 19-22 M-old mice and washed with 5% FBS-PBS. Live cell numbers were counted using the TC20 cell counter (Bio-Rad, Hercules, CA) with trypan blue staining. BM cells (1×10 6 ) were suspended in 0.1 mL of erythroid progenitor culture medium [16]

Immunostaining analysis
BM cells were fixed with 4% paraformaldehyde for 20 min at room temperature and the cells were rinsed and suspended in the culture medium. 1×10 6 cells were pelleted in microfuge and resuspended in 0.2 mL of deionized H 2 O. 0.1 mL of the cell suspension was added to a chamber (48 mm 2 ) in 3-chamber cytofuge cassette (StatSpin Cytofuge 2, Nihon Rufuto, Tokyo, Japan) on MAS-coated slide glass (Matsunami Glass, Osaka, Japan), and settled onto glass surface by centrifugation at 1000 rpm for 4 min. The cells were briefly dried on a hot plate at 30 C. The cells were then immersed in PBS, incubated in a blocking buffer (5% donkey serum, 0.1%TX100, PBS) for 45 min at room temperature, and reacted with goat anti mouse EpoR antibody (AF1390, R&D Systems, 1/200 in the blocking buffer) at 4˚C overnight. The primary antibody was detected by Alexa488-conjugated donkey anti-goat IgG (Jackson ImmunoResearch, West Grove, PA). After mounted in VectaShield with DAPI (Vector Laboratories), images were taken by LSM800 confocal microscope (ZEISS, Oberkochen, Germany). Images were analysed by using Image J (https://imagej.nih.gov/ij/index.html).

Statistical analysis
All data were expressed as the mean ± standard deviation (SD). Statistical analyses of all data were performed using two tailed Student's t-test for two group comparisons, unless otherwise noted. Two-way analysis of variance was used to evaluate the effect of MK-801. Percentage values in the results section indicate the percentage of (KO mean value -WT mean value) when WT mean values are defined as 100%. A P value of less than 0.05 was considered statistically significant.

Hematological abnormalities in Lrfn2 KO mice
To address the role of Lrfn2 in hematopoiesis, we first determined the hematological profiles of Lrfn2 KO mice in the 3 M, 6 M, and 12 M old groups in comparison to littermate WT mice. The results (Fig 1, S1 Table) revealed multilineage abnormalities. With regards to erythrocyterelated indexes, erythrocyte numbers ( Fig 1A), hemoglobin (Fig 1B), and hematocrit ( Fig 1C) were increased in both 6 M (11.3%, 11.2%, and 12.0%, respectively) and 12 M (6.4%, 7.7%, and 6.7%, respectively) old groups; moreover, 3 M old mice showed the same tendency (6.7%, 5.3%, and 6.2%, respectively). No significant difference in the mean corpuscular volume was observed among the groups, indicating normocytic erythrocythemia in Lrfn2 KO mice. However, mean corpuscular hemoglobin level in 3 M old Lrfn2 KO mice was slightly (1.5%) reduced in comparison to WT mice. In the context of platelet-related indexes, reduced platelet volumes were observed in both 6 M (-6.5%) and 12 M (-4.7%) old mice ( Fig 1D). Among leukocyte-related indexes, the total leukocyte number was decreased (-21%) in 6 M old mice ( Fig  1E), with reduced counts of granulocytes and lymphocytes (S1 Table).

Colony forming units were reduced in Lrfn2 KO
To investigate whether any relevant phenotypes exist in the BM of Lrfn2 KO mice, we first performed colony forming unit assay. BM cells were cultured in methyl cellulose medium and the resultant colonies were counted. As shown in Fig We also performed colony forming unit assay using spleen cells and found that CFU-E tended to decrease (-40%, P = 0.18) in Lrfn2 KO mice. Both BFU-E cells and CFU-E cells represent two differentiation stages in early erythropoiesis [pro-erythroblasts, early (basophilic) erythroblasts, mid (polychromatic) erythroblasts, and  [17]. A recent study involving transcriptome analysis showed that BFU-E and CFU-E represent CD45 + GPA -IL-3R -CD34 + CD36 -CD71 low cells and CD45 + GPA -IL-3R -CD34 -CD36 + CD71 high cells, respectively, and suggested that erythroid differentiation progresses in the following order: CD34 + hematopoiesis progenitor cells ! BFU-E cells ! CFU-E cells ! pro-erythroblasts [17]. Thus, the colony forming unit assay demonstrated that Lrfn2 KO mice exhibited either reduced counts of BFU-E cells or accelerated differentiation to pro-erythroblast, or both, irrespective of the increased count of mature erythrocytes. In any case, it was thought that early phase of erythropoiesis was impaired in Lrfn2 KO mice.

Serum EPO level was reduced in Lrfn2 KO mice
As erythrocythemia can be secondarily caused by increased EPO levels [18], we measured the serum EPO levels in Lrfn2 KO mice ( Fig 3A) and found that it was significantly decreased (-6.7%, P = 0.022). Therefore, erythrocytosis in Lrfn2 KO mice was considered to be caused by the alteration of a mechanism intrinsic to erythroid progenitors. Interestingly, 10-12 M-old Lrfn2 KO mice showed significant body weight loss (-14%, P = 0.00054) (Fig 3B), which was severer than that observed in 8 weeks-old mice (-6.0%, P = 0.022) [4]. Because splenomegaly is included in the diagnostic criteria of polycythemia vera [19], we measured the weight of the spleen (Fig 3B). In both the 10-12 M-old and 19-27 M-old age groups, the average weights of spleens were comparable between WT and KO mice (Fig 3B). However, the weights of spleen from KO mice were less divergent than from WT mice in the 10-12 M-old age group (P = 9.0 ×10 −5 in f-test) but more divergent in the 19-27 M-old age group (P = 0.041 in f-test) (Fig 3B), indicating hematological dysregulation that is distinct from polycythemia vera. The increased serum EPO level, together with Lrfn2 overexpression [8], led us to hypothesize that Lrfn2 deficiency directly affects erythropoiesis in vivo.

Abnormalities of the late erythropoiesis in Lrfn2 KO mice
We next examined the erythrocyte progenitors in the BM of Lrfn2 KO mice using flow cytometry. Double labeling with CD71, a transferrin receptor, and TER119, a cell surface antigen specific to mature erythroid cells, enabled us to define subsets of erythroid cell lineage [13]. We investigated Lrfn2 KO and WT BM cells from 6-12 M and 18-22 M old mice (Fig 4, S2  Table). In terms of cell counts, the number of early erythroblasts (EryA) in the Lrfn2 KO group was larger (12%, P = 0.0089) than that in the WT group, whereas the number of total erythroid lineage cells (TER119+ cells) was comparable in both groups (Fig 4B, S2 Table). Furthermore, the expression of CD71 in late erythroblasts (EryC) was higher in the Lrfn2 KO group than in the WT group (21%, P = 0.0037) (Fig 4D, S2 Table). Proerythroblast (proE) side scattering (-8.3%, P = 0.060) tended to be lower in Lrfn2 KO (Fig 4C, S2 Table). These results suggested that the late phase of erythropoiesis was affected by the lack of Lrfn2.

Altered NMDA receptor-mediated calcium influx in Lrfn2 KO erythroid progenitors
We further investigated the mechanism by which Lrfn2 controls erythropoiesis. We hypothesized that NMDA receptor was involved in Lrfn2-regulated erythropoiesis because previous studies have demonstrated the NMDAR1-binding ability of Lrfn2 [1] and the essential roles of NMDA receptors in human erythroid precursor cells [15]. To examine the function of NMDA receptor in Lrfn2 KO BM cells, we measured the intracellular Ca 2+ level using the membrane permeable Ca 2+ -sensitive dye Fluo-4 AM. BM cells were loaded with Fluo-4 AM, labeled with TER119 and CD71 antibodies, and stimulated with the NMDA receptor agonists NMDA and glycine (NMDA/GLY) immediately before flow cytometry (Fig 5A).
NMDA receptor-induced changes in Ca 2+ level and the reduction of Ca 2+ level caused by pretreatment with the NMDA receptor non-competitive antagonist MK-801 were measured for two age groups In addition, Ca 2+ analysis uncovered an aging effect on the NMDA receptor functions in mouse erythroid progenitors (Fig 5B). When compared between 6-12 M and 19-22 M WT erythroblasts, NMDA/GLY-induced ΔF/F was found to be lower in the 19-22 M group (mean ± SD, 26 ± 31%) than in the 6-12 M group (mean ± SD, 304 ± 158%, P = 0.0015) whereas the corresponding difference in the KO group was not so clear (P = 0.057).

Discussion
This study revealed the importance of Lrfn2 in the regulation of erythropoiesis. Hemogram and BM cell analyses of Lrfn2 KO mice revealed multilineage abnormalities. In erythroid cells, normocytic erythrocythemia, decreased CFU-E, increased early erythroblast populations, and impaired late erythroblast differentiation were observed. In addition, we observed dysregulation of NMDA receptors involved in erythropoiesis.
Previous studies have elucidated the role of NMDA receptors in erythropoiesis. NMDA receptors have been detected in rat erythrocytes [20] and human erythroid precursor cells [21]. Moreover, NMDA receptors play a dual role in erythropoiesis, supporting survival of mid (EryB) erythroblasts and contributing to the Ca 2+ homeostasis in late (EryC) erythroblasts and circulating erythrocytes [15]. In a recent study using Meg-01 cells (models of human megakaryocytic and erythroid progenitors (MEP)), NMDAR1 KO (Meg-01-GRIN1 -/-) cells underwent atypical differentiation toward erythropoiesis, which is associated with increased basal ER stress and cell death [10]. Thus, NMDAR1 function may be involved in erythropoiesis at multiple stages, from the lineage determination to the late differentiation stage. In addition, calcium signaling, such as EPO signaling [22][23][24], PIEZO1 signaling [25], and erythroid enucleation [26], plays essential roles during erythroid differentiation.
The roles of NMDA receptors in erythropoiesis and binding of Lrfn2 to NMDAR1 [1] led us to hypothesize that the erythropoietic abnormalities in Lrfn2 KO were caused by alterations in NMDA receptor function or calcium signaling (Fig 7). First, the decreased CFU-E may reflect the impaired role of NMDAR1 in MEP [10]. This hypothesis is supported by increased BFU-E in Lrfn2-overexpressing BM cells [8]. Together with the BFU-E decrease tendency in Lrfn2 KO, CFU-E decrease might reflect the impaired differentiation from MEP to BFU-E. Otherwise, CFU-E decrease could be associated with intracellular calcium signaling after stimulation by interleukin-3 and granulocyte-macrophage colony-stimulating factor [27]. Second, increased early erythroblasts (EryA) may reflect impaired differentiation into EryB and/or EryC. Third, increased CD71 level in EryC may reflect impaired differentiation into EryC, because the expression of CD71 on the cell surface decreases during the differentiation of EryB to EryC in murine erythropoiesis [28]. NMDA receptor plays a dual role in the survival of EryB and maintenance of calcium homeostasis during the differentiation of EryC into circulating erythrocytes [15]. Collectively, erythropoietic abnormalities in Lrfn2 KO can be explained measurement. (A) Outline of the assay. (B) %ΔF/F values between no treatment and NMDA/glycine treatment (NMDA-untreated) groups or those between NMDA/glycine treatment and NMDA/glycine treatment groups with MK-801 pretreatment (MK801-NMDA) are shown for total erythroblasts (TER119+), pro-erythroblasts (proE), early erythroblasts (EryA), mid erythroblasts (EryB), and late erythroblasts (EryC). Open bar, WT; closed bar, KO; error bar, SD. Each value from a mouse is indicated by circles. P values were obtained by two tailed t-tests between WT and KO groups. P values (f-test) were obtained by f-test between WT and KO. � , P < 0.05; ��� , P < 0.001 in one tailed t-tests for null hypothesis denote mean values equal to zero.
At the individual level, there was no difference between the numbers of cells in femur and erythroid (TER119 + ) cells in BM. This can explain the counteracting effects during MEP (decreased counts of CFU-E and BFU-E cells) and EryB and EryC stages (apparently increased counts of erythroid cells in BM due to delayed differentiation). However, we did not obtain direct evidence to explain the increased circulating erythrocytes number in Lrfn2 KO mice in this study. Nevertheless, increased EpoR expression in BM cells may be associated with erythrocytosis because EpoR is required to suppress apoptosis in later stages of erythroblast maturation [29]. Moreover, NMDA receptors in mature erythrocytes may affect the life span of erythrocytes. In mature human and rat erythrocytes, NMDA receptors impact cellular nitric oxide synthase activity [9,20]. It is possible that altered NMDA receptor function affects the life span of erythrocytes.
The decreased platelet volume in Lrfn2 KO mice indicates the involvement of Lrfn2 in thrombopoiesis, in addition to its role in erythropoiesis. NMDAR1 is expressed in human BM megakaryocytes [30], and MK-801 inhibits not only proplatelet formation in megakaryocytes but also megakaryocytic differentiation from hematopoietic stem cells [31]. In mice, thrombocytopenia has been described as a phenotype of NMDAR1 KO heterozygotes (Grin1 -/+ ) (Mouse Genome Informatics and the Wellcome Trust Sanger Institute Mouse Genetics Project, http://www.informatics.jax.org/allele/genoview/MGI:5781633). Therefore, the decreased platelet volume may also be explained by the altered function of NMDA receptor.
Although the above discussion focused on NMDA receptors to elucidate the hematopoietic role of Lrfn2, we cannot exclude other possibilities at this point. Considering the low abundance of both Lrfn2 and NMDA receptors in hematopoietic cells in comparison to those of brains, it would be necessary to further address the hematopoietic role of NMDA receptors using genetically modified mice for better understanding. To our knowledge, very few studies have addressed the role of the so-called synapse adhesion molecules in hematopoiesis. It would be interesting to investigate the roles of synaptic molecular complexes in hematopoiesis and physiological regulation of blood cells.