The inner nuclear membrane protein NEMP1 supports nuclear envelope openings and enucleation of erythroblasts

Nuclear envelope membrane proteins (NEMPs) are a conserved family of nuclear envelope (NE) proteins that reside within the inner nuclear membrane (INM). Even though Nemp1 knockout (KO) mice are overtly normal, they display a pronounced splenomegaly. This phenotype and recent reports describing a requirement for NE openings during erythroblasts terminal maturation led us to examine a potential role for Nemp1 in erythropoiesis. Here, we report that Nemp1 KO mice show peripheral blood defects, anemia in neonates, ineffective erythropoiesis, splenomegaly, and stress erythropoiesis. The erythroid lineage of Nemp1 KO mice is overrepresented until the pronounced apoptosis of polychromatophilic erythroblasts. We show that NEMP1 localizes to the NE of erythroblasts and their progenitors. Mechanistically, we discovered that NEMP1 accumulates into aggregates that localize near or at the edge of NE openings and Nemp1 deficiency leads to a marked decrease of both NE openings and ensuing enucleation. Together, our results for the first time demonstrate that NEMP1 is essential for NE openings and erythropoietic maturation in vivo and provide the first mouse model of defective erythropoiesis directly linked to the loss of an INM protein.


Introduction
Nuclear envelope membrane protein1 (NEMP1, encoded by Tmem194a) is a highly conserved multipass transmembrane protein that resides within the inner nuclear membrane (INM) of the nuclear envelope (NE). We recently showed that genetic inactivation of Nemp1 leads to a loss of fertility in worm, fish, and flies [1]. In mice, Nemp1 is required for female fertility but dispensable for male fertility [1]. Except for the presence of a conserved domain of unknown function (DUF 2215) that encompasses its transmembrane and proximal nucleoplasmic C-terminal region, NEMP1 does not harbor any known functional motif. However, its nucleoplasmic region has been shown to interact with barrier-to-autointegration factor (BAF) and RAN GTPase to mediate Xenopus eye development [2,3]. NEMP orthologs have recently been identified in plants (PNET2a, b, and c) where they play an essential role in chromatin architecture [4]. Using BioID as well as affinity purification followed by mass spectrometry, we recently showed that NEMP1 interacts with LEM domain proteins EMERIN, MAN1, and LAP2, known to physically link the NE to chromatin and support mechanical stiffness. Accordingly, we showed that loss of Nemp1 expression drastically affects NE mechanical stiffness in cultured cells and oocytes [1]. Mammalian erythropoiesis consists of the differentiation of hematopoietic stem cells into megakaryocyte-erythrocyte progenitors (MEPs) that generate burst-forming unit-erythroid (BFU-E) that in turn differentiate into colony-forming unit-erythroid (CFU-E). The latter generate proerythroblasts (ProE) that correspond to the first recognizable erythroid cell. During terminal erythropoiesis, ProE undergo 4 to 5 mitoses that generate basophilic (EryA), polychromatophilic (EryB), and orthochromatic (EryC) erythroblasts. Erythroblast differentiation is characterized by chromatin condensation that is required for enucleation, the ultimate step of erythropoiesis that generates pyrenocytes and reticulocytes [5][6][7][8]. Interestingly, recent studies have established that recurrent NE openings in maturing erythroblasts allow for the partial and selective release of histones in the cytoplasm, a biological process that is essential for chromatin condensation and final enucleation [9][10][11][12]. However, the role of NE proteins in this remarkable biological process remains to be established.
Adult Nemp1 knockout (KO) mice are overtly normal. However, both Nemp1 KO males and females display strikingly enlarged spleens. This phenotype and the involvement of NE openings in terminal erythropoiesis led us to examine the biological function of Nemp1 in erythropoiesis. We show that Nemp1 KO mice display erythroid lineage differentiation defects. Polychromatophilic erythroblasts displayed reduced frequencies of NE openings and of enucleation as well as increased apoptosis, leading to erythroid maturation defects. These data show that NEMP1 supports NE openings and enucleation during the late stages of erythroblast maturation.

Nemp1 KO mice have splenomegaly and abnormal erythropoiesis
Nemp1 KO mice displayed significantly enlarged spleens with increased cellularity compared to heterozygous Nemp1 or wild-type (WT) mice (Fig 1A-1C). Wright-Giemsa staining of Nemp1 KO blood smears showed red blood cells (RBCs) with irregular shapes and spiky membranes ( Fig 1D, red arrows). Complete blood count (CBC) analyses of peripheral blood (PB) using a Hemavet revealed decreased RBC counts in neonates. A decreased hemoglobin (Hb) content persisting throughout life and decreased mean corpuscular hemoglobin (MCH) and mean corpuscular volumes (MCV) appearing at 2 months of age indicated the occurrence of anemia ( Fig 1E). Higher red cell distribution widths (RDWs) values that reflect irregular RBC membrane shapes also persisted throughout life ( Fig 1E). Finally, FACS analysis of PB also showed an increased percentage of nucleated RBCs (nRBCs) in the circulation (Fig 1F). The bone marrow (BM) of Nemp1 KO mice also appeared more densely packed and displayed increased cellularity by comparison to WT BM (Fig 1G).
To understand the nature and origin of erythroid defects, we quantified the erythroid population of BM and spleens from 2-to 4-month-old WT and Nemp1 KO mice by using Ter119 and CD71 markers. Compared to WT BM, Nemp1 KO BM showed increased Ter119+ cell population ( Fig 1H) that was mostly accounted for by a significantly increased population of Ter119+CD71+ erythroblasts (Fig 1I). The increase in erythroblasts was also detected by flow imaging of Hoechst+Ter119+ populations (Fig 1J). In the adult spleen, FACS and flow imaging data showed similar increases in the representation of erythroblast population (Fig 1K-1M). Taken together, these data show that loss of Nemp1 leads to a significant increase of the erythroid lineage in BM and spleens.
To better understand the erythroid differentiation defects in Nemp1 KO mice, we next examined the representation of ProE, EryA, EryB, and EryC erythroblasts using the gating strategy shown in Fig 2E. Consistent with increased MEP population and higher BFU formation capacity in Nemp1 KO BM (Fig 2C and 2D), the ProE and EryA populations were increased, with EryA showing a more significant increase (Fig 2F and 2G). By contrast, the EryB and EryC were decreased, with EryB showing a more significant decrease, in the Nemp1 KO BM. Whereas apoptosis was mildly reduced in EryA, the apoptotic EryB population was significantly higher in Nemp1 KO BM ( Fig 2H). Similar trends were also observed in the spleen (Fig 2I-2K). Collectively, these data indicate that genetic ablation of Nemp1 leads to the expansion of MEPs to EryA populations and to a decrease of EryB and EryC populations. Increased apoptosis in EryB probably accounts for the loss of RBC.
Given the compromised erythropoiesis in the BM, we suspected that spleen enlargement in Nemp1 KO mice might be due to stress erythropoiesis. Accordingly, c-kit+/CD71+/Ter119stress erythroid progenitors (SEPs) (Fig 2L and 2M) and Epo-responsive BFU-E ( Fig 2N) and CFU-E ( Fig 2O) were significantly increased in Nemp1 KO spleen. Splenectomy experiments also pointed to a significant contribution of splenic SEPs to ongoing erythropoiesis in Nemp1 KO mice (S1 Fig and S1 Text). Collectively, these data show that genetic ablation of NEMP1 affects erythroid maturation, especially from EryA to EryB, and elicit stress erythropoiesis.

NEMP1 supports erythroblast NE openings, nuclear compaction, and nuclear extrusion
Using an NEMP1 antibody directed towards a stretch of 15 amino acids from the C-terminal region of NEMP1 (S2A Fig Data are shown as mean ± SD. Student t test. ns: not significant, � p < 0.05. �� p < 0.01. ��� p < 0.001. ���� p < 0.0001. Data underlying the graphs shown can be found in S1 Data. Hb, hemoglobin; KO, knockout; MCH, mean corpuscular hemoglobin; MCV, mean corpuscular volume; PB, peripheral blood; RBC, red blood cell; RDW, red cell distribution width; WT, wild type. CD11b+ myeloid and Ter119+ erythroid cells but not in their KO counterparts ( Fig 3A). In immunofluorescence confocal microscopy, NEMP1 was detected at the NE of ProE, EryA, B, and C erythroblasts where it colocalized with LAP2, a well-established NE marker ( Fig 3B). However, NEMP1 noticeably formed occasional NE puncta of higher intensity whereas LAP2 was homogenously distributed on the NE of erythroblasts ( Fig 3B). NEMP1 was also detected at the NE of cKit+/Ter119-progenitors (S2B Fig). NEMP1 was undetectable at the NE of BM cells isolated from Nemp1 KO mice demonstrating the specificity of the NEMP1 antibody (S2C Fig). Taken together, these results show that NEMP1 is ubiquitously expressed at the NE of the erythroid lineage. Late stages of erythroblast maturation are characterized by the progressive compaction of chromatin and its partial release into the cytoplasm via transient openings of the NE, which is most prominent in the polychromatophilic stage [9,10]. As shown in Fig 3C, NE openings were clearly identified with NEMP1 and LAP2 antibodies in Ter119+ erythroblasts. This phenomenon is distinct from nuclear extrusion as the cytoplasmic membrane labeled with Ter119 remains intact (Fig 3C). 3D reconstruction of confocal Z-stacks clearly emphasized NE openings delineated by NEMP1 through which chromatin protrudes into the cytoplasm (Fig 3D  and S1 Video).
Interestingly, the close examination of successive confocal planes encompassing NE openings revealed the accumulation of NEMP1 into higher intensity aggregates (Fig 3E). Maximum intensity projections also showed that NEMP1 aggregates preferentially accumulated near or close to NE openings (Fig 3E, bottom panel). In contrast, the localization of LAP2 remained uniform and did not accumulate into NEMP1 aggregates. Intensity profiles showed that NEMP1 was 2 to 3 times more abundant in aggregates at NE openings by comparison to intact NE (S2D Fig). To better appreciate the spatial distribution of Nemp1 aggregates relative to NE openings in 3 dimensions (3D), we performed intensity thresholding of confocal slices followed by 3D reconstruction and α-blending. As shown in Fig 3F,  To determine the role of NEMP1 in NE openings, WT and Nemp1 KO BM were immunostained with Lamin B1. Loss of Nemp1 expression led to a significant decrease of NE opening frequencies suggesting a role for NEMP1 in NE openings (Fig 3G). We also observed a significant increase of nuclear size in Nemp1 KO BM cells that may be indicative of decreased chromatin compaction (Fig 3H).
Because chromatin compaction is required for enucleation [9], we next examined whether NEMP1 supports enucleation. As shown in Fig 4A, NEMP1 decorated the NE with occasional puncta at all described stages of enucleation [13]. We used 2 approaches to determine whether genetic ablation of Nemp1 affects nuclear extrusion. First, lineage negative cells from WT and Nemp1 KO mice were purified from BM, cultured for 2 days in erythropoietin-containing medium, and analyzed to distinguish nucleated versus non-nucleated Ter119+ cells. As shown in Fig 4B, although erythroblast differentiation was not affected, the ratio of nucleated erythroblasts (DNA+/Ter119+) versus RBC (DNA-/Ter119+) was increased in cultures derived from Nemp1 KO BM. In agreement with these data and using flow imaging as a second approach, can be found in S2 Data. BFU-E, burst-forming unit-erythroid; BM, bone marrow; CFU-E, colony-forming unit-erythroid; CMP, common myeloid progenitor; GMP, granulocyte-macrophage progenitor; HSPC, hematopoietic stem and progenitor cell; KO, knockout; LT-HSC, long-term hematopoietic stem cell; MEP, megakaryocyte-erythrocyte progenitor; SEP, stress erythroid progenitor; WT, wild type.

Discussion
In this work, we show that loss of the integral transmembrane NE protein NEMP1 results in erythropoietic defects consisting of the expansion of the erythroid lineage in BM and spleens and erythroid maturation defects during terminal erythropoiesis ( Fig 4D). As a consequence, Nemp1 KO mice display anemia and splenomegaly associated with stress erythropoiesis in adult mice.
The higher capacity of Nemp1 KO BM and spleens to generate BFU-E and CFU-E and the significant expansion of ProE and EryA populations together show that erythroid expansion is taking place throughout the EryA stage in Nemp1 KO mice. MEP expansion was notably associated with a significant decrease in GMPs, suggesting that Nemp1 may have an additional biological function in non-erythroid lineages. Alternatively, GMP decrease may reflect compensatory mechanisms in response to erythroid lineage expansion.
We speculate that the overrepresentation of erythroid progenitors and EryA as well as the stress erythropoiesis we observed in the spleen originate from a feedback loop due to the massive apoptotic loss of EryB erythroblasts. Indeed, in contrast to their progenitors and EryA precursors, EryB and EryC populations were markedly decreased in Nemp1 KO BM and RBC counts were significantly lower in the PB of neonate and young Nemp1 KO mice. This decrease of EryB and EryC populations was accompanied by a marked increase of apoptosis that was especially pronounced in EryB from BM. We reason that this apoptosis stems from key biological functions of NEMP1 in NE openings during terminal erythropoiesis. First, Nemp1 transcripts are significantly up-regulated during terminal differentiation [14] and Nemp1 expression level peaks in EryB/polychromatophilic erythroblasts [15]. Second, in erythroblasts undergoing NE openings, NEMP1 accumulates into aggregates that preferentially localized near or at NE openings. Importantly, this aggregation is specific to NEMP1 as LAP2 remains uniformly distributed at NE openings. It is possible that increased levels of NEMP1 expression reported in proteomic screens at that differentiation stage [15] reflects the accumulation of Nemp1 into aggregates. Finally, we directly show that lack of Nemp1 expression in erythroblasts is linked to reduced frequencies of NE openings. Taken together, we propose that the increased apoptosis measured in Nemp1 KO EryB and EryC stems from the requirement of NEMP1 aggregates for efficient NE openings. In support of this idea, inhibition of NE openings is directly linked to induction of cell death in G1ER cells [9].
An alternative but not exclusive origin of elevated apoptosis in EryB erythroblasts may also stem from a biological function of NEMP1 in chromatin organization. Indeed, we recently reported the interaction of NEMP1 with LEM domain-containing proteins EMERIN, LAP2, and MAN1 [1] that are also expressed in erythroblasts [15] and play essential roles in chromatin organization in cooperation with BAF [16]. In addition, the genetic ablation of NEMP   [4] and Nemp1 KO erythroblasts displayed larger nuclear sizes. To that regard, it is interesting to note that the genetic inactivation of LAP2α, a LAP2 isoform devoid of transmembrane domain and involved in the stabilization of higher-order chromatin [17], also leads to the overrepresentation of the erythroid lineage [18]. Impaired nuclear extrusion in Nemp1 KO erythroblasts may also contribute to increased apoptosis levels but future studies are needed to determine whether NEMP1 intrinsically affects nuclear extrusion or if decreased enucleation in Nemp1 KO BM is a mere consequence of impaired NE openings.
We detected the expression of NEMP1 at the NE of erythroblasts and cKit+ progenitors by immunofluorescence microscopy and in purified Ter119+ erythroid cells by immunoblotting which is in agreement with its identification in proteomic analyses of the human erythroid cells [15]. By contrast, the same proteomic analyses show that human erythroblasts do not express Nemp2, a homolog of Nemp1 [15]. Taken together, we conclude that Nemp1 specifically supports the normal homeostasis of the erythroid lineage.
The precise function of NEMP1 aggregates at NE openings requires more investigation. Biochemical studies showing that Nemp1 oligomerizes via its transmembrane domains [3] suggest that NEMP1 may aggregate through multimerization upon increased expression levels detected in transcriptomic and proteomic studies [14,15]. Because we know that NEMP1 supports NE stiffness in the germline and cultured cells [1], it is possible that Nemp1 aggregates mechanically support the NE during the physical stresses of NE openings.
To our knowledge, this is the first report describing the requirement for an integral transmembrane protein of the NE in erythropoiesis. Indeed, pathologies linked to mutations of NE proteins or the nuclear lamina, globally termed "nuclear envelopathies" and "laminopathies," have not previously been associated to erythropoietic defects [19,20]. Similarly to the involvement of multiple NE proteins in specific pathologies, Nemp1 deficiency specifically underlies female infertility and erythropoietic defects in mice. This further fuels the concept of tissue-specific composition of NE proteins whose mutations results in tissue-specific pathologies [21].
Repetitive NE openings during terminal differentiation provide a relatively little known yet powerful opportunity to study NE dynamics and remodeling in vivo. To that regard, Zhao and colleagues [7,9] have shown that NE openings are blocked by inhibition of caspase-3 or through the expression of a caspase-3 non-cleavable Lamin B1 mutant. As a result, histone release from the nucleus, chromatin condensation, and the terminal differentiation of erythroid cells are also affected in vitro. These data and our current findings therefore further stress the biological relevance of NE openings during terminal erythropoiesis. Finally, because caspase-3 KO mice show relatively mild erythroid defects most likely due to in vivo compensatory pathways [9], Nemp1 KO mice provide the first mouse model of acute erythropoietic defects linked to NE openings deficiency. In conclusion, our results uncovered the involvement of Nemp1 in NE openings and enucleation in erythroblasts and its requirement for normal erythropoiesis.

Animals
Animal protocols used in this study strictly adhered to the ethical and sensitive care and use of animals in research and were approved by the Washington University School of Medicine the lack of expression of NEMP1. Blue arrows depict increased apoptosis. See text for details. Data underlying the graphs shown can be found in S4 Data. BM, bone marrow; KO, knockout; WT, wild type. https://doi.org/10.1371/journal.pbio.3001811.g004 Animal Studies Committee (Animal Welfare Insurance Permit #A-3381-01, protocol#21-0206). mNemp1 (Nemp1 em# (TCP) McNeill ) CRISPR KO allele was obtained by CRISPR-Cas9mediated deletion of exon3 that is present in all mNemp1 transcripts (Toronto Center for Phenogenomics). Mice were generated and maintained on a C57Bl6N background [1].

Bone marrow and spleen cells collection
Two-to 4-month-old mice were euthanized by CO 2 inhalation. Legs were separated at the pelvic-hip joint and femurs and tibia cleaned off from tendons and muscle tissues in cold PBS. Bones were cut on their extremities and transferred into collection units consisting of 0.5 ml tubes with pre-perforated (18 gauge needle) bottoms inserted in 1.5 ml collection tubes. Collections units were centrifuged 4 min at 6,000 RPM at 4˚C. Pellets of BM cells were then resuspended in 1 ml of FACS buffer (0.5% BSA and 2 mM EDTA in PBS), transferred to a 40 μm cell strainer and washed with 10 ml of FACS buffer. Cell suspensions were then spinned down for 3 min at 3,000 RPM. Pellets were resuspended in 2 ml of FACS buffer and fixed by adding 3.5 ml of fixation/permeabilization buffer (BD Biosciences) and rocking overnight at 4˚C. Cells were then washed in permeabilization buffer (BD Biosciences) through 3 cycles of centrifugation for 3 min at 850 g and stored at 4˚C for further use.

Flow imaging
One million fixed BM and spleen cells were immunolabeled for 1 h at room temperature with Ter119-PE (1:200, Biolegend) and 1 μg/ml Hoechst 33342 (Thermo Fisher) in permeabilization buffer and then washed in PBS through 3 cycles of centrifugation for 3 min at 850 g. The final pellet was resuspended in 40 μl PBS for image flow analysis. Data were acquired on an AMNIS ImageStreamX multispectral imaging flow cytometer (Luminex) using the Inspire software package. All images were acquired with the 60× objective with Hoechst (405 laser line), Ter119PE (488 laser line), and brightfield imaged in channels 1, 3, and 4, respectively. Laser intensities were adjusted to avoid signal saturation. Single fluorophore labeling were used to build a compensation matrix. Post-acquisition data analyses were performed with the IDEAS software package. For measurements of Ter119+ erythroid populations, cells in focus (gradient RMS) were gated on Hoechst+ cells that were subsequently plotted for Ter119 intensity.

Immunofluorescence confocal microscopy
For immunostaining of intracellular epitopes, fixed BM cells from 2-to 4-month-old mice were washed and permeabilized 3 times in Perm/wash buffer (BD Biosciences) supplemented with 0.1% TritonX100 and further incubated in the same buffer with primary antibodies overnight at 4˚C. After 3 washes, cells were incubated with secondary antibodies and with fluorescently labeled antibodies against extracellular epitopes for 2 h at room temperature and then washed 3 times with 3 cycles of centrifugation for 3 min at 850 g. Cells pellet were resuspended in 20 μl of PBS. Approximately 3 μl of cell suspension were mixed with 10 μl of fluorescence mounting medium (Dako) and mounted for downstream confocal imaging. All images were acquired on a Nikon confocal microscope with a 1.4 NA 100× objective. Images denoising and 3D reconstruction were performed with the NIS-Element software package suite.

Analysis of mouse peripheral blood
Whole blood from neonates, young, and older mice was collected by venipuncture of the facial vein and immediately transferred in blood collection tubes (BD Microtainer). Blood samples were mixed and placed under the Hemavet HV950 probe (Drew Scientific) for analysis using reagents from the LV-PAK (Drew Scientific). Multi-Trol mouse serum controls (Drew Scientific) were used for calibration of the Hemavet HV950. Collected blood was also spread on glass slides for Wright-Giemsa staining according to the manufacturer's protocol (Wright-Giemsa Stain Modified, Sigma-Aldrich).

In vitro colony-forming assay
Methylcellulose colony-forming assay were performed using Epo-only MethoCult 3334 (Stem Cell Technologies) according to the manufacturer's instructions. BM (6 × 10 4 cells/ml) or spleen (1 × 10 5 cells/ml) cells were mixed with M3334 methylcellulose and plated in triplicates using 35-mm Petri dishes. Cultures were maintained in a humidified incubator at 37˚C, 5% CO 2 . CFU-E colonies were counted after 2 to 3 days of culture. BFU-E colonies were counted 5 days after culturing.

Acute anemia and splenectomy
Acute hemolytic anemia was induced in 2-to 4-month-old mice by intraperitoneal injection of phenylhydrazine (PHZ) (Sigma) with a single dose of 100 mg/kg. Splenectomy was performed at the Hope Center for Neurological Disorder at Washington University in St. Louis. Approximately 1 month later, the splenectomized mice were induced by PHZ for hemolytic anemia. Blood samples were collected by venipuncture of the facial vein at different time point and hematologic parameters measured on a Hemavet HV950 complete blood count instrument.

Flow cytometry and cell sorting
BM and spleen cells from WT or Nemp1 KO mice (2 to 4 months old) were dissociated, resuspended in PBS/0.5% BSA and passed through a 40 μm cell strainer to obtain single-cell suspension before antibody staining. Analysis of erythroid maturation using CD71 and Ter119 was conducted as previously described [22]. Freshly isolated cells were stained at 4˚C in PBS/0.5% BSA with purified anti-mouse CD16/32 to block Fc receptors and incubated with PE-Ter119 (TER-119) and APC-CD71 (R17217) antibodies for 30 min at 4˚C. DAPI was used to exclude dead cells from analysis. Where apoptosis was measured, immunostaining for Ter119 and CD71 was followed by a 15-min incubation with FITC-conjugated Annexin-V and propidium iodide following the manufacturer's protocol (BD Biosciences). Flow cytometry was carried out on BD LSR II machine and BD FACSAria II was used for cell sorting. Gate strategy was performed as previously described [22]. The ProE gate contains CD71 high Ter119 intermediate . The Ter119 high cells are further analyzed. Here, CD71 high cells are subdivided into less mature, large EryA erythroblasts (CD71 high Ter119 high FSC high ) and smaller, more mature EryB erythroblasts (CD71 high Ter119 high FSC low ). The most mature erythroblasts subset is EryC (CD71 low Ter119 high FSC low ).