The Endoplasmic Reticulum Chaperone Protein GRP94 Is Required for Maintaining Hematopoietic Stem Cell Interactions with the Adult Bone Marrow Niche

Hematopoietic stem cell (HSC) homeostasis in the adult bone marrow (BM) is regulated by both intrinsic gene expression products and interactions with extrinsic factors in the HSC niche. GRP94, an endoplasmic reticulum chaperone, has been reported to be essential for the expression of specific integrins and to selectively regulate early T and B lymphopoiesis. In GRP94 deficient BM chimeras, multipotent hematopoietic progenitors persisted and even increased, however, the mechanism is not well understood. Here we employed a conditional knockout (KO) strategy to acutely eliminate GRP94 in the hematopoietic system. We observed an increase in HSCs and granulocyte-monocyte progenitors in the Grp94 KO BM, correlating with an increased number of colony forming units. Cell cycle analysis revealed that a loss of quiescence and an increase in proliferation led to an increase in Grp94 KO HSCs. This expansion of the HSC pool can be attributed to the impaired interaction of HSCs with the niche, evidenced by enhanced HSC mobilization and severely compromised homing and lodging ability of primitive hematopoietic cells. Transplanting wild-type (WT) hematopoietic cells into a GRP94 null microenvironment yielded a normal hematology profile and comparable numbers of HSCs as compared to WT control, suggesting that GRP94 in HSCs, but not niche cells, is required for maintaining HSC homeostasis. Investigating this, we further determined that there was a near complete loss of integrin α4 expression on the cell surface of Grp94 KO HSCs, which showed impaired binding with fibronectin, an extracellular matrix molecule known to play a role in mediating HSC-niche interactions. Furthermore, the Grp94 KO mice displayed altered myeloid and lymphoid differentiation. Collectively, our studies establish GRP94 as a novel cell intrinsic factor required to maintain the interaction of HSCs with their niche, and thus regulate their physiology.


Introduction
In the adult hematopoietic system, hematopoietic stem cell (HSC) regulation of self-renewal and differentiation is at both the intrinsic and extrinsic level, and is tightly regulated under physiological conditions [1,2]. HSCs reside in a specific anatomic location in the bone marrow (BM) known as the stem cell niche [3,4]. Signaling cues from neighboring cells in the niche are key in dictating the function of the cell to maintain the hematopoietic system of the individual [5][6][7]. At the endosteal surface [8], which is the interface between bone and the BM, specialized osteoblasts represent the main components of HSC niche [9,10]. It has been proposed that this heterogeneous group of cells regulates HSC survival, selfrenewal, migration, differentiation, and quiescence [11,12] through several soluble factors and their receptors such as angiopoietin/Tie2 [13], osteopontin [14,15] and Ca 2+ /CaR [16], as well as direct contact through extracellular matrix and cell surface proteins [17,18], such as integrins [19,20]. HSC engraftment is a multistep process involving homing, transmarrow migration, and lodging within the BM niche [15], all of which is controlled by adhesion molecules, soluble ligands and their receptors [3].
It is also clear that the decision of the HSC to self-renew or differentiate is dependent upon the extrinsic signaling mechanisms controlling the expression of intrinsic determinants of HSC function. Previous studies have determined that a number of cell cycle regulators [1,21] such as the early G1-phase checkpoint regulator p18 INK4C [22], and the later G1-phase checkpoint regulator P21 cip1/waf1 [23], are critical to maintain HSC quiescence. Retinoblastoma family protein [24] and PTEN [25,26] also play crucial roles in maintaining HSC homeostasis. Transcription factors such as Zinc-finger repressor Gfi1 [27,28] and chromatin-associated factors like Bmi1 [29] have been implicated as key regulators in maintaining HSC self-renewal.
Glucose-regulated protein 94 (GRP94; also referred to as gp96, CaBP4, endoplasmin, Tra-1) is a metazoan-restricted member of the HSP90 protein family. It is traditionally regarded as an endoplasmic reticulum (ER) localized chaperone assisting in protein folding, assembly and secretion [30]. Due to its client selectivity and interactions with late folding intermediates, GRP94 is postulated to perform unique chaperone functions in the ER, and control specific pathways critical for cell growth and differentiation [31]. In Drosophila, Gp93 (ortholog of GRP94) is required for gut epithelial homeostasis and nutrient assimilationcoupled growth control, suggesting an essential role in the functional expression of specific secretory/integral membrane proteins in tissue specialization [32]. Loss of GRP94 function in mouse models revealed that it is required for mouse embryonic development [33] and plays an important role in immune functions [34][35][36]. For example, GRP94 is required for the expression of a large number of integrins, as well as Toll-like receptors and selectively regulates innate immunity of macrophages [37,38], and early T and B lymphopoiesis [39,40]. Furthermore, using a BM chimeric approach, an increase in HSCs and multipotent progenitors was observed in BM devoid of GRP94 [40], raising the interesting issue of how GRP94 deficiency could lead to HSC expansion.
Recently, we described the creation of a floxed mouse model of GRP94 targeting the exon 2 of the Grp94 allele (Grp94 F/F ) for examining GRP94 function in specific tissues [41,42]. We report here that by breeding the Grp94 F/F mice with the transgenic mice expressing Mx-1 promoted Cre recombinase [43], we generated Grp94 F/F ; Mx-1-Cre (referred to below as cKO) mice that allows inducible, acute homozygous knockout (KO) of GRP94 in the hematopoietic system through injection of polyinosinic-polycytidylic acid (pI.pC). GRP94 deficiency results in a striking 2-fold expansion of the HSC-enriched Lin 2 c-Kit + Sca-1 + (LSK) cell population (p,0.001) with increases in both the long-term (LT) and short-term (ST) HSC populations. Cell cycle analysis showed loss of quiescence and lodgment assays revealed impaired HSCniche interaction as a major contributing factor for this increase. Through transplantation assays we further determined that this is a HSC cell autonomous effect, which can be mediated by the loss of integrin a4 expression on the cell surface of Grp94 KO HSCs, leading to impaired HSC retention in the niche. Collectively, our studies establish GRP94 as a novel cell intrinsic factor required to maintain the interaction of HSCs with their niche, and thus regulate their physiology.

Results
Creation of a mouse model with inducible knockout of Grp94 in the hematopoietic system To study the in vivo function of GRP94, we created Grp94 mutant mice with the floxed or knockout allele ( Figure 1A). Deletion of exon 2 flanked by the loxP-FRT sites led to an early frameshift mutation resulting in the inactivation of the Grp94 allele.
To determine the role of GRP94 in the hematopoietic system, we bred Grp94 F/F mice with transgenic mice bearing the Mx-1-Cre transgene that allows GRP94 to be acutely eliminated in the hematopoietic system in an inducible manner by administration of pI.pC. In these studies, Grp94 was deleted in 5.5 to 6.5 week old Grp94 F/F ; Mx-1-Cre (referred to below as cKO) mice. Littermates lacking the Cre transgene (Grp94 F/F ), which are phenotypically equivalent to animals with wild-type (WT) Grp94 alleles, were also injected with pI.pC and served as controls for any side effects of pI.pC injection. These control mouse cohorts are referred to below as WT mice. The status of the Grp94 allele deletion was validated by PCR in isolated BM cells ( Figure 1B). Real-time quantitative PCR analysis with mouse BM cells 11 days after 7 shots of pI.pC injection showed that Grp94 transcripts were mostly depleted from the BM cells of the cKO mice ( Figure 1C).

GRP94 deficiency expands primitive cell pool through loss of quiescence
In analyzing the effect of acute Grp94 deletion on primitive hematopoietic cells, we observed that within 3 weeks of GRP94 depletion there was an expansion of the most primitive Lin 2 c-Kit + Sca-1 + Flk2 2 CD34 2 (LSKFlk2 2 CD34 2 ) LT-HSC-enriched cell population [44] (2-fold, p,0.01) in the BM. Correlating with this was an expansion of the ST-HSC enriched LSKFlk2 2 CD34 + cells, MPP-enriched LSKFlk2 + and LSK populations of cells (p,0.001) (Figure 2A and 2B), indicating that all primitive hematopoietic cells were significantly increased upon acute hematopoietic Grp94 deletion. These findings, coupled with the observation that the total BM cell number was not significantly different in the mutant mice (p = 0.373) ( Figure 2C), reveal that the total number of primitive hematopoietic cells in cKO mice BM is higher compared to controls. We also analyzed the progenitor cell populations in the BM. While the common myeloid progenitor (CMP) enriched Lin 2 c-Kit + Sca-1 2 IL-7Ra 2 CD34 + FccII/ IIIR lo cells, the common lymphoid progenitor (CLP) enriched Lin 2 c-Kit lo Sca-1 lo IL-7Ra + cells, and the megakaryocyteerythroid progenitor (MEP) enriched Lin 2 c-Kit + Sca-1 2 IL-7Ra 2 CD34 2 FccII/IIIR lo cells were all comparable between the cKO mice and their WT littermates, there were significantly more granulocyte-monocyte progenitor (GMP) enriched Lin 2 c-Kit + Sca-1 2 IL-7Ra 2 CD34 + FccII/IIIR high cells in the cKO mice ( Figure 3A). This correlates with the increased number of colonies (p,0.01 for total colony number; p,0.05 for CFU-G and CFU-GM; p = 0.07 for CFU-M) formed by unfractioned cKO BM cells in CFC assays ( Figure 3B).
To examine the underlying cause for the increase in primitive hematopoietic cells in the BM of the cKO mice, we analyzed the cell cycle distribution in LSK cells by Hoechst/Pyronin Y staining   Figure 4A). While approximately 75% of the LSK cells in both genotypes were in G1 phase, the proportion of cells in G0 decreased from 15% in the control to 7.7% in the cKO mice (p,0.001), whereas the amount of S+G2+M cells increased from 10% in the control to 17% for cKO mice (p,0.01) ( Figure 4B). Furthermore, the expansion of the Grp94 KO primitive hematopoietic cell pool was not a consequence of altered cell survival as analysis of the level of apoptotic cells using Annexin V did not reveal significant differences between WT and cKO LSK cells (p = 0.324) ( Figure 4C). These findings suggest a loss of quiescence and increase in proliferation account at least in part for the primitive hematopoietic cell expansion observed in cKO mice, similar to what has been shown using other models of HSC function [45].

Grp94 KO HSCs display increased mobilization and impaired HSC-niche interaction
We next examined whether the alteration in the cell cycle status of the primitive hematopoietic cells was accompanied by any changes in their localization [46]. Histological analysis showed that the spleen size was significantly enlarged in cKO mice ( Figure 5A and 5B) with signs of increased extramedullary hematopoiesis as indicated by the increased hematopoietic cells in the red pulps ( Figure 5C). Flow cytometric analysis showed that there was a significant increase (2.5-fold, p,0.001) in the percentage of LSK cells in the spleen of cKO mice compared to WT mice ( Figure 6A). The number of circulating LSK cells in the peripheral blood of cKO mice were also greatly increased (17-fold, p,0.001) compared to control ( Figure 6B). We also observed increased Lin 2 Sca-1 2 c-Kit + progenitor cells in the peripheral circulation. These results suggest that primitive cell mobilization was increased as a consequence of Grp94 deletion.
Since the HSC niche has been proposed to maintain stem cells in the BM in a quiescent state, we hypothesized that the increase in proliferation of the Grp94 KO HSCs may have been mediated by a loss of interaction with the niche, which thus also led to increased HSC mobilization and enlargement of the spleen by extramedullary hematopoiesis in cKO mice. Therefore, we directly examined the capability of Grp94 KO progenitors to home and lodge in the endosteal niche. To do this, we performed a competitive lodgment assay ( Figure 7A), in which the lodging of primitive cells is dependent upon their functional interaction with the niche in vivo. An equal number of WT and cKO Lin 2 c-Kit + (LK) cells were labeled with fluorescent vital dyes CFSE and SNARF respectively, and co-injected into non-irradiated WT littermates. In these experiments we used LK cells, as in agreement with previous studies, the use of a purified HSC population would result in too few cells to be reliably detected [47]. We observed that fewer labeled cells from mutant mice homed to BM, while more of them migrated to the spleen of recipient mice (p,0.05) ( Figure 7B and 7C). We cut a total of 180 sections of femurs and tibias from a total of three recipient mice in three independent experiments, and visualized the localization of cells as shown in Figure 7D. The endosteal region is defined as the area within 2 cell diameters from endosteal surface. Among all the 400-1000 cells that homed to the BM (numbers varied each time according to the number of cells injected), about 13.5% of WT LK cells lodged at the endosteal region, while only 5% of the cKO cells that homed to the BM were found at the endosteal surface (p,0.05) ( Figure 7E). These findings suggest that acute deletion of Grp94 in primitive hematopoietic cells impairs the ability of HSCs to home to BM and lodge at the endosteal niche. As a consequence, in the competitive repopulation assay ( Figure 8A) cKO LSK cells failed to engraft and reconstitute the hematopoietic system in lethally irradiated recipient mice in the presence of WT competitors ( Figure 8B and 8C).

GRP94 in HSCs, but not niche cells, is necessary for maintaining HSC homeostasis
In the competitive repopulation and competitive lodgment assay, where cKO cells are transplanted into a WT microenvironment, we observed impairment in the function of the cells indicating that Grp94 deletion in HSCs at least partially contributes to the disruption of the HSC-niche interaction. However, in our model system Grp94 was knocked out in the whole BM, therefore the disruption of HSC-niche interactions could also be attributed to the niche cells. To further explore a possible extrinsic role of GRP94 null niche cells in HSC homeostasis disturbance in cKO mice, we created a chimeric mouse model with WT hematopoietic cells and a GRP94 null microenvironment. WT BM cells were transplanted into lethally irradiated WT or cKO recipient mice. Following 8 weeks to allow for full reconstitution of WT hematopoietic system, Grp94 deletion was induced by pI.pC administration ( Figure 9A). According to previous reports, BM stromal cells do not engraft after BM transplantation [48], therefore, Grp94 KO only occurred in the hematopoietic cells. Chimeras with WT hematopoietic cells and cKO microenvironment showed a normal hematology profile and spleen size as WT control chimeras ( Figure 9B and 9C). Flow cytometric analysis revealed a similar LSK percentage in BM from cKO chimeras and WT controls ( Figure 9D and 9E). These findings provide direct evidence that GRP94 in HSCs is required for their interaction with the niche, and depleting GRP94 in BM microenvironment itself is not sufficient to affect the regulation of HSCs.
Grp94 KO HSCs lack cell surface integrin a4 expression and exhibit impaired binding to fibronectin To investigate the potential mechanism by which GRP94 maintains the interaction between HSCs and the niche, we examined the cell surface expression of integrin a4 and a5, which are known to be important for the homing and adhesion of HSCs to the endosteal niche [49,50]. Flow cytometric analysis with mouse BM LSKFlk2 2 and LSKFlk2 + populations demonstrated significantly lower expression of cell surface CD49d (integrin a4/ b1) on cKO than WT cells, whereas CD49e (integrin a5/b1) expression was comparable between the two groups ( Figure 10A). These results establish that GRP94 is required for the expression of integrin a4 on the cell surface of hematopoietic stem and progenitor cells, which is consistent with the recent finding that loss of GRP94 abrogates integrin a4 but not a5 expression on BM cells [40].
To test specific interactions of the primitive hematopoietic cells with adhesion molecules known to be present at the endosteal surface and play a role in mediating HSC-niche interactions, we examined the ability of the cells to bind to fibronectin in vitro, the binding partner of integrin a4, with binding to BSA as a negative control. While WT LSK cells were able to bind to fibronectin, we did not detect binding of the cKO cells to fibronectin above the negative control level ( Figure 10B). Collectively, these findings demonstrate that acute depletion of GRP94 in HSCs disrupts the interaction of HSCs with niche, possibly through the disruption of adhesion to key extracellular matrix molecules mediated by integrin a4.
Deletion of Grp94 in the hematopoietic system leads to altered myeloid and lymphoid differentiation To determine the consequence of acute depletion of GRP94 following Mx-1-Cre induction, we examined the hematopoietic phenotypes of cKO mice compared to control siblings. Complete blood count analysis demonstrated a general leukocytosis and severe thrombocytopenia, while red blood cell and hemoglobin counts were normal. Monocytes and granulocytes in peripheral blood were increased 8-fold and 10-fold, respectively, in cKO mice compared to control mice ( Figure 11A), which is consistent with the increased GMP in the cKO BM ( Figure 3A). Analysis of blood smears further confirmed the leukocytosis in peripheral blood ( Figure 11B). Compared with WT littermates, cKO mice showed significantly lower thymus and lymph node cellularities ( Figure 11C), consistent with smaller thymus and peripheral lymph nodes (data not shown). The increase in myeloid cells and decrease in lymphoid cells corresponded with a higher proportion of granulocytes and macrophages and a lower proportion of Tcells and B-cells in the spleen of the cKO mice ( Figure 11D and E). The expansion of myeloid cells was also evident in the BM, where Gr-1 + cells were increased and B220 + cells were decreased in cKO mice. The proportion of CD4 + and CD8 + cells in the BM of cKO mice were comparable to the control mice ( Figure 11F and G).

Discussion
GRP94 is the most abundant glycoprotein in the ER and is emerging as an important player in protein processing. Its unique interaction with specific client proteins may allow it to specifically regulate cellular processes distinct from other major chaperones in the ER. In particular, recent model studies suggested that GRP94 is critically required for the functional expression of secretory and/ or membrane proteins that enable the integration of cells into tissues [32]. Here we examined the role of GRP94 in the HSC interactions with the adult BM niche. The model system that we Figure 6. Increased mobilization of Grp94 KO HSCs. A) Representative flow cytometric analysis of splenocytes from WT (n = 9) and cKO (n = 9) mice using Lin, c-Kit and Sca-1 (left) and quantitation (right). In these studies, splenocytes were extracted from WT and cKO mice 5 days after 7 injections of pI.pC (to examine HSC mobilization before the spleen was enlarged). B) Representative flow cytometric analysis of blood MNCs from WT (n = 11) and cKO (n = 10) mice using Lin, c-Kit and Sca-1 (left) and quantitation (right). All data are presented as mean 6 s.e., ***p,0.001. doi:10.1371/journal.pone.0020364.g006 employed is the Grp94; Mx-1-Cre mouse model whereby GRP94 can be acutely eliminated largely in the hematopoietic system in a temporally controlled manner in adult mice. Our results showed that within 3 weeks of depletion of GRP94 in the hematopoietic cells, there was a 2-fold increase in primitive HSCs in the mutant mice. The Grp94 KO LSK cells are more proliferative, likely as a consequence of loss of niche interaction. The latter is supported by our observation that the mutant LSK cells are more mobilized to the spleen and peripheral blood. The mutant LSK cells lack cell surface integrin a4 expression and do not bind to fibronectin as efficiently as the WT LSK cells in vitro, are less able to lodge in the BM niche in vivo, and as a consequence showed no engraftment in competitive repopulation assays.
The maintenance of HSCs involves the balance between selfrenewal and differentiation, proliferation and quiescence, mobilization and homing, as well as their capacity to provide lifelong reconstitution of all blood-cell lineages after transplantation into lethally irradiated recipients [3,12]. Our study demonstrates the unique function of GRP94 in regulating the homeostasis of HSCs and their interaction with the stem cell niche. Currently, the regulation of HSC physiology is mostly attributed to transcription factors, cell-cycle regulators, chromatin-associated proteins, extracellular signaling molecules and their receptors [1]. Here we explore the notion that an ER chaperone protein may play a role in the regulation of HSC-niche interactions. The HSC niche is believed to house HSCs, maintain their quiescence and prevent PLoS ONE | www.plosone.org them from differentiation [3]. Here we determined that Grp94 KO primitive cells are less efficient in contacting with BM niche, with the consequence that the mutant LSK cells can exit quiescence and enter cell cycle. While the precise mechanisms by which GRP94 affects this awaits future investigation, one plausible mechanism may be attributed to the requirement of GRP94 to facilitate processing of factors such as integrins specifically required for HSC-niche adhesion [19]. Secretory factors and cell-surface proteins are synthesized in the rough ER, and translocated into the ER where they are folded into their functional conformation by ER chaperone proteins [30]. GRP94 has been reported to be essential for the proper folding of several integrins, including a4 [40]. This has been reported to mediate the interaction of HSCs with niche cells and the extracellular matrix, as loss of integrin a4 results in impaired homing and reduced efficiency of HSCs to lodge to the endosteal surface [19,49]. While integrin a4 and a5 are both important for HSC retention in the niche [17,51], GRP94 is specifically required for processing integrin a4 but not a5 in HSCs. The failure of the Grp94 KO HSCs to express integrin a4 on the cell surface and its inability to bind to fibronectin are supportive of this notion. It is also possible that GRP94 is required for the proper folding of soluble ligands mediating the cross talk from HSCs to the niche. Other mechanisms may include cell cycle regulation by GRP94 and the ability of GRP94 to regulate ER Ca 2+ efflux and signaling. As GRP94 is also recently localized to the cell surface and also identified as a transmembrane ER protein [30], it may also control signaling pathways initiated from the cell surface and/or the ER and cytosol interphase affecting stem cell proliferation.
Our studies raise the general issue concerning the roles of protein folding chaperones in hematopoiesis and its relationship to the unfolded protein response (UPR). Hematopoiesis is tightly regulated by growth factors, cytokines and their receptors, which are secretory and cell membrane proteins [52]. Secretory factors and cell surface proteins are synthesized in the rough ER where they are folded into their functional conformation by ER chaperone proteins. Therefore, ER chaperone proteins are likely to be important for maintaining hematopoiesis, and their function depends on the role of specific client proteins in hematopoiesis. Other than synthesizing client proteins that are regulating hematopoiesis, the ER is also involved in hematopoiesis through UPR signaling pathways. For example, UPR sensor IRE1 ribonuclease activity for XBP-1 splicing is required for both the early differentiation of T and B lymphocytes and the terminal differentiation of activated B cells to immunoglobin-secreted plasma cells [53,54]. Whether UPR signaling is involved in regulating HSC homeostasis is not known, and awaits further investigation.
In comparing our Mx-1-Cre-mediated acute GRP94 KO model with the chimera model using tamoxifen-inducible GRP94 KO following BM transplantation [40], similar hematopoietic phenotypes, including general leukocytosis, increased spleen cellularity, decreased thymus and lymph node cellularities, increased myeloid cells and decreased lymphoid cells in the BM and spleen were observed in both models. The increase of LSK cells in the BM, spleen and peripheral blood and loss of integrin a4 expression upon GRP94 depletion are also in agreement with the previous report [40], which focuses on the role of GRP94 in lymphogenesis, specifically pro-B to pre-B transition in B cell development, as well as T cell development beyond CD4 2 CD8 2 double negative stage. Our work here investigates the unique function of GRP94 in the homeostasis of HSCs and their interaction with the stem cell niche. With our model, we provide new evidence that the expansion of primitive hematopoietic cells is attributed to the change of cell cycle status upon Grp94 deletion and that the change of HSC cell cycle status and localization is due to loss of interaction with the niche, revealed by the impaired homing and lodging ability of GRP94-deficient HSCs in vivo. We further determined that GRP94 in the niche cells is not required for maintaining the interaction with HSCs.
In summary, we identified GRP94 as a novel regulator of HSCs physiology and their interaction with the niche. It is tempting to speculate that agents that can alter GRP94 expression or activity may be explored for therapeutic intervention, such as increasing the proportion of HSCs in peripheral blood as a source for blood stem cell therapy, or as an alternative method for nonmyeloablative conditioning through emptying the HSC niche [55] and these warrant future investigation. Furthermore, overexpression of GRP94 has been associated with cellular transformation, tumorigenicity and decreased sensitivity to anticancer treatment [56]. With cancer stem cells implicated as the cancer initiating cells responsible for tumorigenesis and contribute to cancer resistance [57], it would be interesting to determine in future studies whether GRP94 regulates cancer stem cell niche interaction and thus contributes to tumorigenicity and self renewal of resistant cancer cells.

Ethics statement
All protocols for animal use were reviewed and approved by the USC Institutional Animal Care and Use Committee. The animal assurance number is A3518-01. The protocol number is 9964.

Mice
Grp94 F/F mice in a mixed C57BL/6; 129/Sv background were generated as previously described [42]. Grp94 F/F mice were crossed with the transgenic Mx-1-Cre mice on a C57BL/6 background (Jackson Laboratory) to generate Grp94 F/F ; Mx-1-Cre mice. Littermates that were negative for Cre transgene were used as controls. Genotyping was performed by PCR using genomic DNA extracted from mouse tail biopsies as previously described [58]. To induce the genomic deletion, 5.5 to 6.5 week-old male and female mice were injected intraperitoneally with pI.pC (25 mg/g mouse body weight) every other day. Mice were analyzed 11 days post 7 injections, except with modifications as specifically mentioned in the figure legends.

Complete blood count
Peripheral blood was collected via tail bleeding and analyzed using an auto hematology analyzer BC-2800 vet (Mindray, Redmond, WA) according to manufacturer's instructions.

Fluorescence activated cell sorting
Mouse whole BM cells resuspended in DPBS were enriched for c-Kit + cells using CD117 MicroBeads (Miltenyi Biotec, Auburn, CA) according to manufacturer's instructions. c-Kit + cells were stained with anti-mouse c-Kit (Miltenyi Biotec), anti-mouse Sca-1, and Lin antibody cocktail as described above. Primitive hematopoietic stem and progenitor cells were purified using a FACSAria flow cytometer (Becton Dickinson) based on established cell surface phenotypes. LSK cells were sorted for the competitive repopulation and adhesion assays; LK cells were purified for the in vivo homing and lodgment assay.
Cell cycle analysis 1610 7 whole BM cells were incubated with 10 mg/ml Hoechst 33342 (Sigma-Aldrich) at 37uC for 45 min, then stained with primitive hematopoietic cell antibodies (Lin, c-Kit and Sca-1) as described above. The stained cells were resuspended in 10% neutral buffered formalin (BDH Chemicals) and incubated at 4uC overnight. To stain for RNA content, Pyronin Y (Polysciences Figure 11. Grp94 KO mice displayed altered myeloid and lymphoid differentiation. A). Complete blood count of peripheral blood from WT (n = 31) and cKO (n = 37) mice. B) Representative Wright-Giemsa staining of blood smear with tail peripheral blood from WT and cKO mice. Scale bar represents 500 mm. C) Total thymus cell number (left) and total left and right axillary lymph nodes cell number (right) from WT and cKO mice (n = 7 for each group). D) Representative flow cytometric analysis of splenocytes from WT and cKO mice using lineage markers Gr-1 and CD3 (left), F4/80 and B220 (right). E) Quantitation of (D) from WT (n = 4) and cKO (n = 7) mice. F) Representative flow cytometric analysis with BM cells using lineage markers Gr-1 and B220 (left) and CD4 and CD8a (right). G) Quantitation of (F). Gr-1 and B220 (n = 7 for WT and n = 9 for cKO mice); CD4 and CD8a (n = 7 for each genotype). All data are presented as mean 6 s.e., ***p,0.001. doi:10.1371/journal.pone.0020364.g011 Inc., Warrington, PA) was added to the cells at a final concentration of 0.75 mg/ml and incubated at 4uC for 30 min. Cell cycle status was examined using a BD LSR II flow cytometer.
Apoptosis assay 1610 6 whole BM cells were stained with primitive hematopoietic cell antibodies (Lin, c-Kit and Sca-1) as described above. The stained cells were resuspended in 150 mL 16Annexin V binding buffer (BD Pharmingen) and incubated with Annexin V and 7AAD (both from BD Pharmingen) for 15 min. Cell apoptosis was examined within an hour using a BD LSR II flow cytometer.

Colony-forming cell (CFC) assay
Freshly isolated BM cells were mixed thoroughly with 300 mL Iscove's MDM with 2% FBS and 4 mL defined methylcellulose medium M3434 (StemCell Technologies). Cells were then plated in triplicate in six-well plates with 1.1 mL volume at a density of 1610 4 BM cells per well. The colonies were then scored at day 10 under an inverted microscope and reported as colony number in specific lineages according to the morphology of colonies described in the manufacturer's instructions.

Real-time quantitative RT-PCR
To detect Grp94 knockout efficiency, RNA was extracted from mouse whole BM cells and reverse-transcription was performed as previously described [59]. cDNA samples were analyzed in triplicate with the SYBR Green Supermix (Quanta Biosciences, Gaithersburg, MD) according to manufacturer's instructions. The following primers were used: Grp94, 59-TGG GCC TCT GCT  GTG TCC TGC-39  Competitive repopulation assay LSK cells were sorted using a FACSAria flow cytometer and 1000 WT or cKO LSK cells (CD45.2) were mixed with 250,000 BM MNCs (CD45.1) which is equivalent to approximately 500 LSK cells, and injected into the tail vein of B6.SJL mice that were lethally irradiated at 9.5Gy approximately 24 hr prior to transplantation. Engraftment levels and multilineage reconstitution were measured in peripheral blood samples obtained from the tail vein of hosts starting week 4.

In vivo homing
Sorted WT and cKO Lin 2 c-Kit + (LK) cells (,5610 5 to 1610 6 ) were labeled with a green fluorescent dye, carboxyfluorescein diacetate succinimidyl ester (CFSE), and a red fluorescent dye, SNARF-1 (both from Invitrogen), respectively, according to manufacturer's instructions. Labeled cells were injected into the tail vein of non-irradiated Grp94 F/F C57BL6; 129/Sv background littermates. Mice were then sacrificed after 16 hr, and the number of labeled cells was measured in the BM and spleen through the detection of CFSE + and SNARF + cells by flow cytometry.
In vivo lodgment LK cells were labeled with 5 mM CFSE or 5 mM SNARF celllabeling solutions and injected into the tail vein of non-irradiated mice, as described above. Tibias and femurs were dissected from recipient mice 16 hr after injection, decalcified for 3 days in Immunocal (Decal Chemical Corporation, Tallman, NY), and embedded in paraffin blocks after processing. A total of 180 bone sections of 5 mm thickness from three recipients were cut and mounted with Vectashield containing 49,6-diamidino-2-phenylindole (DAPI) (Vector Laboratories, Burlingame, CA). To assess the lodgment of injected cells to the endosteal niche, the numbers of CFSE + (green) or SNARF + (red) cells within two cell diameters from the endosteal surface were counted.
Adhesion assay 5610 2 LSK cells were added to wells coated with fibronectin (1.5 mg/cm 2 , BD Biosciences, Bedford, MA) in cell culture treated 96-well plates and incubated in minimum essential medium (MEM) alpha medium with 10% FBS for 3 hr at 37uC/5% CO 2 in a humidified atmosphere. Non-adherent cells were washed off with DPBS and adherent cells were visually counted microscopically. To control for nonspecific binding, adhesion to 1% bovine serum albumin (Sigma-Aldrich) was quantified.

Statistics
Statistical significance was assayed by Student's t test.