Targeted Mutation of the Mouse Grp94 Gene Disrupts Development and Perturbs Endoplasmic Reticulum Stress Signaling

Glucose-regulated protein 94 (GRP94) is one of the most abundant endoplasmic reticulum (ER) resident proteins and is the ER counterpart of the cytoplasmic heat shock protein 90 (HSP90). GRP94, a component of the GRP78 chaperone system in protein processing, has pro-survival properties with implicated function in cancer progression and autoimmune disease. Previous studies on the loss of GRP94 function showed that it is required for embryonic development, regulation of toll-like receptors and innate immunity of macrophages. Here we report the creation of mouse models targeting exon 2 of the Grp94 allele that allows both traditional and conditional knockout (KO) of Grp94. In this study, we utilized the viable Grp94+/+ and +/− mice, as well as primary mouse embryonic fibroblasts generated from them as experimental tools to study its role in ER chaperone balance and ER stress signaling. Our studies reveal that while Grp94 heterozygosity reduces GRP94 level it does not alter ER chaperone levels or the ER stress response. To study the effect of complete loss of GRP94 function, since homozygous GRP94 KO leads to embryonic lethality, we generated Grp94−/− embryonic stem cells. In contrast to Grp94 heterozygosity, complete knockout of GRP94 leads to compensatory upregulation of the ER chaperones GRP78, calnexin and calreticulin but not protein disulphide isomerase. Unexpectedly, loss of GRP94 leads to significant decrease in the level of ER-stress induced spliced form of XBP-1 protein, a downstream target of the IRE1 signaling pathway. Furthermore, from analysis of microarray database and immunohistochemical staining, we present predictions where GRP94 may play an important role in specific adult organ homeostasis and function.


Introduction
Glucose-regulated protein 94 (GRP94), also referred to as endoplasmin, CaBP4, ERp99 and gp96, is the most abundant endoplasmic reticulum (ER) resident glycoprotein accounting for 5 to 10% of the ER content [1][2][3][4][5][6]. Interestingly, while GRP94 is highly conserved in all vertebrates, some invertebrates such as C. elegans, Drosophila and in plants, it has not been found in yeast [2,7,8]. Recently, a GRP94-like protein was found in E. coli as the receptor of ompA protein [9]. GRP94 is the ER counterpart of heat shock protein 90 (HSP90) and shares about 50% amino acid homology and similar structural organization. GRP94 exists as homodimers containing an N-terminal ATP-binding domain, peptide-binding domain and C-terminal dimerization domain [2,[10][11][12]. The N-terminus of GRP94 also contains a peptide binding site [13] and ATP binding is reported to dictate the conformation of the N-terminus and regulate its ability to form quaternary structural interactions [14,15]. Recent mutational analysis further established that the ATPase activity of GRP94 is essential for chaperone activity [16].
GRP94 is coordinately regulated with GRP78, a major ER chaperone that also acts as a master regulator of ER stress signaling [17][18][19]. GRP94 and GRP78 are components of an ER chaperone system [20,21]; they share common promoter regulatory sequences and are inducible by a variety of ER stress conditions at the transcriptional level [22][23][24]. Similar to GRP78, GRP94 has been implicated as an anti-apoptotic protein conferring survival advantage to cells subjected to ER or chemotoxic stress [1,25,26]. However, unlike GRP78 which associates with nascent polypeptides and early folding intermediates, GRP94 binds substrates after they have been released from GRP78 [27] and appears to have a more limited number protein clients including some toll-like receptors and integrins [27][28][29].
During mouse embryonic development both GRP94 and GRP78 transcripts are expressed in very early embryos [30] and both proteins are detected at high levels at later stages during organogenesis [31,32]. Interestingly, heterozygous knockout of Grp78 triggers upregulation of GRP94 and other ER chaperones but not activation of other unfolded protein response (UPR) targets such as CHOP and XBP-1 splicing. Although Grp78+/2 heterozygotes are viable and exhibit no phenotypical abnormality, homozygous knockout of Grp78 leads to lethality at embryonic day 3.5 (E3.5) [33]. Null mutation of Grp94 in mice also results in embryonic lethality, but later in development around E7 [28,34]. Grp942/2 embryonic stem cells (ESCs) are hypersensitive to serum deprivation and not able to differentiate into muscle cells due to deficiency in secretion of insulin-like growth factor (IGF) II [34,35]. Despite these advances, the in vivo function of GRP94, which is not only a major ER chaperone but also an ER calcium binding protein, is just emerging. Here we report the creation of mouse models targeting exon 2 of the Grp94 allele that allow both traditional and conditional knockout (KO) of GRP94. We examined the requirement of GRP94 in mouse embryonic development and the experimental conditions required for establishment of Grp942/2 mouse ESCs. Using primary mouse embryonic fibroblasts (MEFs) and ESCs derived from the Grp94+/+, +/2 and 2/2 embryos, we discovered that in response to ER stress, GRP94 depletion results in compensatory upregulation of specific ER chaperones and suppression of ER stress-induced splicing of XBP-1, a downstream target of IRE1 signaling pathway. Furthermore, from analysis of microarray database and immunohistochemical staining of GRP94 expression in mouse adult organs, we present predictions where GRP94 may play an important role in specific adult organ homeostasis and function.

Creation of Grp94 Floxed and Knockout Alleles for Targeted Mutation of GRP94
To elucidate the in vivo function of GRP94, we created targeted mutation of Grp94 taking advantage of the cre-loxP and the flp-FRT systems, which allows both traditional as well as conditional knockout of the targeted allele [36,37]. In our mouse models, inactivation of the Grp94 gene was achieved through deletion of exon 2, which contains 103 base pairs (bp). Since exon 1 only has 49 bp and the deletion of exon 2 will cause an early frameshift mutation, at most a truncated peptide of 26 amino acids is generated if the mRNA escapes the premature termination codon mediated mRNA degradation system [38]. This abolishes the expression of GRP94 without generating any residual functional domain.
To achieve this, the targeting vector for Grp94 was created by inserting a neo cassette flanked by a pair of loxP sequences into intron 1, and a third loxP into intron 2. The neo cassette was inserted in a reverse orientation with respect to the Grp94 allele. Upon homologous recombination by expression of cre recombinase, exon 2 was deleted ( Figure 1A). To avoid the complication of removing the neo cassette, a pair of FRT sequences was also added adjacent to the loxP sites, flanking the neo cassette in a tandem manner [39]. Removal of the neo cassette by expressing b-actindriven flp recombinase generated a floxed allele containing two loxP sites flanking exon 2 ( Figure 1A). Subsequent removal of exon 2 by cre recombinase generated the knockout (KO) allele ( Figure 1A).
The targeting vector was electroporated into ES cells and subjected to G418 and ganciclovir selection. Two out of 600 positive clones were found to contain homologous recombination into the wild-type (WT) allele. One of the clones was able to generate two viable chimeric founder mice, as confirmed by both genotyping and Southern blotting. The chimeric founder mice were mated with WT female C57BL/6J mice and F1 mice with germline transmission of the loxP-FRT targeted allele were generated. To generate the Grp94 conventional KO allele, the loxP-FRT Grp94 mice were first mated with an EIIA-cre transgenic mouse. The chimeric progeny with the recombination of the first and third loxP were mated with WT female C57BL/6J mice to generate the pure KO allele, and to segregate the KO allele from the EIIA-cre transgene. The genotype of the Grp94+/2 mice was validated by Southern blot of tail DNA, using a DNA probe outside the targeting vector ( Figure 1A). As expected, after BamH1 digestion, the WT mice yielded a single band of 7 kilobase pairs (kb), and the Grp94+/2 mice yielded two bands (7 kb and 6 kb) of equal intensity ( Figure 1B). The genotype was further confirmed by PCR with specific primer sets designed around exon 1, 2 and 3 to differentiate between the WT allele (336 bp) and the KO allele (593 bp) ( Figure 1C). To generate the Grp94 floxed allele, the loxP-FRT Grp94 mice were mated with b-actin-driven flp transgenic mouse. The genotype was confirmed by PCR with specific primer sets designed to flank part of intron 1 to differentiate between the WT allele (336 bp) and the floxed allele (420 bp) ( Figure 1D).

Grp94 Heterozygosity Reduces GRP94 Level but Does Not Alter Basal ER Chaperone Level or Response to ER Stress
Mice bearing the Grp94+/+ and +/2 genotypes were generated and subjected to analysis. The Grp94+/2 mice are phenotypically normal and grow at the same rate as WT siblings (data not shown). The Grp94+/2 mouse embryos are viable and in Western blot analysis, GRP94 protein level in both the brain and liver of the E14.5 embryos was about 50% compared to WT embryos (Figure 2A, B, C). These results are consistent with both the inactivation of one allele of Grp94 in the heterozygotes and the absence of feedback mechanisms in these embryonic organs to upregulate the remaining WT Grp94 allele to compensate for the partial loss of GRP94. Previous studies showed that a similar level of reduction of GRP78 in the Grp78+/2 cells led to a 1.7-and 2-fold increase in the basal level of GRP94 and PDI respectively, but with no change in calnexin (CNX) and calreticulin (CRT) levels [33]. In examining whether there is reciprocal upregulation of ER chaperones in Grp94 heterozygotes, we observed no change in basal level of GRP78, CNX or CRT, suggesting that in contrast to GRP78, partial GRP94 reduction in mouse embryonic organs does not elicit compensatory increase in basal ER chaperone levels (Figure 2A, B).
To test whether cells expressing GRP94 at 50% of the WT level altered their response to ER stress, MEFs isolated from E13.5 Grp94+/+ and+/2 embryos were subjected to various ER stress conditions. We observed that GRP78 induction by all three ER stress inducers, thapsigargin (Tg), tunicamycin (Tu) and azetidine (AzC), was not affected in Grp94+/2 MEFs ( Figure 3A). Further, Grp94+/2 MEFs did not spontaneously trigger the UPR signaling pathways such as XBP-1 splicing and CHOP induction ( Figure 3B). Upon treatment with Tg and Tu, both WT and heterozygous MEFs showed XBP-1 splicing and CHOP induction. Interestingly, in AzC treated cells, the spliced form of XBP-1 [XBP-1(s)] was not evident in either Grp94+/+ or +/2 MEFs and CHOP induction was milder compared to Tg and Tu treated cells.

Homozygous Grp94 Knockout Results in Embryonic Lethality and Growth Factor Dependency of ESCs
When the Grp94+/2 mice were intercrossed, out of the 80 pups born, the ratio of +/+, +/2 and 2/2 genotype was 25:55:0 respectively ( Figure 4A). This suggests that Grp942/2 embryos were not viable, whereas Grp94+/2 mice were born in the expected Mendelian ratio. Further analysis showed that the Grp942/2 embryos were detectable by nested PCR at E3.5 up to E8.5. By E10.5 and later, no such genotype of embryos was detected and resorbed embryos were evident. Morphological examination of the embryos revealed that at E7.5 and E8.5, the Grp942/2 embryos were much reduced in size with no recognizable structure, compared to normally developing Grp94+/+ and +/2 embryos ( Figure 4B). The survival of the Grp942/2 embryos past blastocyst stage afforded the opportunity to establish Grp942/2 ESC line for ascertaining GRP94 function in vitro. Two different culture mediums were used in our attempts to derive Grp942/2 ESC lines. The results were summarized in  The quantitation was performed using Quantity One software with the level of GRP94 in Grp94+/+ embryonic brain and liver setting as 1. The data are presented as mean6s.e.m. n = 2,3 per group. These 5 embryos (2 +/+ and 3 +/2) were from one litter. **p#0.01 (Student's t-test). doi:10.1371/journal.pone.0010852.g002 Table 1. Using 44 blastocysts which generated 35 outgrowths, 18 ESC lines were established when they were cultured in the Knockout (KO) DMEM with 15% knockout serum replacement and 5% fetal bovine serum with standard supplements. Surprisingly, none of the 18 lines contained the Grp942/2 genotype. After insulin-transferrin-selenium (ITS) was added to the same medium only a single Grp942/2 ESC line was established among 10 ESC lines derived from 24 outgrowths from 30 blastocysts. This implies that exogenous growth factors are uniquely required for Grp942/2 ESC establishment. Once the lines were established, we observed that ITS was not required for the maintenance of the Grp942/2 ESCs, although upon freezing and thawing, the survival rate of the Grp942/2 ESCs was lower compared to wildtype cells. In order to investigate the differentiation ability of the GRP94 null ESCs, paired Grp94+/+ and 2/2 ESCs were utilized. Our results showed that both genotypes could be induced to differentiate into adipocytes (oil red positive cells), hepatocytes (indocyanine green positive cells) and neurons ( Figure 5). The results imply that Grp942/2 ESCs can differentiate into cells of all three germ layers: ectoderm (neurons), mesoderm (adipocytes) and endoderm (hepatocytes). However, the Grp942/2 ESCs are unable to differentiate into cardiomyocyte-like cells in contrast to Grp94+/+ ESCs which can form contracting cells similar to cardiomyocytes (Supplementary Video S1).

Compensatory Upregulation of Specific ER Chaperones in GRP94 Null ESCs
The consequence of homozygous knockout of GRP94 on basal and ER stress induced chaperone levels was studied utilizing the ESCs. First, we confirmed by Western blot that the Grp942/2 cells are devoid of GRP94 ( Figure 6A). Interestingly, compensatory upregulation of basal level of GRP78 and CNX was evident in the Grp942/2 cells and these increases were further enhanced in Tg stressed cells ( Figure 6A, B). For CRT, the increase was also apparent and reached statistical significance upon Tg treatment ( Figure 6A, B). Interestingly, for PDI, the increase, if any, was minor; and may even slightly decrease upon 16 hr of Tg treatment in the Grp94 null cells ( Figure 6A, B). These results show that specific elimination of GRP94 triggers a regulatory circuit to enhance expression of specific ER chaperones which include GRP78, CNX and CRT.

Suppression of ER-stress Induced XBP-1 Splicing in GRP94 Null ESCs
Since GRP78 and GRP94 are components of chaperone complex and GRP78 is known to regulate UPR signaling through interaction and inhibition of transmembrane ER signaling molecules, we investigated whether complete depletion of GRP94 will affect these pathways. Grp94+/+ and 2/2 ESCs were subjected to Tg treatment and the induction of UPR targets were examined. XBP-1 splicing is a major downstream target of the IRE1 signaling. In Grp94+/+ ESCs, Tg treatment for 4 hr induced a high level of Xbp-1 mRNA splicing, which persisted but at a lower level at 16 hr ( Figure 7A). For Grp942/2 ESCs, Tginduced Xbp-1 mRNA splicing was also observed, however, quantitation of the spliced products showed that at the 4 hr time point, the level of spliced Xbp-1 mRNA was reduced by about 30% compared with Grp94+/+ ESCs. At the XBP-1 protein level, compared to the Grp94+/+ ESCs which showed robust induction of XBP-1(s) at 4 hr and residual amount still present at 16 hr, Grp942/2 ESCs showed a 75% reduction of XBP-1(s) at 4 hr and non-detectable amount at 16 hr ( Figure 7B). The protein level of the unspliced form of XBP-1 [XBP-1(u)] was relatively constant throughout the Tg treatment period for the Grp942/2 ESCs, compared to a slight elevation in the Tg treated Grp94+/+ ESCs ( Figure 7B).
As a downstream target of XBP-1, Tg-induction of EDEM was also reduced in Grp942/2 ESCs ( Figure 8A). In contrast, other UPR targets and cellular proteins were only mildly affected or completely unaffected by GRP94 depletion. For example, Tg-induced phosphorylation of PERK was only slightly increased in Grp942/2 ESCs at 16 hr, with no corresponding increase of CHOP ( Figure 8A). The level of p62, a cytosolic protein which binds to polyubiquitinated proteins and targets them to the autophagy machinery for degradation [40] was also not affected in the Tg-treated cells of both genotypes ( Figure 8A). There were substantial variations on the Tginduced phosphorylation of eIF2a in the Grp942/2 ESCs however on the average there was no increase ( Figure 8B and data not shown). Tg-induction of ATF4 was mildly increased while the level of HSP70, a major heat shock protein related to GRP78, was not affected ( Figure 8B). Quantitation of p-PERK and p-eIF2a levels from several independent experiments showed no statistically significant change between the two genotypes ( Figure 8C). Thus, the minor differences observed are likely due to experimental variations in the cultured ESC system. Upon Tg treatment, Grp94+/+ and 2/2 ESCs showed similar levels of proliferating cell nuclear antigen (PCNA), suggesting no major alteration in the proliferative properties ( Figure 9A). However, there was a substantial increase in the level of cleaved caspase-7 (C-7), implying that GRP94 depletion in the Grp942/2 ESCs sensitizes the cells to Tg-induced apoptosis ( Figure 9A). This result was further confirmed by caspase-3 (C-3) activation. Consistent with C-7, the level of cleaved C-3 was also elevated in Grp942/2 ESCs, following treatment with ER stress-inducer Tu ( Figure 9B).

Differential Expression of GRP94 in Adult Mouse Tissues
Analysis of a microarray database (Amazonia) revealed that Grp94 mRNA is expressed at relatively low levels in most mouse adult tissues ( Figure 10). Strikingly, the level is about 20-fold higher in dendritic cells, 12-fold higher in both smooth muscle and lung bronchial epithelium and 10-fold higher in pancreatic islet, with the Grp94 mRNA level in whole blood setting as 1. To correlate these mRNA data with protein expression level, we examined the level of GRP94 in mouse adult organs such as lung, pancreas and kidney by immunohistochemistry. The results showed that while GRP94 protein was expressed at low basal level in the lung, pancreas and kidney, it is strikingly high in the lung bronchial epithelium, pancreatic islets of Langerhans and kidney cuboidal epithelium ( Figure 11). For comparison, GRP78 staining was elevated in the lung bronchial epithelium and kidney cuboidal epithelium, however, it was not as prominent as GRP94 in the pancreatic islets of Langerhans ( Figure 11A). The immunohistochemical staining results for GRP94 and GRP78 were consistent with relative levels of Grp94 and Grp78 mRNA from the microarray database in the specific tissues being examined ( Figure 11B). Thus, high level expression of GRP94 in specific cell types of the lung, pancreas and kidney suggests that GRP94 could play an important role in homeostasis in these cell types, which may represent excellent targets for conditional knockout of GRP94 in mouse models for further probing of GRP94 function in vivo.

Discussion
Endoplasmic reticulum stress has emerged to play important roles in mammalian development and human diseases. In order to better understand the molecular pathways that allow cells to adapt to ER homeostasis, we focus on ER chaperones which are critical not only for quality control of proteins processed in the ER, but may also directly or indirectly regulate ER signaling in response to ER stress and maintenance of calcium homeostasis [19]. GRP94 is highly conserved in vertebrates but not in yeast and has been well documented to be inducible by a variety of metabolic stresses, suggesting that it is uniquely required for functions in vertebrates not shared by yeast. As documented in earlier reports [28,34,41], creation of model organisms with targeted mutation of GRP94 allows comprehensive analysis of its in vivo function.
Based on our finding that knockout of GRP78, another major ER chaperone closely associated with GRP94 in function and regulation, results in embryonic lethality around peri-implantation (E3.5) stage [33], mice deficient in GRP94 may also die during embryonic development. Hence, using the cre-loxP system, we designed a mutated allele of Grp94 that allows both the traditional knockout and conditional knockout in a tissue or time specific manner. Since exon 1 of Grp94 only contains 49 bp, the positioning of the neo cassette into intron 1 coupled with expected frame shift ensures early disruption of the GRP94 of the coding sequence and elimination of GRP94 with any functional domain. Utilizing heterozygous and homozygous knockout Grp94 cells and mice derived from this strategy, we examined how complete and partial GRP94 depletion altered development and the ER stress response. We discovered that deficiency of GRP94 is distinct in many different aspects from deficiency in GRP78, which has an established role in regulating ER signaling in addition to being a major chaperone.
While both heterozygous Grp78 and Grp94 mice show no obvious abnormal phenotype, cells with partial GRP78 knockdown upregulate GRP94 and PDI; however no such compensation is observed for partial GRP94 deficiency. Nonetheless, in cells completely devoid of GRP94, we discovered that not only is GRP78 upregulated, but also CNX and CRT. Interestingly, PDI appears to be substantially less affected or unaffected. Thus, it is possible that feedback mechanisms exist to compensate the complete loss of GRP94 by increasing GRP78, a partner protein of GRP94 in the GRP78/GRP94 chaperone system, and also of the CNX/CRT chaperone system, but apparently not the thiol oxidoreductases such as PDI. Collectively, these results imply that cells have abilities to adjust the ER chaperone balance in response to depletion of key chaperones such as GRP94 and GRP78.
During the course of creation of our mouse models and analysis of the phenotypes, it has been reported that homozygous knockout of Grp94 through insertion of a neo resistance cassette into the coding region at the end of exon 3 (61 amino acids into the mature protein) of the murine Grp94 gene leads to embryonic lethality around day 7 of gestation [34]. The embryos failed to develop mesoderm, primitive streak, or proaminotic cavity, and GRP94 is essential for mesoderm induction and muscle development due to deficiency in IGF-II secretion. Our Grp94 knockout mouse model targets exon 2 of the Grp94 allele. Additionally, our model differs from the previous Grp94 KO model [34] in that in our model the neo resistance cassette was inserted into intron 1, and exon 2 was completely eliminated through flox recombination. While we also observed embryonic lethality around E7, the ratio of pups born for the +/+, +/2 and 2/2 genotype was 25:55:0 for our model eliminating exon 2, whereas the model disrupting exon 3 [34] reported a ratio of 414:515:0. Since the Grp94+/2 pups in our model were born in the expected Mendelian ratio of 1:2 with respect to Grp94+/+ pups, our results suggest that the Grp94+/2 embryos in our model were fully viable. This is consistent with our observation that the growth and biochemical properties of Grp94+/+ and +/2 ESCs and MEFs isolated from our Grp94   [42]. Our conditional deletion model created here is distinct from another conditional deletion model of Grp94 which targets exon 1, which has been used to achieve specific deletion of Grp94 in macrophages and determining its function in innate immunity and macrophage homeostasis [28]. Our difficulty in establishing Grp942/2 ESC lines from E3.5 blastocysts in routinely used ESC culture medium suggests that some critical component is missing or not of sufficient quantity. The recent discovery that Grp94 deficiency blocks IGFII secretion  [34,35] and our observation that addition of ITS supplement yielded successful ESC line derivation supports the stringent requirement of exogenous growth factors for Grp94 null ESC establishment. Another contributing factor may be the genetic background of the Grp94 blastocysts as the mice used in our study have been backcrossed to the C57BL/6 background, and establishment of C57BL/6 mouse ESCs has been reported to be more difficult [43]. Finally, with the GRP94 null ESCs independently derived from our Grp94 KO mice, we directly confirmed the requirement of GRP94 for ESC differentiation into cardiac muscle cells.
The ESC lines with complete elimination of GRP94 also provide us with a unique opportunity to study whether GRP94, as a partner protein of GRP78, contributes to ER stress signaling. In the case of GRP78 knockdown or knockout, UPR signaling targets such as p-eIF2a, CHOP and GADD34 are activated in the absence of stress, clearly showing GRP78 suppresses their activation under non-stress conditions [42,44]. Here we report the novel observation that the XBP-1(s) protein level was substantially reduced following Tg treatment in Grp94 null ESCs compared to wild-type control. Interestingly, for both Grp94+/+ and 2/2 ESCs, PERK phosphorylation persisted up to 16 hr after Tg treatment, whereas in most other cell lines, ER stress-induced PERK phosphorylation was transient and subsided within a few hours of ER stress. On the other hand, in contrast to GRP78 depletion, no spontaneous induction of CHOP is observed in cells depleted of GRP94 and the kinetics of its induction by Tg is unaffected. This suggests that while GRP94 may act in concert with GRP78 in folding protein intermediates, its depletion results in de-regulation of the specific UPR targets quite distinct from those affected by GRP78 deficiency.
How might GRP94 depletion suppresses ER-stress induced XBP-1 splicing? Both GRP94 and XBP-1 have been implicated in immune function [28,[45][46][47]. Using mouse models of specific knockout of Grp94 in macrophages, it was demonstrated that GRP94 is the master chaperone for Toll-like receptors (TLR) and is important in the innate function of macrophages [28]. Analysis of B-cell specific GRP94 null mice further revealed that GRP94 optimizes B-cell function serving as chaperones for integrin and TLR but not immunoglobulins [29]. Recently, it was discovered that TLR4 and TLR2 specifically activated the UPR sensor IRE1a and its downstream target XBP-1 in macrophages [48]. Since GRP94 is required for maturation of TLRs, including TLR4 and TLR2, it may regulate XBP-1 activation through the TLRs. Future studies will be required to test this and other mechanisms. The differential effect of GRP94 depletion on XBP-1 splicing but not other UPR signaling pathways also supports the notion that UPR sensors can be dissociated and selectively regulated [49].
Through analysis of a microarray database, we discovered that GRP94 expression is strikingly high in dendritic cells, smooth muscle and lung bronchial epithelium. The microarray data, which is supported by tissue staining results, predicts potential role of GRP94 in immunity and function/survival of specific cell types in the lung, pancreas and kidney. Future studies with conditional KO of Grp94 in these cell types will address these important issues. Furthermore, the role of GRP94 is human diseases such as cancer, diabetes and neurodegeneration can be also achieved through the use of the mutant mouse models.

Materials and Methods
Generation of the loxP-FRT Grp94 Targeting Vector An 8 kb DNA fragment for Grp94 gene (spanning from the promoter region to intron 5) was isolated from a 129/Sv-derived lambda FixII mouse genomic library (a gift from Dr. Robert Maxson, University of Southern California (USC) Keck School of Medicine). Using this as a template, subcloning and PCR reactions were performed and fragments were subcloned into the pBluescript KS vector to create the Grp94 loxP-FRT targeting vector. This vector was constructed by inserting a neo cassette, flanked by a pair of loxP sites and a pair of FRT sites (kindly provided by Dr. Hualin Fu, USC), into a Stu I site in intron 1, a third loxP site into intron 2 and a pgk-TK (thymidine kinase) expression cassette at the 39 end of the construct as a negative selection marker ( Figure 1A). The targeting vector was linearized by NotI and was Generation of Grp94+/2 and F/F Mice All animals were housed under pathogen-free conditions and experiments were performed in accordance with the USDA Animal Welfare Regulations on Humane Care and Use of Laboratory Animals. All animal experimental protocols were approved by the University of Southern California Institutional Animal Care and Use Committee. The selected ESC clone was subjected to subcloning to achieve a higher purity of targeted ESCs and injected to blastocysts from the C57BL/6J female mouse. Chimeric founder mice were genotyped by PCR. The founder mouse with the correct genotype was mated with C57BL/6J female mice to generate the Grp94 loxP-FRT mice. The progenies were subjected to PCR genotyping. The primers for identifying loxP-FRT allele were FL-F: 59-CTCCTGAGACCGAAAAGGACT-39 and FL-R: 59-AG-GATTGGGAAGACAATAG CAG-39, with a 440 base pair (bp) product. To generate the Grp94 conventional knockout (KO) (2) allele, the loxP-FRT Grp94 mice were mated with an EIIA-Cre transgenic mouse (Jackson Laboratory) and progenies were genotyped by PCR. To generate the Grp94 floxed (F) allele, the loxP-FRT mice were mated with b-actin-driven flp transgenic mice (Jackson Laboratory) and PCR was used to confirm the successful generation of the F allele. The primers used to differentiate WT, KO and F alleles were 94KOniL: 59-GCTGTGTCCTG-CTGACCTTCG-39, 94KOniR: 59-TACCTCACCGATTGAA-AAGC-39 and PAS3: 59-TGATCAGCGATCGCCAAAAG-TCCTTAGGGAGG-39. 94KOniL and PAS3 were used to detect both the WT allele with a product of 336 bp and the F allele with a product of 420 bp. 94KOniL and 94KOniR were used to detect the KO allele with a product of 593 bp. The Grp94+/2 progeny bearing a KO allele was mated to C57BL/6J female mice to segregate the KO allele from the EIIA-Cre transgene. The Grp94+/2 mice used in this study were backcrossed to C57BL/6 background for 3 to 4 generations.

Isolation, Culture and Drug Treatment of MEFs
Mouse embryonic fibroblasts (MEFs) were isolated from E13.5 embryos. Briefly, the embryo was washed twice in PBS after removal of head and internal organs, disaggregated with an 18-gauge needle in 0.25% trypsin-EDTA, and were incubated at 37uC for 2-3 min before adding DMEM with 10% FBS and 1% penicillin and streptomycin. The cells were split after 2 days in culture and MEFs were used for experiments. For drug treatment, the cells were treated with 300 nM thapsigargin (Tg), 1.5 mg/ml tunicamycin (Tu), or 5 mM azetidine (AzC) for 16 hr prior to harvest.

Generation of ESCs and Differentiation Analysis
ESCs were derived from interbred Grp94+/2 mice (backcrossed to C57BL/6 background for 3 or 4 generations). The blastocysts from E3.5 pregnant female mice were recovered and cultured on MEFs. One Grp942/2 and multiple +/+ and +/2 ESC lines were established. Grp94+/+ and 2/2 ESCs were used to evaluate the role of GRP94 in ESC differentiation potential. The detailed protocol for ESC differentiation has been described previously [34]. In brief, both lines were differentiated via embryoid body (EB) formation and treatment with or without retinoic acid. In two independent experiments, EBs were generated from both lines. Oil red and indocyanine green were used to identify adipocytes and hepatocyte-like cells, respectively. Neuronal cells were identified by the morphology. No beating foci were observed in EBs derived from Grp942/2 ESCs while the EBs derived from Grp94+/+ ESCs produced many such foci.

Southern Blotting
The BamHI digested DNA (from tail biopsy or ESCs) was run on a 1% agarose gel at 12 volt overnight. The DNA was then transferred to a nylon membrane for Southern blotting using the 59 external probe described above for ESCs screening.
Phospho-PERK was quantitated against total PERK, and phospho-eIF2a was quantitated against total eIF2a. Both spliced and unspliced XBP-1 were quantitated against b-actin. All the other proteins were quantitated against b-actin.

RT-PCR Analysis of Xbp-1 mRNA Splicing
The cells were either untreated or treated with 300 nM ER stress inducer thapsigargin (Tg) for 4 hr or 16 hr. After that, total RNA was extracted using TRI reagent (Sigma-Aldrich) following the manufacturer's instructions. First-strand cDNA was synthesized with SuperScript II (Invitrogen). To detect both unspliced and spliced Xbp-1 mRNA, PCR was performed as previously [51]. The primers for PCR of Xbp-1 and b-actin were described previously [42]. These experiments were repeated two to three times.

Immunohistochemical Staining
Mouse tissues were fixed overnight in 10% buffered formalin and embedded in paraffin using standard protocols. Immunohis- Figure 11. GRP94 and GRP78 staining and mRNA expression in adult mouse tissues. (A) Immunohistochemical staining for GRP94 and GRP78 in adult mouse lung, pancreas and kidney tissues. The GRP78 staining and control staining (without primary antibody) were presented to show the specificity of the GRP94 antibody. The sections were counterstained with hematoxylin. Brown color depicts GRP94 and GRP78 staining. White arrowheads indicate lung bronchial epithelium. Black arrowheads indicate pancreatic islet. Black arrows indicate renal cuboidal epithelium. Scale bars represent 50 mm. (B) Relative levels of Grp94 and Grp78 mRNA in adult mouse lung, broncho epithelium, pancreas, islet and kidney, with the level of Grp94 and Grp78 mRNA in whole blood set as 1. doi:10.1371/journal.pone.0010852.g011 tochemical staining was carried out as described previously [52]. In brief, Vectastain Elite avidin-biotin complex kit (Vector Laboratories) was used for staining. Paraffin sections were incubated with rat anti-GRP94 antibody (1:1000) (Stressgen), mouse anti-BiP/GRP78 antibody (1:200) (BD Biosciences) or without primary antibody (control staining) in blocking solution (1.5% serum in PBS) at 4uC overnight after antigen retrieval with retrivagen A (pH 6.0) (BD Biosciences).