IRE1α Disruption Causes Histological Abnormality of Exocrine Tissues, Increase of Blood Glucose Level, and Decrease of Serum Immunoglobulin Level

Accumulation of unfolded proteins in the endoplasmic reticulum (ER) causes ER stress. As a cellular adaptive response to ER stress, unfolded protein response (UPR) activates molecules for the quality control of ER proteins. One enzyme that plays an important role in UPR is Inositol Requiring Enzyme-1 (IRE1), which is highly conserved from yeast to humans. In particular, mammalian IRE1α activates X-box-binding protein 1 (XBP1) by unconventional splicing of XBP1 mRNA during ER stress. From analysis of knockout mice, both IRE1α and XBP1 have been shown to be essential for development and that XBP1 is necessary for the secretory machinery of exocrine glands, plasma cell differentiation, and hepatic lipogenesis. However, the essentiality of IRE1α in specific organs and tissues remains incompletely understood. Here, we analyzed the phenotype of IRE1α conditional knockout mice and found that IRE1α-deficient mice exhibit mild hypoinsulinemia, hyperglycemia, and a low-weight trend. Moreover, IRE1α disruption causes histological abnormality of the pancreatic acinar and salivary serous tissues and decrease of serum level of immunoglobulin produced in the plasma cells, but not dysfunction of liver. Comparison of this report with previous reports regarding XBP1 conditional knockout mice might provide some clues for the discovery of the novel functions of IRE1α and XBP1. (196 words)


Introduction
Since the majority of secretory proteins, such as antibodies, digestive enzymes, and hormones, are synthesized in the cytoplasm and are cotranslationally translocated into the lumen of the endoplasmic reticulum (ER) through a narrow channel called translocon on the ER membrane, they are initially located in the ER as unfolded and unmodified nascent polypeptides. These proteins then undergo meticulous folding by molecular chaperones, correct disulfide bond formation by protein disulfide isomerases, and proper oligosaccharide modification by the oligosaccharyltransferase complex, sugar trimming enzymes, and calnexin/calreticulin cycle in the ER [1,2]. Therefore, when cells produce these proteins in large amounts, the ER is thought to be liable to become overloaded for the maturation of these proteins. Accumulation of unfolded proteins in the ER also causes ER stress. To adaptively respond to ER stress, the cell induces the transcriptional activation of molecules for the maturation of proteins in the ER. This response is called unfolded protein response (UPR) [3]. Thus, UPR is an important cellular response for the mass production of functional secretory proteins from unfolded proteins in cells which produce them in large amounts.
To date, several molecules have been reported to play important roles in UPR. One of these molecules, IRE1, is an ER-located type I transmembrane protein with a kinase domain and RNase domain in the cytosolic region. When exposed to ER stress, via trans-autophosphorylation and activation of its RNase domain, IRE1 induces unconventional splicing of an mRNA encoding a specific transcription factor responsible for UPR activation [4][5][6][7][8]. IRE1 is highly conserved from yeast to humans, and two IRE1 paralogues have been reported in mammals: IRE1a and IRE1b [9][10][11]. Specifically, IRE1a induces the unconventional splicing of XBP1 mRNA under ER stress condition [12]. The spliced XBP1 mRNA is then translated into a functional transcription factor to induce UPR. Besides IRE1, two ER-located transmembrane proteins, PERK and ATF6, play important roles in mammalian UPR [13,14]. On sensing ER stress, PERK induces the phosphorylation of eIF2a and the translational activation of ATF4 [15]. On the other hand, under ER stress condition, ATF6 is cleaved by Site-1 and Site-2 proteases, and its cytoplasmic domain is translocated to the nucleus [16,17]. Both ATF4 and the cleaved ATF6 work as transcription factors in UPR induction, as well as XBP1 which is activated by IRE1a.
As described above, IRE1a directly catalyzes the cleavage of XBP1 mRNA in the splicing reaction under ER stress condition [12]. To our knowledge, this reaction is exclusively dependent on IRE1a activity and is not detected in IRE1a-deficient cells [18]. This implies that IRE1a and XBP1 function on the same signal transduction pathway in ER stress response. Also, IRE1a knockout (KO) mice and XBP1 KO mice commonly have embryonic lethality and that both IRE1a and XBP1 play an essential role in mammalian development [19][20][21]. However, although embryonic lethality of XBP1 KO mice is rescued with an XBP1 transgene specifically expressed in the liver [22], that of IRE1a KO mice is rescued with endogenous IRE1a specifically expressed in the extra-embryonic tissues and not in the liver [18]. This suggests that not only a known IRE1a-dependent XBP1 function but also an XBP1-independent IRE1a function(s) may exists in extra-embryonic tissues and that an IRE1a-independent XBP1 function(s) may exists in the fetal liver. Thus, a comparison analysis of conventional and conditional KO mice in terms of IRE1a and XBP1 may further provide some clues for the discovery of additional tissue-specific functions of each molecule. Analysis of XBP1 conditional KO mice, including XBP1 KO mice rescued with an XBP1 transgene specifically expressed in the liver, previously demonstrated that XBP1 is required for the secretory machinery of exocrine glands, plasma cell differentiation, and hepatic lipogenesis [22][23][24]. However, it remains unclear whether IRE1a plays an essential function for these biological phenomena.
To elucidate this, we analyzed the phenotype of IRE1a conditional KO mice in this study.

Measurement of blood glucose and insulin
Blood glucose level was measured using a portable glucose measuring device (Arkray). Insulin level was determined by enzyme linked immunosorbent assay (ELISA) using mouse insulin as a standard (Shibayagi). Glucose tolerance tests were performed on 20-week-old IRE1a conditional KO and control mice that had been fasted for 16 hours. Mice were orally administered with 2 mg/g body weight glucose. Blood glucose level and serum insulin level were measured at indicated intervals.

Histological analysis
Each tissue was fixed in 10% formalin and then embedded in paraffin. Paraffin blocks were sliced into 5-mm-thick sections and stained with hematoxylin and eosin for general histopathological analysis. Immunohistochemical analysis was performed using 6-mm-thick paraffin sections. Immunoreactivity of insulin and glucagons was detected using guinea pig polyclonal antiinsulin and rabbit polyclonal anti-glucagon antibodies (both from Dako), respectively. Immunoreaction with secondary antibodies and chromogenic reaction with 3,39-diaminobenzidine tetrahydrochloride were performed according to standard procedures.

Western blot analysis
The tissues were lysed in SDS sample buffer (50 mM Tris-HCl pH 6.8, 2% SDS, 50 mM DTT, 10% glycerol, and 1 mg/ml bromophenol blue), and the lysate was heated to 98uC for 10 min. SDS-PAGE (10% gel) was used to resolve the proteins in the lysate. After electrophoresis, the proteins were electrotransferred onto polyvinylidene fluoride microporous membranes. Immunodetection of amylase and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was performed using rabbit polyclonal antiamylase antibody (Sigma) and mouse monoclonal anti-GAPDH antibody (Cell signaling), respectively.

Analysis of B cell differentiation
Serum levels of IgM and IgG1 were measured by ELISA (Bethyl Laboratories Inc.) according to the manufacturer's instruction. Flow cytometry analysis was performed with FACS vantage SE (Becton Dickinson) and FlowJo ver. 6.3.2 software. For the analysis, single-cell suspensions from the spleen were prepared according to the protocol of Beckman Coulter Japan. Splenocytes were then stained with PE-conjugated IgD antibody (Becton Dickinson), FITC-conjugated IgM antibody (eBioscience), and APC-conjugated B220 antibody (Becton Dickinson).

Measurement of transaminase activity in serum
Both aspartate aminotransferase (AST) and alanine aminotransferase (ALT) activities in serum were measured using the transaminase assay kit (Wako).

Results
IRE1a-deficient mice exhibit mild hypoinsulinemia, hyperglycemia, and low-weight trend Although IRE1a conventional KO mice show embryonic lethality as described earlier, embryo proper-restricted IRE1a conditional KO mice, which specifically express IRE1a in the extra-embryonic tissues, can avoid embryonic lethality [18,20,21]. We initially measured body weight and blood glucose level from age 4 weeks to age 16 weeks to examine the growth process and health condition of embryo proper-restricted IRE1a conditional KO mice (hereinafter IRE1a CKO mice). IRE1a CKO mice exhibited mild hyperglycemia and a low-weight trend compared with control mice under free feeding conditions ( Figure 1A), although there was no difference in body length, amount of feed and water intake per day between IRE1a CKO and control mice (Table S1). To investigate the blood glucose regulatory ability of IRE1a CKO mice, we performed oral glucose tolerance test (OGTT). The measurement of blood insulin level during OGTT revealed that insulin level at its peak was decreased by 30,50% in IRE1a CKO mice compared with control mice. On the other hand, the measurement of blood glucose level during OGTT showed that IRE1a CKO mice have a significantly higher blood glucose level at its peak than control mice and that IRE1a CKO mice required a longer time for blood glucose level to recover than control mice ( Figure 1B). Moreover, we measured the level of insulin extracted from the pancreas and islets of IRE1a CKO mice by ELISA in which the amount of total proteins was used as an internal standard. In addition, we performed immunohistochemical analysis of insulin and glucagons in the tissue sections of the pancreas of IRE1a CKO mice. The level of insulin extracted from the pancreas and islets of IRE1a CKO mice was quite similar to that of control mice (Table S2). The signals of insulin and glucagons in the pancreas of IRE1a CKO mice were normally detected (Figure 2A). Number and size of the islets and ratio of insulin + cells to glucagons + cells per islet were also nearly equal between IRE1a CKO and control mice ( Figure 2B). Thus, there is no histological difference in the islets between IRE1a CKO and control mice.
IRE1a disruption causes histological abnormality of the pancreatic acinar and salivary gland serous tissues As described above, the embryonic lethality of XBP1 KO mice could be rescued by the expression of a liver-specific XBP1 transgene; however, the mice die a few days after birth. This death is thought to be caused by the dysfunction of digestion-associated exocrine glands, such as the pancreas and salivary glands [22]. Therefore, we investigated the size and histological phenotype of the pancreas and salivary glands of IRE1a CKO mice. We found that the size of the salivary glands of the IRE1a CKO mice was normal, but their pancreas was 30,50% smaller than that of control mice ( Figure 3A and 3B and Figure S1). HE-stained sections revealed that the salivary gland serous acini of the IRE1a CKO mice were half the size compared with those of the control mice; however, the size of the salivary gland mucous acini of the IRE1a CKO mice was similar to that of the control mice ( Figure 3C and 3D and Figure S2 and S3). On the other hand, the pancreatic acinar tissue of the IRE1a CKO mice showed local spongy lesion with loss of acinar cells ( Figure 3E and Figure S4), which started occurring four weeks after birth. We suspected that this pathological change was caused by apoptosis; however, TUNEL analysis showed no difference between the IRE1a CKO mice and the control mice (Table S3). Next, we examined the effects of IRE1aon the production of digestive enzymes in the pancreas and salivary glands. Amylase is one of the key digestive enzymes produced both in the pancreatic acinar and salivary gland serous cells. Western blot analysis performed using the tissue lysate and saliva revealed no difference in the amount of amylase expression and secretion between the IRE1a CKO mice and the control mice ( Figure 3F).

IRE1a is necessary for the differentiation of B cells into plasma cells
Previous studies have demonstrated that IRE1a was activated during the differentiation of B cells into plasma cells and that XBP1 played an essential role during the same process [21,23,25].
To investigate the roles of IRE1a in the differentiation of B cells in mice, serum IgM and IgG1 as well as the populations of IgM + , IgD + and B220 + cells in splenocytes were analyzed in the IRE1a CKO mice. Serum IgM and IgG1 serve as markers of plasma cell differentiation, and IgM, IgD and B220 serve as markers of mature B cells. ELISA revealed that serum levels of IgM and IgG1 were reduced by half in the IRE1a CKO mice compared with the control mice. Lipopolysaccharide (LPS) stimulation increases the concentration of these immunoglobulins two-to three-fold; however, as in the case of no LPS stimulation, the levels of IgM and IgG1 were approximately half in the IRE1a CKO mice compared with the control mice ( Figure 4A). Flow cytometry analysis of splenocytes showed that the populations of positive or negative cells for IgM, IgD and B220 markers were similar between the IRE1a CKO mice and the control mice ( Figure 4B). This suggests that IRE1a disruption causes decrease of serum level of immunoglobulin produced in the plasma cells, but does not have any effect on the differentiation into mature B cells.

Hepatic function of IRE1a-deficient mice is normal
Hepatic abnormality has been reported in the embryo of both IRE1a KO mice and XBP1 KO mice [19,21]. Moreover, XBP1 has been shown as necessary for lipid metabolism in the hepatocytes of mature mice [24]. On the other hand, since histological analysis of the embryonic liver of IRE1a CKO mice previously failed to find abnormality [18], we performed phenotype analysis of the liver of IRE1a CKO mature mice by measurement of serum AST and ALT, observation of HE-stained sections, and analysis of the expression of lipid metabolism-related genes. However, no clear differences were found between the IRE1a CKO mice and the control mice ( Figure 5).

Functional IRE1a mRNA is significantly disrupted in IRE1a CKO mice
Quantitative PCR analysis of IRE1a and XBP1 in various tissues was performed to determine the degree of functional disruption of IRE1a mRNA in the IRE1a CKO mice. In all the tissues tested, PCR signals of IRE1a were detected in the control mice as previously reported [18], but weakly or not detected in the IRE1a CKO mice ( Figure 6A). Although PCR signals of total XBP1 were similarly detected in both the control and IRE1a CKO mice ( Figure 6B), those of spliced XBP1 were slightly or obviously detected in the control mice, but little in the IRE1a CKO mice ( Figure 6C and 6D). In the case that mice were injected with tunicamycin, an ER streesor, the spliced XBP1 level of IRE1a CKO mice were hardly detected in all the tested tissues, but that of control mice strongly increased in the liver and kidney and slightly increased in other tissues ( Figure 6C and 6D). In addition, although IRE1b is specifically expressed in the gastrointestinal tract [26], IRE1b expression was examined in the same tissues by quantitative PCR. As expected, PCR signals of IRE1b were predominantly detected in the gastrointestinal tract in both the control and IRE1a CKO mice, but slightly in the other tissues ( Figure 6E). These results confirm that functional IRE1a mRNA and spliced XBP1 mRNA are significantly disrupted in the IRE1a CKO mice and suggest that IRE1b may not complement the role of IRE1a in the IRE1a CKO mice. On the other hand, IRE1a mRNA and spliced XBP1 mRNA were slightly detected in some tissues of the IRE1a CKO mice. Thus, IRE1a is not completely disrupted in the IRE1a CKO mice, which may be the reason why the phenotype of IRE1a CKO mice is not clear-cut.

Discussion
ER stress is associated with various diseases [27], and the in vivo functions of ER stress-related molecules are presently receiving attention. With regard to IRE1a the functional analysis of this molecule in adult KO mice has been lagging because the KO mice suffer embryonic lethality; however, we could show here using the conditional KO method that IRE1a-deficient mice exhibit mild hypoinsulinemia, hyperglycemia, and a low-weight trend, and that IRE1a disruption causes histological abnormality of the pancreatic acinar and salivary serous tissues and decrease of serum level of immunoglobulin produced in the plasma cells.
In the present study, the level of insulin protein extracted from the pancreas and islets was similar between the IRE1a CKO mice and the control mice. Presumably, insulin expression level is normal in IRE1a-deficient b cells. Therefore, the lower level of blood insulin in the IRE1a CKO mice may be due to an abnormality of insulin secretion. Insulin is prevented from entering its secretory pathway by the quality control of ER proteins if it does not receive proper modifications (e.g., folding and disulfide bond formation) during its biosynthesis and secretion. It is possible that IRE1a is indirectly involved with the maturation process of insulin in the ER by inducing the expression of chaperones, disulfide isomerase and other enzymes. On the other hand, based on in vitro analysis of the effect of IRE1a on insulin, suppression of IRE1a has been reported to cause low insulin biosynthesis at the translation level but has little impact on insulin secretion [28]. The downregulation of IRE1a has also been reported to suppress insulin mRNA degradation that occurs under a chronic highglucose environment [29]. However, we have not yet performed experiments to confirm these findings, because it is not easy to prepare many samples and mice for detailed experiments on insulin biosynthesis; in theory, the IRE1a CKO mice that we used here have only a 25% chance of being born. In the future, we should aim to analyze the functions of IRE1a in association with insulin biosynthesis in the b cells of the islets by generating mice in which IRE1a is disrupted specifically in b cells. It was previously reported that loss of XBP1 had no effect on the development of the islets [22] and that XBP1 +/2 mice tended to have high blood glucose levels [30]. However, it remains unclear how IRE1a and XBP1 function in terms of insulin biosynthesis/secretion and the control of blood glucose level. In other words, it is uncertain whether IRE1a and XBP1 function on the same pathway or independently. To answer these questions, we need to generate mice expressing the active (spliced) form of XBP1. We believe that valuable information on the functional relationship between IRE1a and XBP1 can be obtained by analysis of insulin biosynthesis/secretion and blood glucose level in the offspring produced by crossing IRE1a KO mice and XBP1 KO mice to the mice expressing the spliced XBP1.
Size reduction of acinar tissue and partial deletion of acinar cells in the pancreas and size reduction of serous acini in the salivary gland were observed in the IRE1a CKO mice. The pancreatic acinar tissue and salivary gland serous tissue are both representative organs for the secretion of digestive enzymes and produce a large amount of proteins such as amylase. As with insulin, amylase is prevented from entering the secretory pathway by quality control of ER proteins if it does not receive proper modifications in the ER (e.g., folding and disulfide bond formation) during its biosynthesis and secretion. IRE1a is activated in the pancreatic acinar tissue and salivary gland even under physiological conditions and excessive expression of amylase causes ER stress and IRE1a activation [31,32]. Based on these previous findings, it is possible that IRE1a is involved in homeostasis in the ER and the development and/or survival of exocrine cells in the pancreatic acinar tissue and salivary gland by constantly inducing the expression of chaperones, disulfide isomerase and other enzymes. As described above, a developmental disorder of the pancreatic acinar tissue and salivary gland is also observed in XBP1 CKO mice [22]. Therefore, signal transduction from IRE1a to XBP1 is important for the development and/or maintenance of the pancreatic acinar tissue and salivary gland. However, unlike in IRE1a CKO mice, phenotypes are more severe in XBP1 CKO mice as exemplified by observation of apoptotic signals in the pancreatic acinar tissue, reduced production of digestive enzymes in the pancreatic acinar tissue and salivary gland, indigestion of food, and death shortly after birth [22]. This may indicate the importance of IRE1a-independent XBP1 functions for the development and/or maintenance of the pancreatic acinar tissue and salivary gland. Like hormones (e.g., insulin and glucagons) and digestive enzymes, antibodies are also secretory proteins and are prevented from entering their secretory pathway by quality control of ER proteins if they do not receive proper modifications in the ER (e.g., folding and disulfide bond formation) during their biosynthesis and secretion. Therefore, IRE1a is believed to play an important role in plasma cells in terms of the production and quality of antibodies. In the present study, the serum levels of IgM and IgG1 in the IRE1a CKO mice were half of those in the control mice, and this trend was observed regardless of the presence of LPS stimulation. Previous studies also revealed IRE1a activation during the differentiation of B cells into plasma cells and that XBP1 played an essential role in the differentiation process [21,23,25]. Based on these findings, IRE1a may contribute to the normal production of antibodies by activating XBP1 on known UPR pathways, thereby inducing the expression of chaperones, disulfide isomerase and other enzymes in the ER.
Hepatic abnormalities in the embryos of IRE1a KO and XBP1 KO mice have been reported [19,21], while the absence of histological abnormalities in the embryonic liver of IRE1a CKO mice has been described [18]. Also, XBP1 has been shown to be essential for lipid metabolism in the hepatic cells of mature mice [24]. In the present study, hepatic abnormalities were not found on analysis of IRE1a CKO mice. This may suggest the existence of IRE1a-independent XBP1 functions. If XBP1 exerts functions depending on IRE1a, splicing of XBP1 mRNA by IRE1a would be necessary. Also, splicing of XBP1 mRNA is blocked by inactivation of IRE1a [18]. Given these observations, the functions of IRE1a-independent XBP1 may be brought about by unspliced XBP1.
Phenotype analysis of IRE1a CKO mice in this study provided important information regarding the physiological functions of IRE1a. Comparison of our results with the phenotypes of XBP1 KO mice previously reported indicates the possibilities that IRE1a has functions independent of XBP1 and XBP1 has functions independent of IRE1a. This study therefore provided findings that are important for revealing new roles played by IRE1a and XBP1 in vivo. In this regard, however it is also possible that the discrepancies between the present data and those from Glimcher's [22,24] and Kaufman's [21] groups may depends on difference among each experimental condition. For example, in conditional KO mice with the exogenous target gene [21,22], the exogenous target gene may be different from the endogenous target gene in expression level and have any unexpected dominant effects, unlikely in conditional KO mice with flox/Cre system. Further, difference in mouse genetic background may raise distinct compensatory mechanism by other UPR pathways between IRE1a and XBP1 CKO mice even if these mice are commonly generated with flox/ Cre system. Therefore we need to carefully analyze the phenotypes of IRE1a and XBP1 disruption under an identical experimental condition in the future.       . Tun 2 or + indicates i.p. injection with saline or tunicamycin (500 ng/g body weight) 16 before tissue collection. Li; liver, K; kidney, Sp; spleen, H; heart, Lu; lung, B; brain, Sa; salivary gland, St; stomach, In; intestine, Co; colon, Pa; pancreas, Is; islet. All the data were obtained from 20 weeks old male mice. doi:10.1371/journal.pone.0013052.g006