Tissue Transglutaminase Directly Regulates Adenylyl Cyclase Resulting in Enhanced cAMP-response Element-binding Protein (CREB) Activation*

Tissue transglutaminase (tTG) is present in the human nervous system and is predominantly localized to neurons. Treatment of human neuroblastoma SH-SY5Y cells with retinoic acid results in increased tTG expression, which is both necessary and sufficient for differentiation. The goal of the present study was to determine whether tTG modulates the activation of the cyclic AMP-response element (CRE)-binding protein, CREB, an event that likely plays a central role in the differentiation of SH-SY5Y cells. SH-SY5Y cells stably transfected with active wild type tTG, tTG without transamidating activity (C277S), an antisense tTG construct that depleted the endogenous levels of tTG, or vector only were used for the study. Treatment with forskolin, an adenylyl cyclase activator, increased that activation-as-sociated phosphorylation of CREB, which was pro-longed by tTG overexpression. CRE-reporter gene activity was also significantly elevated in the tTG cells compared with the other cells. The enhancement of CREB phosphorylation/activation in the tTG cells is likely due to the fact that tTG significantly potentiates cAMP production, and our findings indicate that tTG enhances adenylyl cyclase activity by modulating the conformation state of adenylyl cyclase. This is the first study to provide evidence of the mechanism by which tTG may contribute to neuronal differentiation.

Tissue transglutaminase (tTG) 1 is the most ubiquitously expressed member of the transglutaminase family of proteins (1). This protein exhibits at least two distinct enzymatic activities as follows: (i) a Ca 2ϩ -dependent transamidating activity that cross-links proteins or incorporates polyamines into protein substrates (2,3); and (ii) GTPase activity (4 -6). Tissue TG is present in the human peripheral and central nervous system and is localized predominantly in neurons (7,8). In rat brain and spinal cord, TG activity is highest in the late fetal stage (8,9). Additionally, tTG protein levels and TG activity have been shown to be elevated in neurodegenerative conditions such as Alzheimer's disease (10,11) and Huntington's disease (12,13).
However, the role of tTG in the pathological processes of these diseases is not clear. In the nervous system, tTG has been postulated to play a role in many different processes such as synaptic plasticity (14,15), release of neurotransmitters (16), long-term potentiation (17), axonal regeneration (9), and neuronal death and/or survival (18). Tissue TG also clearly plays a key role in neuronal differentiation (19,20). Indeed, tTG is both necessary and sufficient for the neuronal differentiation of human neuroblastoma SH-SY5Y cells (20).
The cyclic AMP-response element (CRE)-binding protein, CREB, was originally identified as a transcriptional activator that responds to cAMP and is phosphorylated and activated by protein kinase A (21). CREB belongs to the CREB transcription factor family that, besides CREB, consists of the cAMP modulatory protein (CREM) and activating transcription factor 1 (ATF1) (22,23). CREB has been postulated to play a key role in the modulation of neuronal differentiation, neuronal survival, and synaptic plasticity in the vertebrate nervous system (24). Indeed, CREB phosphorylation and CRE reporter gene activity are increased in cortical neurons during specific periods of nervous system development (25). Overexpression of CREB in SK-N-BE cells leads to morphological alterations typical of neuronal differentiation (26). In contrast, overexpression of a dominant negative CREB construct induced a round, poorly differentiated phenotype (26). Furthermore, CREB has been described as the first transcription factor to be activated in response to an elevation in intracellular cAMP levels (27).
Cyclic AMP has multiple, diverse effects on cellular processes such as gene expression, cell cycle control, and cell survival-death decisions (for recent reviews, see Refs. 28 -31). Cyclic AMP also promotes neuronal survival in a manner independent of neurotrophic factors (32)(33)(34). Although cAMP has been shown to have other functions in the cell, the primary mechanism by which it acts as a second messenger is by activation of protein kinase A (35), and CREB is a primary target of protein kinase A (21). In a previous study (20), we demonstrated that tTG is required for development of the neuronal morphology of human neuroblastoma SH-SY5Y cells. Indeed, treatment of SH-SY5Y cells with the transglutaminase inhibitor monodansylcadaverine inhibited the retinoic acid-induced differentiation of SH-SY5Y cells (36). Although it is clear that tTG is necessary for neuronal differentiation, the molecular mechanisms by which tTG contributes to this process are unclear. Therefore, the goal of the present study was to determine whether tTG modulates CREB phosphorylation/activation, which likely plays a significant role in the differentiation of SH-SY5Y cells (37,38). To address this question, SH-SY5Y cells stably transfected with active wild type tTG, tTG without transamidating activity (C277S), an antisense tTG construct that depletes endogenous tTG, or vector only (18,20) were treated with forskolin, an adenylyl cyclase activator (39). For-skolin treatment resulted in a significant enhancement of CREB phosphorylation that lasted longer in the cells overexpressing tTG compared with the other cell lines. Furthermore, CRE-reporter gene activity was also significantly elevated in the tTG-overexpressing cells compared with the other cells. The enhancement of CREB phosphorylation/activation in the tTG-overexpressing cells is likely due to the fact that tTG significantly potentiates cAMP production, and our findings indicate that tTG directly modulates adenylyl cyclase activity, most likely by facilitating the conversion of adenylyl cyclase into a more active conformation. This is the first study to provide evidence of the mechanism by which tTG may contribute to neuronal differentiation.

EXPERIMENTAL PROCEDURES
Cell Culture and Treatments-Human neuroblastoma SH-SY5Y stably expressing wild type tTG (tTG), tTG without transamidating activity due to a point mutation within the active site (C277S), and an antisense tTG DNA construct (anti-tTG) along with cells stably transfected with vector-only pcDNA3.1 (SH/pc) were described previously (18,20). In some experiments we used two different tTG cell lines, tTG-1 and tTG-2, and two different C277S cell lines, C277S-1 and C277S-2. The majority of experiments were carried out in the tTG-2 and C277S-2 cell lines, which are referred to as tTG cells and C277S cells, respectively, throughout.
Dual Luciferase Assay-Cells were grown in low serum conditions with 1% Fetal Clone II (HyClone) and 4% horse serum to ϳ30% confluency prior to transient transfection with the pCRE-Luc reporter construct (Stratagene) and pRL-TK vector as an internal control (Promega, Madison, WI) at a 10:1 ratio. Thirty-six hours later, the cells were transferred to serum free media for 3 h subsequent to incubation for 4 h with 10 M forskolin. Luciferase levels were measured using the Dual-Luciferase reporter assay system kit (Promega) and a TD-2021 luminometer (Turner Designs, Sunnyvale, CA). The Firefly luciferase data was normalized to the Renilla luciferase values for each sample.
Measurement of cAMP Accumulation-The levels of cAMP in control and stimulated cells were determined by using an enzyme immunoassay kit according to the manufacturer's protocol (Amersham Biosciences). Briefly, 10 4 cells/well were plated in low serum conditions with 1% Fetal Clone II and 4% horse serum 12 h before being transferred to serum-free media for 3 h prior to treatment. After each treatment, the cells were rinsed and then incubated with the lysis reagent (0.5% dodecyltrimethylammonium bromide) for 15 min at room temperature. Cellular extracts were then transferred to individual wells of a 96-well plate pre-coated with donkey anti-rabbit antibody, and then rabbit antiserum against cAMP was added, and the plates were incubated for 2 h. cAMP-horseradish peroxidase conjugates were added to each individual well and incubated at 4°C for 1 h. After extensive washing, peroxidase substrate was added, incubation was carried out at room temperature for 1 h, and the plate was read at 450 nm after stopping reaction with 50 l of 1 M sulfuric acid.
Statistics-Data were analyzed using analysis of variance (ANOVA), and values were considered significantly different when p Ͻ 0.05. Results were expressed as mean Ϯ S.E.

Elevated tTG Expression Leads to Enhanced CREB Phosphorylation and Its Transcriptional Activation in Response to
Forskolin-Tissue TG is essential for retinoic acid-induced neuronal differentiation of SH-SY5Y cells (20,36). One of the proteins that likely plays a key role in differentiation and survival processes in neuronal cells is the CREB protein (24,41). Phosphorylation of Ser 133 , located within a 60-amino acid kinase-inducible domain, is a prerequisite for CREB transcriptional activation (21). To enhance CREB phosphorylation on Ser 133 , cells were treated with forskolin, a direct adenylyl cyclase activator (39). As expected, treatment with 10 M forskolin resulted in a significant and transient increase in CREB phosphorylation on Ser 133 in all of the human neuroblastoma SH-SY5Y cell lines (Fig. 1). However, in the tTG cells, forskolin treatment resulted in a more robust and sustained CREB phosphorylation (Fig. 1). To determine whether the differences in CREB phosphorylation in response to forskolin treatment in the different SH-SY5Y cell lines resulted in differential CREB transcriptional activity, studies were carried out with a lucif-FIG. 1. Forskolin-induced CREB phosphorylation at Ser 133 is enhanced in cells overexpressing wild type tTG. Human neuroblastoma cells stably transfected with vector only (SH/pc), wild type tTG (tTG), transglutaminase-inactive tTG mutant (C277S), or antisense tTG (anti-tTG) were exposed to 10 M forskolin for the indicated time periods. A, cell lysates were immunoblotted for phosphorylated CREB (p-CREB) or total CREB. B, quantitation of the increase in CREB phosphorylation at Ser 133 in response to 10 M forskolin. Values were normalized to the total amount of CREB in each fraction. Enhancement of CREB phosphorylation was observed in the tTG-overexpressing cells compared with the other cells. erase CRE-reporter vector. In the control SH/pc cells, forskolin treatment resulted in a significant increase in CRE-luciferase activity. CRE-luciferase was also elevated in the tTG cells in response to forskolin, although the increase was significantly greater then it was in the SH/pc cells (6.9-fold increase for the tTG cells compared with a 3.7-fold increase for the SH/pc cells) (Fig. 2). In contrast, forskolin-stimulated CRE activity was significantly attenuated in the anti-tTG cells compared with the SH/pc or tTG cells (Fig. 2).
cAMP Accumulation Is Directly Modulated by tTG Expression in SH-SY5Y Cells-The data described above suggest that tTG may increase forskolin-stimulated cAMP production, which, in turn, leads to enhanced CREB activation. Therefore, basal and forskolin-stimulated cAMP levels were measured in two different tTG cell lines, tTG-1 and tTG-2, and two different cell lines overexpressing transglutaminase inactive C277S-tTG, C277S-1 and C277S-2, along with the control SH/pc cells and the anti-tTG cells. Basal cAMP levels in all cell lines were in the range of 30 -70 fmol/well (Fig. 3). However, after stimulation with 10 M forskolin, the different cell lines exhibited significantly different increases in the levels of cAMP. In both wild type tTG-expressing cell lines, cAMP levels were elevated to a significantly greater extent compared with the SH/pc cells. In contrast, forskolin-stimulated cAMP levels in the anti-tTG and the C277S cells were significantly lower than what was observed in the SH/pc cells (Fig. 3). Given the finding that both tTG cell lines and both C277S cell lines responded similarly, in all subsequent experiments just the tTG-2 and C277S-2 cells lines were used and referred to as tTG cells and C277S cells, respectively. These were also the cell lines used for Figs. 1 and 2.
The level of intracellular cAMP is regulated by the balance between the activities of adenylyl cyclase and cyclic nucleotide phosphodiesterase (PDE) (42,43). To test whether the observed enhancement of cAMP accumulation in the tTG cells is due to a decrease in the levels and/or activity of PDE, cells were pretreated for 3 h with 30 M rolipram, a specific inhibitor of the high-affinity cAMP PDE (44), prior to treatment with 10 M forskolin for 5 min. Rolipram treatment significantly increased cAMP levels only in the tTG cells from a basal level of 46 Ϯ 12 fmol cAMP/well to 281 Ϯ 69 fmol cAMP/well after rolipram treatment (Fig. 4A). Treatment with rolipram and forskolin together resulted in a dramatic synergistic increase of cAMP in all the cell lines (Fig. 4B). Furthermore, the extent of cAMP accumulation in the tTG cells was significantly greater then that observed in the other cell lines (Fig. 4B). These data indicate that decreases in PDE are not responsible for enhanced cAMP production in the tTG cells, as inhibiting PDE did not prevent the enhanced forskolin-stimulated cAMP accumulation in the tTG cells compared with the other cell lines.
Several studies indicate that there are two conformations of adenylyl cyclase for forskolin binding. There is a lower affinity binding state, wherein forskolin binds and activates adenylyl cyclase at concentrations of Ն10 M, and a higher affinity binding state, in which forskolin can activate adenylyl cyclase at a concentration of ϳ0.1 M (45). Furthermore, the higher affinity forskolin binding state of adenylyl cyclase is usually due to the binding of G␣ s (46), which results in a conformational change in adenylyl cyclase. Treatment of the cells with 0.1 M forskolin for 5 min significantly increased cAMP levels only in the tTG cells from a basal level of 30 Ϯ 20 fmol cAMP/ well to 192 Ϯ 26 fmol cAMP/well after forskolin treatment (Fig.  5A). In contrast, no changes in cAMP levels were observed in either the SH/pc or C277S cells (Fig. 5A). This suggests that adenylyl cyclase in the tTG cells may be in a conformation that allows forskolin to bind with a high affinity.
To assess whether the G␣ s subunit may be involved in the observed enhancement of forskolin-stimulated adenylyl cyclase activity in the tTG cells, cholera toxin (CTX) was used in the next set of experiments. CTX catalyzes the NAD-dependent ADP-ribosylation of the G␣ s subunit (47), which results in inhibition of the intrinsic GTPase activity of G␣ s and stabilization of the activated conformation (48). CTX-induced stabilization of the activated conformation of G␣ s results in a longlived activation of adenylyl cyclase (49,50). Treatment with 20 M CTX alone for 4 h did not result in any significant increases in cAMP levels in either the SH/pc or C277S cells (Fig. 5B). However, CTX treatment of the tTG cells increased cAMP levels from 30 Ϯ 20 fmol cAMP/well under basal conditions to 331 Ϯ 83 fmol cAMP/well (Fig. 5B). Treatment with CTX and 0.1 M forskolin together resulted in a dramatic synergistic increase in cAMP levels in all the cell lines (Fig. 5C); however, the extent of cAMP accumulation in the tTG cells was significantly greater then than that observed in the other cell lines (Fig. 5C). In the next set of experiments, cAMP accumulation was measured in the different cells lines after treatment with CTX alone for 4 h, 10 M forskolin for 5 min, or CTX for 4 h and, subsequently, 10 M forskolin for an additional 5 min (Fig. 5D). CTX treatment alone had no effect on the cAMP levels in the SH/pc and C277S cells; however, as indicated above, CTX treatment did significantly increase cAMP levels in the tTG cells. In the SH/pc cells, treatment with 10 M forskolin increased cAMP levels to 1088 Ϯ 62 fmol cAMP/well, and pretreatment with CTX followed by treatment with 10 M forskolin resulted in a synergistic increase in cAMP levels to 2570 Ϯ 165 fmol cAMP/well. Similarly, in the C277S cells, cAMP levels increased from 452 Ϯ 102 fmol cAMP/well in the presence of 10 M forskolin to 1174 Ϯ 24 fmol cAMP/well after treatment with CTX and 10 M forskolin. Surprisingly though, a similar synergistic enhancement of cAMP accumulation in the tTG cells was not observed. In the tTG cells, cAMP levels were 2348 Ϯ 198 fmol cAMP/well in the presence of 10 M forskolin and 3351 Ϯ 67 fmol cAMP/well in the presence of CTX and forskolin (Fig. 5D). This indicates that adenylyl cyclase in the tTG cells is already in a more activated state, and, thus, pretreatment with the CTX is not as effective at increasing forskolin-induced cAMP production as it is in the other cell lines.
Expression of Adenylyl Cyclase, G␣ s , and G␣ i Are Unaltered by tTG Expression-To evaluate whether the observed differences in the rate of cAMP accumulation may be associated with differences in the expression of adenylyl cyclase, G␣ s , and/or G␣ i , an immunoblot analysis was performed. The results indicate that there are no significant differences in the protein expression levels of these proteins in SH/pc, tTG, anti-tTG, and C277S cells that correlate with the enhanced cAMP production in the tTG cells (Fig. 6).

DISCUSSION
The aim of this study was to gain insight into the mechanism by which tTG contributes to neuronal differentiation processes (19,20). The results presented in this study clearly demonstrate that tTG greatly facilitates adenylyl cyclase activation, resulting in enhanced cAMP production and subsequent CREB phosphorylation and activation. Furthermore, the data indicate that tTG directly or indirectly facilitates the transition of adenylyl cyclase into an activated state. These findings are significant given the fact that the cAMP-CREB pathway has been shown to play a critical role in neuronal differentiation processes (33,51).
It is well established that treatment of human neuroblastoma SH-SY5Y cells with retinoic acid results in differentiation (52). The retinoic acid-induced neuronal-like phenotype is accompanied by significant increases in tTG expression (3,20). This characteristic of SH-SY5Y cells makes them an ideal model for investigating the role of tTG in neuronal differentiation. Previously, we demonstrated that tTG appears to be necessary and sufficient for the development of the neuronal morphology of SH-SY5Y cells. Overexpressing wild type tTG in SH-SY5Y cells results in spontaneous differentiation in 5% serum conditions. In contrast, overexpression of an antisense tTG construct that suppresses the expression of endogenous tTG results in cells that are unresponsive to retinoic acid in that they continue to proliferate and do not extend neurites even after extended retinoic acid treatment (20). This fact clearly indicates that tTG is an essential component of the differentiation process of human neuroblastoma SH-SY5Y cells. The finding that treatment of SH-SY5Y cells with the transglutaminase inhibitor monodansylcadaverine inhibited retinoic acid-induced differentiation has further substantiated this conclusion (36). Although it is clear that tTG is essential for differentiation to a neuronal phenotype, the cell signaling events modulated by tTG that regulate this process are not known.
Previous studies have provided evidence that the cAMP-CREB pathway is of a critical importance for appropriate neuronal differentiation (26,53). Indeed, increasing cAMP levels have been shown to induce neuronal differentiation in several cell models (33,51,54). Furthermore, a critical role for CREB in neuronal differentiation has been clearly illustrated by the fact that overexpression of CREB in SK-N-BE cells leads to morphological alterations typical of neuronal differentiation (26). In contrast, overexpression of a dominant negative CREB construct results in a poorly differentiated phenotype (26). Moreover, in another neuronal-like cell system, PC12 cells, neuronal differentiation has been documented to be tightly associated with CREB transcriptional activation (53), and elevation of cAMP in PC12 cells results in morphological changes similar to those observed in response to nerve growth factor treatment (53). Additionally it has been shown that, in CREB null mice, axonal growth is severely compromised (55,56). Thus, it is readily evident that any processes that result in sustained CREB activation are likely to contribute significantly to the neuronal differentiation process. Therefore, it is reasonable to hypothesize that the tTG-induced potentiation of CREB activation may be one mechanism by which tTG facilitates neuronal differentiation.
Because both CREB and the CREB-binding protein have glutamine-rich regions (22,57), we originally postulated that these proteins were substrates of tTG. However, to date we have not been able to demonstrate that either CREB or the CREB-binding protein are substrates of tTG in situ. Given this situation, we focused on signaling events upstream of CREB activation. The data in this study strongly suggest that tTG is either directly or indirectly modulating adenylyl cyclase. Even though the molecular mechanism of activation of adenylyl cyclase by tTG remains to be elucidated, our findings indicate that tTG probably affects the conformational state of adenylyl cyclase, leading to a more responsive state. This mechanism appears to be most probable given the activation of adenylyl cyclase in the tTG cells with a very low forskolin concentration (0.1 M) and the fact that there was the fact that there was a significant accumulation of cAMP upon treatment with rolipram only in the tTG cells. Indeed, it is clear that adenylyl cyclase exists in multiple conformational states, elicited by binding of effectors such as G␣ subunits, forskolin, or by occupancy of the ATP binding site (42,58). Therefore, we postulate that increased tTG expression in SH-SY5Y cells results in a change in the conformational state of adenylyl cyclase, which leads to an increased sensitivity to the activating agents.
Our findings as to the role of tTG in the modulation of adenylyl cyclase activity and the accumulation of intracellular cAMP are in contrast to the findings of Gentile et al. (59). In this previous study it was postulated that wild type tTG or its TG-inactive mutant, C277S, inhibits adenylyl cyclase activity in human fibroblast Balb-C 3T3 and bovine aorta endothelial cells. The apparent discrepancy between the present study and FIG. 6. The expression of adenylyl cyclase, G␣ s , and G␣ i are not affected by tTG overexpression. Immunoblot analysis for adenylyl cyclase (AC) was carried out using a membrane enriched fraction, while whole cell lysates were probed for either G␣ s or G␣ i . Actin was used a loading control. These data indicate that alterations in the expression levels of these proteins do not contribute to the enhanced cAMP production in the tTG cells.
the study by Gentile et al. (59) may be associated with different cell models and experimental approaches. One the other hand, the present finding may emphasize the complexity of tTG function in different cell lineages. In conclusion, we believe that the present findings add a new dimension to our understanding of the functions of tTG in the nervous system and clearly indicate that tTG plays a central role in regulating neuronal signaling processes.