Regulation of the glutamine synthetase gene expression in the liver - Stanulovic proefschrift

UvA-DARE (Digital Academic Repository)

Regulation of the glutamine synthetase gene expression in the liver

Stanulovic, V.

Publication date

2007

Link to publication

Citation for published version (APA):

Stanulovic, V. (2007). Regulation of the glutamine synthetase gene expression in the liver.

General rights

It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).

Disclaimer/Complaints regulations

If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask the Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible.

Regulation of the Glutamine Synthetase

Gene Expression in the Liver

ISBN/EAN: 978-90-9022444-2 Printed by Grafomed-trejd, Bor, Serbia

© 2006 by Vesna Stanulović

No parts of this thesis may be reproduced, stored in a retrieval system, or transmitted in any form or by any meands without permission of the author.

REGULATION OF THE GLUTAMINE SYNTHETASE GENE EXPRESSION IN THE LIVER

ACADEMISCH PROEFSCHRIFT

ter verkrijging van de graad van doctor aan de Universiteit van Amsterdam op gezag van de Rector Magnificus

prof.dr. D.C. van den Boom

ten overstaan van een door het college voor promoties ingestelde commissie, in het openbaar te verdedigen in de Agnietenkapel

op donderdag 20 december 2007, te 12.00 uur door

Vesna Stanulovic

Promotiecommissie:

Promotor: Prof.dr. W.H. Lamers Co-promotor: Dr. T.B.M. Hakvoort Overige leden: Prof.dr. A.F.M. Moorman

Prof.dr. F. Baas Prof.dr. R. van Driel Dr. G.T.J. van der Horst

Dr. A.K. Groen

Contents

Scope

9

Chapter 1

13

Introduction

Chapter 2

30

The 3'-UTR of the Glutamine Synthetase Gene Interacts Specifically with

Upstream Regulatory Elements, Contains mRNA-Instability Elements and

is Involved in Glutamine Sensing

Chapter 3

52

DBP- and REV-ERB-Binding Sites in the First Intron of the Glutamine

Synthetase Regulate Its Glutamine-Dependent Expression

Chapter 4

80

Hepatic HNF4α Deficiency Induces Periportal Expression of Glutamine

Synthetase and Other Pericentral Enzymes

Chapter 5

108

β-Catenin Suppresses HNF4α Expression in the Liver

Scope

Glutamine synthetase (GS) catalyzes the ATP-dependent conversion of glutamate and ammonia to glutamine. The gene is expressed at various levels in a wide range of tissues. The expression of GS is higher in cells that are involved in glutamate-removal and ammonia-detoxification (e.g. astrocytes and pericentral hepatocytes) than in glutamine-producing cells (e.g. myocytes and adipocytes), suggesting a complex regulatory network that determines the localization and level of GS expression.

Our interest focuses on the regulation of GS expression in the liver, where it is expressed in a narrow, 2-3 cell-layers thick rim of hepatocytes around the central veins. This pattern of expression is very stable and only modified under extreme metabolic or hormonal conditions. Glucocorticoids and cAMP and the ambient concentration of glutamine influence GS mRNA and protein levels. The pericentral position of GS is at least partly regulated by activated β-catenin via the -2.5kb upstream enhancer element (UE). The intention of this thesis is to deliver a better understanding of the regulatory mechanism maintaining pericentral pattern of GS expression. Chapter1 reviews the influence of amino acid depletion in general and glutamine in particular on gene expression and especially on transcription. Chapter 2 investigates the interactions between the upstream enhancer, regulatory regions in the first intron, and the untranslated region of the GS gene (GS 3’-UTR) and its immediate downstream genomic sequences (collectively dubbed the GS "tail"). We report that the GS "tail" contains elements that destabilize GS mRNA. Additionally, the "tail", in conjunction with the upstream enhancer and the 3'-part of the first intron, mediates an increase in GS expression upon glutamine depletion. Chapter 3 explores the contribution of the 3'-part of the first GS intron to glutamine sensitivity. Subsequently, we show that glutamine sensitivity is dependent on transcription factors DBP and Rev-erb and is associated with circadian GS expression. In chapter 4 we show that HNF4α binds to the UE and inhibits GS expression in the periportal hepatocytes, thereby confining GS expression to pericentral hepatocytes. Chapter 5 investigates the dynamics of GS, HNF4α and β-catenin expression during the embryonic and early postnatal development, showing that the cessation of expression of all three factors at ED17 is followed by the first appearance of the pericentral GS and β-catenin expression. At this

point the pericentral gradient is very wide and its constriction to central veins coincides with the HNF4α reappearance and β-catenin translocation from the nucleus to membrane. Additionally, inhibitory interactions between activated β-catenin and HNF4α are revealed.

Chapter 1

Regulation of gene expression by amino acids

Amino acids are the building blocks of proteins. Depending on whether the body can synthesize a specific amino acid, they are divided in essential, conditionally essential, and nonessential amino acids. Essential amino acids cannot be synthesized and must, therefore, be obtained via the diet, while conditionally essential amino acids become essential only when bodily requirements exceed the endogenous capacity to synthesize them e.g. during rapid growth or strongly catabolic conditions. Certain amino acids are, in addition to their major function as protein building blocks, essential precursors of important biomolecules, including nucleotides and nucleotide coenzymes, heme, various hormones and neurotransmitters, and glutathione (54).

For a long time it was thought that only prokaryotes and unicellular eukaryotes could accommodate their gene expression pattern to amino acid deprivation (54). Inhibition of protein synthesis and arresting the cell cycle are the most obvious consequences of amino acid depletion that allow prolonged cell survival under non-permissive conditions (52). Culturing cells in the glutamine-depleted medium causes gradual cell death in the course of 5 days. Depletion of a single amino acid leads to a decrease in total protein levels of 62-78% in the case of essential amino acids and to 7-46% in the case of nonessential amino acids (2;47). Surprisingly, the overall level of transcription remains undisturbed. However, transcription of tissue-specific genes is suppressed, whereas that of genes that regulate cell adaptation and the stress response is upregulated (47). The increase in gene expression upon amino acid depletion results from an altered rate of transcription, mRNA splicing, nuclear export, stability and translation (21;27;36;37;39;41;43;44;49;55). Genes reported to be responsive to amino acid depletion and the mechanisms governing this responsiveness are listed in table 1. CCAAT/enhancer-binding protein-related gene (CHOP), asparagine synthetase (ASNS) and cationic amino acid transporter (CAT-1) have been used as model genes for understanding amino acid sensitivity, as they are all regulated in a similar fashion (for recent reviews, see (28;33)). This review provides a brief overview of the different mechanisms involved, with an emphasis on the mechanism of action of glutamine.

1. Amino acid-sensing pathways that affect mRNA translation

The regulation of protein synthesis is virtually synonymous with the regulation of translation. Growth-factor receptors (GFRs, e.g. the insulin receptor), amino acid availability and the activation of stress kinases influences translation. Excess amino acids, in particular leucine and signaling from growth factors is transferred via PI3K, PKB, TSC1/2 and Rheb to mTOR kinase that controls translation machinery and therefore protein production. The activation of mTOR by amino acids is still incompletely understood. The development of a deficiency in any of the amino acids activates the General Control Non-derepressible-2 (GCN2) stress kinase. Activation of GCN2 results in phosphorylation of the translation-initiation factor eIF2α and inhibition of protein synthesis.

1.a. The mTOR signalling pathway

The level of amino acids is detected by intracellular or membrane-bound sensors, with amino acid transporters being likely candidates. Signaling from growth factors and amino acids is transferred to mTOR kinase (for recent reviews, see (15;25)). Downstream targets of mTOR include elongation initiation factor 4E binding protein-1 (4E-BP1) and ribosomal protein S6 kinase-1 (S6K1). Sources of free amino acids in the body are autophagic-lysosomal and ubiquitin-proteasome pathways. Several amino acids (Leu, Gln, Tyr, Phe, Pro, Met, Trp and His) regulates autophagic proteolysis via mTOR, signaling cascade (30). In the presence of amino acids, mTOR phosphorylates 4E-BP1 and prevents its binding to elongation initiation factor 4E (eIF4E). Free eIF4E facilitates the initiation of translation and protein synthesis by connecting mRNAs with the 40S ribosomal subunit. Second substrate of mTOR is S6K1 that by phosphorylating the ribosomal protein S6 (rpS6) enhances translation of a subset of mRNAs. These mRNAs are termed terminal oligopyrimidine tract (TOP)mRNA since they contain an uninterruptedstretch of 7–15 pyrimidines, adjacent to the 5' cap. TOP mRNA code proteins involved in mRNA translation. Therefore mTOR regulates protein synthesis and consequently cell size and cell-cycle progression (15;25).

1.b. The GCN2 stress kinase signalling pathway

The GCN2 stress kinase detects a decreased availability of amino acids by sensing an increased concentration of uncharged tRNA. eIF2α facilitates initiation of mRNA translation by delivering the initiator Met-tRNA to the 40S ribosome.When the ribosome reaches the initiation codon, the GTP associated witheIF2α is hydrolyzed and eIF2α is released from the ribosome.Binding of a new GTP to eIF2α is attended by a guanine nucleotide exchange factor, eIF2B. Phosphorylation of eIF2α by GCN2 inhibits the guanine nucleotide exchange process and, thereby, the assembly of the translational machinery (46). This response to amino acid deprivation becomes fully active during the first hour of starvation (19).

1.b.1. The role of uORFs

Whereas the translation of most mRNAs is inhibited by phosphorylation of eIF2α, several mRNAs are preferentially translated under these circumstances to produce proteins that ensure prolonged cell survival. Examples are ATF4 and NFκB (27;39;53). Translational control of ATF4 is mediated via the 5' end of the ATF4 mRNA, which contains two conserved upstream open reading frames (uORFs). While the first uORF is only 9nt long, the second uORF is 177nt in length and overlaps with the first 83nt of the normal ATF4-coding region. Unstressed cells, with low levels of eIF2α phosphorylation, start translation from the second uORF, which inhibits subsequent translation of ATF4. In stressed cells, high levels of phospho-eIF2α increase the time required for the scanning ribosomes to initiate translation and therefore results in a shift from the inhibitory uORF2 towards the ATF4-encoding ORF (39;53). A similar regulation of protein translation by upstream ORFs is found for other transcription factors, such as the glucocorticoid receptor, the retinoic acid receptor, C/EBPα and CEBPβ (25).

1.b.2. The role of IRES

The conventional initiation of translation occurs when the 40S ribosomal subunit binds to the 5'm7G cap and then moves along the mRNA until an initiation codon is encountered

However, certain mRNAs contain an internal ribosome entry site (IRES) within their 5'-untranslated region, which allows initiation independently of the 5'-cap through a mechanism that also requires the phosphorylation of eIF2α (46). Examples are the enhanced translation of the system A-neutral amino acids transporter (SNAT-2) and cationic amino acid transporter-1 (Cat-1) upon amino-acid starvation (19;20).

2. Mechanisms for increased gene transcription during amino-acid deprivation.

As much as one third of genes may exhibit increased expression in response to amino-acid limitation (41;47). Unaltered gene expression was found for ubiquitously expressed "housekeeping" genes, whereas genes with a tissue-specific expression were downregulated (41;47). Amino-acid deprivation influences both the rate of transcription and the stability of mRNAs, but the amount of total poly(A)+ mRNAs stays the same (47). Specific inductive and repressive effects on gene expression are, therefore, quite common in amino acid-deprived cells. Unfortunately, genome-wide screens for genes that respond to changing amino-acid availability are not yet available.

Asparagine synthetase (ASNS) was the first reported example of this group of genes (22). The study of its transcriptional regulators has identified several transcription factors with a basic leucine zipper (bZIP) DNA-binding domain, in particular the activating transcription factors (ATFs) and CCAAT/enhancer-binding protein-beta (C/EBPβ), as the most important mediators of the response to amino-acid deprivation. In addition to ASNS, ATF 2, 3, and 4, and C/EBPβ increase the rate of transcription of CHOP (CCAAT/enhancer-binding protein-related gene), cationic amino acid transporter-1 (CAT-1), sodium-coupled neutral amino acid transporter 2 (SNAT2), as well as their own expression (3;20;38;49).

A cross-regulatory cascade between bZIP transcription factors appears to be largely responsible for the amino-acid responsiveness of gene expression (10;13;18;45;50). ATFs and C/EBP regulate transcription by binding to specific composite sequences termed amino

consensus sequence. Such elements mediate amino acid sensitivity of CHOP, CAT-1 and ATF3 expression (18;19;38). The NSRE is comprised of two elements (NSRE-1 and NSRE-2) and is required for increased ASNS transcription following stress-kinase activation (5;23). The NSRE-1 differs from the AARE that regulates CHOP transcription by onlytwo nucleotides. This small difference makes the CHOP AARE an independently functioning response element, whereas the NSRE-1 at the ASNS gene requires the presence of the NSRE-2 for full function (50). The ASNS NSRE-2 is not completely specific and can confer ER stress responsiveness to the CHOP AARE.

2.a. ATF2

Together with ATF3 and ATF4, ATF2 regulates amino acid sensitivity of the CHOP and ATF3 genes (10-12). ATF2 is necessary to confer amino acid sensitivity upon the expression of CHOP and ATF3 (3;11). ATF2 binds constitutively to the CHOP AARE and becomes phosphorylated within 15 minutes after amino acid depletion (3;10;11). Phosphorylation of ATF2 activates its intrinsic histone acetyltransferase (HAT) activity (32) and causes the acetylation of histones H4 and H2B at the CHOP promoter (10). Histone acetylation and binding of RNA polymerase II precede the binding of ATF4 and the production of CHOP and ATF3 mRNA, showing that phosphorylation of the prebound ATF2 is an early step in amino acid-dependent upregulation of gene transcription (9). However, ATF2 is not equally crucial for all amino acid-sensitive genes, since ATF2 deletion reduces, but does not abolish ASNS responsiveness (9).

2.b. ATF4

ATF4 is also necessary for the amino acid sensitive expression of its target genes, such as CHOP, ASNS, CAT-1 and ATF3 (3;10;38). Upon starvation, ATF4 binds to both NSREs and AAREs. This interaction is weak, if it binds as a homodimer and 3-fold stronger if it binds as a heterodimer with C/EBPβ (3;38;49). Little is known about the amino acid-dependent ATF4 transcription and mRNA stability, but eIF2/GCN2-mediated translation of the ATF4 uORF contributes most to the elevation of the ATF4 protein level (39;53). The level of ATF4 protein increases to reach a plateau between 1 and 2 fours after the onset of amino-acid starvation and then declines back to baseline with the next 3h (38). Recruitment

of ATF4 protein to already open chromatin of responsive promoters follows the dynamics of ATF4 protein synthesis (10).

2.c. ATF3

ATF3 binds the promoters of the CHOP, ASNS, CAT-1, C/EBPβ, VEGF and SNAT-2 genes as a homodimer or as a heterodimer with C/EBPβ (3;26;38;43;44) and regulates the late phase of the starvation response (3;38;43). Transcriptional activation of the ATF3 gene following amino acid limitation is mediated by an AARE that is positioned between 23 to -15 bp relative to its transcription start site (45). Binding of phospho-ATF2 and H4 acetylation at the ATF3 promoter suggests similar early activation steps as observed for the CHOP gene (10). Phosphorylation of the constitutively bound ATF2 is followed by histone H2B, H3 and H4 acetylation, enhanced ATF4 and RNA polymerase II binding, and a 15-fold increase in ATF3 mRNA (10;45). The rate of ATF3 transcription increases ~2-15-fold within 2h after the challenge (45). However, the newly produced ATF3 increasingly binds to its own promoter, which curtails the increase in the ATF3 and CAT-1 transcription rates (45).

2.d. C/EBPβ

Amino acid sensitivity of the expression of C/EBPβ is mediated by a 93 bp enhancer that is located downstream of the ORF (13). This enhancer is bound by ATF4, ATF3 and C/EBPβ. The sequence of activation of C/EBPβ transcription is similar to that described for other genes regulated by ATF transcription factors (13). C/EBPβ is not a critical component of the amino acid response pathway, but its heterodimers with ATF2, ATF3 and ATF4 bind with increased affinity to the AAREs and NSREs of CHOP, ASNS, CAT-1 and ATF3 (11;13;38;50).

Depressed protein production due to the inhibition of translation is expected consequence of the limited amino acid availability. This is the main reason why influence of amino acid starvation on gene expression is not sufficiently explored area. The transcription of hepatocyte-specific genes, including serum albumin, is selectively downregulated in H4-II-E rat hepatoma cells that are grown in a medium depleted of tryptophan, methionine, leucine, or phenylalanine (41). The amino acid sensitivity of serum albumin transcription depends on a HNF-1α binding site adjacent to the TATA box and an upstream regulatory element that contains a functional FoxA (formerly known as HNF-3) binding site (41). Similarly, mRNA levels of HNF-1 and the inhibitory truncated form of C/EBPβ (LIP) were increased and the DNA-binding activity of HNF-1, FoxA1, FoxA2, C/EBPα and C/EBPβ was increased in protein-deprived animals, whereas the DNA-binding activity of HNF-4α was decreased (42). These findings suggest that the decrease in transcription of the liver-specific genes is the result of an altered activity of liver-enriched transcription factors.

4. Mechanisms for increased mRNA stability during amino-acid deprivation

Increased mRNA stability in amino acid depleted medium contributes to the increased levels of GS, CAT-1, p21, ATF3, ATF4 and C/EBPβ mRNAs (37;44;49;55). RNA stabilisation ranges from a modest 1.2-fold (C/EBPβ mRNA) to 8-fold (ATF3 mRNA) (44). An increased binding of HuR, an RNA-binding protein that binds to AU-rich elements (AREs) in the 3'UTR of mRNAs, mediates this stabilisation for CAT-1 and ATF3 (24;44;55). At the same time, ATF3 mRNA associates less with AUF1, an mRNA-binding protein that enhances degradation (6).

The mRNA concentration of a key enzyme in polyamine synthesis, ornithine decarboxylase (ODC), as well as that of c-myc, c-fos, c-jun and junB, that is, of cell proliferation-enhancing factors, were all greatly increased in rats fed with protein-free diet or by starvation (31). Increased mRNA stability contributed to the elevated expression of c-jun and c-myc (47).

Table 1. Genes upregulated by amino acid limitation.

GENE REGULATION REGULATORS REFERENCE

eIF4E binding protein 1

(4EBP1) transcription ATF2, ATF4 (3)

4F2 antigen heavy chain transcription ATF2, ATF4 (3)

β-actin (41) Asparagine Synthetase transcription, mRNA _{stability translation} ATF3, ATF4, _C/EBPβ (5;9;13;14;21-_23;43;49;50)

ATF2 phosphorylation (3;10) ATF3 transcription, mRNA stability, alternative splicing, translation GCN2, eIF2, HuR, AuF1, ATF4, C/EBPβ (43-45)

ATF4 transcription, mRNA _{stability, translation} (3;39;45;49;53)

CAT-1 transcription, mRNA _{stability, translation}

GCN2, eIF2, HuR, ATF3,

ATF4, C/EBPβ (19;24;38;55)

C/EBPα (41)

C/EBPβ transcription, mRNA stability, translation

ATF3, ATF4,

C/EBPβ (3;13;41;42) CHOP transcription _{mRNA stability} ATF2, ATF3, _{ATF4, C/EBPβ} (2;3;9-12;18)

c-jun mRNA stability (47)

c-myc transcription _{mRNA stability} (47)

CXCL1 NFκB (7)

Dihydrodiol

dehydrogenase-1 NFκB (7)

Cycline-D1 NFκB (7)

Glutamine synthetase transcription, protein _stability DBP, Rev-Erb (34;35;51) IκBα NFκB (7) IL-8 (CXCL8) transcription _{mRNA stability} NFκB, JunD, _Fra1 (7;40)

IL-15 NFκB (7;40) Insulin-like growth factor

I (IGF-I) (29)

Insulin-like growth factor binding protein 1

Insulin-like growth factor binding protein 2 (IGFBP2) NFκB (7) Intracellulare adhesion molecule-1 (ICAM-1) NFκB (7) IFN-β NFκB (7)

IFN regulatory factor-2 NFκB (7)

jun-B NFκB (7)

L17 transcription _{nuclear retention} (7;36)

NFκB2 NFκB (7;36) Ornithine decarboxylase

(ODC) (47)

p53 NFκB (7)

S25 transcription _{nuclear retention} (36)

Seryl tRNA synthetase

(SARS) ATF2, ATF4 (3)

Solute carrier family 1 (neutral amino acid transporter) member 5 (SLC)

transcription ATF4 (3)

Sodium-coupled neutral amino acid transporter (SNAT2)

transcription ATF3, ATF4, _C/EBPβ (45) Thrombospondin-2

(THBS-2) NFκB (7)

TNF-β NFκB (7) Ubiquitin (2.8 mRNA

form) (7;41)

VEGF transcription _translation

GCN2, eIF2,ATF3, ATF4, C/EBPβ (1;7;16;40;41;4 5) Vimentin NFκB (1;7;16;40)

Wilm's tumor suppresor

gene (WT1) NFκB (7)

5. Specific effects of glutamine on gene expression 5.a. Effect of glutamine on transcription

Glutamine stimulates the expression of the argininosuccinate synthetase (ASS) gene at both the level of enzyme activity and mRNA in Caco-2 cells. For the transcriptional stimulation by glutamine, metabolism of glutamine to glucosamine 6-phosphate is required. Glutamine and glucosamine increase cytosolic Sp1 and NFκB O-glycosylation that leads to its translocation into nucleus, increased DNA binding and enhanced ASS transcription (8).

The mRNA levels of vascular endothelial growth factor-A (VEGF), a potent proangiogenic protein, and interleukin-8 (IL-8, or CXCL8), a leukocyte chemoattractant, are strongly induced, when ambient glutamineconcentrations are decreased (7). Although glutamine deprivation increases VEGF and IL-8 mRNAstability to some extent, its major effect is a strong increase in the transcription rate of both genes (7). Nuclearfactor-κB (NFκB) and activating protein-1 (AP-1), comprised of Fra-1 and JunD, induce IL-8 but not VEGF transcription. NFκB and AP-1 show increased DNA binding upon glutaminestarvation. At present, it is not known whether glutamine acts via the glycosylation of these transcription factors. Despite the increased binding of NFκB to its response elements during glutamine starvation, only a minority of NFκB target genes (~15%) are upregulated by > 2.0-fold (7).

Another gene that is influenced by glutamine availability is glutamine synthetase (GS). GS catalyses the production of glutamine from glutamate, ammonia and ATP. When the concentration of glutamine in muscle declines to one-third, GS mRNA levels increase ~26-fold (34;35;51). Glutamine-dependent repression of GS transcription is mediated by a polyA tract in the 3’ boundary of the GS 3'UTR and by the 3’ part of the first intron of the GS gene (51). The glutamine-sensing region in the 1st _{intron of the GS gene contains}

composite glutamine response sequences that are analyzed in detail in chapter 2.

5.b. Effects of glutamine on increased mRNA stability

In critically ill patients, there is an inverse correlation between survival and plasma glutamine concentration. Glutamine is an energy substrate for enterocytes and immune

cells. Its deprivation results in an increased susceptibility to cell stress and apoptosis, and a reduced responsiveness to pro-inflammatory stimuli. Glutamine depletion is associated with specific changes in the cellular concentration of 23 proteins, including metabolic enzymes, proteins involved in synthesis and degradation of RNA and protein, and stress proteins. The heat shock protein (Hsp) 70 showed the highest reduction in synthesis due to an enhanced decay of its mRNA (17).

5.c. Mechanisms for specific effects of glutamine on protein stability

Autophagic-lysosomal proteolysis is responsiblefor bulk proteolysis, whereas the ubiquitin-proteasome pathwayplays a significant role in the control of the degradationof specific proteins. The absence of glutamine inhibits the 26S ubiquitin-proteosomal degradation of the GS protein and, thereby, increases its half-life by 5-fold (35). In contrast, excess glutamine, at 10-fold the plasma level, inhibits hepatic proteolysis due to the lysosomotropictoxicity of ammonia derived from glutamine degradation (48).

Reference List

1. Abcouwer SF, Marjon PL, Loper RK, Vander Jagt DL. Response of VEGF expression to amino acid deprivation and inducers of endoplasmic reticulum stress. Invest Ophthalmol Vis Sci 2002; 43(8):2791-2798.

2. Abcouwer SF, Schwarz C, Meguid RA. Glutamine deprivation induces the expression of GADD45 and GADD153 primarily by mRNA stabilization. J Biol Chem 1999; 274(40):28645-28651.

3. Averous J, Bruhat A, Jousse C, Carraro V, Thiel G, Fafournoux P. Induction of CHOP expression by amino acid limitation requires both ATF4 expression and ATF2 phosphorylation. J Biol Chem 2004; 279(7):5288-5297.

4. Averous J, Maurin AC, Bruhat A, Jousse C, Arliguie C, Fafournoux P. Induction of IGFBP-1 expression by amino acid deprivation of HepG2 human hepatoma cells involves both a transcriptional activation and an mRNA stabilization due to its 3'UTR. FEBS Lett 2005; 579(12):2609-2614. 5. Barbosa-Tessmann IP, Chen C, Zhong C, Siu F, Schuster SM, Nick HS et al. Activation of the human

asparagine synthetase gene by the amino acid response and the endoplasmic reticulum stress response pathways occurs by common genomic elements. J Biol Chem 2000; 275(35):26976-26985.

6. Barreau C, Paillard L, Osborne HB. AU-rich elements and associated factors: are there unifying principles? Nucleic Acids Res 2005; 33(22):7138-7150.

7. Bobrovnikova-Marjon EV, Marjon PL, Barbash O, Vander Jagt DL, Abcouwer SF. Expression of angiogenic factors vascular endothelial growth factor and interleukin-8/CXCL8 is highly responsive to ambient glutamine availability: role of nuclear factor-kappaB and activating protein-1. Cancer Res 2004; 64(14):4858-4869.

8. Brasse-Lagnel C, Fairand A, Lavoinne A, Husson A. Glutamine stimulates argininosuccinate synthetase gene expression through cytosolic O-glycosylation of Sp1 in Caco-2 cells. J Biol Chem 2003; 278(52):52504-52510.

9. Bruhat A, Averous J, Carraro V, Zhong C, Reimold AM, Kilberg MS et al. Differences in the molecular mechanisms involved in the transcriptional activation of the CHOP and asparagine synthetase genes in response to amino acid deprivation or activation of the unfolded protein response. J Biol Chem 2002; 277(50):48107-48114.

10. Bruhat A, Cherasse Y, Maurin AC, Breitwieser W, Parry L, Deval C et al. ATF2 is required for amino acid-regulated transcription by orchestrating specific histone acetylation. Nucleic Acids Res 2007; 35(4):1312-1321.

11. Bruhat A, Jousse C, Carraro V, Reimold AM, Ferrara M, Fafournoux P. Amino acids control mammalian gene transcription: activating transcription factor 2 is essential for the amino acid responsiveness of the CHOP promoter. Mol Cell Biol 2000; 20(19):7192-7204.

12. Bruhat A, Jousse C, Wang XZ, Ron D, Ferrara M, Fafournoux P. Amino acid limitation induces expression of CHOP, a CCAAT/enhancer binding protein-related gene, at both transcriptional and post-transcriptional levels. J Biol Chem 1997; 272(28):17588-17593.

13. Chen C, Dudenhausen E, Chen H, Pan YX, Gjymishka A, Kilberg MS. Amino-acid limitation induces transcription from the human C/EBPbeta gene via an enhancer activity located downstream of the protein coding sequence. Biochem J 2005; 391(Pt 3):649-658.

14. Chen H, Pan YX, Dudenhausen EE, Kilberg MS. Amino acid deprivation induces the transcription rate of the human asparagine synthetase gene through a timed program of expression and promoter binding of nutrient-responsive basic region/leucine zipper transcription factors as well as localized histone acetylation. J Biol Chem 2004; 279(49):50829-50839.

15. Dann SG, Thomas G. The amino acid sensitive TOR pathway from yeast to mammals. FEBS Lett 2006; 580(12):2821-2829.

16. Drogat B, Bouchecareilh M, North S, Petibois C, Deleris G, Chevet E et al. Acute L-glutamine deprivation compromises VEGF-a upregulation in A549/8 human carcinoma cells. J Cell Physiol 2007; 212(2):463-472.

17. Eliasen MM, Brabec M, Gerner C, Pollheimer J, Auer H, Zellner M et al. Reduced stress tolerance of glutamine-deprived human monocytic cells is associated with selective down-regulation of Hsp70 by decreased mRNA stability. J Mol Med 2006; 84(2):147-158.

18. Fawcett TW, Martindale JL, Guyton KZ, Hai T, Holbrook NJ. Complexes containing activating transcription factor (ATF)/cAMP-responsive-element-binding protein (CREB) interact with the CCAAT/enhancer-binding protein (C/EBP)-ATF composite site to regulate Gadd153 expression during the stress response. Biochem J 1999; 339 ( Pt 1):135-141.

19. Fernandez J, Yaman I, Sarnow P, Snider MD, Hatzoglou M. Regulation of internal ribosomal entry site-mediated translation by phosphorylation of the translation initiation factor eIF2alpha. J Biol Chem 2002; 277(21):19198-19205.

20. Gaccioli F, Huang CC, Wang C, Bevilacqua E, Franchi-Gazzola R, Gazzola GC et al. Amino acid starvation induces the SNAT2 neutral amino acid transporter by a mechanism that involves eukaryotic initiation factor 2alpha phosphorylation and cap-independent translation. J Biol Chem 2006; 281(26):17929-17940.

21. Gong SS, Guerrini L, Basilico C. Regulation of asparagine synthetase gene expression by amino acid starvation. Mol Cell Biol 1991; 11(12):6059-6066.

22. Greco A, Gong SS, Ittmann M, Basilico C. Organization and expression of the cell cycle gene, ts11, that encodes asparagine synthetase. Mol Cell Biol 1989; 9(6):2350-2359.

23. Guerrini L, Gong SS, Mangasarian K, Basilico C. Cis- and trans-acting elements involved in amino acid regulation of asparagine synthetase gene expression. Mol Cell Biol 1993; 13(6):3202-3212. 24. Hyatt SL, Aulak KS, Malandro M, Kilberg MS, Hatzoglou M. Adaptive regulation of the cationic

amino acid transporter-1 (Cat-1) in Fao cells. J Biol Chem 1997; 272(32):19951-19957. 25. Jefferson LS, Kimball SR. Amino acids as regulators of gene expression at the level of mRNA

translation. J Nutr 2003; 133(6 Suppl 1):2046S-2051S.

26. Jiang HY, Wek SA, McGrath BC, Lu D, Hai T, Harding HP et al. Activating transcription factor 3 is integral to the eukaryotic initiation factor 2 kinase stress response. Mol Cell Biol 2004; 24(3):1365-1377.

27. Jiang HY, Wek SA, McGrath BC, Scheuner D, Kaufman RJ, Cavener DR et al. Phosphorylation of the alpha subunit of eukaryotic initiation factor 2 is required for activation of NF-kappaB in response to diverse cellular stresses. Mol Cell Biol 2003; 23(16):5651-5663.

28. Jousse C, Averous J, Bruhat A, Carraro V, Mordier S, Fafournoux P. Amino acids as regulators of gene expression: molecular mechanisms. Biochem Biophys Res Commun 2004; 313(2):447-452.

29. Jousse C, Bruhat A, Ferrara M, Fafournoux P. Physiological concentration of amino acids regulates insulin-like-growth-factor-binding protein 1 expression. Biochem J 1998; 334 ( Pt 1):147-153. 30. Kadowaki M, Kanazawa T. Amino acids as regulators of proteolysis. J Nutr 2003; 133(6 Suppl

1):2052S-2056S.

31. Kanamoto R, Yokota T, Hayashi SI. Expressions of c-myc and insulin-like growth factor-1 mRNA in the liver of growing rats vary reciprocally in response to changes in dietary protein. J Nutr 1994; 124(12):2329-2334.

32. Kawasaki H, Schiltz L, Chiu R, Itakura K, Taira K, Nakatani Y et al. ATF-2 has intrinsic histone acetyltransferase activity which is modulated by phosphorylation. Nature 2000; 405(6783):195-200. 33. Kilberg MS, Pan YX, Chen H, Leung-Pineda V. Nutritional control of gene expression: how

mammalian cells respond to amino acid limitation. Annu Rev Nutr 2005; 25:59-85.

34. Labow BI, Souba WW, Abcouwer SF. Glutamine synthetase expression in muscle is regulated by transcriptional and posttranscriptional mechanisms. Am J Physiol 1999; 276(6 Pt 1):E1136-E1145. 35. Labow BI, Souba WW, Abcouwer SF. Mechanisms governing the expression of the enzymes of

glutamine metabolism--glutaminase and glutamine synthetase. J Nutr 2001; 131(9 Suppl):2467S-2474S.

36. Laine RO, Shay NF, Kilberg MS. Nuclear retention of the induced mRNA following amino acid-dependent transcriptional regulation of mammalian ribosomal proteins L17 and S25. J Biol Chem 1994; 269(13):9693-9697.

37. Leung-Pineda V, Pan Y, Chen H, Kilberg MS. Induction of p21 and p27 expression by amino acid deprivation of HepG2 human hepatoma cells involves mRNA stabilization. Biochem J 2004; 379(Pt 1):79-88.

38. Lopez AB, Wang C, Huang CC, Yaman I, Li Y, Chakravarty K et al. A feedback transcriptional mechanism controls the level of the arginine/lysine transporter cat-1 during amino acid starvation. Biochem J 2007; 402(1):163-173.

39. Lu PD, Harding HP, Ron D. Translation reinitiation at alternative open reading frames regulates gene expression in an integrated stress response. J Cell Biol 2004; 167(1):27-33.

40. Marjon PL, Bobrovnikova-Marjon EV, Abcouwer SF. Expression of the pro-angiogenic factors vascular endothelial growth factor and interleukin-8/CXCL8 by human breast carcinomas is responsive to nutrient deprivation and endoplasmic reticulum stress. Mol Cancer 2004; 3:4.

43. Pan Y, Chen H, Siu F, Kilberg MS. Amino acid deprivation and endoplasmic reticulum stress induce expression of multiple activating transcription factor-3 mRNA species that, when overexpressed in HepG2 cells, modulate transcription by the human asparagine synthetase promoter. J Biol Chem 2003; 278(40):38402-38412.

44. Pan YX, Chen H, Kilberg MS. Interaction of RNA-binding proteins HuR and AUF1 with the human ATF3 mRNA 3'-untranslated region regulates its amino acid limitation-induced stabilization. J Biol Chem 2005; 280(41):34609-34616.

45. Pan YX, Chen H, Thiaville MM, Kilberg MS. Activation of the ATF3 gene through a co-ordinated amino acid-sensing response programme that controls transcriptional regulation of responsive genes following amino acid limitation. Biochem J 2007; 401(1):299-307.

46. Pestova TV, Kolupaeva VG, Lomakin IB, Pilipenko EV, Shatsky IN, Agol VI et al. Molecular mechanisms of translation initiation in eukaryotes. Proc Natl Acad Sci U S A 2001; 98(13):7029-7036. 47. Pohjanpelto P, Holtta E. Deprivation of a single amino acid induces protein synthesis-dependent

increases in c-jun, c-myc, and ornithine decarboxylase mRNAs in Chinese hamster ovary cells. Mol Cell Biol 1990; 10(11):5814-5821.

48. Poso AR, Schworer CM, Mortimore GE. Acceleration of proteolysis in perfused rat liver by deletion of glucogenic amino acids: regulatory role of glutamine. Biochem Biophys Res Commun 1982; 107(4):1433-1439.

49. Siu F, Bain PJ, LeBlanc-Chaffin R, Chen H, Kilberg MS. ATF4 is a mediator of the nutrient-sensing response pathway that activates the human asparagine synthetase gene. J Biol Chem 2002; 277(27):24120-24127.

50. Siu F, Chen C, Zhong C, Kilberg MS. CCAAT/enhancer-binding protein-beta is a mediator of the nutrient-sensing response pathway that activates the human asparagine synthetase gene. J Biol Chem 2001; 276(51):48100-48107.

51. Stanulovic VS, Garcia de Veas Lovillo RM, Labruyere WT, Ruijter JM, Hakvoort TB, Lamers WH. The 3'-UTR of the glutamine-synthetase gene interacts specifically with upstream regulatory elements, contains mRNA-instability elements and is involved in glutamine sensing. Biochimie 2006;

88(9):1255-1264.

52. van Venrooij WJ, Henshaw EC, Hirsch CA. Effects of deprival of glucose or individual amino acids on polyribosome distribution and rate of protein synthesis in cultured mammalian cells. Biochim Biophys Acta 1972; 259(1):127-137.

53. Vattem KM, Wek RC. Reinitiation involving upstream ORFs regulates ATF4 mRNA translation in mammalian cells. Proc Natl Acad Sci U S A 2004; 101(31):11269-11274.

54. Voet D, Voet JG. Boiochemistry. 3rd Edition ed. Wiley, 2006.

55. Yaman I, Fernandez J, Sarkar B, Schneider RJ, Snider MD, Nagy LE et al. Nutritional control of mRNA stability is mediated by a conserved AU-rich element that binds the cytoplasmic shuttling protein HuR. J Biol Chem 2002; 277(44):41539-41546.

Chapter 2

The 3'-UTR of the Glutamine Synthetase Gene

Interacts Specifically with Upstream Regulatory

Elements, Contains mRNA-Instability Elements

and is Involved in Glutamine Sensing

Vesna S. Stanulović, Rocio M. Garcia de Veas Lovillo,

Wil T. Labruyère, Jan M. Ruijter, Theodorus B. M. Hakvoort and

Wouter H. Lamers

Abstract

Glutamine synthetase (GS) is expressed at various levels in a wide range of tissues, suggesting that a complex network of modules regulates its expression. We explored the interactions between the upstream enhancer, regulatory regions in the first intron, and the 3’-untranslated region and immediate downstream genomic sequences of the GS gene (the GS “tail”), and compared the results with those obtained previously in conjunction with the bovine Growth Hormone (bGH) tail. The statistical analysis of these interactions revealed that the GS tail was required for full enhancer activity of the combination of the upstream enhancer and either the middle or the 3'-intron element. The GS tail also prevented a productive interaction between the upstream enhancer and the 5'-intron element, whereas the bGH tail did not, suggesting that the 5'-intron element is a regulatory element that needs to be silenced for full GS expression.

Using the CMV promoter/enhancer and transfection experiments, we established that the 2.8kb GS mRNA polyadenylation signal is ~10-fold more efficient than the 1.4kb mRNA signal. Because the steady-state levels of both mRNAs are similar, the intervening conserved elements destabilize the long mRNA. Indeed, one but not all constructs containing these elements had a shorter half life in FTO-2B cells. A construct containing only 300 bases before and 100 bases after the 2.8kb mRNA polyadenylation site sufficed for maximal expression. A stretch of 21 adenines inside this fragment conferred, in conjunction with the upstream enhancer and the 3'-part of the first intron, sensitivity of GS expression to ambient glutamine.

Introduction

Glutamine synthetase (GS) catalyzes ATP-dependent conversion of glutamate and ammonia to glutamine. The gene is expressed at various levels in a wide range of tissues. The actual concentration of GS per cell is very high in tissues in which GS is present in a subset of the cells, such as the pericentral hepatocytes of the liver or the astrocytes of the central nervous system. On the other hand, the cellular concentration of GS is very low in tissues in which GS is present in the majority of the cells, such as muscle (21). The cellular concentration of GS appears to correlate with the primary function of the enzyme, cellular levels of GS being 10-100-fold higher in glutamate-removing and ammonia-detoxifying cells (astrocytes and pericentral hepatocytes, respectively) than in glutamine-producing cells (myocytes and adipocytes) (21). Collectively, these observations suggest that the GS expression in different tissues is subjected to a complex regulatory network.

Our interest focuses on the regulation of GS expression in the liver, where it is expressed in a narrow, 2-3 cell-layers thick rim of hepatocytes around the central veins (22). This pattern of expression is very stable and only modified under extreme metabolic and hormonal conditions (5;13;18;19;21). Hormones like glucocorticoids and cAMP regulate GS expression at the transcriptional level (11;12;21), whereas the ambient concentration of glutamine influences GS mRNA and protein levels (7;18). Most likely, the occupation of the glutamine-binding site of the enzyme sensitizes the protein to degradation by the 26S-proteasome proteolytic pathway (8).

Several regulatory regions have been implicated in the transcriptional regulation of the GS gene. These include a farupstream enhancer localized at 6kb, an upstream enhancer at -2.5kb, and the 5’ and middle segments of the first intron (6;10;12). The upstream enhancer at -2.5kb confers pericentral expression upon a reporter gene in the liver of transgenic mice (22). Recently, our group reported synergistic interactions between the upstream enhancer and regulatory regions in the first intron, demonstrating that the effect of each of the regulatory elements on transcription depended on the presence or absence of another regulatory segment (9). This paper further explores the interactions between the -2.5kb upstream enhancer element, regulatory regions from the first intron, and the 3’-untranslated region of the GS gene (GS 3’-UTR) and its immediate downstream genomic sequences (collectively dubbed the GS “tail”).

The GS tail is structurally complex. The longest GS mRNA (2.8kb) has a 1544-nucleotides long 3'-UTR, which encompasses several highly conserved regions, including AU-rich elements (ARE) that may regulate mRNA stability (21). In addition, mRNAs of 1.4, 2.0, and 2.3kb were reported (13;20), which differ in their 3’-UTR and therefore contain only some of the conserved elements. None of the reported mRNA contains a consensus polyadenylation signal (AAUAAA). Such a consensus polyadenylation signal only exists at the site that would create a 2.6kb mRNA. Indeed, a BLAST search of EST libraries revealed that this 2.6kb mRNA exists in several tissues.

We identified regulatory elements in the GS 3'-UTR using transfection experiments with constructs containing the CMV enhancer/promoter. These findings were then used to characterize the interactions between the elements in the GS tail and sequences positioned upstream of the GS open reading frame. We observed that the main effects of the regulatory sequences in the first intron were more strongly positive when the GS tail was present than when the bovine growth-hormone (bGH) tail was present, as in our previous study (9). Furthermore, the interactions between the regulatory sequences in the first intron and those upstream of the structural gene differed in a systematic way between the two tails. Finally, we identified a functional ARE and a sequence that confers glutamine sensitivity upon GS expression.

Materials and Methods

Materials

Media, sera and reagents used for cell culture and transfection were purchased from Gibco Invitrogen Corporation, except actinomycin D, which was obtained from Sigma-Aldrich Chemie.

Plasmid construction

The rat GS mRNA (NM_017073) and 300bp of the GS downstream region (M29599) were fully sequenced. Constructs A-Q (Figure 1A) contained modules of the GS genomic fragment from -2520bp (relative to the transcription start site) to +2774bp (start of the open reading frame in the second exon) that drive expression of the firefly luciferase gene from the pSPluc+ plasmid (Promega). These constructs were prepared as described in detail (9), except that the bGH tail was exchanged for the GS tail. This GS tail contained the 1527bp 3’-untranslated region (UR; TaqI-end) and the 291bp long genomic sequence positioned immediately downstream of the GS gene (to EcoRI). These sequences were inserted into the EcoRV site of the downstream polylinker of pSPluc+.

Five starting constructs were used to delineate the active regions in the GS tail: constructs D, J, K and Q (Figure 1A), and construct R (Figure 2C), in which the GS promoter region was replaced by the BglII-HindIII cytomegalovirus (CMV) promoter/enhancer, as present in the pcDNA3 vector (Invitrogen). Ten modifications were made in the GS tail by modular deletions (Figures 2 and 3). Restriction sites used were: XbaI and EcoRV from the pSPluc+ downstream polylinker; TaqI, BstEII, PstI, PvuII, DraI, SacI present in the rat genomic sequence at 328, 565, 1150, 1675, 1949, and 2042bp from the start of the GS seventh exon; and the artificial, PCR-introduced sites XbaI*1, XbaI*2 and EcoRI* at 1332, 1510, and

1801bp. The constructs with deletions in the tail were suffixed to constructs D, J, K, and R as follows: 1: 1581bp BstEII-EcoRV deletion; 2: 247bp TaqI-BstEII deletion; 3: 830bp

TaqI-PstI deletion; 4: 1011bp TaqI-XbaI1 deletion; 5: 1199bp TaqI-XbaI2 deletion; 6:

1353bp TaqI-PvuII deletion; 7 1479bp: TaqI-EcoRI* deletion; 8: 1479bp TaqI-EcoRI* and 99bp SacI-EcoRV deletions; 9: 1479bp TaqI-EcoRI* and 194bp DraI-EcoRV deletions; 10: 1823bp TaqI-EcoRV deletion; and 11: 1637bp TaqI-DraI deletion. 7x: is the same as 7

except for the mutation of the (A)21 stretch at position 1829-1850 into 5’-

AGATCTGCGCGCGCAGATCT.

Cell culture

FTO-2B rat hepatoma cells (15) were maintained in DMEM/F-12 medium containing 2mM GlutaMax (L-Alanyl-L-Glutamine) at 37°C in a humidified atmosphere containing 5% CO₂. One passage prior to transfection, the medium was changed to DMEM/F-12 with or without GlutaMax. All media were supplemented with 10% foetal bovine serum, 10U/mL penicillin G and 10µg/mL streptomycin. The glutamine concentration in the FCS used was 1mM, so that the culture medium initially contained 0.1mM glutamine. After 2 days of culture, glutamine was no longer detectable.

DNA transfection and luciferase assay

Plasmids used for transfection were isolated on the Nucleobond G-500 columns (Macherey-Nagel). Ten million FTO-2B cells were harvested at 60-70% confluence for electroporation with 20µg of supercoiled plasmid (9). To correct for differences in transfection efficiency, the cells were co-transfected with 2µg of the Renilla luciferase expression vector, pRL-CMV (Promega). The culture medium was changed 24 hours after transfection. The cells were lysed 48 hours after transfection in “Passive Lysis Buffer” (Promega). Both luciferase activities were measured with the Dual-Luciferase® Reporter-Assay System (Promega) and the AutoLumat plus LB953 luminometer (Berthold Technologies).

mRNA stability assay

After transfection of constructs R, R3, R4, R5, R6, or R7, and pRSVcat as reference (9), the co-transfected cells were distributed over 3 wells of a six-well plate and cultured for 48h, with a medium change at 24h. Forty-eight hours after transfection, 10µg/mL actinomycin D was added for 0, 1 or 4 hours. Thereafter, cells were harvested and stored at -70°C. mRNA was isolated from the cells with the QuickPrep micro mRNA purification Kit (Amersham Biosciences). The mRNA pellet was dissolved in 10µL diethylpyrocarbonate-treated water containing 2U/µL of the SUPERase•In RNnase inhibitor (Ambion).

Reverse transcription

RNA was reverse transcribed with specific antisense primers for rat GAPDH (5’-GTGAGCTTCCCGTTCAGCTC), firefly luciferase (5’-TCGACGCAAGAAAAATC AGAG) and chloramphenicol acetyltransferase (5’-TAGAAACTGCCGGAAATCG TC). Each primer (25pmol) was incubated with 20% of the isolated mRNA for 10min at 70°C in a total volume of 10µL and then cooled to 4°C. Reverse transcription was carried out for 1h at 42°C with 100U SuperScript™II RNase H- _{reverse transcriptase (Gibco Invitrogen) in a}

total volume of 25µL, containing 8mM MgCl2, 0.5mM dNTP, 50mM Tris-HCl, 75mM

KCl. The reaction was stopped by heating for 15min at 70°C.

Quantitative PCR

For cDNA quantification, the LightCycler FastStart DNA Master SYBR Green I (Roche) assay kit was used in conjunction with the LightCycler System (Roche). The antisense primers were the same as those used for the first strand cDNA synthesis. Sense primers were GACCCCTTCATTGACCTCAAC for GAPDH, 5’-CAGTCAAGTAACAACCGCGA for firefly luciferase and 5’-GAGGCA TTTCAGTCAGTTGCT for chloramphenicol acetyltransferase. The real-time PCR reaction mixture (total volume: 10µL) contained 5 pmol sense and antisense primers, 4mM MgCl2,

1µL of FastStart DNA Master SYBR green-1 mix and, as template, 1 out of 25µL cDNA produced during reverse transcription. The samples were preheated at 95°C for 10min and for 15sec between cycles, followed by annealing for 10sec at 55°C for GAPDH, and at 58°C for firefly luciferase and chloramphenicol acetyltransferase, and elongation for 15sec at 72°C. The purity of the reaction product was confirmed by denaturing the samples at 95°C, followed by renaturing for 15sec at 50°C and denaturation by heating at 0.1°C/sec to 95°C for 10min. Data were processed with the LightCycler Software Version 3.5 systems (Roche) and the LinRegPCR program (25).

Correction for experimental variation and statistics

The transactivation potential of the tested DNA constructs was expressed as the ratio between their firefly luciferase activity and the renilla luciferase activity of the co-transfected pRL-CMV construct. The data were collected from 46 measurement sessions

with FTO-2B cells. In each session, different combinations of the various constructs were tested. The resulting number of observations per construct was 8-20. Between-session variation in reporter-gene activity was removed by calculating correction factors for each session (26). Thereafter, the mean activity of reference construct A was set to 100. For constructs A to Q (Figure 1), the significance of the main effects and of the interactions between upstream and intron modules was tested with a two-way analysis of variance (ANOVA). Figure 1B gives the test results as well as the values of the main and interaction effects calculated from the ANOVA results (9). For each of these series of constructs, a significant construct and/or GlutaMax effect was found with a one-way (series D, J and R) or two-way (series K) ANOVA. For the construct series derived from constructs D, J and K, and R, the difference between the starting construct and each of the other constructs was then tested with Student’s t-test. The set of K constructs was tested similarly for the difference between activities in medium containing 0 or 2mM GlutaMax.

In the mRNA stability experiment, the concentration of Luc mRNA at different time points was corrected for the number of cells assayed by relating it to the concentration of GAPDH mRNA. Three independent experiments in duplicate were done. The steady-state mRNA levels were determined as the ratio between the luciferase mRNA levels of the respective R constructs and the chloramphenicol-acetyltransferase mRNA level of the pRSVcat construct at 0h. Between-session variation was removed based on the GAPDH values (26). Every experimental point is presented as a mean ± SEM. Time-dependent changes in cellular mRNA concentrations for each construct were used for calculation of the rate of degradation and half-life (T½) using linear regression. For each half-life, the 95%

Results

In our previous report (9), the interaction between the GS promoter, the -2.5kb upstream region (UR) and the 2.6kb first intron were analyzed in the context of the bovine growth hormone 3'-UTR and immediate downstream region (bGH “tail”). In the present study, we analyzed the interactions between the same upstream and intronic regions in the context of the GS 3'-UTR and immediate downstream region (designated the GS “tail”). The mean activity of construct A, containing the minimal promoter and minimized 1st _{intron, was set}

to 100 (Figure 1A, panel I). Elongation of the upstream region to the PstI (-965) and

EcoRV (-2148) sites (constructs B and C) reduced luciferase activity to 40 and 10,

respectively (Figure 1A, panel I). The further addition of the 372bp HindIII-EcoRV element reactivated the expression to 60. Previously, the HindIII-EcoRV element was identified as an upstream enhancer (UE) (6;9;12;22). When this UE was connected directly to the minimal promoter, the resulting construct (H; panel III) nevertheless demonstrated only an activity similar to that of construct A.

The effects of the first intron on the promoter activity, in the context of the GS tail, were tested in the second group of constructs (Figure 1A, panel II). The 5’-part of the intron (construct D) strongly stimulated promoter activity (2.6-fold). In contrast, the middle (E) and 3’-intron (F) modules decreased reporter gene activity to 10 and 60, respectively. When the entire first GS intron sequence was present (G), the activity was still only 70% of that of construct A.

Interactions between the upstream enhancer or the upstream region and the first intron modules were tested in constructs D to Q (Figure 1A, panels II-IV) and expressed as ANOVA-derived main effects and interactions (Figure 1B). The UE, and the 5'-, 3'- and entire first intron each had a significant positive effect on reporter-gene activity. Furthermore, the promoter had a positive interaction with the 5'-intron element and negative interactions with the middle, 3'- and entire first intron. In conjunction with the upstream enhancer or entire upstream region, the promoter had a negative interaction with the 5'-intron element. Finally, the promoter and upstream enhancer had a positive interaction with the entire first intron. From this analysis, the picture emerges that the 5'-intron element activates the promoter in the absence of the UE, whereas the middle and/or 3'-intron elements interact with the UE to effectively activate the promoter (cf. Figure 1A).

Figure 1. Schematic representation of the GS regulatory sequences used in constructs A-Q and their reporter gene activities.

Panel A. At the top, a restriction map of the rat genomic GS region analyzed in these experiments is shown. The

exons are shown as boxes, with the arrows indicating the start sites of transcription (0) and translation (+2774). Polyadenylation sites of the shortest, 1.4kb and the longest 2.8kb mRNAs are labelled with pA. The upstream boundary (-2520), the transcription start site (0) and the downstream boundary (+2774) of the genomic DNA segment that was analyzed, are indicated. The sequences present in the respective constructs are shown in the left portion of the figure as solid black lines. The upper line shows the linkage of the 5.3kb genomic GS segment with the luciferase reporter gene at the NcoI site (translation start site) in the second exon, and with the GS transcription termination and polyadenylation signal (GS 3'-UTR) and immediate downstream genomic sequences. The first and the second exon up to the translation start site are represented as black boxes. The right portion of the figure shows

upstream region, respectively, in conjunction with the minimal promoter, the minimized first intron and the respective intron elements. Luciferase activity (Mean ± SEM) is expressed relative to construct A, containing only the GS promoter and minimized first intron. The mean activity of construct A was set to 100. The result of the statistical analysis between constructs is given in panel B. Restriction sites: H-HindIII, E-EcoRV, P-PstI,

Bg-BglII, S-SmaI, B-BamHI, EI-EcoRI, N-NcoI.

Panel B. Results of the statistical analysis of the main and interaction effects of GS upstream and intron modules.

All effects (±95% confidence interval) were calculated from the activities of constructs A and D-Q, using the ANOVA technique (for details, see material and methods). Main effects were expressed as increase or decrease in activity upon addition to reference construct A, which was set at 100. Interaction effects were expressed as the difference between the observed activity of a construct and the main effects of its components. Main activities and interactions marked in bold differ significantly (P<0.05) from construct A and from 0. The activity of a construct (e.g. that of I (= 139; Figure 1A)) can be calculated from the Figure as the sum of the basal activity (= 100), the main activity of the upstream element (UE = 57), the intron element (5’ = 125) and their interaction (= -143).

Panel C. Scatter plot of the luciferase activities of constructs A to Q in conjunction with the bovine growth

hormone tail (bGH) (9) on the X-axis and that of the same constructs in conjunction with the GS tail on the Y-axis. The bold line is the best-fitting line for all measurement points, the two other lines surrounding the 95% confidence interval of this fit. Luciferase activity (Mean ± SEM) is expressed relative to construct A.

A direct comparison of the effects of the bGH (9) and GS tails in the corresponding constructs is shown in Figure 1C. The bold line represents the best fit for all experimental points, while the dashed lines enclose the 95% confidence interval of this fit. Only 4 constructs (C, D, J, and K) fall outside of this area. The activities and interactions of the fragments present in constructs D, J and K (the 5’-intron fragment alone, and the upstream enhancer in conjunction with either the middle or the 3’-intron fragment) were higher in the presence of the GS tail than in the presence of the heterologous bGH tail. For construct C, containing the upstream region up to the upstream enhancer, a strong negative influence of the GS tail compared to the bGH tail was found.

The GS tail is not well characterized. It does not contain a standard AAUAAA polyadenylation signal in any of the reported mRNA species. However, it does contain several highly conserved segments (Figure 2A) (21), the function of which is unknown. The positions of the reported polyadenylation sites and the highly conserved sequences are shown in Figure 2B. To identify the elements in the GS tail that are important for proper expression of the GS gene, constructs were initially tested in the context of the heterologous CMV early promoter/enhancer and the firefly luciferase ORF. The starting, reference

construct in this series is construct R, which contains the complete GS tail (Figure 2C). The shortest construct (R1) only contains the conserved U-rich element “a” and the polyadenylation site of the short 1.4kb GS mRNA (21). This construct is able to confer luciferase expression, albeit at only 25% of that of construct R. The deletions in constructs R2, R3, R4 and R5, encompassing the 1.4kb polyadenylation signal, AU-rich elements “b” and “e”, GU-rich element “c”, and U-rich element “d”, increased expression ~1.5-fold relative to construct R (Figure 2C). Reporter-gene expression of construct 6, in which the GU-rich element “f” was disrupted, was 2-fold higher than that of construct R. However, a further deletion of 126bp (construct R7) reduced reporter-gene expression to 50% of that in reference construct R. Construct R7 was delineated further through truncations from the 3’-end (constructs R8 and R9), leaving a fragment of 144bp surrounding the polyadenylation site of the long 2.8kb GS mRNA that still allowed expression at 50% of the reference level. When the EcoRI*-DraI fragment present in R9 was deleted (generating construct R10), reporter-gene activity no longer exceeded background levels (Figure 2C). These findings show that sequences up to the PvuII site (325 bases from the end of the 2.8kb mRNA) prevent maximal expression, whereas the sequence between the PvuII and the DraI sites (100 bases downstream of the polyadenylation site of the 2.8kb mRNA) contain the sites necessary for optimal transcription termination and polyadenylation.

Figure 2. Schematic representation of the GS 3'-UTR and downstream sequences ("GS tail") used in constructs R to R10 and their luciferase activities. Panel A. Position and sequence of conserved and other

potentially functional regions in the GS tail. The names and the fill patterns used in panel C for their representation are also indicated. Panel B. Map of the GS tail, in which the position of the polyadenylation signals for all GS mRNAs known are presented as pA 1.4, pA 2.0, pA 2.3, pA 2.6 and pA 2.8. Conserved and other potentially functional regions are marked as blocks a-i. The fill patterns represent different types of elements mentioned in panel A. Restriction sites: T–TaqI, Xb-XbaI, B-BstEII, P-PstI, Xb1*-XbaI1*, Xb2*-XbaI2*, Pv-PvuII, EI*-EcoRI*,

D-DraI, Sc-SacI, E-EcoRV. Panel C. R constructs and their luciferase activities. The left part provides a

schematic presentation of the sequences present in the respective R constructs. The CMV promoter/intron, luciferase open-reading frame and GS tail are indicated. Solid black lines represent the sequences present in a specific R construct. Reporter gene activity is presented as mean ± SEM. Asterisks indicate activities significantly different from R with P<0.05.

The interactions between the upstream and the intron sequences on the one hand and the GS tail on the other hand were further investigated in the constructs that showed specific enhancement of the reporter-gene activity in combination with the GS tail compared to the bGH tail (constructs D, J and K; Figure 1C). The upstream and intron sequences present in constructs D and J were combined with GS tails 7, 8 and 9 (Figure 3B), whereas in construct

K, tails 4, 5, 6, 11, and 7x were additionally tested (Figure 3C, no GlutaMax). In the context of either the 5’-intron region (construct D) or the UE and the middle intron region (construct J), the activity of GS tails 7 and 9 was similar to that of the complete GS tail, whereas GS tail 8 caused the activity to drop to 50% (Figure 3A and B). For these constructs, the short

EcoRI*-DraI region, therefore, sufficed for full expression. However, in the context of the

upstream enhancer and the 3’-part of intron 1 (construct K), the entire downstream element of 290 bases was necessary for full activity (Figure 3C, constructs K7-K11). The comparison of constructs K4-K7 shows that, in combination with the GS upstream region, the last 325 bases of the 7th _{exon are necessary for full expression of the reporter constructs. This conclusion}

concurs with the findings in conjunction with the CMV promoter (Figure 2). When searching for a possible explanation of the very high activity of K6 and the still high activity of K7, we noticed a stretch of 21 adenines at the very end of the 7th _{exon (referred to as (A)}

21). Searching

the BLAST database for EST sequences containing TTC(A)21CCTC we found a number of

cDNA molecules, in which this sequence was also present at the very end of the cDNA. Mutation of the (A)21 to a GC-rich sequence (K7x) decreased K7 activity to the level of

construct K (Figure 3C), identifying this adenine-rich sequence as a strong regulatory element.

Figure 3. Schematic representation of constructs D, J and K and their derivates and their associated reporter-gene activities. At the top, a restriction map of the rat genomic GS regions analyzed in these

experiments is shown. The variants of constructs D, J and K differ in the sequences present downstream of the GS ORF and are labelled with numbers that correspond with the numbers of the R construct variants shown in Figure 2. The right portion of each panel shows the luciferase activity of the respective constructs after transfection to FTO-2B hepatoma cells. Panels A and B show constructs D and J, and their derivates, while panel C shows construct K and its derivates cultured in the absence (grey bars) and presence (white bars) of 2mM GlutaMAX. Reporter-gene activity is presented as mean ± SEM. Asterisks indicate activities significantly different from that of the main construct (P<0.05). For the K series differences were tested for the 0 and 2mM GlutaMAX series separately. Restriction sites are given in Figures 1 and 2.

Many metabolic genes, such as glutamine synthetase, are regulated by the concentrations of their substrate(s) and product(s). We therefore tested the expression of the D, J, K and R constructs also in the presence of GlutaMax, containing the glutamine-alanine dipeptide (only results for the K constructs are presented in Figure 3C). In the presence of 2mM GlutaMax, only the luciferase activity of construct K7 decreased to the level found for basic construct K. Since construct K7x was not influenced by glutamine availability, we conclude that the (A)21 stretch is

crucial for glutamine sensing. This K7-specific responsiveness to glutamine was only seen between 0-1mM GlutaMax. Further increasing GlutaMax to 2.5mM did not influence GS gene expression.

To test whether sequences residing inside the constructs R, R3, R4, R5, R6 and R7 affected reporter-gene expression by stabilization of the luciferase-GS chimeric mRNA, these constructs were transfected into FTO-2B cells. pRSVcat was co-transfected in all experiments to serve as the common denumerator of the experiments. Forty-eight hours after transfection, the cells were exposed to actinomycin D and harvested 0, 1 or 4h later. The cellular steady-state concentration of the constructs was related to the cellular GAPDH levels and set to 100 at 0h. The luciferase mRNA concentration of constructs R, R3, R5, R6 and R7 did not significantly change within 4 hours in 3 independent experiments carried out in duplicate, but that of construct R4 showed a significantly shorter half-life (T½ = 1.4 hours; P = 0.005)(Table 1). Construct K [min-1] Lower CI [min-1] Upper CI [min-1] P-value T1/2 [min] R 18.6 -19.8 57 0.294 - R3 1.8 -30 33.6 0.892 - R4 44.4 18.6 69.6 0.005 84 R5 -5.4 -22.2 10.8 0.461 - R6 9.6 -14.4 33 0.384 - R7 -4.2 -31.8 22.8 0.715 -

Table 1. Stability (T1/2) of the R, R3, R4, R5, R6 and R7 mRNAs. FTO-2B cells were transfected and cultured

for 48h before treatment with actinomycin D for 0, 1, or 4 hours. The mRNA half-life (T1/2) was calculated according to the equation

N

=

N

⋅

e

Discussion

The present study has shown that the behaviour of the sequences that determine the transcriptional activity of the GS gene differs depending on the identity of the transcription termination and polyadenylation signal. We further confirmed that highly conserved sequences in the GS 3'-UTR do indeed exert major effects on gene expression.

Interaction between GS regulatory elements.

To explain the strict and very stable pericentral expression of GS in the liver, we postulated the “double-lock” model of transcriptional regulation (3). In this model, two or more interacting regulatory elements determine gene expression. Because experimental conditions almost never change the activity of these elements simultaneously, their combined effect hardly ever changes. We recently showed the existence of such interactions between regulatory elements in the upstream region and first intron of the GS gene in vitro (9). In that study, the activity of the GS regulatory sequences was tested in the context of the luciferase ORF and the bovine growth hormone (bGH) tail. In the present study, we replaced the bGH tail by the GS tail to explore specific interactions between regulatory elements upstream and downstream of the GS open reading frame. The comparison (Figure 1C) showed that across all constructs, both tails did not differ with respect to their effect on reporter-gene expression. However, the finding that constructs D, J, and K had a significantly higher than expected reporter-gene expression when combined with the GS tail, suggested that this tail enhanced the outcome of the interaction between the promoter and 5'-segment of the first intron (construct D) and between the UE, promoter and middle or 3'-intron part (constructs J and K). A similar, highly productive interaction between upstream sequences, intron 1, and highly conserved 3'-UTR plus genomic flanking region was demonstrated for the human β-actin gene (24). The statistical analysis of the interactions of the GS and bGH tails with the upstream regulatory elements (Figure 1B and Figure 2 of reference (9)) revealed, indeed, that the stimulatory effect of the first intron on the promoter is much stronger in conjunction with the GS tail, whereas that of the upstream enhancer is much stronger in conjunction with the bGH tail. In more detail, the comparison shows that the GS tail increases the positive interaction between the 5'-intron element and the promoter, whereas the upstream enhancer prevents this productive interaction. In