Structural Studies of Human 5’-Nucleotidases

Karin Walldén

Doctoral thesis at Stockholm University Department of Biochemistry and Biophysics © Karin Walldén, Stockholm 2008 ISBN 978-91-7155-718-6 Printed in Sweden by Universitetsservice AB, Stockholm 2008 Distributor: Department of Biochemistry and Biophysics, Stockholm University Papers I-III are reprinted with permission from the publisher.

Abstract

5’-Nucleotidases (5’NTs) are catabolic enzymes of the nucleotide metabo-lism. They catalyze dephosphorylation of deoxyribo- and ribonucleoside monophosphates and constitute an important control point in the regulation of intracellular nucleotide pools for the maintenance of correct DNA and RNA synthesis.

By removing the α-phosphate group from a nucleotide, the 5’NTs release the nucleoside to pass the plasma membrane by facilitated diffusion. De-pending on the cellular need for nucleotides, the nucleosides can either exit the cell for reuse elsewhere or be imported and subsequently phosphorylated by nucleoside and nucleotide kinases.

The knowledge of how nucleotides are metabolized has been used for ra-tional design of nucleoside analogues that are used in treatment of cancer and viral diseases. These drugs are phosphorylated within the cell to become active. Their dephosphorylation by 5’NTs might be one of the mechanisms behind the resistance experienced by patients towards such drugs.

This thesis describes structure-function studies on four of the seven

known human 5’-NTs. The focus of the work is on the substrate specificity and regulation of these enzymes. Inactive variants of the mitochondrial and cytosolic deoxynucleotidases and the cytosolic 5’-nucleotidase II were used to characterize the structural basis for their substrate specificity in high de-tail.

Based on structures of the apoprotein and activator/activator+substrate complexes of cytosolic 5’-nucleotidase II, a mechanism for the allosteric activation of this enzyme was presented. In this mechanism, the activator induces a conformational change that involves conserved residues of the active site. The conformational change drastically increases the enzyme af-finity for the phosphate moiety of the substrate.

List of publications

I Wallden K., Ruzzenente B., Rinaldo-Matthis A., Bianchi V. and Nordlund P. (2005) Structural basis for substrate specificity of the human mitochondrial deoxyribonucleoti-dase, Structure 13, 1081-1088.

II Wallden K., Rinaldo-Matthis A., Ruzzenente B., Ram-

pazzo C., Bianchi V. and Nordlund P. (2007) Crystal struc-tures of human and murine deoxyribonucleotidases: In-sights into recognition of substrates and nucleotide ana-logues, Biochemistry 46, 13809-13818.

III Wallden K., Stenmark P., Nyman T., Flodin S., Graslund

S., Loppnau P., Bianchi V. and Nordlund P. (2007) Crystal structure of human cytosolic 5 '-nucleotidase II - Insights into allosteric regulation and substrate recognition, Journal of Biological Chemistry 282, 17828-17836.

IV Wallden K. and Nordlund P. Mechanism for allosteric ac-

tivation and substrate recognition of human cytosolic 5'-nucleotidase II, Manuscript

Additional publication

Kosinska U., Wallden K., Flodin S., Nyman T., Stenmark P., Eklund H. and Nordlund P. Structure of human uridine cytidine kinase 1 in ligand free and ADP bound states, Manuscript.

Contents

Introduction .....................................................................................................1 Nucleotide Metabolism ....................................................................................................1

De novo pathway .......................................................................................................3 Salvage pathway........................................................................................................4

Regulation of nucleotide metabolism ..............................................................................4 Types of regulatory mechanisms...............................................................................4 Regulation of de novo synthesis................................................................................5 Regulation of the salvage pathway............................................................................6 Cell cycle dependence...............................................................................................6

5’-Nucleotidases (5’NTs) .................................................................................................7 Overview ....................................................................................................................7 Cytosolic 5’(3’)-deoxynucleotidase (cdN) ..................................................................9 Mitochondrial 5’(3’)-deoxynucleotidase (mdN) ........................................................10 Cytosolic 5’-nucleotidase IA (cN-IA) ........................................................................10 Cytosolic 5’-nucleotidase IB (cN-IB) ........................................................................11 Cytosolic 5’-nucleotidase II (cN-II) ...........................................................................11 Cytosolic 5’-nucleotidase III (cN-III) .........................................................................12 Ecto-5’-nucleotidase (eN) ........................................................................................13

Nucleotide pools and disease .......................................................................................13 Nucleotide pools ............................................................................................................14 Nucleoside/nucleotide transporters...............................................................................16 Anti-cancer/viral nucleoside analogues.........................................................................17 Evolution and catalysis of intracellular 5’NTs................................................................20

Overall structure of intracellular 5’NTs.....................................................................20 Catalytic mechanism................................................................................................23 Catalytic mechanism of eN ......................................................................................25

Present Investigation.....................................................................................27 Aim.................................................................................................................................27 Strategy .........................................................................................................................27

Fluorometallic complexes as models for enzyme intermediates .............................28 Substitutions to trap substrate in active site ............................................................28

Structural basis for substrate specificity of mdN and cdN (Paper I and II) ...................30 Pyrimidine base specificity.......................................................................................32 Deoxy/ribo specificity ...............................................................................................33

2’-,3’- and 5’-phosphate specificity ..........................................................................34 Purine/pyrimidine specificity.....................................................................................35 Nucleoside analogue recognition in mdN ................................................................36 Implications for the catalytic mechanism .................................................................38

Structure of human cN-II (Paper III and IV)...................................................................39 Characterization of 2 effector sites ..........................................................................39 Activator recognition in effector site 1......................................................................40 Mechanism for activation of cN-II ............................................................................42 Covalent modification on Asn52 ..............................................................................45 Substrate recognition in cN-II...................................................................................47

Structure of human cN-III ..............................................................................................49 Mapping of cN-III deficiency substitutions ...............................................................49 Phosphotransferase activity of cN-II and cN-III .......................................................53

Future perspectives.......................................................................................54

Acknowledgments .........................................................................................55

References....................................................................................................56

Abbreviations

Enzymes ADA Adenosine deaminase AK/ UCK Adenosine kinase/uridine cytidine kinase ANC Adenine nucleotide carrier APRT Adenosine phosphoribosyl transferase cdN Cytosolic 5’(3’)-deoxynucleotidase cN-IA Cytosolic 5’-nucleotidase IA cN-IB Cytosolic 5’-nucleotidase IB cN-II Cytosolic 5’-nucleotidase II cN-III Cytosolic 5’-nucleotidase III CNT Concentrative nucleoside transporter DNC Deoxynucleotide carrier dCK/ dGK Deoxycytidine kinase/deoxyguanosine kinase eN Ecto 5’-nucleotidase ENT Equilibrative nucleoside transporter HAD Haloacid dehalogenase HGPRT Hypoxanthine guanosine phosphoribosyl transferase mdN Mitochondrial 5’(3’)-deoxynucleotidase (d)NK (Deoxy)nucleoside kinase NMPK/NDPK Nucleoside mono-/diphosphate kinase 5’NT 5’-Nucleotidase β−PGM β-Phosphoglucomutase PNP Purine nucleoside phosphorylase PSP Phosphoserine phosphatase RNR Ribonucleotide reductase SERCA Sarcoplasmic Ca2+ ATPase TK1/TK2 Thymidine kinase 1/ thymidine kinase 2 TP/UP Thymidine/uridine phosphorylase Analogues NA (-MP) Nucleoside analogue (5’-monophosphate) d4T 2’,3’-didehydro-2’,3’-dideoxythymidine AZT 3’-Azidothymidine BVdU E)-5-(2-bromovinyl)-2’-deoxyuridine FdU 5-fluoro-2’-deoxyuridine AraT 1-β-D –arabinosylthymine ddC 2’,3’-dideoxycytidine

dFdC 2’,2’-difluoro-2’-deoxycytidine (Gemcitabine) araC 1-β-D –arabinofuranosylcytosine 3TC 2’-deoxy-3’-thiacytidine CdA 2-chloro-2’-deoxyadenosine (cladribine) araG 9-β-D –arabinosylguanine DPB-T (s)—1-[2’-deoxy-3’,5’-O-(1-phosphono) benzylidene-b-

D-threo-pento-furanosyl]-thymine PMcP-U (+)-1-trans-(2-Phosphonomethoxycyclopentyl)uracil Diseases AIRP autoimmune infertility protein AML acute myeloid leukemia MDS Mitochondrial DNA depletion syndrome MNGIE Mitochondrial neurogastrointestinal encephalomyopathy Other ApnA Diadenosine 5’,5’-polyphosphate GPI glycosyl phosphatidylinositol Pi Inorganic phosphate Amino acids Alanine Ala A Arginine Arg R Asparagine Asn N Aspartic acid Asp D Cysteine Cys C Glutamic acid Glu E Glutamine Gln Q Glycine Gly G Histidine His H Isoleucine Ile I Leucine Leu L Lysine Lys K Methionine Met M Phenylalanine Phe F Proline Pro P Serine Ser S Threonine Thr T Tryptophan Trp W Tyrosine Tyr Y Valine Val V

1

Introduction

Living organisms share the capacity to replicate themselves. To succeed with this and to survive, they must take up energy from their surroundings and converge this energy to synthesize organic material. This is mediated by enzymes, which are the catalysts of life.

Humans contain thousands of different enzymes that catalyze specific chemical reactions. Some enzymes degrade components of food and channel the energy released by this process into chemical energy in the form of for instance the nucleotide ATP. Other enzymes can then use ATP as energy source for synthesis of various components of the cell.

The degradation (catabolism) and biosynthesis (anabolism) of an organic compound usually involve several sequential reactions catalyzed by different enzymes. These form together a “pathway”, in which one reactant (substrate) enters and a product from the first enzyme will become a substrate for the second enzyme and so forth.

Most such pathways are regulated, so that the amount of product is fine-tuned to the need of the cell. For instance, when exercising you activate catabolic pathways that release sufficient amount of chemical energy needed for the exercise. Enzymes can be regulated in several ways, in which either the amount, localization or catalytic efficiency of the enzyme is changed.

The focus of this thesis is on a family of enzymes called 5’-nucleotidases (5’NTs), which are catabolic enzymes that participate in the metabolism of nucleotides.

Nucleotide Metabolism Nucleotides (Fig. 1) play several important roles in all cells. ATP and to a lower extent GTP are essential chemical energy carriers that transfer energy liberated by catabolism to anabolic processes. Nucleotides are also used as building blocks in DNA and RNA that carry the genetic information of which proteins that can be produced by the cell. Nucleotides are also ele-ments in cofactors, such as NAD, FAD and coenzyme A, and activated bio-synthetic intermediates such as UDP-glucose (1).

2

Fig. 1. Composition of nucleotides. (a) The base: Adenine, guanine and hypoxan-thine are modifications of purine, and uracil, thymine and cytosine are modifications of pyrimidine. (b) The sugar ring: β−D-Ribose and β−D-deoxyribose. (c) Nucleo-side: The four deoxynucleosides. d) Nucleotide: (Deoxy)nucleoside 5’-mono-, di- or triphosphates, (d)NMPs, (d)NDPs and (d)NTPs.

Nucleotides are synthesized either de novo from low molecular mass pre-cursors or from recycled nucleosides and bases. These two pathways are called the de novo pathway and the salvage pathway, respectively (Fig. 2). The 5’NTs are active in the salvage pathway.

Fig. 2. De novo pathway and salvage pathway.

3

De novo pathway Nucleotides are composed of a nitrogen-rich base, a sugar ring and phos-phate group(s) (Fig.1). The base is either a modified purine or a modified pyrimidine that are synthesized onto a ribose 5’-phosphate by multi-step enzymatic pathways from amino acids, CO2 and NH3 (1).

The de novo synthesis of pyrimidine nucleotides goes through the forma-tion of uridine monophosphate (UMP) that is further converted to all other pyrimidine nucleotides. The corresponding synthesis of purine nucleotides leads to inosine monophosphate (IMP), which is the precursor to all other purine nucleotides (Fig. 3).

Fig. 3. De novo synthesis of nucleotides goes through UMP and IMP.

Fig. 4 shows a scheme of the de novo pathway going from NDPs to dNTPs for synthesis of DNA. Nucleoside monophosphate kinases (NMPKs) phosphorylate the NMPs to NDPs, which are then reduced to dNDPs by a single enzyme, ribonucleotide reductase (RNR) (2-4). Then nucleoside di-phosphate kinases (NDPKs) phosphorylate dNDPs into dNTPs. An excep-tion is the synthesis of dTTP, which is formed from dCDP through several enzymatic steps (Fig. 4).

Fig. 4. De novo pathway: (1) RNR, (2) NDPKs, (3) dCMP deaminase, (4) Thymidy-late synthase and (5) NMPKs

4

Salvage pathway Nucleosides and bases are not directly synthesized de novo but are degrada-tion products of nucleotides originating from for instance DNA and RNA. The cell can reuse (salvage) nucleosides and bases, which can be transported across the plasma membrane by facilitated diffusion or by concentrative cotransport into the cell together with sodium (see “Nucleoside transporters” below).

Nucleosides can be degraded to bases by purine nucleoside phosphorylase (PNP), uridine phosphorylase or thymidine phosphorylase (TP). Purine bases can be converted to NMPs by reactions catalyzed by adenine phosphoribo-syltransferase (APRT) and hypoxanthine/guanine phosphoribosyltransferase (HGPRT) (4).

In mammals, deoxynucleosides can be phosphorylated into dNMPs by the deoxynucleoside kinases (dNKs) thymidine kinase 1 (TK1), thymidine kinase 2 (TK2), deoxycytidine kinase (dCK) and deoxyguanosine kinase (dGK) (5). Of these, TK2 and dGK are mitochondrial enzymes. The dNKs are named from their best substrate but they also take other substrates. How-ever they do not take ribonucleosides as substrates. Two mammalian ribonu-cleoside kinases (NKs) have been found in the cytosol, adenosine kinase (AK) (6) and uridine/cytidine kinase (UCK) (7). No mammalian NK has been found for guanosine and inosine phosphorylation.

Very importantly, nucleotides appear not to be transported across the plasma membrane at significant rates. Therefore, the addition of the first phosphate onto a nucleoside probably traps it inside the cell.

5’NTs catalyze the dephosphorylation of (d)NMPs. This reaction releases the nucleoside to allow it to be transported out of the cell by facilitated diffu-sion. Seven human 5’NTs have been characterized so far (see “5’-Nucleotidases” below).

Regulation of nucleotide metabolism

Types of regulatory mechanisms The activity of some enzymes of the nucleotide metabolism is under tight control. In general, enzymes can be regulated so that their catalytic rate, localization or amount in the cell is changed. Often the first enzyme of a pathway is regulated so that the cell can “save resources”.

However, as described above, metabolic pathways are often complicated and branch at specific points into several pathways. Sometimes the ratio between different end products has to be regulated, as with the dNTPs (de-scribed below). Many enzymes of nucleotide metabolism are regulated and

5

can together fine-tune the level of the various nucleotides in response to the cell’s need.

All enzymes have an active site where they perform their catalysis. A ba-sic form of regulation is by substrate availability. All enzymes need a mini-mum concentration of substrate in order to perform the catalysis, which is dependent on how “tight” the substrate binds in the active site (the substrate affinity).

Often enzymes are “feedback” inhibited by the end product of their path-way. This inhibition generally occurs when the end product reaches a con-centration that exceeds the cell’s need. (d)NKs are feedback inhibited by (d)NTPs (see below), which exert their inhibitory effect by binding in the active site. Feedback inhibition can also occur through “allosteric” mecha-nisms.

Allosteric enzymes are regulated by specific “effector molecules” that bind at a site that is separate from the active site. The binding of an effector induces a conformational change in the enzyme that changes its catalytic rate. This adds a way for nature to “turn on/off” enzymes. The cytosolic 5’-nucleotidase II (cN-II) that is discussed in this thesis is an allosteric enzyme. The mechanism for its allosteric activation is presented in Paper IV.

Some enzymes are also synthesized and degraded in a regulated manner, as in the case of RNR (see below). There exist several more types of enzyme regulation but that is beyond the scope of this thesis.

Regulation of de novo synthesis The de novo synthesis of purine NMPs is largely regulated by allosteric feedback inhibition by IMP, AMP and GMP. This regulates both the overall rate of purine nucleotide synthesis but also the relative amounts of GMP and AMP. The de novo synthesis of pyrimidine nucleotides is feedback inhibited by CTP (1).

The de novo synthesis of dNTPs is mainly regulated by the allosteric en-zymes RNR and dCMP deaminase. They control the ratios between the four dNTPs (3, 4). RNR is allosterically regulated in a complex manner, in which both its activity and substrate specificity is regulated through separate effec-tor sites. It is activated by ATP and inhibited by dATP. Its specificity is turned towards GDP, by the binding of dTTP; to ADP by the binding of dGTP; and to CDP and UDP by the binding of dATP. Thus, it regulates the ratios between dATP, dGTP and pyrimidine nucleotides (3, 8).

The ratio between dCTP and dTTP is regulated by dCMP deaminase that converts dCMP to dUMP providing substrate for thymidylate synthase that converts dUMP to dTMP (Fig. 4). dCMP deaminase is allosterically acti-vated by dCTP and feedback inhibited by dTTP (3, 4).

6

Regulation of the salvage pathway The salvage pathway is also regulated by feedback inhibition and allosteric activation, which indicate its importance. dNKs and NKs are feedback inhib-ited by their corresponding dNTP, e.g. dCK is inhibited by dCTP and TK1 is inhibited by dTTP (9-12).

The activities of the intracellular 5’NTs are probably largely dependent on their substrate concentrations, since they have relatively low affinities for their substrates. The 5’NTs cytosolic 5’-nucleotidase IA and IB (cN-IA and cNI-B) are allosterically activated by preferably ADP, and cN-II is allosteri-cally activated by preferably dATP and ATP. This reflects that they are sen-sitive to the energy state of the cell.

The 5’NTs and (d)NKs co-exist in the cell, catalyzing opposite and irre-versible reactions. Thus they form “futile” substrate cycles together and can thereby fine-tune the levels of phosphorylated nucleosides within the cell (13) (Fig. 5). A single reversible enzyme would probably be less versatile in this respect. In case of purine ribonucleotides, the 5’NTs indirectly oppose the action of APRT or HGPRT.

Fig. 5. Substrate cycle between 5’NTs and (d)NKs. Nucleoside transporters (NTs) mediate (deoxy)nucleoside transport across membranes.

Cell cycle dependence The concentration of dNTPs in the cell varies during the cell cycle, being highest during the DNA synthesis phase (S-phase). In resting cells, low

7

amounts of dNTPs are needed for DNA repair and mitochondrial DNA syn-thesis, which occur independently throughout the cell cycle. Many of the anabolic enzymes of nucleotide metabolism are up-regulated before and during the S-phase of the cell cycle and down-regulated in resting cells.

RNR of the de novo pathway consists of two subunits (R1 and R2). The R2 is known to be degraded during mitosis and to be absent in resting cells (14). Because of this it was also believed that the de novo pathway is primar-ily responsible for dNTP synthesis in proliferating cells and that the supply of dNTPs in resting cells for cell cycle independent events such as DNA repair and mitochondrial DNA synthesis is provided through the salvage pathway. The salvage enzyme TK1 is also specific for the S-phase suggest-ing that the mitochondrial TK2 and dGK would be important in resting cells (15).

Recently, a second form of R2 has been discovered that is expressed in low amounts independently of the cell cycle. Since it can be induced by p53, it was named p53R2 and proposed to provide the cell with dNTPs for DNA repair (16). Mutation of p53R2 causes severe mitochondrial DNA depletion, which indicates its importance for mitochondrial DNA synthesis (17). I was shown that it is active in resting cells (18) and that resting cells have a com-plete de novo pathway (19). dTTP was formed in resting cells through the action of an RNR consisting of R1 and p53R2 in the cytosol and by the sal-vage enzyme TK2 in mitochondria (20). The differential contributions to the dNTP pools of dNTPs from the de novo pathway and the salvage pathway in resting cells are not yet clear.

5’-Nucleotidases (5’NTs)

Overview The first report of a 5’NT (EC 3.1.3.5) was made 1934 (21) and this family of enzymes has been reviewed several times since then (22-24). They all catalyze the hydrolysis of (d)NMPs into nucleoside and inorganic phosphate (Pi) (Fig. 6). The human 5’NTs are listed in Table 1, assigned with aliases, tissue specific expression and preferred substrates.

Fig. 6. 5’NTs dephosphorylates (d)NMPs.

8

Six of the seven characterized mammalian 5’NTs are intracellular en-zymes that have similar fold and catalytic residues. These are the mitochon-drial and cytosolic 5’(3’)-deoxynucleotidases (mdN and cdN), cN-IA, cN-IB, cN-II and cytosolic 5’-nucleotidase III (cN-III). They work by a similar catalytic mechanism, are completely dependent on Mg2+ for activity and are inhibited by Pi.

The intracellular 5’NTs share the role of regulating intracellular nucleo-tide pools for maintaining balanced DNA and RNA synthesis, although they also have more tissue-specific functions. Their Km values are very high rela-tive to the substrate concentrations. However, their biological roles in for instance regulation of dNTP pools are to some extent verified (see “Nucleo-tide pools” below).

The extracellular ecto 5’-nucleotidase (eN) has a different fold and cata-lytic mechanism relative the intracellular 5’NTs. It is anchored to the plasma membrane via a glycosyl phosphatidylinositol (GPI) anchor. eN exhibits high affinity for its substrates, especially AMP, and is transcriptionally regu-lated and participates in adenosine signaling.

9

Table 1. Properties of human 5’NTs 5’NT Alias Expression Preferred substrates, those

verified in vivo in bold

Intracellular mdN dNT-2 Ubiquitous dUMP, dTMP, 3’dTMP,

3’UMP cdN dNT-1; PN-II;

UMPH-2 Ubiquitous dUMP, dTMP, dGMP,

dIMP, 3’dTMP, 3’UMP

cN-IA AMP-specific 5’NT; cN-I

Heart, skele-tal muscle

AMP, pyrimidine-dNMPs

cN-IB cN-IA homo-logue; AIRP

AMP

cN-II High Km 5’NT; purine 5’NT; GMP-IMP specific NT

Ubiquitous GMP, IMP, dIMP, dGMP, XMP Phosphotransfer: (deoxy)inosine

cN-III PN-I, P5’NT, UMPH, UMPH-1

Erythrocytes CMP, UMP, dUMP, dCMP, dTMP Phosphotransfer: Uridine, cytidine, deoxy-cytidine

Extracellular eN Ecto-5’NT,

NT5, eNT, low 5’NT, CD73

Ubiquitous, GPI an-chored to plasma membrane

AMP, broad specificity, both pyrimidine and purine (d)NMPs

Cytosolic 5’(3’)-deoxynucleotidase (cdN) cdN was first purified from human placenta (25) and later from erythrocytes (26). It is ubiquitously expressed and shows highest catalytic efficiency (Vmax/Km) for 3’dUMP, 3’dTMP, 3’UMP and 2’UMP, although the enzyme also has high activity for all 5’dNMPs except dCMP (25, 26). Recombinant murine cdN shows similar specificity as the human enzyme but also exhibits some activity towards dCMP (27). cdN purified from erythrocytes was re-ported to have phosphotransferase activity (26), whereas cdN purified from human placenta and the recombinant mouse cdN do not show phosphotrans-ferase activity (25, 27).

10

Human cdN has a molecular mass of 23.4 kDa and is functional as a dimer. The crystal structures of both human and murine cdN were solved by us and are presented in Paper II.

cdN has mainly been studied for its role in regulating the cytosolic dTTP pool, by dephosphorylation of dUMP and dTMP (see “Nucleotide pools” below). It might also be important for catabolizing 3’- and 2’(d)NMPs pro-duced by nucleases.

Mitochondrial 5’(3’)-deoxynucleotidase (mdN) mdN shows 52% sequence identity to cdN and has similar size and oli-gomerization state as cdN. It is ubiquitously expressed and is targeted to the mitochondria by a signal sequence. Recombinant human mdN prefers dUMP, dTMP and 3’dTMP as substrates. However, it is not active with purine nucleotides (28).

The crystal structure of human mdN was published together with a pro-posed catalytic mechanism for intracellular 5’NTs (29). Further structural studies of this enzyme that involve its substrate specificity are presented in Paper I and II.

Two inhibitors for mdN were identified (30): (+)-1-trans-(2-Phosphonomethoxycyclopentyl)uracil (PMcP-U) a competi-tive inhibitor with a Ki of 40 μM, and (s)—1-[2’-deoxy-3’,5’-O-(1-phosphono)benzylidene-b-D-threo-pento-furanosyl]-thymine (DPB-T), an inhibitor with mixed-type inhibition and a Ki of 70 μM (30). PMcP-U is also a competitive inhibitor of cdN with a Ki of 26 μM, while DPB-T is not. Crystal structures of mdN in complex with PMcP-U and DPB-T showed the structural basis for the inhibition and pro-posed why DPB-T does not inhibit cdN (31). DPB-T can be used to distin-guish between mdN and cdN and PMcP-U to distinguish between cytosolic nucleotidases in enzyme assays (30).

The proposed function of mdN is to regulate the mitochondrial dTTP pool by counteracting TK2 in a substrate cycle (see “Nucleotide pools” below).

Cytosolic 5’-nucleotidase IA (cN-IA) cN-IA was first purified from pigeon heart where it shows high specific ac-tivity for AMP at millimolar concentrations (32). The enzyme is ~40 kDa and appears to function as a tetramer (32). It is abundant in heart and skeletal muscle (33, 34) but is also expressed in other tissues (35). Its structure has not yet been solved.

Comparing catalytic efficiencies (Vmax/Km) the enzyme prefers de-oxypyrimidine substrates over AMP (36). It is allosterically activated by ADP and to a lesser extent by GTP (35, 37). A selective inhibitor for cN-IA has been developed, 5-Ethynyl-dideoxyuridine (5-EddU), which is >1000-

11

fold more selective for cN-IA than cN-II or eN, suggesting it is a useful tool for specific inhibition of cN-IA (36, 38).

cN-IA might have a physiological function in heart in the generation of signaling adenosine during ischemia (33, 39). It is activated by ADP and not by ATP, which supports a role in AMP breakdown during ATP catabolism (35-37). The high affinity towards dNMPs suggests a role in regulating the dNTP pools.

Cytosolic 5’-nucleotidase IB (cN-IB) cN-IB has four possible isoforms produced by alternative splicing, of which the largest isoform has 610 amino acids. Counting 323 amino acids (identi-cal between isoform 1-3), cN-IB has 67% sequence identity with cN-IA. It is almost identical to the autoimmune infertility protein (AIRP) (40). It has the highest expression in testis and its oligomerization state is not known. Its structure has not yet been solved.

Overexpression of murine cN-IB in COS-7 cells showed that the enzyme hydrolyzes AMP and is activated by ADP (40), but its substrate specificity is still poorly characterized.

Cytosolic 5’-nucleotidase II (cN-II) cN-II was first purified from chicken liver (41) and has been intensively investigated in vertebrates for its specificity and regulation, but also for its involvement in resistance towards anticancer drugs (see “Anti-viral/cancer Nucleoside analogues” below).

It is ubiquitously expressed and functions as a tetramer (42-45). Recom-binant human cN-II consists of 561 amino acids although the reported mo-lecular mass of purified human cN-II varies. Its crystal structure was solved by us and is presented in Paper III. A structural study of its substrate speci-ficity and allosteric regulation is presented in Paper IV.

It prefers 6-hydroxypurine NMPs as substrates with the highest catalytic efficiency for IMP and GMP followed by dIMP, dGMP and XMP (41, 44-48). It possesses phosphotransferase activity (46-50), by which it can trans-fer a phosphate from a “donor” NMP (any substrate of the nucleotidase reac-tion) to an “acceptor” nucleoside, preferably inosine of deoxyinosine (51).

cN-II is allosterically activated by dATP, ATP, 2,3-bisphosphoglycerate (2,3-BPG), polyphosphates and diadenosine polyphosphates (ApnAs) (43-45, 47, 52). The ApnAs are composed of two adenosines that are linked together at the 5’-ends by a number of phosphates. Ap4-6As activate cN-II at micro-molar concentrations. These are formed as byproducts of aminoacyl tRNA synthetases and are removed by ApnA hydrolases before they reach toxic levels (53, 54). They have been reported as both intra- and extracellular sig-naling molecules (55).

12

The enzyme might have a complex regulation involving a second effector site (56, 57) (Paper III) and its 13 residue C-terminal acidic stretch that might mediate regulation via subunit association/dissociation (45).

cN-II appears to be involved in regulating the IMP and GMP pools within the cell (see “Nucleotide pools” below), and indirectly also participate in regulating the AMP pool.

Increased cN-II levels were found in patients suffering from Lesch-Nyhan syndrome, which is caused by a deficient HGPRT (see “Nucleotide pools and disease” below). Possibly cells with deficient HGPRT then instead use the phosphotransferase activity of cN-II to produce GMP.

Recently, knockdown of cN-II by RNA interference was shown to lead to apoptosis in astrocytoma brain tumor cells (58).

Cytosolic 5’-nucleotidase III (cN-III) cN-III has mainly been studied in red blood cells but has also been found in other tissues (59-61). It functions as a monomer (26, 61). The cN-III gene gives rise to four alternatively spliced mRNA transcripts, predicted to pro-duce enzymes of the lengths 297 (isoform 1), 286 (isoform 2), 285 (isoform 3) and 336 (isoform 4) amino acids. Isoform 2 (Q9H0P0-2) is identical to the p36 protein that has been found in inclusion bodies in lupus and AIDS (60).

cN-III shows highest catalytic efficiency (Vmax/Km) towards CMP fol-lowed by UMP, dCMP, dUMP and dTMP, and does not take purine nucleo-tides as substrates (62). Similar to cN-II, this enzyme has phosphotransferase activity. The transferred phosphate can originate from any monophospate substrate, whereas uridine and cytidine are shown to be the best phosphate acceptors (26).

cN-III is believed to catabolize UMP and CMP produced by RNA degra-dation, in red blood cells during the final steps of erythroid differentiation (63). During these steps, the RNA content of red blood cells is drastically reduced and cN-III appears to be up-regulated (63, 64).

A reduction of cN-III activity due to genetic defects or lead poisoning causes hemolytic anemia in humans (65). Here the circulating red blood cells show significant accumulation of pyrimidine nucleotides as well as incom-plete degradation of RNA and ribosomes, which indicate the important role of cN-III in red blood cell (63).

19 different mutations related to hemolytic anemia have been identified so far (66, 67). Many of these, e.g. the missense mutations giving D87V, L131P, N179S, G230R, C63R, G157R and I247T reduce the activity or sta-bility of the enzyme (61, 64, 67-70) (isoform 2: Q9H0P0-2).

The crystal structure of murine cN-III has been published (71) and the structure of human cN-III was solved by us and is discussed in this thesis. These structures display where the substitutions are located, which enlights how they could affect protein function.

13

Ecto-5’-nucleotidase (eN) eN has a different fold and catalysis relative the intracellular 5’NTs, with a catalysis depending on a dimetal center (see “Catalytic mechanism of eN” below). It is attached extracellularly to the plasma membrane by a GPI an-chor. The mature enzyme is of 548 residues and has a calculated molecular mass of 61 kDa and functions as a dimer.

The crystal structure of an E. coli homologue of human eN has been solved (72). A homology model was made of the human eN based on the E. coli structure (73).

eN has a broad substrate specificity, although AMP is the most efficient substrate considering Vmax/Km values. The Km values for AMP are in the low micromolar range and ADP and ATP are competitive inhibitors. Because of the high affinity for AMP and the fact that extracellular adenine nucleotides are relatively abundant, AMP has been considered as the major substrate.

Extracellular adenosine, ADP, ATP and some other nucleotides function as signaling molecules for purinergic receptors. Adenosine acts on adenosine (P1) receptors and eN probably has a biological role in releasing adenosine for adenosine receptor activation (74-78).

eN is expressed in most tissues although only in certain cell types within each tissue (79). This is controlled at the transcriptional level and several transcription factor-binding sites are present in the promoter region that may mediate both positive and negative regulation. Ischemia and hypoxia has been shown to induce eN expression (80-82), indicating a role for eN in sig-naling events related to these conditions.

eN probably also affects the intracellular nucleotide pools, since it’s nu-cleoside products can pass the plasma membrane through nucleoside trans-porters.

Nucleotide pools and disease The size of the dNTP pools and their relative amounts are important. It was early demonstrated in E. coli that depletion of the dTTP pool induces cell death (83). The consequences of imbalanced dNTP pools are also reflected in several genetic diseases that are associated with deficiencies in enzymes of nucleotide metabolism.

Deficiencies in adenosine deaminase (ADA; that converts de-oxyadenosine and adenosine to deoxyinosine and inosine) or in PNP lead to sever immune deficiency due to defects in T and B cell function. These de-fects are associated with accumulation of dATP and dGMP, respectively (84, 85). Imbalance in dATP or dGTP pools affect the activity of RNR, thus af-fecting also the other dNTP pools. Elevated levels of dATP might trigger

14

apoptosis, since dATP participates together with cytochrome c in triggering apoptosis through Apaf-1 mediated activation of procaspase 9 (86, 87).

Deficiency in the salvage enzyme HGPRT, which causes Lesch-Nyhan syndrome, leads to accumulation of hypoxanthine and guanine followed by overproduction of uric acid, which causes problems such as severe gout, poor muscle control, and moderate mental retardation (88).

Deficiency of TP that degrades thymidine and deoxyuridine has been found in patients with the mitochondrial neurogastrointestinal encephalo-myopathy (MNGIE), which is a disease linked to e.g. deletions in mitochon-drial DNA (89). This causes mitochondrial dysfunction that gives severe symptoms (90). Increased levels of thymidine and deoxyuridine (from <0.5 μM to 10-20 μM) were found in blood plasma of patients with MNGIE (91). This leads to elevation of cytosolic and mitochondrial dTTP and to other effects that cause imbalance in dNTP pools.

Loss-of-function mutations in TK2, dGK and in the RNR subunit p53R2 (see below) have been found in patients suffering from mitochondrial DNA depletion syndrome (MDS) (17, 92-94). MDS causes quantitative mitochon-drial DNA depletion. The concomitant low amount of mitochondrial DNA is presumably causing insufficient synthesis of the mitochondrial DNA-encoded proteins involved in oxidative phosphorylation, causing loss of muscle activity and death in early childhood (90, 92).

The only 5’NT deficiency known is that of cN-III, which has been associ-ated with hemolytic anemia (discussed above) (65).

Nucleotide pools The dNTP pool sizes within the cell are rather small and vary during the cell cycle (Table 2). During the S-phase of the cell cycle these pools would be diminished within a few minutes without biosynthesis. The genetic diseases discussed above underline the importance of maintaining balanced nucleo-tide pools. The dNTP pools are much smaller in resting cells than in differ-entiating cells (95). The cytosolic and mitochondrial pools of dTTP and dGTP are exchanged rapidly and share a dynamic equilibrium (96).

15

Table 2. Average dNTP (97) and NTP (98) pool sizes in mammal cells (d)NTP Mitochondria Cytosol pmole/106 cells (μM if

106 cells is 0.02 μl) pmole/106 cells (μM if 106 cells is 0.2 μl)

dTTP 0.88 (44) 32 (160) dATP 0.75 (37.5) 17.5 (87.5) dCTP 0.74 (37) 27.5 (137,5) dGTP 0.40 (20) 5.9 (29.5)

Concentrations below are shown in μM UTP 480 ATP 2800 CTP 210 GTP 480

In situ studies on the substrate cycles between dNKs and 5’NTs have pro-

vided evidence supporting that dNKs and 5’NTs are involved in the regula-tion of specific nucleotide pools. Mammal cell lines lacking a dNK or over-expressing a 5’NT were produced (99, 100). By measuring excretion of nu-cleosides from these cell lines by isotope flow experiments, the involvement of dNKs and 5’NKs could be measured (97, 100-102).

This provided evidence for anabolic roles of TK1 in thymidine/TMP and deoxyuridine/dUMP substrate cycles and of dCK in deoxycytidine/dCMP and deoxyadenosine/dAMP substrate cycles (99, 103).

cN-II was indicated not to be involved in the regulation of the deoxynu-cleotide pools but instead affect the IMP and GMP pools (100). Recombi-nant murine cdN gave evidence for a catabolic role of cdN in substrate cy-cles regulating pyrimidine nucleotide pools (100). Early studies on human cN-III deficient erythrocytes showed that dTMP and dUMP, but not UMP, CMP and dCMP were degraded in these cells. This indicated the existence of another pyrimidine specific cytosolic 5’NT, besides cN-III, that took dTMP and dUMP, but not UMP, CMP or dCMP as substrates (59, 104). This en-zyme was later identified as cdN.

It was also demonstrated that mdN participates in a substrate cycle to-gether with TK2 in regulating the mitochondrial dTMP pool (97).

There are still many gaps in the understanding of the cellular functions of the different 5’NTs. For instance, it is yet not clear which 5’NTs regulate the cytosolic or mitochondrial dGTP or dATP pools. Although for regulation in the cytosol, in vitro kinetic studies suggest cdN and cN-II for dGTP regulta-tion and cN-IA for dAMP regulation.

16

Nucleoside/nucleotide transporters Two types of mammalian nucleoside transporters have been characterized. These are the equilibrative nucleoside transporters (ENTs) that transport nucleosides and bases through the plasma membrane bidirectionally down their concentration gradient, and the concentrative nucleoside transporters (CNTs) that cotransport nucleosides against their concentration gradient into the cell together with sodium. The CNTs use the force derived from trans-porting sodium down its concentration gradient to concentrate nucleosides inside the cell (105).

Of the four known human ENTs (hENT1 (106), hENT2 (107), hENT3 (108) and hENT4 (109), hENT1 and hENT2 are mainly found in the plasma membrane and have broad tissue distribution. They transport both pyrimidine and purine nucleosides and ENT2 can also transport some bases. These two can be distinguished by their sensitivity to inhibition by nanomo-lar concentrations of nitrobenzylmercaptopurine ribonucleoside (NBMPR), hENT1 being sensitive and hENT2 being insensitive (105). ENT1 has also been found in membranes of mitochondria, lysosomes, the nuclear envelope and the endoplasmic reticulum (105).

Three human CNTs (hCNT1, hCNT2 and hCNT3) have been character-ized (110-112). The CNTs are highly expressed in intestine, kidney and liver. This implies that they are involved in absorption and excretion of die-tary nucleosides (105). ENTs and CNTs are important for the cellular uptake of drugs used in the treatment of cancer and viral diseases (105).

When nucleosides become phosphorylated into nucleotides, they can no longer be transported across the plasma membrane. However, they can dif-fuse through nuclear pores and there is no concentration gradient of nucleo-tides between the cytosol and the nucleus (113). Nucleotides are transported in a controlled manner in and out of the mitochondria. The adenine nucleo-tide carrier (ANC) catalyzes transport of ADP into mitochondria in exchange for ATP, which is needed for the generation and distribution of ATP, that is formed from ADP by the action of ATP synthase within the mitochondria.

A mitochondrial carrier (SLC25A19) was reported as a deoxynucleotide carrier (DNC) that catalyzes transport of (d)NDPs and (d)NTPs across the mitochondrial inner membrane (114). Later studies indicate that SLC25A19 is not a deoxynucleotide carrier, but might be a carrier for thiamine pyro-phosphate (115-117).

A highly specific import system for dTMP into the mitochondria was demonstrated (118). This could be the main road for dTTP supply to mito-chondria, since dTMP is the first thymidine nucleotide formed during de novo synthesis (118). Evidence for transport of dCTP and UTP in mitochon-dria have also been reported (119, 120).

17

Anti-cancer/viral nucleoside analogues Nucleoside analogues (NAs) are widely used in treatment of cancer and viral diseases. Fig. 7 provides chemical structures and names of some NAs. Some of these have been used in treatment of HIV-1 (e.g. AZT and d4T), or other viral diseases (e.g. BVDU and ribavirin). Others are currently used in treat-ment of cancer (e.g. araC, dFdC and CdA). NAs are prodrugs that need to be phosphorylated within the cell to become active. They are given to the pa-tient as nucleosides since their phosphorylated forms cannot pass through the plasma membrane.

To exert their cytotoxic effect, many NAs are phosphorylated to their triphosphate form by (d)NKs, NMPKs and NDPKs (5, 121, 122) and then presumably incorporated into DNA leading to DNA chain termination. Many NAs also act through enzyme inhibition, e.g. inhibiting DNA poly-merase α or RNR (123). Inhibition of RNR decreases the biological dNTP pools and thereby increases the NA-TP/dNTP ratio.

18

Fig. 7. Nucleoside analogues. Top: AZT: 3’-azidothymidine, BVdU: E)-5-(2-bromovinyl)-2’-deoxyuridine and d4T: 2’,3’-didehydro-2’,3’-dideoxythymidine. Middle: AraC: 1-β-D –arabinofuranosylcytosine, dFdC: 2’,2’-difluoro-2’-deoxycytidine (Gemcitabine) and CdA: 2’-chlorodeoxyadenosine. Bottom: Ri-bavirin.

A successful NA treatment is highly dependent on the level of cell uptake through plasma membrane transporters and on how efficiently they are phosphorylated into their active form (105, 123).

5’NTs can dephosphorylate the monophosphate form of many NAs (NA-MPs). This is not only a step in the inactivation of the drugs, but it also re-leases them to exit the cell.

Patients often evolve resistance to NAs. The mechanism behind this has been studied intensively, e.g. concerning resistance to cytosine and adenine

19

based NAs such as araC, dFdC and CdA in patients suffering from acute myeloid leukemia (AML). 5’NTs, ENT1, RNR, dCK and other enzymes involved in metabolism or uptake of NAs have been the focus in these stud-ies (123).

Several studies link increased cN-II activity or mRNA expression with re-sistance to cytosine and adenine based NAs in cell lines and poor clinical outcome of araC treatment of AML patients (124-126). Inhibition of cN-II resulted in a 2-5 -fold increase in dFdC cytotoxity in lymphocytic B-leukemia cells (127). However, cN-II expression was not linked to dFdC resistance in patients with advanced non-small cell lung cancer: here high levels of cN-II mRNA instead correlated with increased survival (128).

The focus on cN-II in studies of adenine and cytidine based NAs can seem somewhat strange since it has been known for a long time that cN-II has very low activity towards cytosine and adenine based nucleotides. More recent in vitro activity measurements on recombinant cN-II indicated a very low activity towards such NA-MPs (Table 5) (30). This was supported by the observation that increased expression of cN-II in human 293 cells gave no increased resistance towards toxicity of CdA (129). However, dephos-phorylation of the monophosphophate form of fludarabine (an adenine-based NA; F-ara-AMP) by cN-II was measured by HPLC and NMR (130), which suggested a role of cN-II in dephosphorylating this analogue.

A 5’NT with known high activity towards AMP is cN-IA. Overexpression of cN-IA in human cell lines generated resistance towards CdA and dFdC (35, 131). cN-IA dephosphorylated CdAMP with a specific activity of 39% compared to that of AMP (35).

cN-II might activate guanosine and inosine analogues by its phos-photransferase activity. cN-II was suggested as the main enzyme activating the anti-HIV analogue 2’,3’-dideoxyinosine (ddI) (50) . It was also shown to activate the antiviral NA ribarivin significantly better than the human (d)NKs (132).

cdN could in principle be involved in resistance towards thymine, uracil or guanine based NAs, considering its substrate specificity and ubiquitous expression. It also shows some activity towards such NAs in vitro (Table 5) (30).

The mitochondrial mdN is however not likely to be involved in NA resis-tance, since the NAs are mainly targeted to nuclear DNA. Instead mdN might affect the mitochondrial toxicity that is commonly experienced as a side effect of NA treatment. The mechanism behind this toxicity comes from competitive inhibition of mitochondrial enzymes such as the mtDNA poly-merase γ. Such inhibition of the mtDNA polymerase γ, by AZT, was associ-ated with mtDNA depletion in HIV-1-infected patients (133).

20

Evolution and catalysis of intracellular 5’NTs

Overall structure of intracellular 5’NTs The intracellular 5’NTs belong to the haloacid dehalogenase (HAD) super-family, which includes phosphohydrolases such as the 5’NTs, phosphotrans-ferases, P-type ATPases, phosphonatases, dehalogenases and sugar phos-phomutases (134, 135).

The HAD superfamily is characterized by a core domain, consisting of an α/β-Rossmannoid fold, containing a three-layered α/β sandwich built up from repeating β−α units. The β−sheet is parallel and consists most often of at least five strands in a 54123 strand order, which is typical for the α/β-Rossmann fold.

The fold of HAD superfamily enzymes can be distinguished from other α/β-Rossmann-like folds by two structural signatures, a “squiggle” (a single helical turn) and a “flap” (a β-hairpin motif) (135). The “squiggle” is located immediately downstream of strand S1 and the “flap” is located downstream of the “squiggle” (Fig. 8a).

Most HAD superfamily enzymes also contain a cap domain, which is much less conserved than the core domain. This cap domain is inserted ei-ther (1) in the middle of the β-hairpin of the “flap” motif or (2) immediately after strand S3. mdN has the cap domain inserted after the “flap” motif (Fig. 8a). Bioinformatic analysis (135) indicates that the cap domain has been inserted at multiple occasions during evolution, independently of the core domain, and that this has enabled the great diversity in function of these enzymes.

The active site is located in the cleft between the core and the cap do-mains with the catalytic residues in the core domain and most of the residues that determine substrate specificity in the cap domain. In the 5’NTs, the sub-strate binds with its phosphate and ribose moieties in the core domain and its base moiety in the cap domain (Fig. 8b). The cap domain probably deter-mines the base specificity of the 5’NTs.

Most of HAD enzymes share four structurally conserved motifs that take part in forming the catalytic Mg2+-phosphate site (with some exceptions, e.g. dehalogenases). Motif I (DXDX[T/V]) is located on strand S1 and on the downstream “squiggle” loop, and motif II ([T/S]) on strand S2. The loca-tions of motif III ([K/R]) and motif IV (DX0-4D) are less conserved. Motif III is found on the loop downstream of strand S3 and motif IV is located in strand S4 and downstream of it. (motif III and IV are sometimes described as both included in motif III) (Fig. 8b).

21

Fig. 8. HAD superfamily fold and conserved motifs (D41N variant of mdN; Paper I). a) The core domain (light) and cap domain (dark), with strands S1-S5, squiggle (Sq) helical turn, Flap β−hairpin, C-terminal (C) and N-terminal (N). b) Motif I residues: N(D)41, D43 and V45. Motif II residue: T130. Motif III residue: K165. Motif IV residues: D175, D176. dUMP and Mg2+ are also shown.

The annotation of intracellular 5’NTs to the HAD superfamily was first proposed by sequence analysis (136), which was later confirmed by the structure of human mdN (29), followed by those of human cdN (Paper II), human cN-II (Paper III) and human cN-III (unpublished). The murine cdN was also solved by us (Paper II), whereas the murine cN-III structure was published by another group (71).

cN-II has a large cap domain following the “flap” motif and an additional fold following strand S3 (Fig. 9).

cN-III is structurally very similar to phosphoserine phosphatase (PSP), which dephosphorylates phosphorylated L-serine (137). Both cN-III and PSP have cap domains following the “flap” motif, but also small contributions to the cap domain inserted after strand S3 (Fig. 10).

22

Fig. 9. cN-II structure with cap domain (dark) and core domain (light). Squiggle (Sq) motif and Flap motif are marked.

Fig. 10. Overall structure of a) cN-III and b) PSP, with cap domains (dark) and core domain (light). Squiggle (Sq) motif and Flap motif are marked, as well as strands S1-S5.

23

Catalytic mechanism The catalytic mechanism of intracellular 5’NTs (Fig. 11) was proposed based on the structures of mdN that mimic reaction intermediates 3, 4 and 6 of the proposed mechanism (29). It has been supported by a theoretical study (138) and is also very similar to the catalytic mechanism proposed for PSP (139). The catalytic mechanism of 5’NTs starts as follows:

The first aspartate of motif I initiates the catalytic reaction by making a nucleophilic attack on the phosphate moiety of the substrate. Hypothetically, it might form a pentavalent substrate intermediate (Intermediate 2) (140). This nucleophilic attack is probably conserved throughout the HAD super-family.

The second aspartate of motif I appears to be bifunctional. It first acts as a general acid, donating a proton to the leaving nucleoside product. A phos-phoenzyme intermediate probably remains (Intermediate 3 and 4) (136, 141). This Asp then probably acts as a general base, activating the water molecule that then hydrolyses the phosphate (Intermediate 5 and 6).

The second Asp of motif I is present in most HAD superfamily enzymes where there is a need for protonation of the leaving group. It is not present in for instance HAD (in which the first leaving group is a Cl- ion) or in P-type ATPases (where the leaving group is ADP) (Table 3). Neither Cl- nor ADP need to take up a proton. The threonine that substitutes the second Asp in P-type ATPases appears to have a function in stabilizing the phosphoenzyme intermediate, enabling the necessary conformational change that drives the ion transport.

Motif II (Thr/Ser) is very conserved and appears to coordinate the phos-phate moiety in many HAD enzymes. Motif III (K/R) is also very conserved. It probably has a role in stabilizing negative charge in reaction intermediates.

Motif IV contains two aspartates, of which one is coordinating the Mg2+ and the other has an unclear role. The sequential order of these two aspar-tates varies in different enzymes. In Table 3, the Asp with unknown role is marked Asp*.

We speculated in Paper IV over the role of this Asp*. It is clearly in-volved in the mechanism of allosteric activation of cN-II. This Asp* is very conserved among the HAD enzymes. Substitutions of this residue abolish the enzyme activity in mdN, PSP, HAD and Sarcoplasmic Ca2+-ATPase (SERCA) (142-144 and unpublished data on mdN).

24

Fig. 11. Proposed catalytic mechanism of intracellular 5’NTs. The scheme is taken from reference (29).

25

Table 3. Conseved motifs of HAD superfamily enzymes Protein (pdb code) Motif I

DXDX[V/T] Motif II [T/V]

Motif III [K/R]

Motif IV DX0-4D

C–Cl cleavage HAD (1QQ6) 8DAYGT S114 K147 D214, D218* C–P cleavage PALD (1FEZ)1 12DWAGT T126 R160 D186, D190* CO–P cleavage PSP (1F5S) 11DFDST S99 K144 D167, D171* Human mdN 41DMDGV T130 K165 D175*, D176 Human cdN 10DMDGV S99 K134 D144*, D145 Human cN-IA 211DGDAV - - - Human cN-IB2 467DGDAV - - - Human cN-II 52DMDYT T249 K292 D351, D356* Human cN-III3 38DFDMT S153 K202 D227, D231* MPP (1XVI)4 13DLDGT S47 K188 D214, D218* MDP1 (1U7O)5 11DLDYT S69 K100 D122*, D123 CO–P cleavage/formation (mutase) β-PGM (1O03)6 8DLDGV S114 K145 E169*, D170 CO–P cleavage/formation (mutase) SERCA (2ZBD) 351DKTGT T625 K684 D703, D707* H-pump (3B8C)7 329DKTGT T511 K569 D588, D592* Na/K ATPase8 369DKTGT T610 K691 D710, D714* 1Phosphonoacetaldehyde hydrolase 2Isoform 1 (Q96P26-1) 3Isoform 2 (Q9H0P0-2) 4Mannosyl 3-phosphoglycerate phosphatase 5Mg2+-Dependent phosphatase 1 6 β-phosphoglucomutase 7Plasma membrane proton pump from Arabidopsis thaliana 8Na/K ATPase from pig (3B8E)

Catalytic mechanism of eN The extracellular eN dephosphorylates (d)NMPs by a mechanism different from that of the intracellular 5’NTs. The structure of the E. coli homologue of eN showed that it belongs to a large superfamily of metallophos-phoesterases that use a dimetal center for catalysis (72).

Based on the structure of E. coli eN in complex with α,β-methylene-ADP, a catalytic mechanism was proposed (145). The E. coli eN has broader substrate specificity than human eN, but consists of very similar structural motifs participating in the catalysis. The E. coli eN can dephosphorlyate not

26

only monophosphates but also di- and triphosphates, whereas the human enzyme is competitively inhibited by di- and triphosphate nucleosides.

One oxygen atom of the terminal phosphate group of α,β-methylene-ADP coordinates the dimetal centre. A catalytic histidine and an arginine were proposed to polarize the phosphate group to prepare it for the nucleophilic attack, which appears to be carried out by a water molecule (145).

The mechanism of E. coli eN also involves a large domain rotation (146, 147). This might be conserved to human eN, although a homology model of human eN indicates that the active site is more accessible due to a shorter loop (73).

27

Present Investigation

Aim The focus of my PhD project was to study substrate recognition, catalysis and regulation of 5’NTs. The main method used was x-ray crystallography.

The project was initiated with the aim to solve structures of mdN and cdN in complex with substrates. We wanted to analyze their substrate recognition in detail, which would increase the understanding of their substrate specific-ity and their possible involvement in the degradation of NA-MPs.

The project continued with the aim to determine structures of the remain-ing human intracellular 5’NTs. We solved the structures of human cN-II and human cN-III. We aimed at investigating the structural basis for their sub-strate specificity and to study the allosteric regulation of cN-II.

Strategy To achieve our aims, we needed to produce crystals of the enzymes with substrate (or nucleoside product) bound in the active sites.

This might seem straightforward at first glance, but substrate/product complexes cannot be obtained unless the enzyme can be stabilized in a state where the substrate/product can stay bound for a longer time.

A reason for this is that enzymatic reactions such as the hydrolysis of a phosphomonoester bond often are extreamly fast and substrate/product in-termediates very short-lived (148). Crystal structures are built from data that reflect the “average” state of the protein; the structures show the reaction intermediate that dominates at the equilibrium of the crystal. Because of this, it is very unlikely to obtain a structure of a 5’NT in complex with substrate or nucleoside product, without first performing a trick. We used two strate-gies to produce substrate/nucleoside product complexes, see below. Several structures of various complexes were solved (Table 4).

28

Fluorometallic complexes as models for enzyme intermediates Fluoride has long been known to be very toxic. In complex with metals, fluoride is known to inhibit nucleotide-binding proteins, e.g. phosphatases, phosphorylases and many ATPases, and also activate G-proteins participat-ing in signal transduction. It has a preference to form complexes with metals such as aluminium, beryllium or magnesium. These complexes, e.g. BeF3

-, MgF3

- and AlF4-, mimic phosphate and bind with high affinity to phosphate

sites in proteins. They also form tight penta- or hexavalent complexes with reaction products, trapping them in the active site thus locking the enzyme in a state similar to e.g. a substrate intermediate state (149).

This property of fluorometallic complexes has been used extensively by crystallographers to mimic reaction intermediates, in order to study catalytic mechanisms. Such studies were made on HAD superfamily enzymes e.g. mdN (29) and PSP (139) and several other proteins e.g. cAMP-dependent protein kinase (150) and transducin α (151). This strategy was used several times in the present investigation, e.g. we solved the structure of human cdN in complex with the product deoxyuridine together with AlF4

- (Paper II). It is highly unlikely to capture the “real” high-energy reaction intermedi-

ate of a phosho-transfer reaction. In spite of this, a structure of β−PGM in complex with a “real” pentacoordinated phosphorane intermediate was pub-lished (140). This intermediate was later indicated by 19F NMR to be a MgF3

- complex (152). Recently, it was also shown by 19F NMR that tetrahe-dral fluorometallic complexes modeled in electron densities as AlF3

- are probably MgF3

- (153).

Substitutions to trap substrate in active site The proposed catalytic mechanism of 5’NTs helped us to design inactive variants of mdN (D41N), human cdN (D10N), murine cdN (D12N), cN-II (D52N), and cN-III (D38N, isoform 2: Q9H0P0-2), with the purpose of binding substrates in the active site without hydrolysis. In these variants, the first aspartate of motif I (which in the catalytic reaction makes a nucleophilic attack on the phosphate of the substrate) was mutated into an asparagine. This asparagine forms a strong hydrogen bond to the phosphate probably further increasing the affinity for the substrates without hydrolysis.

Several structures of human mdN (D41N) and murine cdN (D12N) in complex with various substrates and NAs were determined (Paper I and II). The cN-II variant D52N enabled us to bind substrates together with activa-tors (Paper IV).

We could not obtain structures of substrate complexes of the D10N vari-ant of human cdN, probably since Pi (present in the crystallization solution)

29

competively inhibited the substrate to bind. The crystals did not survive longer soaking experiments for the necessary removal of Pi.

Table 4. Structures solved Protein Structures (Resolution in Å;pdb code) Wild type mdN FdU-AlF4

- (1.7)1 Ca2+ (1.9)1

D41N variant of mdN dUMP (2.0; 1Z4I)2 dTMP (1.8; 1Z4L)2 dGMP (2.0; 1Z4P)2 UMP (1.7; 1Z4M)2 3’dTMP (1.75; 1Z4K)2 2’UMP (1.8; 1Z4J)2 d4TMP (2.05; 1Z4Q)2 AZTMP (1.8; 2JAU)3 BVdUMP (1.95; 2JAW)3

PO4 (1.4)1 Wild type human cdN Deoxyuridine-AlF4

- (1.2; 1SRW)3 D10N variant of human cdN PO4 (1.05)1 Wild type murine cdN Deoxyuridine-BeF3

- (1.5)1 Thymidine-AlF4

- (1.6)1 D12N variant of murine cdN dUMP (1.9; 2JAR)3

dGMP (2.0; 2JAO)3 Wild type cN-II Ado (1.5; 2JC9)4

BeF3 (2.15; 2JCM)4 SO4 (2.2; 2J2C)4 Ap4A-SO4 (2.2)1

D52N variant of cN-II Apo (2.3)5 ATP (2.0)5 IMP-ATP (1.9)5 IMP-BPG (2.3)5 GMP-Ap4A (2.0)5 dGMP-dATP (2.3)5 UMP-ATP (2.3)5

GMP-BPG (2.5)1 Carboxy-Asp-BPG (2.1)1

Wild type cN-III Apo (2.7; A2CN1)1 BeF3 (2.5; 2VKQ)1 PO4 (3.0; 2JGA)1

1Unpublished, 2Paper I, 3Paper II, 4Paper III, 5Paper IV

30

Structural basis for substrate specificity of mdN and cdN (Paper I and II)

mdN and cdN share 52% sequence identity. They differ in their substrate

specificity (Table 5), mdN being very specific for the uracil and thymine based nucleotides, whereas cdN also takes dGMP and dIMP as substrates. Murine cdN is 85% sequence identical to human cdN. Human and murine cdN have similar specificities although the murine enzyme has some activity also for dCMP (Table 5). We wanted to characterize the structural basis for the substrate specificity of mdN and cdN, and to analyze the basis for the differences between them in substrate specificity.

The three enzymes have very similar overall structures and their active sites are also very similar. Despite this, they have some important differ-ences in the base recognition site of the cap domain: Residues Ile133, Trp76 and Trp96 in mdN correspond to Leu45(47), Tyr45(47) and Leu102(104) in human and murine cdN (the residue numbers in parenthesis refer to murine cdN).

31

Table 5. Substrate Specificities of cdN, mdN and cN-II

Relative enzyme activity, % Substrate Human cdNa Murine cdNb Human mdNc Human cN-IId

dUMP 100 100 100 (100) 23 dTMP 55 65 50 (48) 16 dCMP 0.7 16 0 (0) 1 dAMP 13 9 2 (1) 7.5 dGMP 81 45 6 (2) 72 dIMP 245 96 8 (3) 86 IMP 21 11 Nd 100 GMP 4.5 4 2 (0) 85 AMP 0 1 0 (1) 7 XMP Nd Nd Nd 23 UMP 8.9 16 30 (37) 14 CMP 0 0 0 (0) 6 3’dTMP Nd 58 77 (103) Nd 3’UMP 76* 35 47 (34) Nd 2’UMP 53* 11 18 (15) Nd Nucleoside analogue

d4TMP1 4 3 9 0.6 AZTMP1 22 103 2 6 BvdUMP1 8.5 19 28 1.5 FdUMP1 217 108 82 11 AraTMP1 0.1 Nd 1 0.2 ddCMP1 1 18 0.1 1 dFdCMP1 2 20 0.1 0.6 araCMP1 0.1 0.2 0.1 0.2 3TCMP1 1.5 1.2 0.2 0.2 CdAMP1 0.5 Nd 0.1 4.5 araGMP1 1 Nd 0.3 3.5 The relative enzyme activities are taken from previous publications: 1(30) a(25) (*calculated from the Vmax values) b(27) (in parenthesis is measured at pH 7.5) c(28) which also report Km and Vmax values for the individual enzymes. d(44) Nd: Values not found in literature

32

Pyrimidine base specificity The base recognition site of both mdN and cdN contains two main chain amides that form favorable hydrogen bonds to the 4-carbonyl group of (d)UMP and (d)TMP, but would repel the amino group of (d)CMP. In cdN, the two main chain amides also favorably bind the 6-carbonyl group of (d)GMP and (d)IMP, but would repel the amino group of (d)AMP (Fig. 1, Fig. 12).

In this way, mdN and cdN can discard cytosine and adenine based nucleo-tides as substrates (Paper I and II).

Fig. 12. mdN-dUMP and 4-carbonyl specific hydrogen bonds to main chain amides.

33

Deoxy/ribo specificity Both mdN and cdN prefer the deoxy-form over the ribo-form of nucleotides (Table 5). They accomplish this specificity by having a hydrophobic patch stacking the 2’-C of deoxy-substrates (Fig. 1). Fig. 13 shows the mdN-dUMP structure and the mdN-UMP structure, superimposed.

The 2’C of dUMP is favorably stacked by Phe49, Phe102 and Ile133, while the 2’OH of UMP binds unfavorably close to these residues, with the closest distance (2.7 A) to Ile133. Clearly, the hydrophobic patch of Phe49, Phe102 and Ile133 provides the deoxy specificity of mdN (Paper I). This hydrophobic patch is also conserved in cdN, where the corresponding resi-dues in human cdN are Phe18, Phe71 and Ile102.

Fig. 13. mdN-dUMP structure (light) and mdN-UMP structure (dark) superimposed. Dotted lines indicate the unfavorable interactions between the 2’OH and hydropho-bic/aromatic residues.

34

2’-,3’- and 5’-phosphate specificity The relatively high activity of mdN and cdN towards 2’- and 3’-phosphorylated (d)NMPs made us compare the substrate recognition of them with the 5’dNMPs (Fig. 14) (Paper I).

The phosphate moieties are recognized similarly, whereas the ribose and base moieties of 3’dTMP and 2’UMP are rotated relative in dUMP (Fig. 14).

The bases of 3’dTMP and 2’UMP appear to be more favorably stacked by the aromatic Phe49, than the pyrimidine base of 5’substrates (Fig. 14). This supports the idea that at least 3’-phosphorylated pyrimidine (d)NMPs are biologically relevant substrates.

It becomes clear that 3’- and 2’-purine NMPs would probably not bind with high affinity, since the purine base probably would clash into aromatic residues of the active site.

Fig. 14. mdN in complex with dUMP, 3’dTMP and 2’UMP superimposed.

35

Purine/pyrimidine specificity Human and murine cdN have a wider base recognition site than mdN, which probably is the basis for why they have higher activity for purine nucleotides than mdN. The active site residues Leu45(47), Tyr45(47) and Leu102(104) of human and murine cdN (the residue numbers in parenthesis refer to mur-ine cdN) form a more favorable binding surface for purine nucleotides than the corresponding Ile133, Trp76 and Trp96 of mdN. Fig. 15 shows the mdN-dGMP structure (Paper I) and the murine cdN-dGMP structure (Paper II) superimposed.

dGMP fits nicely in the active site of cdN, with favorable distances to catalytic residues.

In mdN, dGMP binds slightly displaced from catalytic residues. Its phos-phate moiety is displaced from Asp43 (second aspartate of motif I), the resi-due that is proposed to donate a proton to the leaving nucleoside during the catalysis (Fig. 11). This longer distance probably makes the transfer of a proton from Asp43 to the nucleoside less efficient, causing slower catalysis. This, in combination with lower affinity, probably causes the low activity of mdN towards purine nucleotides.

Fig. 15. Superposition of the mdN-dGMP structure (light) and the murine cdN-dGMP structure (dark).

36

Nucleoside analogue recognition in mdN The D41N variant of mdN was also used to study recognition of the mono-phosphorylated NAs d4TMP, AZTMP and BVdUMP (Paper I and Paper II). The chemical structures of these NAs are shown in Fig. 7, and their rela-tive activities with cdN, mdN and cN-II are shown in Table 5.

The anti-HIV analogue d4TMP binds nearly identically to dTMP, with the exception of the missing 3’OH (Fig. 16). The low activity for d4TMP indicates the importance of the 3’OH for retaining enzyme activity.

There is a tight interaction of the 3’OH of dTMP with the backbone of mdN. In all solved structures of 5’-(d)NMP complexes of both mdN and cdN, the 3’OH forms tight interactions with the protein (however, sometimes the 3’OH forms a hydrogen bond with Asp43 (mdN) instead of the main chain). This indicates that most analogues with substituents at the 3’-position would be poor substrates for mdN and cdN.

Fig. 16. Superposition of the mdN-d4TMP structure (dark) and the mdN-dTMP structure (light).

The mdN-AZTMP structure shows another example of how an unfavor-

able binding in the pocket for the 3’OH, negatively affects catalytic effi-cency. The low activity towards AZTMP could be explained by the interac-tions of the azido group with the protein (Fig. 17). The azido group at the 3’-position of AZTMP is too large to fit in the pocket for the 3’OH. This probably greatly reduces substrate affinity.

37

Fig. 17. Superposition of mdN-AZT-MP (dark) and mdN-dTMP (light), the azido group is marked AZ.

The analogue BVdUMP is a significantly better substrate for mdN than AZTMP and d4TMP, probably because BVdUMP is substituted on the base instead of the 3’-position. It fits relatively well in the active site of mdN, although the base is turned 180 degrees relative the other substrates.

However, BVdUMP is a very poor substrate for cdN. The reason for this might be that Arg47 and Tyr65 of human cdN possibly would clash with the bromovinyl group of BVdUMP (Fig 18).

Fig. 18. Superposition of mdN-BVdUMP and human cdN. Residues R47 and Y65 belong to cdN, S78 and W96 to mdN. Clashes indicated by dotted lines.

38

Implications for the catalytic mechanism The substrate complexes of mdN and cdN mimic the reaction intermediate 1 (Fig. 11). Intermediate 2 is supported by the structure of human cdN in com-plex with deoxyuridine and AlF4

- (Paper II). Fig. 19 shows structures that mimic intermediates 1 and 2. Although the deoxyuridine-AlF4

- complex is hexacoordinated, it still to some extent mimics the pentavalent substrate intermediate 2. The deoxyuridine-AlF4

- complex has a somewhat similar charge distribution and geometry as the hypothetical pentavalant substrate intermediate.

Fig. 19. Superposition of human cdN-AlF4--deoxyuridine (light) and mdN (D41N

variant)-3’dTMP (dark) structures.

39

Structure of human cN-II (Paper III and IV)

One of the aims in studying cN-II by x-ray crystallography was to better understand the substrate specificity of this enzyme, which could increase the understanding of cN-II’s role in NA resistance.

cN-II is allosterically regulated and probably has two effector sites, of which one was well characterized (Paper IV). Important aspects of its mechanism of regulation were also elucidated in Paper IV, see below.

Characterization of 2 effector sites cN-II was initially crystallized in ~2 M of MgSO2. The sulfates efficiently bound at several phosphate sites in the protein, probably stabilizing its struc-ture. As a consequence, we could not bind nucleotides (e.g. effectors) or other phosphate ligands to cN-II in the crystal, since the phosphate sites were already occupied by sulfate.

However, we could compete out sulfate from the active site by binding BeF3

- covalently to catalytic Asp52 (first aspartate of motif I). We could also bind adenosine to the protein, which bound with high occupancy to a site that we named effector site 1. We also found less defined density at another location that we interpreted as the adenine base of adenosine. We modeled in adenosine in this site that we named effector site 2 (Paper III). We could not confirm this site in later studies (Paper IV). However, a positive patch (Q420RRIKK) in this site that bound two sulfate ions indicates this being phosphate binding sites. This supports the notion that effector site 2 is a binding site for nucleotides (Fig. 20). Possibly, a structural change is needed to complete this effector site.

The structure shows cN-II as being a tetramer of two dimers (Fig. 21), which is consistent with biochemical studies indicating that it functions as a tetramer.

40

Fig. 20. cN-II in complex with two adenosines. AS=Active site. ES1=Effector site 1. ES2=Effector site 2. Adenosines and sulfates are shown as sticks.

Activator recognition in effector site 1 The D52N variant of cN-II was co-crystallized with activator bound to effec-tor site 1 and substrate and Mg2+ bound in the active site. Effector site 1 is located close to the subunit interface between the dimers of the tetramer. In this interface, the phosphate moieties of activators bind (Fig. 21).

In the case of dATP and ATP, a Mg2+ is coordinated between the phos-phate moieties of the two nucleotides (Fig. 22). This Mg2+ neutralizes two negative charges from the activators. Because of the well-defined electron density and symmetric binding of dATP and ATP, these are likely to be “true” effectors for this site (Fig. 22). Mg2+ binds at the 2–fold axis between the dimers of the tetramer.

41

Fig. 21. Tetrameric structure of cN-II, with activator bound in effector site 1 and substrate bound in the active site, shown as spheres.

Fig. 22. dATP bound in effector site 1. dATP* belongs to the adjacent subunit.

In contrast, the activator Ap4A binds with its phosphates in two alterna-tive conformations. The activator 2,3-BPG also binds in two alternative con-formations in the dimer interface. It traverses the 2-fold axis between the subunits.

42

Mechanism for activation of cN-II One of the first very important observations that we made when solving structures of the D52N variant of cN-II, was that neither substrate nor Mg2+ bound to the active site of cN-II without an activator bound to effector site 1. After soaking the crystals in reservoir solution with 10 mM of substrate and 10 mM of Mg2+, the structure obtained was still of the apoprotein. This sug-gests that the mechanism of activation of cN-II functions through a drastic increase in substrate affinity.

Extensive kinetic data is available on how activators affect Km and Vmax of cN-II (43-45, 47, 52), but the different studies are contradictory and it is difficult to derive a consensus from them.

Table 6 shows how various effectors affect the activity of cN-II purified from human placenta.

Table 6. Activity of cN-II, with/without activator and Pi

Activator Reaction velocity (nmole/min) None 0.6 dATP 26.9 ATP 20.5 GTP 12.0 2,3-BPG 17.8 Pi 0.3 ATP + 1 mM Pi 11.1 ATP +4 mM Pi 1.2 100 μM IMP was used as substrate, the activators were at 3 mM (44)

43

By comparing the apoprotein with activator/activator+substrate com-plexes, we could see significant changes in the structure that enabled us to elucidate a mechanism for the allosteric activation of cN-II.

An α−helix seen in the complex structures is disordered in the apoprotein. We have called this helix, helix A. (Fig. 23).

Fig. 23. IMP and Mg2+ bind in the active site and 2,3-BPG binds in effector site 1. The apoprotein lacks helix A.

Helix A contains Asp356 of motif IV, a residue that is very conserved be-

tween HAD superfamily enzymes. Substitution of this aspartate into an aspa-ragines or serine abolished the activity in mdN, PSP, HAD, and SERCA (142-144 and unpublished data on mdN). The role of this residue is still unknown.

44

The hydrogen bond pattern in the Mg2+-phosphate site is destroyed in the apoprotein, compared to the complex structures (Fig. 24). Upon activator recognition, structural changes occur that stabilize the hydrogen bond pattern in the active site: Phe354 turns from the active site and Asp356 enters in-stead. There is also a conformational change in the conserved Lys292 of motif III.

The activator stabilizes helix A by forming a hydrogen bond to Lys362 and by π−stacking with Phe354. The mechanism of activation of cN-II, is probably initiated by the binding of an activator in effector site 1. This in-duces the formation of helix A containing Asp356 that completes the Mg2+-phosphate binding site, whereafter the Mg2+ and substrate can bind.

The mechanism of activation through effector site 1 is, to our knowledge, a novel type of mechanism. It might be that this mechanism is conserved in other HAD superfamily enzymes that are regulated such as the P-type AT-Pases.

Fig. 24. Mechanism of activation. Left: Active site of apoprotein. Right: Activator Ap4A bound in effector site 1 and GMP bound in active site.

45

Covalent modification on Asn52 Recently we solved a structure of the D52N variant of cN-II with a very surprising covalent modification on Asn52 (corresponding to the first Asp of motif I) (Fig. 14). We obtained this modification only when soaking crystals in reservoir solution and activator, without Mg2+. We modeled this modifica-tion as a carboxylation (Fig. 25).

Fig. 25. Covalent modification of Asn52 modeled as a carboxy-Asn. Dotted lines from Lys292 indicate hydrogen bonds, whereas the dotted line from Asp356 indi-cates the short distance (3.6 A) to the β-C of Asn52. a) Fo-Fc map calculated at 4σ. b) Superposition of cN-II-carboxy-Asp structure and cN-II-GMP-Ap4A structure, with helix A shown.

Having observed this modification, we speculated on the chemical back-

ground for its formation. Candidates as reactants are bicine and OH- from the crystallization solution. One possibility might be that the mechanism for the formation of this carboxylation involves a nucleophilic attack by Asn52 on bicine.

Since asparagine is a poor nucleophile, Asn52 probably has to be highly activated, in order to become sufficiently nucleophilic. One speculative chemical route for this to happen is an enolation of Asn52, by the removal of a proton from the β-C. If this is the case, then something within close prox-imity of the β-C of Asn52 has to be the activating agent. An important ob-servation is that there is no room for a water molecule or OH- close to the β-C of Asn52. The closest candidate for activating Asn52, is Asp356 (second Asp of motif IV). Could Asp356 abstract a proton from the β-C of Asn52?

46

Asp52 of the wild type enzyme is clearly a stronger nucleophile than as-paragine. However, it is not clear if Asp52 needs to be activated or not in order to perform the nucleophilic attack on the substrate. By coordinating the Mg2+, it temporarly loses some of its nucleophilicity, which could be a rea-son for the need of activation. It might, however, be possible that the sub-strate phosphate moiety (which also forms an electrostatic interaction with the Mg2+) replaces Asp52 in neutralizing a charge of the Mg2+. This would free the negative charge of the Asp to preform the nucleophilic attack, with-out activation.

If Asp52 needs to be activated to perform the nucleophilic attack, this might occur through an enolation (similar as described above) (Fig. 26). The enolation would generate two single bonded C-O moieties where one coor-dinates the metal and the other attacks the substrate phosphate.

Fig. 26. Hypothetical activation of D52 by an enole reaction.

47

Substrate recognition in cN-II Several structures of the D52N variant of cN-II were solved in complex with various substrates together with various activators (Table 4). The substrates IMP, GMP, dGMP and UMP were bound to the active site of cN-II together with activator bound in effector site 1. The suboptimal substrate UMP bound with somewhat less defined density than the other substrates. AMP could not be bound to the protein in the crystal.

From the recognition of IMP/GMP/dGMP it becomes clear how cN-II discards AMP as substrate: Arg202 and Asp206 interact favorably with the 6-carbonyl group and NH1 group of IMP/GMP/dGMP, respectively (Fig. 1 and 27a). These two residues probably repel the amino group and N1 group of AMP, respectively.

From the recognition of UMP, we can similarly explain how cN-II dis-cards CMP as substrate: Arg202 that favorably binds the 4-carbonyl group of UMP, would repel the amino group of CMP.

Probable explanations for why cN-II generally prefers purine over pyrimidine nucleotides are (1) the fewer interactions that pyrimidine nucleo-tides have with the protein compared to purine nucleotides, and (2) that the 2-carbonyl group (that is present in all pyrimidine nucleotides) binds unfa-vorably close to Phe157 (Fig. 1 and 27b).

Lys215 might provide the preference of cN-II for ribonucleotides over deoxynucleotides, since it creates a hydrophilic environment suitable for a 2’OH rather than a hydrophobic C2’.

Fig. 27. Substrate recognition in cN-II. a) IMP b) UMP.

48

The above observations strongly suggest that cN-II has a very poor affin-ity for cytosine and adenine based NAs. This further suggests that cN-II does not directly contribute to resistance towards araC, dFdC, CdA and other commonly used cytidine and adenine based NAs in treatment of cancer.

Instead guanine based NAs, such as ribavirin (Fig. 7) are likely to be di-rectly affected by cN-II. Ribavirin was shown to be phosphorylated by cN-II at a higher rate than by the dNKs (132).

49

Structure of human cN-III The structure of cN-III was solved with molecular replacement using the 92% sequence identical murine cN-III structure (pdb code 2BDU) as a tem-plate. It was solved as apoprotein (2.7 Å), in complex with beryllium trifluoride (BeF3

-) (2.5 Å) and in complex with Pi (3.0 Å). Both the BeF3-

complex and the Pi complex has Mg2+ bound in the active site. The numbering of the amino acids in the structure corresponds to isoform

2 (Q9H0P0-2), since the protein was expressed using cDNA of this isoform. In order to get crystals, a construct was made that lacks the first 13 amino acids. All residues expressed (Asn14–Leu286) are well defined in the struc-ture.

Mapping of cN-III deficiency substitutions cN-III is thought to catabolize UMP and CMP originating from RNA in erythrocytes. Mutations in the cN-III gene have been found in patients with hemolytic anemia. Several of these give amino acid changes: D87V, L131P, N179S, G230R, C63R, Q143del, G157R and I247T.

It was shown that the L131P substitution decreased the stability of the en-zyme (69). The G230R substitution gave lower affinity for CMP, suggesting that Gly230 is important for substrate binding (69). The D87V, N179S, G157R and I247T substitutions strongly reduced enzyme activity (67, 68, 70), whereas the L131P and C63R substitutions gave only moderate altera-tions in activity (70). The Q143 deletion did not significantly change enzyme activity although it reduced the stability (66). The G157R variant was very unstable (67).

The cN-III structure enabled mapping of the above-mentioned substitu-tions, providing structural evidence for how they can affect protein function and stability. Four of the substitutions were already discussed based on the murine cN-III structure (71), reaching conclusions to some extent similar to ours. Fig. 28 shows Asp87, L131, N179 and G230, and also those residues that we predict can participate in recognizing the base moiety of the sub-strate. Fig. 29 shows the location of residues C63, Q143, G157 and I247.

Implications based on our human cN-III structure: The D87V substitution might lead to a hydrophobic collapse in the cap

domain, that probably affects residues that take part in recognition of the base moiety of the substrate.

The L131P substitution probably changes the secondary structure of adja-cent residues. Since Leu131 is close in space to motif I, it might affect the structure of the Mg2+-phosphate site.

Asn179 forms tight hydrogen bonds to three main chain carbonyl groups (Ala154, Gly155 and Ile197) and one main chain amide (Ile197). These resi-

50

dues are close in space both to catalytic residues and residues that possibly participate in base recognition. The structure around the N179S substitution is probably destabilized, affecting residues of the active site.

The G230R substitution could affect the fold of the protein, since an ar-ginine at this position would clash into the adjacent secondary structure ele-ment (Val199 or Phe200). This could possibly affect the ability of the cap domain to close over the core domain, thus reducing the substrate affinity. Another possibility is that the arginine adapts its conformation so that it fills up the active site and in that way inhibits substrate binding.

The C63R substitution might affect the structure of the cap domain, but is rather far from the active site, which correlates with its low effect on enzyme activity.

The Q143 deletion is also located rather far from the active site, which correlates with its low effect on enzyme activity.

The G157 residue constitutes the tight turn between strand S2 and the fol-lowing helix. Glycine is often positioned at tight turns in proteins since it has no side chain and is therefore not as sterically restricted as other residues. The G157R substitution probably strongly affects the tertiary structure, which could explain the low stability of this variant.

The I247T substitution is located on strand S5 in the hydrophobic core of the core domain close to motif IV, which probably affects the structure of the catalytic phosphate site, affecting catalysis.

51

Fig. 28. Human cN-III-BeF3-Mg structure. Residues that are substituted are shown as spheres, and residues of the cap domain (light) that might be involved in substrate recognition are shown as sticks. BeF3

-, covalently bound to Asp38 (sticks) and Mg2+ (sphere) are also shown.

52

Fig. 29. Human cN-III-BeF3-Mg structure. Residues that are substituted or deleted are shown as spheres. Residues D227 and D231 of motif IV and BeF3

-, covalently bound to Asp38 are shown as sticks. Mg2+ is shown as a sphere.

53

Phosphotransferase activity of cN-II and cN-III The role of the Thr/Val of motif I has yet not been biochemically investi-gated. We speculate that a Thr at this position might have a role in providing 5’NTs with phosphotransferase activity. Of the intracellular 5’NTs, only cN-II and cN-III have a Thr at this position (Table 4). These two are also the only 5’NTs that possess phosphotransferase activity.

This Thr appears to destabilize the conformation of the first Asp of motif I. In structures of both cN-II and cN-III, we observe that this Asp can flip and form a hydrogen bond to the Thr. This would not be possible with the hydrophobic Val. Fig. 30 shows cN-III in complex with Pi where it has this Asp in a flipped conformation, compared to its conformation with BeF3

- bound covalently.

Fig. 30. Superposition of the cN-III-BeF3 (light) and cN-III-PO4 (dark) structures.

54

Future perspectives

Structural insights into the substrate recognition of cytosolic 5’NTs could assist efforts directed towards the understanding of 5’NTs involvement in NA resistance. It could direct the choice of which 5’NT that should be used for studying a specific NA. For instance, our structures of cN-II suggest that it does not take adenine or cytosine based nucleotides as substrates. This indicates that cN-II is not directly involved in resistance to treatment with cytosine and adenine based NAs.

Characterization of the substrate recognition and implications for the tran-sition states of 5’NT’s could greatly assist in efforts directed at generating inhibitors of 5’NTs. The correct combination of NA with an inhibitor of a 5’NT might reduce NA resistance.

We are currently aiming at determining structures of substrate complexes of cN-III, since the structural basis for its specificity is still unknown. We would also like to solve the structures of cN-IA and cN-IB, to complete the structural view of human intracellular 5’NTs. The structural basis for their substrate specificity would provide a more complete understanding of intra-cellular 5’NT specificity. Although these three 5’NTs appear to have tissue-specific expression, they might be relevant for development of drugs that are directed toward those tissues.

The complicated regulation of cN-II is yet not fully understood. It would be of great interest to solve the structure of the full-length construct to ana-lyze the possible effector site 2 and the role of its C-terminal acidic stretch. Structures of cN-IA and cN-IB would also show the structural basis for their allosteric regulation.

55

Acknowledgments

I would like to express my gratitude to everyone that has encouraged me during my PhD studies, which have been a very exciting journey! Especially I would like to thank:

Pär, for accepting me as a PhD student and for your encouragement, good supervision and trust! I also thank you for providing a very inspiring atmos-phere in the lab!

My collaborators in Padova: Vera Bianchi, for your great support in the process of writing papers and for scientific guidance! Benedetta Ruzzenente and Chiara Rampazzo for the purification of mdN and cdN, and for nice discussions and nice company during meetings.

Agnes, for your support, collaboration, friendship and fun company dur-ing many travel adventures!

Stefan Nordlund for being a good boss! I would also like to thank you for sending me to Zurich as an Erasmus student, and you and Peter Brzezinski for recommending me to ”Forskarskolan”, since those opportunities helped me a great deal!

Pål, Susanne, Tomas and Urszula of Team 1 at SGC, for a nice collabora-tion on the wild type cN-II, cN-III and UCK projects.

Karl-Magnus for always being very helpful and supportive! Daniel G for being a great labasse-kompis! Audur, Sue-Li, Lola and Anna-Karin for friendship and for providing some balance against the boys, Monica V for friendship and all the fun during travels ☺, Damian, Daniel MM, Tobias, Ulrika, Amin, Marie, Marina, Pelle, Maria, Hanna, Said, Anna, Henrik, Heidi, Albert, Martin Hö, the newcomers Christine, Christian and Cedric, all other past and present members of the PN-group, the SGC people and the administrative staff at both SU and KI (especially Elisabeth) for being very nice and helpful!

Jag skulle också vilja tacka min kära familj, mina underbara utanför-

labbet-vänner och min gosiga pojkvän David för det stöd och tålamod som jag fått från er! TACK!

56

References

1. Nelson, D., and Michael, M. (2005) Lehninger Principles of Bio-chemistry, 4 ed., Worth Publishers, New York.

2. Reichard, P. (1962) Enzymatic synthesis of deoxyribonucleotides. I. Formation of deoxycytidine diphosphate from cytidine diphosphate with enzymes from Escherichia coli., J Biol Chem 237, 3513-3519.

3. Nordlund, P., and Reichard, P. (2006) Ribonucleotide reductases., Annu Rev Biochem 75, 681-706.

4. Reichard, P. (1988) Interactions between Deoxyribonucleotide and DNA-Synthesis., Ann Rev Biochem 57, 349-374.

5. Arnér, E., and Eriksson, S. (1995) Mammalian deoxyribonucleoside kinases., Pharmacol Ther 67, 155-186.

6. Spychala, J., Datta, N., Takabayashi, K., Datta, M., Fox, I., Gribbin, T., and Mitchell, B. (1996) Cloning of human adenosine kinase cDNA: sequence similarity to microbial ribokinases and fructo-kinases., Proc Nat Aca Sci U S A 93, 1232-1237.

7. Van Rompay, A. R., Norda, A., Linden, K., Johansson, M., and Karlsson, A. (2001) Phosphorylation of uridine and cytidine nucleo-side analogs by two human uridine-cytidine kinases., Mol Pharma-col 59, 1181-1186.

8. Larsson, K., Jordan, A., Eliasson, R., Reichard, P., Logan, D., and Nordlund, P. (2004) Structural mechanism of allosteric substrate specificity regulation in a ribonucleotide reductase., Nat Struc Mol Biol 11, 1142-1149.

9. Vidair, C., and Rubin, H. (2005) Mg2+ as activator of uridine phos-phorylation in coordination with other cellular responses to growth factors., Proc Nat Aca Sci U S A 102, 662-666.

10. Carnrot, C., Wehelie, R., Eriksson, S., Bölske, G., and Wang, L. (2003) Molecular characterization of thymidine kinase from Urea-plasma urealyticum: nucleoside analogues as potent inhibitors of mycoplasma growth., Mol Microbiol 50, 771-780.

11. Mikkelsen, N., Johansson, K., Karlsson, A., Knecht, W., Andersen, G., Piskur, J., Munch-Petersen, B., and Eklund, H. (2003) Structural basis for feedback inhibition of the deoxyribonucleoside salvage pathway: studies of the Drosophila deoxyribonucleoside kinase., Biochemistry 42, 5706-5712.

12. Knecht, W., Petersen, G., Munch-Petersen, B., and Piskur, J. (2002) Deoxyribonucleoside kinases belonging to the thymidine kinase 2 (TK2)-like group vary significantly in substrate specificity, kinetics and feed-back regulation., J Mol Biol 315, 529-540.

57

13. Bianchi, V. (1998) Regulation of deoxynucleotide pools by substrate cycles., Adv Exp Med Biol 431, 501-506.

14. Chabes, A., and Thelander, L. (2000) Controlled protein degradation regulates ribonucleotide reductase activity in proliferating mammal-ian cells during the normal cell cycle and in response to DNA dam-age and replication blocks., J Biol Chem 275, 17747-17753.

15. Kauffman, M., and Kelly, T. (1991) Cell cycle regulation of thymidine kinase: residues near the carboxyl terminus are essential for the specific degradation of the enzyme at mitosis., Mol Cell Biol 11, 2538-2546.

16. Tanaka, H., Arakawa, H., Yamaguchi, T., Shiraishi, K., Fukuda, S., Matsui, K., Takei, Y., and Nakamura, Y. (2000) A ribonucleotide reductase gene involved in a p53-dependent cell-cycle checkpoint for DNA damage., Nature 404, 42-49.

17. Bourdon, A., Minai, L., Serre, V., Jais, J., Sarzi, E., Aubert, S., Chrétien, D., de Lonlay, P., Paquis-Flucklinger, V., Arakawa, H., Nakamura, Y., Munnich, A., and Rötig, A. (2007) Mutation of RRM2B, encoding p53-controlled ribonucleotide reductase (p53R2), causes severe mitochondrial DNA depletion., Nat Genet 39, 776-780.

18. Håkansson, P., Hofer, A., and Thelander, L. (2006) Regulation of mammalian ribonucleotide reduction and dNTP pools after DNA damage and in resting cells., J Biol Chem 281, 7834-7841.

19. Pontarin, G., Ferraro, P., Håkansson, P., Thelander, L., Reichard, P., and Bianchi, V. (2007) p53R2-dependent ribonucleotide reduction provides deoxyribonucleotides in quiescent human fibroblasts in the absence of induced DNA damage., J Biol Chem 282, 16820-16828.

20. Rampazzo, C., Fabris, S., Franzolin, E., Crovatto, K., Frangini, M., and Bianchi, V. (2007) Mitochondrial thymidine kinase and the en-zymatic network regulating thymidine triphosphate pools in cultured human cells., J Biol Chem 282, 34758-34769.

21. Reis, M.J., R. (1934) La nucleotidase et sa relation avec la desami-nation des nucleotides dans le coeur et dans le muscle., Seance.

22. Zimmermann, H. (1992) 5'-Nucleotidase: molecular structure and functional aspects., Biochem J 285 (2), 345-365.

23. Bianchi, V., and Spychala, J. (2003) Mammalian 5 '-nucleotidases., J Biol Chem 278, 46195-46198.

24. Hunsucker, S. A., Mitchell, B. S., and Spychala, J. (2005) The 5'-nucleotidases as regulators of nucleotide and drug metabolism., Pharmacol Therapeut 107, 1-30.

25. Hoglund, L., and Reichard, P. (1990) Cytoplasmic 5'(3')-Nucleotidase from Human Placenta., J Biol Chem 265, 6589-6595.

26. Amici, A., Emanuelli, M., Magni, G., Raffaelli, N., and Ruggieri, S. (1997) Pyrimidine nucleotidases from human erythrocyte possess phosphotransferase activities specific for pyrimidine nucleotides., FEBS 419, 263-267.

58

27. Rampazzo, C., Johansson, M., Gallinaro, L., Ferraro, P., Hellman, U., Karlsson, A., Reichard, P., and Bianchi, V. (2000) Mammalian 5'(3')-deoxyribonucleotidase, cDNA cloning, and overexpression of the enzyme in Escherichia coli and mammalian cells., J Biol Chem 275, 5409-5415.

28. Rampazzo, C., Gallinaro, L., Milanesi, E., Frigimelica, E., Reichard, P., and Bianchi, V. (2000) A deoxyribonucleotidase in mitochon-dria: involvement in regulation of dNTP pools and possible link to genetic disease., Proc Nat Aca Sci, U S A 97, 8239-8244.

29. Rinaldo-Matthis, A., Rampazzo, C., Reichard, P., Bianchi, V., and Nordlund, P. (2002) Crystal structure of a human mitochondrial de-oxyribonucleotidase., Nat Struc Biol 9, 779-787.

30. Mazzon, C., Rampazzo, C., Scaini, M. C., Gallinaro, L., Karlsson, A., Meier, C., Balzarini, J., Reichard, P., and Bianchi, V. (2003) Cy-tosolic and mitochondrial deoxyribonucleotidases: activity with sub-strate analogs, inhibitors and implications for therapy., Biochem Pharmacol 66, 471-479.

31. Rinaldo-Matthis, A., Rampazzo, C., Balzarini, J., Reichard, P., Bi-anchi, V., and Nordlund, P. (2004) Crystal structures of the mito-chondrial deoxyribonucleotidase in complex with two specific in-hibitors., Mol Pharmacol 65, 860-867.

32. Gibson, W., and Drummond, G. (1972) Properties of 5'-nucleotidase from avian heart., Biochemistry 11, 223-229.

33. Sala-Newby, G. B., Skladanowski, A. C., and Newby, A. C. (1999) The mechanism of adenosine formation in cells - Cloning of cytoso-lic 5 '-nucleotidase-I., J Biol Chem 274, 17789-17793.

34. Tkacz-Stachowska, K., Lechward, K., and Skladanowski, A. (2005) Isolation and characterization of pigeon breast muscle cytosolic 5'-nucleotidase-I (cN-I)., Acta Biochim Pol 52, 789-796.

35. Hunsucker, S. A., Spychala, J., and Mitchell, B. S. (2001) Human cytosolic 5 '-nucleotidase I - Characterization and role in nucleoside analog resistance., J Biol Chem 276, 10498-10504.

36. Garvey, E. P., Lowen, G. T., and Almond, M. R. (1998) Nucleotide and nucleoside analogues as inhibitors of cytosolic 5'- nucleotidase I from heart., Biochemistry 37, 9043-9051.

37. Skladanowski, A., and Newby, A. (1990) Partial purification and properties of an AMP-specific soluble 5'-nucleotidase from pigeon heart., Biochem J 268, 117-122.

38. Garvey, E. P., and Prus, K.L. (1999) A specific inhibitor of heart cytosolic 5'-nucleotidase I attenuates hydrolysis of adenosine 5'-monophosphate in primary rat myocytes., Arch Biochem Biophys 364, 235-240.

39. Sala-Newby, G. B., Freeman, N. V., Skladanowski, A. C., and Newby, A. C. (2000) Distinct roles for recombinant cytosolic 5'-nucleotidase-I and -II in AMP and IMP catabolism in COS-7 and H9c2 rat myoblast cell lines., J Biol Chem 275, 11666-11671.

59

40. Sala-Newby, G. B., and Newby, A. C. (2001) Cloning of a mouse cytosolic 5'-nucleotidase-I identifies a new gene related to human autoimmune infertility-related protein., Biochim Biophys Acta 1521, 12-18.

41. Itoh, R., Mitsui, A., and Tsushima, K. (1967) 5'-Nucleotidase of Chicken Liver., Biochim Biophys Acta 146, 151-159.

42. Itoh, R. (1993) IMP GMP 5'-Nucleotidase., Comp Biochem Physiol Biochem Mol Biol 105, 13-19.

43. Marques, A. F. P., Teixeira, N. A., Gambaretto, C., Sillero, A., and Sillero, M. A. G. (1998) IMP-GMP 5 '-nucleotidase from rat brain: Activation by polyphosphates., J Neurochem 71, 1241-1250.

44. Spychala, J., Madridmarina, V., and Fox, I. H. (1988) High Km Soluble 5'-Nucleotidase from Human-Placenta - Properties and Al-losteric Regulation by IMP and ATP., J Biol Chem 263, 18759-18765.

45. Spychala, J., Chen, V., Oka, J., and Mitchell, B. S. (1999) ATP and phosphate reciprocally affect subunit association of human recombi-nant High Km 5 '-nucleotidase - Role for the C-terminal polyglu-tamic acid tract in subunit association and catalytic activity., Eur J Biochem 259, 851-858.

46. Allegrini, S., Pesi, R., Tozzi, M. G., Fiol, C. J., Johnson, R. B., and Eriksson, S. (1997) Bovine cytosolic IMP/GMP-specific 5'-nucleotidase: cloning and expression of active enzyme in Es-cherichia coli., Biochem J 328, 483-487.

47. Pesi, R., Turriani, M., Allegrini, S., Scolozzi, C., Camici, M., Ipata, P. L., and Tozzi, M. G. (1994) The Bifunctional Cytosolic 5'-Nucleotidase - Regulation of the Phosphotransferase and Nucleoti-dase Activities., Arch Biochem Biophys 312, 75-80.

48. Randitelli, S., Baiocchi, C., Pesi, R., Allegrini, S., Turriani, M., Ipata, P. L., Camici, M., and Tozzi, M. G. (1996) The phosphotrans-ferase activity of cytosolic 5'-nucleotidase; A purine analog phos-phorylating enzyme., Int J Biochem Cell Biol 28, 711-720.

49. Tozzi, M. G., Camici, M., Pesi, R., Allegrini, S., Sgarrella, F., and Ipata, P. L. (1991) Nucleoside Phosphotransferase Activity of Hu-man Colon-Carcinoma Cytosolic 5'-Nucleotidase., Arch Biochem Biophys 291, 212-217.

50. Johnson, M. A., and Fridland, A. (1989) Phosphorylation of 2',3'-Dideoxyinosine by Cytosolic 5'-Nucleotidase of Human Lymphoid-Cells., Mol Pharmacol 36, 291-295.

51. Worku, Y., and Newby, A. C. (1982) Nucleoside Exchange Cata-lyzed by the Cytoplasmic 5'-Nucleotidase., Biochem J 205, 503-510.

52. Pinto, R. M., Canales, J., Sillero, M. A. G., and Sillero, A. (1986) Diadenosine Tetraphosphate Activates Cytosol 5'-Nucleotidase., Biochem Biophys Res Commun 138, 261-267.

53. Wang, Q., Hu, W., Gao, W., and Bi, R. (2006) Crystal structure of the diadenosine tetraphosphate hydrolase from Shigella flexneri 2a., Proteins 65, 1032-1035.

60

54. Lee, Y., Nechushtan, H., Figov, N., and Razin, E. (2004) The func-tion of lysyl-tRNA synthetase and Ap4A as signaling regulators of MITF activity in FcepsilonRI-activated mast cells., Immunity 20, 145-151.

55. Kisselev, L. L., Justesen, J., Wolfson, A. D., and Frolova, L. Y. (1998) Diadenosine oligophosphates (Ap(n)A), a novel class of sig-nalling molecules?, FEBS 427, 157-163.

56. Pesi, R., Baiocchi, C., Tozzi, M. G., and Camici, M. (1996) Syner-gistic action of ADP and 2,3-bisphosphoglycerate on the modulation of cytosolic 5'-nucleotidase., Biochim Biophys Acta-Protein Struc Mol Enzymol 1294, 191-194.

57. Pesi, R., Baiocchi, C., Allegrini, S., Moretti, E., Sgarrella, F., Camici, M., and Tozzi, M. G. (1998) Identification, separation and characterisation of two forms of cytosolic 5 '-nucleotidase/nucleoside phosphotransferase in calf thymus., Biol Chem 379, 699-704.

58. Careddu, M., Allegrini, S., Pesi, R., Camici, M., Garcia-Gil, M., and Tozzi, M. (2008) Knockdown of cytosolic 5'-nucleotidase II (cN-II) reveals that its activity is essential for survival in astrocytoma cells., Biochim Biophys Acta 1783, 1529-1535.

59. Swallow, D. M., Turner, V. S., and Hopkinson, D. A. (1983) Isozymes of Rodent 5'-Nucleotidase - Evidence for 2 Independent Structural Loci Umph-1 and Umph-2., Ann Hum Genet 47, 9-17.

60. Amici, A., Emanuelli, M., Raffaelli, N., Ruggieri, S., Saccucci, F., and Magni, G. (2000) Human erythrocyte pyrimidine 5 '-nucleotidase, PN-I, is identical to p36, a protein associated to lupus inclusion formation in response to alpha-interferon., Blood 96, 1596-1598.

61. Marinaki, A. M., Escuredo, E., Duley, J. A., Simmonds, H. A., Amici, A., Naponelli, V., Magni, G., Seip, M., Ben-Bassat, I., Harley, E. H., Thein, S. L., and Rees, D. C. (2001) Genetic basis of hemolytic anemia caused by pyrimidine 5 ' nucleotidase deficiency., Blood 97, 3327-3332.

62. Amici, A., and Magni, G. (2002) Human erythrocyte pyrimidine 5 '-nucleotidase, PN-1., Arch Biochem Biophys 397, 184-190.

63. Dragon, S., Hille, R., Gotz, R., and Baumann, R. (1998) Adenosine 3 ': 5 '-cyclic monophosphate (cAMP)-inducible pyrimidine 5 '-nucleotidase and pyrimidine nucleotide metabolism of chick embry-onic erythrocytes., Blood 91, 3052-3058.

64. Rees, D. C., Duley, J. A., and Marinaki, A. M. (2003) Pyrimidine 5 ' nucleotidase deficiency., Br J Haematol 120, 375-383.

65. Valentin, W. N., Fink, K., Paglia, D. E., Harris, S. R., and Adams, W. S. (1974) Hereditary Hemolytic-Anemia with Human Erythro-cyte Pyrimidine 5'-Nucleotidase Deficiency., J Clin Invest 54, 866-879.

66. Chiarelli, L., Fermo, E., Abrusci, P., Bianchi, P., Dellacasa, C., Gal-izzi, A., Zanella, A., and Valentini, G. (2006) Two new mutations of

61

the P5'N-1 gene found in Italian patients with hereditary hemolytic anemia: the molecular basis of the red cell enzyme disorder., Haematologica 91, 1244-1247.

67. Chiarelli, L., Morera, S., Galizzi, A., Fermo, E., Zanella, A., and Valentini, G. (2008) Molecular basis of pyrimidine 5'-nucleotidase deficiency caused by 3 newly identified missense mutations (c.187T>C, c.469G>C and c.740T>C) and a tabulation of known mutations., Blood Cells Mol Dis 40, 295-301.

68. Bianchi, P., Fermo, E., Alfinito, F., Vercellati, C., Baserga, M., Ferraro, F., Guzzo, I., Rotoli, B., and Zanella, A. (2003) Molecular characterization of six unrelated Italian patients affected by pyrimidine 5 '-nucleotidase deficiency., Br J Haematol 122, 847-851.

69. Kanno, H., Takizawa, T., Miwa, S., and Fujii, H. (2004) Molecular basis of Japanese variants of pyrimidine 5 '-nucleotidase deficiency., Br J Haematol 126, 265-271.

70. Chiarelli, L., Bianchi, P., Fermo, E., Galizzi, A., Iadarola, P., Mattevi, A., Zanella, A., and Valentini, G. (2005) Functional analy-sis of pyrimidine 5'-nucleotidase mutants causing nonspherocytic hemolytic anemia., Blood 105, 3340-3345.

71. Bitto, E., Bingman, C. A., Wesenberg, G. E., McCoy, J. G., and Phillips, G. N. (2006) Structure of pyrimidine 5 '-nucleotidase type 1 - Insight into mechanism of action and inhibition during lead poison-ing., J Biol Chem 281, 20521-20529.

72. Knofel, T., and Strater, N. (1999) X-ray structure of the Escherichia coli periplasmic 5'-nucleotidase containing a dimetal catalytic site., Nat Struc Biol 6, 448-453.

73. Sträter, N. (2006) Ecto-5'-nucleotidase: Structure function relation-ships., Purinergic Signal 2, 343-350.

74. Napieralski, R., Kempkes, B., and Gutensohn, W. (2003) Evidence for coordinated induction and repression of ecto-5'-nucleotidase (CD73) and the A2a adenosine receptor in a human B cell line., Biol Chem 384, 483-487.

75. Synnestvedt, K., Furuta, G., Comerford, K., Louis, N., Karhausen, J., Eltzschig, H., Hansen, K., Thompson, L., and Colgan, S. (2002) Ecto-5'-nucleotidase (CD73) regulation by hypoxia-inducible factor-1 mediates permeability changes in intestinal epithelia., J Clin Invest 110, 993-1002.

76. Castrop, H., Huang, Y., Hashimoto, S., Mizel, D., Hansen, P., Theilig, F., Bachmann, S., Deng, C., Briggs, J., and Schnermann, J. (2004) Impairment of tubuloglomerular feedback regulation of GFR in ecto-5'-nucleotidase/CD73-deficient mice., J Clin Invest 114, 634-642.

77. Eltzschig, H., Thompson, L., Karhausen, J., Cotta, R., Ibla, J., Robson, S., and Colgan, S. (2004) Endogenous adenosine produced during hypoxia attenuates neutrophil accumulation: coordination by extracellular nucleotide metabolism., Blood 104, 3986-3992.

62

78. Koszalka, P., Ozüyaman, B., Huo, Y., Zernecke, A., Flögel, U., Braun, N., Buchheiser, A., Decking, U., Smith, M., Sévigny, J., Gear, A., Weber, A., Molojavyi, A., Ding, Z., Weber, C., Ley, K., Zimmermann, H., Gödecke, A., and Schrader, J. (2004) Targeted disruption of cd73/ecto-5'-nucleotidase alters thromboregulation and augments vascular inflammatory response., Circ Res 95, 814-821.

79. Resta, R., Hooker, S., Hansen, K., Laurent, A., Park, J., Blackburn, M., Knudsen, T., and Thompson, L. (1993) Murine ecto-5'-nucleotidase (CD73): cDNA cloning and tissue distribution., Gene 133, 171-177.

80. Kitakaze, M., Minamino, T., Node, K., Komamura, K., Inoue, M., Hori, M., and Kamada, T. (1996) Activation of ecto-5'-nucleotidase by protein kinase C attenuates irreversible cellular injury due to hy-poxia and reoxygenation in rat cardiomyocytes., J Mol Cell Cardiol 28, 1945-1955.

81. Braun, N., Lenz, C., Gillardon, F., Zimmermann, M., and Zimmermann, H. (1997) Focal cerebral ischemia enhances glial ex-pression of ecto-5'-nucleotidase., Brain Res 766, 213-226.

82. Ledoux, S., Runembert, I., Koumanov, K., Michel, J., Trugnan, G., and Friedlander, G. (2003) Hypoxia enhances Ecto-5'-Nucleotidase activity and cell surface expression in endothelial cells: role of membrane lipids., Circ Res 92, 848-855.

83. Gallant, J., and Suskind, S. (1961) Relationship between thymineless death and ultraviolet inactivation in Escherichia coli., J Bacteriol 82, 187-194.

84. Cohen, A., Hirschhorn, R., Horowitz, S., Rubinstein, A., Polmar, S., Hong, R., and Martin, D. J. (1978) Deoxyadenosine triphosphate as a potentially toxic metabolite in adenosine deaminase deficiency., Proc Nat Aca Sci U S A 75, 472-476.

85. Cohen, A., Gudas, L., Ammann, A., Staal, G., and Martin, D. J. (1978) Deoxyguanosine triphosphate as a possible toxic metabolite in the immunodeficiency associated with purine nucleoside phos-phorylase deficiency., J Clin Invest 61, 1405-1409.

86. Saleh, A., Srinivasula, S., Acharya, S., Fishel, R., and Alnemri, E. (1999) Cytochrome c and dATP-mediated oligomerization of Apaf-1 is a prerequisite for procaspase-9 activation., J Biol Chem 274, 17941-17945.

87. Van De Wiele, C., Vaughn, J., Blackburn, M., Ledent, C., Jacobson, M., Jiang, H., and Thompson, L. (2002) Adenosine kinase inhibition promotes survival of fetal adenosine deaminase-deficient thymo-cytes by blocking dATP accumulation., J Clin Invest 110, 395-402.

88. Torres, R., and Puig, J. (2007) Hypoxanthine-guanine phosophori-bosyltransferase (HPRT) deficiency: Lesch-Nyhan syndrome., Or-phanet J Rare Dis 2, 48.

89. Nishino, I., Spinazzola, A., and Hirano, M. (1999) Thymidine phos-phorylase gene mutations in MNGIE, a human mitochondrial disor-der., Science 283, 689-692.

63

90. Spinazzola, A., and Zeviani, M. (2005) Disorders of nuclear-mitochondrial intergenomic signaling., Gene 354, 162-168.

91. Spinazzola, A., Marti, R., Nishino, I., Andreu, A., Naini, A., Tadesse, S., Pela, I., Zammarchi, E., Donati, M., Oliver, J., and Hi-rano, M. (2002) Altered thymidine metabolism due to defects of thymidine phosphorylase., J Biol Chem 277, 4128-4133.

92. Saada, A., Shaag, A., Mandel, H., Nevo, Y., Eriksson, S., and Elpe-leg, O. (2001) Mutant mitochondrial thymidine kinase in mitochon-drial DNA depletion myopathy., Nat Genet 29, 342-344.

93. Mandel, H., Szargel, R., Labay, V., Elpeleg, O., Saada, A., Shalata, A., Anbinder, Y., Berkowitz, D., Hartman, C., Barak, M., Eriksson, S., and Cohen, N. (2001) The deoxyguanosine kinase gene is mu-tated in individuals with depleted hepatocerebral mitochondrial DNA., Nat Genet 29, 337-341.

94. Bornstein, B., Area, E., Flanigan, K., Ganesh, J., Jayakar, P., Swoboda, K., Coku, J., Naini, A., Shanske, S., Tanji, K., Hirano, M., and Dimauro, S. (2008) Mitochondrial DNA depletion syndrome due to mutations in the RRM2B gene., Neuromuscul Disord 18, 453-459.

95. Ferraro, P., Pontarin, G., Crocco, L., Fabris, S., Reichard, P., and Bianchi, V. (2005) Mitochondrial deoxynucleotide pools in quies-cent fibroblasts: a possible model for mitochondrial neurogastro-intestinal encephalomyopathy (MNGIE)., J Biol Chem 280, 24472-24480.

96. Leanza, L., Ferraro, P., Reichard, P., and Bianchi, V. (2008) Meta-bolic Interrelations within Guanine Deoxynucleotide Pools for Mito-chondrial and Nuclear DNA Maintenance., J Biol Chem 283, 16437-16445.

97. Rampazzo, C., Ferraro, P., Pontarin, G., Fabris, S., Reichard, P., and Bianchi, V. (2004) Mitochondrial deoxyribonucleotides, pool sizes, synthesis, and regulation., J Biol Chem 279, 17019-17026.

98. Kornberg, A., and Baker, T.A. (1992) DNA Replication, 2 ed., Freeman, W.H. and Company, New York.

99. Bianchi, V., Ferraro, P., Borella, S., Bonvini, P., and Reichard, P. (1994) Effects of mutational loss of nucleoside kinases on de-oxyadenosine 5'-phosphate/deoxyadenosine substrate cycle in cul-tured CEM and V79 cells., J Biol Chem 269, 16677-16683.

100. Gazziola, C., Ferraro P., Moras, M., Reichard, P., and Bianchi, V. (2001) Cytosolic high Km 5'-nucleotidase and 5'(3')-deoxyribonucleotidase in substrate cycles involved in nucleotide me-tabolism., J Biol Chem 276, 6185-6190.

101. Gallinaro, L., Crovatto, K., Rampazzo, C., Pontarin, G., Ferraro, P., Milanesi, E., Reichard, P., and Bianchi, V. (2002) Human mito-chondrial 5 '-deoxyribonucleotidase - Overproduction in cultured cells and functional aspects., J Biol Chem 277, 35080-35087.

102. Pontarin, G., Gallinaro, L., Ferraro, P., Reichard, P., and Bianchi, V. (2003) Origins of mitochondrial thymidine triphosphate: Dynamic

64

relations to cytosolic pools., Proc Nat Aca Sci U S A 100, 12159-12164.

103. Höglund, L., Pontis, E., and Reichard, P. (1988) Effects of deoxy-cytidine and thymidine kinase deficiency on substrate cycles be-tween deoxyribonucleosides and their 5'-phosphates., Cancer Res 48, 3681-3687.

104. Paglia, D. E., Valentine, W. N., and Brockway, R. A. (1984) Identi-fication of Thymidine Nucleotidase and Deoxyribonucleotidase Ac-tivities among Normal Isozymes of 5'-Nucleotidase in Human-Erythrocytes., Proc Nat Aca Sci U S A 81, 588-592.

105. Zhang, J., Visser, F., King, K., Baldwin, S., Young, J., and Cass, C. (2007) The role of nucleoside transporters in cancer chemotherapy with nucleoside drugs., Cancer Metastasis Rev 26, 85-110.

106. Griffiths, M., Beaumont, N., Yao, S., Sundaram, M., Boumah, C., Davies, A., Kwong, F., Coe, I., Cass, C., Young, J., and Baldwin, S. (1997) Cloning of a human nucleoside transporter implicated in the cellular uptake of adenosine and chemotherapeutic drugs., Nat Med 3, 89-93.

107. Griffiths, M., Yao, S., Abidi, F., Phillips, S., Cass, C., Young, J., and Baldwin, S. (1997) Molecular cloning and characterization of a ni-trobenzylthioinosine-insensitive (ei) equilibrative nucleoside trans-porter from human placenta., Biochem J 328, 739-743.

108. Baldwin, S., Yao, S., Hyde, R., Ng, A., Foppolo, S., Barnes, K., Ritzel, M., Cass, C., and Young, J. (2005) Functional characteriza-tion of novel human and mouse equilibrative nucleoside transporters (hENT3 and mENT3) located in intracellular membranes., J Biol Chem 280, 15880-15887.

109. Barnes, K., Dobrzynski, H., Foppolo, S., Beal, P., Ismat, F., Scul-lion, E., Sun, L., Tellez, J., Ritzel, M., Claycomb, W., Cass, C., Young, J., Billeter-Clark, R., Boyett, M., and Baldwin, S. (2006) Distribution and functional characterization of equilibrative nucleo-side transporter-4, a novel cardiac adenosine transporter activated at acidic pH., Circ Res 99, 510-519.

110. Ritzel, M., Yao, S., Huang, M., Elliott, J., Cass, C., and Young, J. (1997) Molecular cloning and functional expression of cDNAs en-coding a human Na+-nucleoside cotransporter (hCNT1)., Am J Physiol 272, 707-714.

111. Ritzel, M., Yao, S., Ng, A., Mackey, J., Cass, C., and Young, J. (1998) Molecular cloning, functional expression and chromosomal localization of a cDNA encoding a human Na+/nucleoside cotrans-porter (hCNT2) selective for purine nucleosides and uridine., Mol Membr Biol 15, 203-211.

112. Ritzel, M., Ng, A., Yao, S., Graham, K., Loewen, S., Smith, K., Ritzel, R., Mowles, D., Carpenter, P., Chen, X., Karpinski, E., Hyde, R., Baldwin, S., Cass, C., and Young, J. (2001) Molecular identifica-tion and characterization of novel human and mouse concentrative Na+-nucleoside cotransporter proteins (hCNT3 and mCNT3)

65

broadly selective for purine and pyrimidine nucleosides (system cib)., J Biol Chem 276, 2914-2927.

113. Leeds, J., Slabaugh, M., and Mathews, C. (1985) DNA precursor pools and ribonucleotide reductase activity: distribution between the nucleus and cytoplasm of mammalian cells., Mol Cell Biol 5, 3443-3450.

114. Dolce, V., Fiermonte, G., Runswick, M.J., Palmieri, F., Walker, J.E. (2001) The human mitochondrial deoxynucleotide carrier and its role in the toxicity of nucleoside antivirals., Proc Nat Aca Sci U S A 98, 2284-2288.

115. Lam, W., Chen, C., Ruan, S., Leung, C., and Cheng, Y. (2005) Ex-pression of deoxynucleotide carrier is not associated with the mito-chondrial DNA depletion caused by anti-HIV dideoxynucleoside analogs and mitochondrial dNTP uptake., Mol Pharmacol 67, 408-416.

116. Lindhurst, M., Fiermonte, G., Song, S., Struys, E., De Leonardis, F., Schwartzberg, P., Chen, A., Castegna, A., Verhoeven, N., Mathews, C., Palmieri, F., and Biesecker, L. (2006) Knockout of Slc25a19 causes mitochondrial thiamine pyrophosphate depletion, embryonic lethality, CNS malformations, and anemia., Proc Nat Aca Sci U S A 103, 15927-15932.

117. Kang, J., and Samuels, D. (2008) The evidence that the DNC (SLC25A19) is not the mitochondrial deoxyribonucleotide carrier., Mitochondrion 8, 103-108.

118. Ferraro, P., Nicolosi, L., Bernardi, P., Reichard, P., and Bianchi, V. (2006) Mitochondrial deoxynucleotide pool sizes in mouse liver and evidence for a transport mechanism for thymidine monophosphate., Proc Nat Aca Sci U S A 103, 18586-18591.

119. Bridges, E., Jiang, Z., and Cheng, Y. (1999) Characterization of a dCTP transport activity reconstituted from human mitochondria., J Biol Chem 274, 4620-4625.

120. Floyd, S., Favre, C., Lasorsa, F., Leahy, M., Trigiante, G., Stroebel, P., Marx, A., Loughran, G., O'Callaghan, K., Marobbio, C., Slot-boom, D., Kunji, E., Palmieri, F., and O'Connor, R. (2007) The insu-lin-like growth factor-I-mTOR signaling pathway induces the mito-chondrial pyrimidine nucleotide carrier to promote cell growth., Mol Biol Cell 18, 3545-3555.

121. Munchpetersen, B., Cloos, L., Tyrsted, G., and Eriksson, S. (1991) Diverging Substrate-Specificity of Pure Human Thymidine Kinase-1 and Kinase-2 against Antiviral Dideoxynucleosides., J Biol Chem 266, 9032-9038.

122. Wang, L., Munch-Petersen, B., Sjoberg, A., Hellman, U., Bergman, T., Jornvall, H., and Eriksson, S. (1999) Human thymidine kinase 2: molecular cloning and characterisation of the enzyme activity with antiviral and cytostatic nucleoside substrates., FEBS 443, 170-174.

66

123. Jordheim, L., and Dumontet, C. (2007) Review of recent studies on resistance to cytotoxic deoxynucleoside analogues., Biochim Bio-phys Acta 1776, 138-159.

124. Dumontet, C., Fabianowska-Majewska, K., Mantincic, D., Callet Bauchu, E., Tigaud, I., Gandhi, V., Lepoivre, M., Peters, G., Rol-land, M., Wyczechowska, D., Fang, X., Gazzo, S., Voorn, D., Vanier-Viornery, A., and MacKey, J. (1999) Common resistance mechanisms to deoxynucleoside analogues in variants of the human erythroleukaemic line K562., Br J Haematol 106, 78-85.

125. Galmarini, C., Graham, K., Thomas, X., Calvo, F., Rousselot, P., El Jafaari, A., Cros, E., Mackey, J., and Dumontet, C. (2001) Expres-sion of high Km 5'-nucleotidase in leukemic blasts is an independent prognostic factor in adults with acute myeloid leukemia., Blood 98, 1922-1926.

126. Galmarini, C., Thomas, X., Calvo, F., Rousselot, P., El Jafaari, A., Cros, E., and Dumontet, C. (2002) Potential mechanisms of resis-tance to cytarabine in AML patients., Leuk Res 26, 621-629.

127. Giovannetti, E., Mey, V., Loni, L., Nannizzi, S., Barsanti, G., Savarino, G., Ricciardi, S., Del Tacca, M., and Danesi, R. (2007) Cytotoxic activity of gemcitabine and correlation with expression profile of drug-related genes in human lymphoid cells., Pharmacol Res 55, 343-349.

128. Sève, P., Mackey, J., Isaac, S., Trédan, O., Souquet, P., Pérol, M., Cass, C., and Dumontet, C. (2005) cN-II expression predicts sur-vival in patients receiving gemcitabine for advanced non-small cell lung cancer., Lung Cancer 49, 363-370.

129. Gazziola, C., Moras, M., Ferraro, P., Gallinaro, L., Verin, R., Ram-pazzo, C., Reichard, P., and Bianchi, V. (1999) Induction of human high K-m 5 '-nucleotidase in cultured 293 cells., Exp Cell Res 253, 474-482.

130. Jordheim, L., Cros, E., Galmarini, C., Dumontet, C., Bretonnet, A., Krimm, I., Lancelin, J., and Gagnieu, M. (2006) F-ara-AMP is a substrate of cytoplasmic 5'-nucleotidase II (cN-II): HPLC and NMR studies of enzymatic dephosphorylation., Nucleosides Nucleotides Nucleic Acids 25, 289-297.

131. Gray, T., Morrey, E., Gangadharan, B., Sumter, T., Spychala, J., Archer, D., and Spencer, H. (2006) Enforced expression of cytosolic 5'-nucleotidase I confers resistance to nucleoside analogues in vitro but systemic chemotherapy toxicity precludes in vivo selection., Cancer Chemother Pharmacol 58, 117-128.

132. Wu, J., Larson, G., Walker, H., Shim, J., and Hong, Z. (2005) Phos-phorylation of ribavirin and viramidine by adenosine kinase and cy-tosolic 5'-nucleotidase II: Implications for ribavirin metabolism in erythrocytes., Antimicrob Agents Chemother 49, 2164-2171.

133. Lewis, W., and Dalakas, M. C. (1995) Mitochondrial toxicity of antiviral drugs, Nat Med 1, 417-422.

67

134. Koonin, E., and Tatusov, R. (1994) Computer analysis of bacterial haloacid dehalogenases defines a large superfamily of hydrolases with diverse specificity. Application of an iterative approach to da-tabase search., J Mol Biol 244, 125-132.

135. Burroughs, A., Allen, K., Dunaway-Mariano, D., and Aravind, L. (2006) Evolutionary genomics of the HAD superfamily: understand-ing the structural adaptations and catalytic diversity in a superfamily of phosphoesterases and allied enzymes., J Mol Biol 361, 1003-1034.

136. Allegrini, S., Scaloni, A., Ferrara, L., Pesi, R., Pinna, P., Sgarrella, F., Camici, M., Eriksson, S., and Tozzi, M. G. (2001) Bovine cyto-solic 5'-nucleotidase acts through the formation of an aspartate 52-phosphoenzyme intermediate, J Biol Chem 276, 33526-33532.

137. Wang, W., Kim, R., Jancarik, J., Yokota, H., and Kim, S. H. (2001) Crystal structure of phosphoserine phosphatase from Methanococcus jannaschii, a hyperthermophile, at 1.8 A resolution., Structure 9, 65-71.

138. Himo, F., Guo, J. D., Rinaldo-Matthis, A., and Nordlund, P. (2005) Reaction mechanism of deoxyribonucleotidase: A theoretical study., J Phys Chem 109, 20004-20008.

139. Wang, W. R., Cho, H. S., Kim, R., Jancarik, J., Yokota, H., Nguyen, H. H., Grigoriev, I. V., Wemmer, D. E., and Kim, S. H. (2002) Structural characterization of the reaction pathway in phosphoserine phosphatase: Crystallographic "snapshots" of intermediate states., J Mol Biol 319, 421-431.

140. Lahiri, S. D., Zhang, G. F., Dunaway-Mariano, D., and Allen, K. N. (2003) The pentacovalent phosphorus intermediate of a phosphoryl transfer reaction., Science 299, 2067-2071.

141. Collet, J. F., Stroobant, V., Pirard, M., Delpierre, G., and Van Shaft-ingen, E. (1998) A new class of phosphotransferases phosphorylated on an aspartate residue in an amino-terminal DXDX(T/V) motif., J Biol Chem 273, 14107-14112.

142. Collet, J. F., Stroobant, V., and Van Schaftingen, E. (1999) Mecha-nistic studies of phosphoserine phosphatase, an enzyme related to P-type ATPases., J Biol Chem 274, 33985-33990.

143. Kurihara, T., Liu, J., Nardi-Dei, V., Koshikawa, H., Esaki, N., and Soda, K. (1995) Comprehensive site-directed mutagenesis of L-2-halo acid dehalogenase to probe catalytic amino acid residues., J Biochem 117, 1317-1322.

144. Clarke, D., Loo, T., and MacLennan, D. (1990) Functional conse-quences of alterations to amino acids located in the nucleotide bind-ing domain of the Ca2+-ATPase of sarcoplasmic reticulum., J Biol Chem 265, 22223-22227.

145. Knöfel, T., and Sträter, N. (2001) Mechanism of hydrolysis of phos-phate esters by the dimetal center of 5'-nucleotidase based on crystal structures., J Mol Biol 309, 239-254.

68

146. Knöfel, T., and Sträter, N. (2001) E. coli 5'-nucleotidase undergoes a hinge-bending domain rotation resembling a ball-and-socket mo-tion., J Mol Biol 309, 255-266.

147. Schultz-Heienbrok, R., Maier, T., and Sträter, N. (2005) A large hinge bending domain rotation is necessary for the catalytic function of Escherichia coli 5'-nucleotidase., Biochemistry 44, 2244-2252.

148. Lad, C., Williams, N., and Wolfenden, R. (2003) The rate of hy-drolysis of phosphomonoester dianions and the exceptional catalytic proficiencies of protein and inositol phosphatases., Proc Nat Aca Sci U S A 100, 5607-5610.

149. Chabre, M. (1990) Aluminofluoride and beryllofluoride complexes: a new phosphate analogs in enzymology., Trends Biochem Sci 15, 6-10.

150. Madhusudan, Akamine, P., Xuong, N.H., Taylor, S.S. (2002) Crystal structure of a transition state mimic of the catalytic subunit of cAMP-dependent protein kinase., Nat Struct Biol 9, 273-277.

151. Sondek, J., Lambright, D., Noel, J., Hamm, H., and Sigler, P. (1994) GTPase mechanism of Gproteins from the 1.7-A crystal structure of transducin alpha-GDP-AIF-4., Nature 372, 276-279.

152. Baxter, N., Olguin, L., Golicnik, M., Feng, G., Hounslow, A., Ber-mel, W., Blackburn, G., Hollfelder, F., Waltho, J., and Williams, N. (2006) A Trojan horse transition state analogue generated by MgF3

- formation in an enzyme active site., Proc Nat Aca Sci U S A 103, 14732-14737.

153. Baxter, N., Blackburn, G., Marston, J., Hounslow, A., Cliff, M., Bermel, W., Williams, N., Hollfelder, F., Wemmer, D., and Waltho, J. (2008) Anionic charge is prioritized over geometry in aluminum and magnesium fluoride transition state analogs of phosphoryl trans-fer enzymes., J Am Chem Soc 130, 3952-3958.