5-Aminolevulinic Acid and Derivatives Thereof “Erwin with His Psi Can Do Örebro Studies in Life Science 6 Calculations Quite a Few

5-Aminolevulinic acid and derivatives thereof “Erwin with his psi can do Örebro Studies in Life Science 6 Calculations quite a few. But one thing has not been seen: Just what does psi really mean?”

– Erich Hückel, translated by Felix Bloch

Edvin Erdtman

5-Aminolevulinic acid and derivatives thereof Properties, lipid permeability and enzymatic reactions “Erwin with his psi can do Örebro Studies in Life Science 6 Calculations quite a few. But one thing has not been seen: Just what does psi really mean?”

– Erich Hückel, translated by Felix Bloch

Edvin Erdtman

5-Aminolevulinic acid and derivatives thereof Properties, lipid permeability and enzymatic reactions Abstract

Edvin Erdtman (2010): 5-Aminolevulinic acid and derivatives thereof Properties, lipid permeability and enzymatic reactions. Örebro Studies in Life Science 6, 76 pp.

5-aminolevulinic acid (5-ALA) and derivatives thereof are widely used prodrugs in treatment of pre-malignant skin diseases of the cancer treatment method photodynamic therapy (PDT). The target molecule in 5-ALA- PDT is protoporphyrin IX (PpIX), which is synthesized endogenously from 5-ALA via the heme pathway in the cell. This thesis is focused on 5-ALA, which is studied in different perspectives and with a variety of computational methods. The structural and energetic properties of 5-ALA, its methyl-, ethyl- and hexyl esters, four different 5-ALA enols, and hydrated 5-ALA have been investigated using Quantum Mechanical (QM) first principles density functional theory (DFT) calculations. 5-ALA is found to be more stable than its isomers and the hydrolysations of the esters are more spontaneous for longer 5-ALA ester chains than shorter. The keto-enol tautomerization mechanism of 5-ALA has been studied, and a self-catalysis mechanism has been proposed to be the most probable. Molecular Dynam- ics (MD) simulations of a lipid bilayer have been performed to study the membrane permeability of 5-ALA and its esters. The methyl ester of 5-ALA

was found to have the highest permeability constant (PMe-5-ALA = 52.8 cm/s). The mechanism of the two heme pathway enzymes; Porphobilinogen synthase (PBGS) and Uroporphyrinogen III decarboxylase (UROD), have been studied by DFT calculations and QM/MM methodology. The rate-limiting © Edvin Erdtman, 2010 step is found to have a barrier of 19.4 kcal/mol for PBGS and 13.7 kcal/mol for the first decarboxylation step in UROD. Generally, the results Title: 5-Aminolevulinic acid and derivatives thereof. are in good agreement with experimental results available to date. Properties, lipid permeability and enzymatic reactions.

Publisher: Örebro University 2010 www.publications.oru.se

Editor: Maria Alsbjer [email protected]

Print: Intellecta Infolog, Kållered 03/2010 Keywords: 5-Aminolevulinic acid, tautomerization, PDT, DFT, MM, QM/MM, Porphobilinogen synthase, Uroporphyrinogen III decarboxylase, issn 1653-3100 membrane penetration, enzyme mechanism. isbn 978-91-7668-718-5

EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 5

Abstract

Edvin Erdtman (2010): 5-Aminolevulinic acid and derivatives thereof Properties, lipid permeability and enzymatic reactions. Örebro Studies in Life Science 6, 76 pp.

Publisher: Örebro University 2010 www.publications.oru.se

Editor: Maria Alsbjer [email protected]

EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 5

List of papers

This thesis is based on following papers:

I. Erdtman, Edvin and Eriksson, Leif A., Theoretical study of 5- aminolevulinic acid (5-ALA) and some pharmaceutically important derivatives. Chem. Phys. Lett., 2007. 434(1-3): p. 101-106.

II. Erdtman, Edvin and Eriksson, Leif A., Theoretical study of 5- aminolevulinic acid tautomerization: a novel self-catalyzed mechanism. J. Phys. Chem. A, 2008. 112(18): p. 4367-4374.

III. Erdtman, Edvin; dos Santos, Daniel J. V. A.; Löfgren, Lennart and Eriksson, Leif A., Modelling the behavior of 5-aminolevulinic acid and its alkyl esters in a lipid bilayer. Chem. Phys. Lett., 2008. 463(1-3): p. 178-182. Erratum in Chem. Phys. Lett., 2009. 470(4-6): p. 369.

IV. Erdtman, Edvin; Gauld, James W. and Eriksson, Leif A., Compu- tational insights into the mechanism of substrate binding in porphobilinogen synthase, (submitted to Phys. Chem. Chem. Phys.), 2010.

V. Erdtman, Edvin; Gauld, James W. and Eriksson, Leif A., Model- ling the mechanism of porphobilinogen synthase, (manuscript, will be submitted to J. Phys. Chem. B), 2010.

VI. Bushnell, Eric A. C.; Erdtman, Edvin; Llano, Jorge; Eriksson, Leif A.; Gauld, James W., A computational study into the first branch- ing point in porphyrin biosynthesis; decarboxylation of ring D in URO-III by uroporphyrinogen-III decarboxylase (submitted to Biochemistry), 2010.

All published papers are printed with permission from the journals.

6 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 7

List of papers

This thesis is based on following papers:

I. Erdtman, Edvin and Eriksson, Leif A., Theoretical study of 5- aminolevulinic acid (5-ALA) and some pharmaceutically important derivatives. Chem. Phys. Lett., 2007. 434(1-3): p. 101-106.

II. Erdtman, Edvin and Eriksson, Leif A., Theoretical study of 5- aminolevulinic acid tautomerization: a novel self-catalyzed mechanism. J. Phys. Chem. A, 2008. 112(18): p. 4367-4374.

V. Erdtman, Edvin; Gauld, James W. and Eriksson, Leif A., Model- ling the mechanism of porphobilinogen synthase, (manuscript, will be submitted to J. Phys. Chem. B), 2010.

All published papers are printed with permission from the journals.

6 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 7

My contributions to the papers are: Abbreviations, symbols and units I-V All calculations and analysis, writing of the first drafts and revi- 5-ALA 5-Aminolevulinic acid sions of the papers. 5-ALA-hyd 5-Amino-4,4-dihydroxy-pentanoic acid VI Supervision and discussions with the first author related to the (hydrated 5-aminolevulinic acid) docking, MM optimizations and MD simulations. Some of the Me-5-ALA 5-Aminolevulinic acid methyl ester MD simulations and the average distance calculations were per- Et-5-ALA 5-Aminolevulinic acid ethyl ester formed by me. He-5-ALA 5-Aminolevulinic acid hexyl ester 5-CLA 5-Clorolevulinic acid ALAS 5-Aminolevulinic acid synthase ALAD 5-Aminolevulinic acid dehydratase, synonym to PBGS AO Atomic orbital B3LYP Becke 3-Parameter (Exchange), Lee, Yang and Parr (Corre- lation) CP-III Coproporphyrinogen III DFT Density functional theory DNA Deoxyribonucleic acid DPPC Dipalmitoylphosphatidylcholine (a phospholipid) EC Enzyme commission number (classification number for enzymes) FC Ferrochelatase GTO Gaussian type orbital HF Hartree-Fock IEFPCM Integral equation formalism of the polarizable continuum model LA Levulinic acid MD Molecular dynamics MM Molecular mechanics MO Molecular orbital MO-LCAO Molecular orbital linear combination of atomic orbitals PI, PII, … Paper I, Paper II etc. PA Proton affinity PBG Porphobilinogen PBGS Porphobilinogen synthase, also called ALAD PDT Photodynamic therapy PMB Photodynamic molecular beacon PpIX Protoporphyrin IX QM Quantum Mechanics QM/MM Combined calculation method, with a QM and a MM part.

8 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 9

RHF Restricted Hartree-Fock Contents ROS Reactive Oxygen Species ROHF Restricted Open-shell Hartree-Fock 1 Introduction ...... 13 SCF Self consistent field 1.1 Photodynamic therapy ...... 13 sCoA succinyl-Coenzyme A STO Slater type orbital 1.1.1 The PDT approach ...... 14 TCA cycle Tricarboxylic acid cycle or Citric acid cycle 1.1.1.1 Photosensitizers ...... 14 UHF Unrestricted Hartree-Fock 1.1.1.2 Light ...... 16 URO-III Uroporphyrinogen-III 1.1.1.3 Oxygen ...... 16 UROD Uroporphyrinogen-III decarboxylase 1.1.2 The PDT mechanism ...... 17 ZPE Zero-point vibrational energy 1.1.3 Cellular mechanisms ...... 18 1.1.4 PDT vs other treatments ...... 19 Ψ psi, wave function 1.2 5-Aminolevulinic acid ...... 19 λ lambda, unit for wavelength 1.2.1 5-ALA metabolism ...... 19 ν nu, unit for frequency. hν represents the energy of a pho- 1.2.1.1 Porphobilinogen synthase ...... 21 ton. 1.2.1.2 Uroporphyrinogen III decarboxylase...... 23 h Planck constant = 6.626 × 10-34 Js 1.2.2 5-ALA-PDT ...... 25 S0 Ground singlet state 1.2.2.1 Fluorescence ...... 26 S1 First excited singlet state 1.2.2.2 Photobleaching ...... 26 Sn Higher (n th) excited singlet state 1.2.2.3 Limitations ...... 26 T1 First excited triplet state 1.3 Tautomerism ...... 27

Å 1 Ångström = 10-10 m 2 Computational Methods ...... 29 -9 nm 1 nanometre = 10 m 2.1 Quantum Mechanics ...... 29 ns 1 nanosecond = 10-9 s 2.1.1 Hartree-Fock ...... 30 ps 1 picosecond = 10-12 s 2.1.2 Basis sets ...... 31 fs 1 femtosecond = 10-15 s 2.1.3 Density Functional Theory ...... 33 kcal/mol 1 kilo calorie per mol = 0.239 kJ/mol 2.1.4 Hybrid methods ...... 34

2.2 Molecular Mechanics & Molecular Dynamics ...... 34 2.3 QM/MM method ...... 36 2.4 Computational methods in the current studies ...... 37 2.4.1 Paper I and II ...... 37 2.4.2 Paper III ...... 37 2.4.3 Paper IV and V ...... 39 2.4.4 Paper VI ...... 39 2.5 Computational facilities ...... 40

3 Summary of results ...... 41

10 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 11

3 Summary of results ...... 41

10 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 11

3.1 QM calculations of 5-ALA and its derivatives (P I & P II) ..... 41 CHAPTER 1 3.1.1 Structural properties ...... 42 1 Introduction 3.1.2 Free energies ...... 43 The focus of this thesis is the drug 5-aminolevulinic acid (5-ALA), which is 3.1.3 Proton affinities ...... 45 used to treat pre-malignant skin disorders with the treatment modality 3.1.4 Tautomerization mechanism ...... 47 photodynamic therapy (PDT). The aims are to with various computational 3.2 MD simulations of 5-ALA and its esters in membrane (P III) 50 methods explore 5-ALA’s properties, understand how it behaves in cellular 3.3 Enzymatic reactions (P IV-VI) ...... 52 environments, and to get more insight into its metabolism. This knowledge 3.3.1 The mechanism of PBGS (PIV-V) ...... 52 can lead to further improvement of the treatment and the development of 3.3.1.1 The Schiff base formation ...... 52 new drugs based on 5-ALA derivatives. 3.3.1.2 Schiff base transfer ...... 55 3.3.1.3 Cyclization reaction mechanism ...... 57 1.1 Photodynamic therapy 3.3.2 The mechanism of UROD (PVI) ...... 60 Photodynamic therapy (PDT) is a treatment modality for primary cancer- 4 Conclusions and future perspectives ...... 63 ous lesions, but also pre-malignant and non-malignant diseases. In the late 19th century Niels Ryberg Finsen began to take an interest in the healing Acknowledgements ...... 65 effect of sunlight on the skin. Finsen, who received the Nobel Prize for his findings in 1903, was able to treat skin disorders such as smallpox and References ...... 67 Lupus vulgaris.1 Inspired by Finsen’s publicity, a lot of research was started in this field in the beginning of the 20th century. Raab, Jesionek and von Tappeiner found that sunlight combined with a photosensitizer and oxygen could destroy cells, whereupon von Tappeiner coined the term photodynamic therapy. The dye eosin was then used to treat both epilepsy and cancer in conjunction with sunlight.2 The curing capability of light was however not a new knowledge. The use of sun treatment was known thousands of years earlier, back in the ancient Egypt, India, China and Greece. They had found that by eating various plants in combination with exposure to sunlight, they could treat skin lesions as vitiligo, cancer, psoriasis and infections. Later these plants were found to contain psoralen compounds, which absorb the light of the sun.2,3 In the last few decades, the number of scientific studies and the usage of PDT have enormously increased. Large steps have been taken in the improvement of photosensitizers and illumination techniques. Due to en- hancement of the illumination there are now a variety of diseases that can be treated with PDT. In the beginning PDT was used mainly to treat skin disorders. However, endoscopes with lasers made it possible to bring the light to the lesion, even within the body. Therefore, is it now possible to treat cancer in for example: the bladder, lungs and the organs in the gastro- intestinal tract (i.e. the part of the digestive system consisting of the stom- ach, small intestine and large intestine). In addition, in conjunction with surgery even brain cancer could be treated by PDT.3,4 Methods are under

12 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 13

12 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 13 development, where light is delivered via optical fibres through needles a high dosage was needed. One of these derivatives is Photofrin – the most stuck into the tumour, which make it possible to treat larger tumours.5,6 clinically used photosensitizer (also known as Porfimer sodium). Photofrin Alongside cancerous treatment, PDT is used to treat some non-malignant is a mixture of oligomers ranging from two to nine porphyrin units linked and pre-malignant skin diseases, such as psoriasis, actinic keratosis, acne, together by primarily ether bonds. Photofrin has been approved for use in age-related macular degeneration and blood sterilization.7,8 treatment of many cancer diseases, such as lung-, oesophageal-, bladder- and cervical cancer as well as malignant and non-malignant skin diseases. 1.1.1 The PDT approach However, Photofrin has serious disadvantages. First of all the clearance of PDT is a three component method, where all of its three components need Photofrin in the body is very slow. It stays photoactive for weeks after the to be present simultaneously. These three essential components are a pho- treatment, and during this time the patient is very sensitive to light and is tosensitizer, light and oxygen. When all these components are present, the not able to stay in the sunlight for longer periods of time. Secondly Pho- light excites the photosensitizer, which reacts with oxygen to form reactive tofrin has a weak absorption peak above 600 nm (630 nm), why the dos- 8,12 oxygen species (ROS), e.g. singlet oxygen and/or peroxide radicals. The age needs to be quite high instead. ROS are very reactive, and destroy the cancer cells by oxidation of cellular To solve these problems, a number of second generation porphyrin components. Normally there is oxygen present in tissue, whereas the other based photosensitizers have been developed, which are more swiftly de- two components have to be added. Each of these components is discussed graded in the body and absorbs at higher wavelengths. Various substitu- more in detail below. ents are attached to the porphyrin ring to get a larger system of conjugated double bonds, which will red-shift the absorption maxima. Two examples 1.1.1.1 Photosensitizers are the drug Foscan, which is applied for clinical use in head and neck cancer, and Tookad, which is applied for prostate cancer treatment. A A photosensitizer is a compound that can be excited to a higher energy drawback of building these large molecules could be that the drugs become level upon illumination by a specific wavelength. It is preferably built up by very lipophilic and may accumulate in the cell membrane. a conjugated π-electron system. Another approach is to apply a prodrug, which is metabolized in situ to There are a number of criteria for a good photosensitizer. First of all, it a photosensitizer that is naturally present in the body. 5-aminolevulinic must be chemically and physically stable, and also chemically pure. It is an acid (5-ALA) is the precursor to heme and other porphyrins in living or- advantage if it is water soluble, but it must also be able to penetrate the ganisms, and by excess of 5-ALA the photosensitizer protoporphyrin IX is lipophilic cell membrane. It should be nontoxic in the absence of light, and accumulated. This will be discussed later in chapter 1.2.2 about 5-ALA- preferably become photoactive in the red to near IR region (i.e. it should PDT. have a high molar absorption coefficient at λ = 600-900 nm). Another A third generation of porphyrin photosensitizers has also begun to be important factor is that the photosensitizer should not become photoactive examined. Beyond the second generation, these photosensitizers are de- upon UV-radiation. Furthermore, the photosensitizer should accumulate signed to have more specific affinity to the tumour tissue, and are built up more selectively in tumour cells than healthy tissue, and reach its max con- by second generation photosensitizers bound to carriers, such as antibodies centration there relatively fast. Finally, it should also leave the body rapidly or liposomes.8,13 to prevent sensitivity to light after the treatment.9-11 Besides the porphyrin derivatives there are a couple of other drugs in use Photosensitizers are divided into porphyrins and non-porphyrins. The and in development; such as metal complexes and dyes like the an- porphyrins are in turn classified as first, second and third generation pho- thraquinone-derivative hypericin and Methylene Blue.10,12 tosensitizers.8 Another very recent method to get more specific treatment to the tu- The first generation of photosensitizers are based on a compound called mour tissue is the use of photodynamic molecular beacons (PMBs). A PMB hematoporphyrin (Hp). Hp is a tetrapyrrole extracted from blood, in consists of a photosensitizer which is combined with a linker to a ROS which the iron has been removed.2 This compound was found to specifi- quencher. The linker is designed to bind to a cancer cell-specific biomarker, cally accumulate in cancerous tissue. A lot of effort was taken into devel- and when bound the quencher is cut off. This means that healthy tissue opment of superior derivatives of Hp, since it was not effective enough and will not get injured, since the ROS produced of the photosensitizer are

14 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 15 development, where light is delivered via optical fibres through needles a high dosage was needed. One of these derivatives is Photofrin – the most stuck into the tumour, which make it possible to treat larger tumours.5,6 clinically used photosensitizer (also known as Porfimer sodium). Photofrin Alongside cancerous treatment, PDT is used to treat some non-malignant is a mixture of oligomers ranging from two to nine porphyrin units linked and pre-malignant skin diseases, such as psoriasis, actinic keratosis, acne, together by primarily ether bonds. Photofrin has been approved for use in age-related macular degeneration and blood sterilization.7,8 treatment of many cancer diseases, such as lung-, oesophageal-, bladder- and cervical cancer as well as malignant and non-malignant skin diseases. 1.1.1 The PDT approach However, Photofrin has serious disadvantages. First of all the clearance of PDT is a three component method, where all of its three components need Photofrin in the body is very slow. It stays photoactive for weeks after the to be present simultaneously. These three essential components are a pho- treatment, and during this time the patient is very sensitive to light and is tosensitizer, light and oxygen. When all these components are present, the not able to stay in the sunlight for longer periods of time. Secondly Pho- light excites the photosensitizer, which reacts with oxygen to form reactive tofrin has a weak absorption peak above 600 nm (630 nm), why the dos- 8,12 oxygen species (ROS), e.g. singlet oxygen and/or peroxide radicals. The age needs to be quite high instead. ROS are very reactive, and destroy the cancer cells by oxidation of cellular To solve these problems, a number of second generation porphyrin components. Normally there is oxygen present in tissue, whereas the other based photosensitizers have been developed, which are more swiftly de- two components have to be added. Each of these components is discussed graded in the body and absorbs at higher wavelengths. Various substitu- more in detail below. ents are attached to the porphyrin ring to get a larger system of conjugated double bonds, which will red-shift the absorption maxima. Two examples 1.1.1.1 Photosensitizers are the drug Foscan, which is applied for clinical use in head and neck cancer, and Tookad, which is applied for prostate cancer treatment. A A photosensitizer is a compound that can be excited to a higher energy drawback of building these large molecules could be that the drugs become level upon illumination by a specific wavelength. It is preferably built up by very lipophilic and may accumulate in the cell membrane. a conjugated π-electron system. Another approach is to apply a prodrug, which is metabolized in situ to There are a number of criteria for a good photosensitizer. First of all, it a photosensitizer that is naturally present in the body. 5-aminolevulinic must be chemically and physically stable, and also chemically pure. It is an acid (5-ALA) is the precursor to heme and other porphyrins in living or- advantage if it is water soluble, but it must also be able to penetrate the ganisms, and by excess of 5-ALA the photosensitizer protoporphyrin IX is lipophilic cell membrane. It should be nontoxic in the absence of light, and accumulated. This will be discussed later in chapter 1.2.2 about 5-ALA- preferably become photoactive in the red to near IR region (i.e. it should PDT. have a high molar absorption coefficient at λ = 600-900 nm). Another A third generation of porphyrin photosensitizers has also begun to be important factor is that the photosensitizer should not become photoactive examined. Beyond the second generation, these photosensitizers are de- upon UV-radiation. Furthermore, the photosensitizer should accumulate signed to have more specific affinity to the tumour tissue, and are built up more selectively in tumour cells than healthy tissue, and reach its max con- by second generation photosensitizers bound to carriers, such as antibodies centration there relatively fast. Finally, it should also leave the body rapidly or liposomes.8,13 to prevent sensitivity to light after the treatment.9-11 Besides the porphyrin derivatives there are a couple of other drugs in use Photosensitizers are divided into porphyrins and non-porphyrins. The and in development; such as metal complexes and dyes like the an- porphyrins are in turn classified as first, second and third generation pho- thraquinone-derivative hypericin and Methylene Blue.10,12 tosensitizers.8 Another very recent method to get more specific treatment to the tu- The first generation of photosensitizers are based on a compound called mour tissue is the use of photodynamic molecular beacons (PMBs). A PMB hematoporphyrin (Hp). Hp is a tetrapyrrole extracted from blood, in consists of a photosensitizer which is combined with a linker to a ROS which the iron has been removed.2 This compound was found to specifi- quencher. The linker is designed to bind to a cancer cell-specific biomarker, cally accumulate in cancerous tissue. A lot of effort was taken into devel- and when bound the quencher is cut off. This means that healthy tissue opment of superior derivatives of Hp, since it was not effective enough and will not get injured, since the ROS produced of the photosensitizer are

14 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 15 quenched by the quencher. However, in cancer cells the quencher will no There are however non-oxygen dependent techniques, which are not longer be proximate to the photosensitizer and the ROS are free to destroy strictly speaking PDT-methods (however, the wider concept photochemo- the cell.8,14 therapy also involves these techniques). Without the involvement of oxygen the photosensitizer is excited to higher triplet states, and is quenched di- 1.1.1.2 Light rectly by the tissue.20 These techniques are useful in tissues with low oxy- The light used in PDT is photosensitizer specific, since each photosensitizer gen levels, for example in the middle of larger tumours. has its own absorption maxima. Even though it is possible to use white light, which consists of a wide spectrum of photons, better results have 1.1.2 The PDT mechanism been found by using monochromatic coherent light. The general mechanism of PDT can be explained as follows; a photosensi- The most relevant light used in PDT is roughly visible light (400-700 nm tizer is excited by light, followed by the reaction of the excited photosensi- cf. Figure 1.1), but in practice the most used light ranges from 600 to 900 tizer with molecular oxygen to produce ROS, such as singlet oxygen, hy- nm. Wavelengths shorter than 600 nm are not suitable, since there is an droxyl- or superoxide radicals. There are two types of photoreaction elevated risk for sunlight photosensitivity (sunlight contains radiation with mechanisms; type I and type II. The first steps are the same in both mecha- wavelength λ < 600 nm). Furthermore, hemoglobin absorbs most of the nisms: incoming photons at these wavelengths.7,15 The penetration depth into P(S ) hν P(S ) P(S ) ISC P(T ) (1.1) tissue is also a limiting factor of illumination. Red light (630-710 nm) for 0 ⎯→⎯ n ⎯→⎯ 1 ⎯→⎯ 1 example has a penetration depth of 2.0-4.5 mm in tumours, whereas near The photosensitizer (P) is excited by a photon (hν) from its ground state 16 IR (1060 nm) light may penetrate up to 6.5 mm. Unfortunately, the pho- (S0) to a singlet excited state (Sn). The photosensitizer is then relaxed to the tons in the IR region do not have enough energy to excite most photosensi- lowest singlet excited state (S1), followed by an intersystem crossing to the tizers and generate singlet oxygen. first excited triplet state (T1). Triplet states are relatively more stable than excited singlet states. Triplets have therefore more time to undergo further

reactions. However, competing reactions from T1 are fluorescence (1.2) and radiation-less relaxation (1.3).21

P(T1) ⎯→⎯ P(S0) + hν' (1.2)

P(T1) ⎯→⎯ P(S0) + heat (1.3) Generally the dominating mechanism is determined by the concentration of oxygen. If the oxygen concentration is large, the type II mechanism is most probable, while the type I mechanism is predominant if there is a lower oxygen concentration.22 The definitions of how to distinguish between the Type I and Type II reaction types diverge. One definition is based on the primary interaction of the photosensitizer. If this first reacts with the solvent or a biological substrate, it is a Type I, but if the photo- Figure 1.1 The radiation spectrum.17 sensitizer first reacts with oxygen it is a Type II process.23 Another classification is based on whether oxygen radicals are formed via electron-transfer 1.1.1.3 Oxygen or singlet oxygen via energy-transfer.21 A simplified scheme over Type I / Molecular oxygen is by definition mandatory for PDT. Moan et al. ob- Type II reactions are shown in Figure 1.2. served the reactive singlet oxygen during PDT and found a connection between low oxygen concentration and less PDT effect.18,19

16 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 17 quenched by the quencher. However, in cancer cells the quencher will no There are however non-oxygen dependent techniques, which are not longer be proximate to the photosensitizer and the ROS are free to destroy strictly speaking PDT-methods (however, the wider concept photochemo- the cell.8,14 therapy also involves these techniques). Without the involvement of oxygen the photosensitizer is excited to higher triplet states, and is quenched di- 1.1.1.2 Light rectly by the tissue.20 These techniques are useful in tissues with low oxy- The light used in PDT is photosensitizer specific, since each photosensitizer gen levels, for example in the middle of larger tumours. has its own absorption maxima. Even though it is possible to use white light, which consists of a wide spectrum of photons, better results have 1.1.2 The PDT mechanism been found by using monochromatic coherent light. The general mechanism of PDT can be explained as follows; a photosensi- The most relevant light used in PDT is roughly visible light (400-700 nm tizer is excited by light, followed by the reaction of the excited photosensi- cf. Figure 1.1), but in practice the most used light ranges from 600 to 900 tizer with molecular oxygen to produce ROS, such as singlet oxygen, hy- nm. Wavelengths shorter than 600 nm are not suitable, since there is an droxyl- or superoxide radicals. There are two types of photoreaction elevated risk for sunlight photosensitivity (sunlight contains radiation with mechanisms; type I and type II. The first steps are the same in both mecha- wavelength λ < 600 nm). Furthermore, hemoglobin absorbs most of the nisms: incoming photons at these wavelengths.7,15 The penetration depth into P(S ) hν P(S ) P(S ) ISC P(T ) (1.1) tissue is also a limiting factor of illumination. Red light (630-710 nm) for 0 ⎯⎯→ n ⎯⎯→ 1 ⎯⎯→ 1 example has a penetration depth of 2.0-4.5 mm in tumours, whereas near The photosensitizer (P) is excited by a photon (hν) from its ground state 16 IR (1060 nm) light may penetrate up to 6.5 mm. Unfortunately, the pho- (S0) to a singlet excited state (Sn). The photosensitizer is then relaxed to the tons in the IR region do not have enough energy to excite most photosensi- lowest singlet excited state (S1), followed by an intersystem crossing to the tizers and generate singlet oxygen. first excited triplet state (T1). Triplet states are relatively more stable than excited singlet states. Triplets have therefore more time to undergo further

reactions. However, competing reactions from T1 are fluorescence (1.2) and radiation-less relaxation (1.3).21

P(T1) ⎯⎯→ P(S0) + hν' (1.2)

P(T1) ⎯⎯→ P(S0) + heat (1.3) Generally the dominating mechanism is determined by the concentration of oxygen. If the oxygen concentration is large, the type II mechanism is most probable, while the type I mechanism is predominant if there is a lower oxygen concentration.22 The definitions of how to distinguish between the Type I and Type II reaction types diverge. One definition is based on the primary interaction of the photosensitizer. If this first reacts with the solvent or a biological substrate, it is a Type I, but if the photo- Figure 1.1 The radiation spectrum.17 sensitizer first reacts with oxygen it is a Type II process.23 Another classification is based on whether oxygen radicals are formed via electron-transfer 1.1.1.3 Oxygen or singlet oxygen via energy-transfer.21 A simplified scheme over Type I / Molecular oxygen is by definition mandatory for PDT. Moan et al. ob- Type II reactions are shown in Figure 1.2. served the reactive singlet oxygen during PDT and found a connection between low oxygen concentration and less PDT effect.18,19

16 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 17

the other hand is a controlled cell death that is naturally taking place so that the organelles of the dead cell can be recycled. Whether the cell death is caused by necrosis or apoptosis depends on the location of the photosensitizer when it is illuminated. It is found that if the photosensitizer is illuminated in the mitochondria, the cell predominantly undergoes apoptosis. However, if the photosensitizer is settled in the cell membrane necrosis is more predominant. Other factors which play an important role are the cell line and the dosage of light and photosensitizer. Generally low doses of PDT lead to apoptosis, while higher doses increase the possibility for necrosis.11,15,25

1.1.4 PDT vs other treatments Compared with other cancer treatment techniques PDT has several advantages. Besides killing the cancer cells directly, PDT can also damage the Figure 1.2 A simplified scheme describing the Type I and Type II photoreactions. P tumour’s associated vasculature. Thus, the blood transfer to the tumours is represents the photosensitizer and A the substrate; a molecule in the cancerous affected, which suffocates the tumour. Another important issue is the im- tissue, for example a membranal phospholipid or solvent. mune system’s response to the treatment. While surgery, ionizing radiation and chemotherapy suppress the immune system, PDT stimulates it. When A type I photoreaction is a hydrogen abstraction or an electron-transfer these three mechanisms; cell death, vasculature destruction and immune reaction between a photosensitizer and a substrate (A in Figure 1.2), which response, can be controlled to act together, a long-term tumour regression can either be the solvent, another photosensitizer or a biological molecule. is performed by PDT.11 Free radicals or radical ions are formed, which are very reactive and react with oxygen to produce superoxide radical anions or hydroxyl radicals. 1.2 5-Aminolevulinic acid These radicals then cause oxidative damage to the cell.12,21 Type I reactions may also be independent of oxygen, as for psoralens reaction with DNA. 5-ALA (Figure 1.3) represents a completely different aspect of PDT. It is These reactions are also sometimes called Type III reactions.10,24 not in itself photosensible but with an excess of 5-ALA, Protoporphyrin IX In a type II reaction, the photosensitizer transfers its excitation energy (PpIX) is produced in situ. In the following section the metabolism of 5- directly into the oxygen molecule. Singlet oxygen is generated via energy ALA will be discussed. transfer from the excited photosensitizer to triplet oxygen when they col- O lide. Singlet oxygen will then cause oxidative damage to the tissue (Figure 2 + O 1 4 NH3 1.2). 3 5 The ROS do also oxidize and degrade the photosensitizer; a process called photobleaching. In average each photosensitizer molecule can cata- O lyse the production of 103-105 singlet oxygen molecules before it is de- Figure 1.3 5-Aminolevulinic acid in its zwitterionic form, with the numbering of stroyed by photobleaching or other processes.21 the carbon atoms. 1.2.1 5-ALA metabolism 1.1.3 Cellular mechanisms 5-ALA is a delta amino acid that has a carbonyl group at the fourth carbon The PDT treatment affects the cells in different ways, and causes cell death by either necrosis or apoptosis. Necrosis on one hand is a sudden cell (systematic name: 5-amino-4-oxopentanoic acid, Figure 1.3). 5-ALA is death, where organelles and membranes are damaged, while apoptosis on present in all kinds of organisms. There are two distinct pathways in which 5-ALA is biosynthesized; from glutamate (the C5 or Beale pathway) or

18 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 19

18 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 19 from succinyl-Coenzyme A (sCoA) and glycine (the C4 or Shemin path- There are seven further enzymes involved in the formation of heme; way). In plants, algae, cyanobacteria, most other bacteria and archaea the porphobilinogen synthase, porphobilinogen deaminase, uroporphyrinogen multistep C5 pathway is used to synthesize 5-ALA, whereas the one-step C4 III synthase, uroporphyrinogen III decarboxylase, coproporphyrinogen III pathway is found in humans, animals, yeasts, and a few bacteria.26 oxidase, protoporphyrinogen IX oxidase and ferrochelatase. The latter Since we are more interested in the 5-ALA mechanisms in humans, we three are located in the mitochondria and the others in the cytosol (see will not go into detail of the C3 pathway. The C4 pathway is in eukaryotes Figure 1.4). In the current study porphobilinogen synthase (PBGS) and combined with the tricarboxylic acid cycle (TCA cycle) by sCoA. sCoA is uroporphyrinogen III decarboxylase (UROD) have been considered with together with glycine the substrates of the mitochondria located enzyme special interest (marked in bold in Figure 1.4). aminolevulinic acid synthase (ALAS; EC: 2.3.1.37). ALAS is a homodimer with the active site located in the subunit interface, in which two pyridoxal 1.2.1.1 Porphobilinogen synthase 5´-phosphate cofactors are symmetrically bound. ALAS catalyses the de- The second enzyme in the heme pathway is porphobilinogen synthase carboxylative condensation of glycine and sCoA, where the release of 5- (PBGS), also called 5-ALA dehydratase (ALAD; EC 4.2.1.24). Two 5-ALA ALA is the rate-determining step. ALAS is considered as the first enzyme in molecules are combined to the pyrrole porphobilinogen (PBG). PBGS is 26,27 the heme biosynthesis (see Figure 1.4). located in the cytosol in contrast to ALAS which is operational in the mitochondria. PBGS is a metalloenzyme, which is most active in a homo- octameric form. By the natural single mutation of phenylalanine to leucine (F12L) in human PBGS, a hexamer structure can be formed; however, with 28,29 Cytosol Mitochondrion a much lower activity (~12% of the wild type enzyme). The active site of PBGS has been found to be highly conserved amongst ALA synthase different species. All residues of PBGS are hereafter identified according to their yeast numbering (PDB ID 1H7O30). In particular, the active site con- sCoA + Gly 5ALA tains two lysine residues, Lys210 and Lys263, in the A- and P-site respectively (Figure 1.5). The sites are named after the acid group (acetyl- and Porphobilinogen propionyl-) of the product PBG derived from the carboxylate moieties of synthase Heme the 5-ALA substrates. Experimental mutagenesis studies have suggested that the latter lysine is essential for enzyme catalysis, and the former is Porphobilinogen essential for the binding of the first substrate.31 Each of the two 5-ALA deaminase Ferrochelatase substrates is found to bind to a lysine with a so-called Schiff base. A Schiff

base is an imine with a hydrocarbyl group on the nitrogen atom (R2C= Uroporphyrinogen PpIX NR'). III synthase Furthermore, the active site consists of several polar groups which form Protoporphyrin- hydrogen bonds to the carboxylate moieties of the P- (Ser290 and Tyr329) ogen IX oxidase and A-site (Gln236) bound 5-ALAs. Several residues (Ser179, Asp131 and Uroporphyrinogen III decarboxylase Tyr 207) form a polar pocket around or hydrogen bond to the terminal Corpoporphyrinogen amino group of the 5-ALA substrates. However, at least for the P-site; III oxidase substrate analogs without the terminal amino group have been found to be good competitive inhibitors.32 Therefore it is suggested that the interactions between the 5-ALA amino group and the enzyme are not essential, at least Figure 1.4 A simplified scheme of the heme biosynthesis, which is taking place in both the cytoplasm and in the mitochondria. The currently studied enzymes are not for the binding. A flexible segment of PBGS is also believed to seal the 33 marked in bold. active site when the 5-ALAs are bound . When this ‘lid’ is closed there are

20 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 21 from succinyl-Coenzyme A (sCoA) and glycine (the C4 or Shemin path- There are seven further enzymes involved in the formation of heme; way). In plants, algae, cyanobacteria, most other bacteria and archaea the porphobilinogen synthase, porphobilinogen deaminase, uroporphyrinogen multistep C5 pathway is used to synthesize 5-ALA, whereas the one-step C4 III synthase, uroporphyrinogen III decarboxylase, coproporphyrinogen III pathway is found in humans, animals, yeasts, and a few bacteria.26 oxidase, protoporphyrinogen IX oxidase and ferrochelatase. The latter Since we are more interested in the 5-ALA mechanisms in humans, we three are located in the mitochondria and the others in the cytosol (see will not go into detail of the C3 pathway. The C4 pathway is in eukaryotes Figure 1.4). In the current study porphobilinogen synthase (PBGS) and combined with the tricarboxylic acid cycle (TCA cycle) by sCoA. sCoA is uroporphyrinogen III decarboxylase (UROD) have been considered with together with glycine the substrates of the mitochondria located enzyme special interest (marked in bold in Figure 1.4). aminolevulinic acid synthase (ALAS; EC: 2.3.1.37). ALAS is a homodimer with the active site located in the subunit interface, in which two pyridoxal 1.2.1.1 Porphobilinogen synthase 5´-phosphate cofactors are symmetrically bound. ALAS catalyses the de- The second enzyme in the heme pathway is porphobilinogen synthase carboxylative condensation of glycine and sCoA, where the release of 5- (PBGS), also called 5-ALA dehydratase (ALAD; EC 4.2.1.24). Two 5-ALA ALA is the rate-determining step. ALAS is considered as the first enzyme in molecules are combined to the pyrrole porphobilinogen (PBG). PBGS is 26,27 the heme biosynthesis (see Figure 1.4). located in the cytosol in contrast to ALAS which is operational in the mitochondria. PBGS is a metalloenzyme, which is most active in a homo- octameric form. By the natural single mutation of phenylalanine to leucine (F12L) in human PBGS, a hexamer structure can be formed; however, with 28,29 Cytosol Mitochondrion a much lower activity (~12% of the wild type enzyme). The active site of PBGS has been found to be highly conserved amongst ALA synthase different species. All residues of PBGS are hereafter identified according to their yeast numbering (PDB ID 1H7O30). In particular, the active site con- sCoA + Gly 5ALA tains two lysine residues, Lys210 and Lys263, in the A- and P-site respectively (Figure 1.5). The sites are named after the acid group (acetyl- and Porphobilinogen propionyl-) of the product PBG derived from the carboxylate moieties of synthase Heme the 5-ALA substrates. Experimental mutagenesis studies have suggested that the latter lysine is essential for enzyme catalysis, and the former is Porphobilinogen essential for the binding of the first substrate.31 Each of the two 5-ALA deaminase Ferrochelatase substrates is found to bind to a lysine with a so-called Schiff base. A Schiff

20 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 21 in for example humans and yeast two arginine residues (Arg220 and five ligands, it is possibly involved in the reaction mechanism, forming 34 35 Arg232) of the lid, that form hydrogen bonds to the carboxylate of the A- bonds to either H2O, or the substrates/product. Experimental pH, site 5-ALA (see Figure 1.5).34,35 mutagenesis and kinetic studies have suggested that the zinc ion plays an important role in substrate binding at the A-site and in stabilizing intermediates and transition structures during the catalytic mechanism, but however not the binding of the first (P-site) 5-ALA molecule.38-41 This has been further supported by experimental NMR studies42 on the enzyme-bound product complex and crystal structures obtained from human and yeast PBGS with an ‘almost product’ intermediate bound within their active sites (PDB ID: 1E5143 and 1OHL35). In both these crystal structures, the terminal amino group corresponding to the A-site bound 5-ALA was found to be neutral and coordinated to the Zn2+ ion. It has also been proposed that P-site the carbonyl of A-site 5-ALA also coordinates the zinc ion; however this has not been observed in any crystal structures.44 The second sequence for metal binding includes a glutamine, two invari-

ant aspartates and one arginine RX~164DX~65EXXXD. This site coordinates A-site an allosteric octahedral Mg2+ ion, with the glutamine and seven water molecules in the first coordination sphere. The aspartate and the arginine residues are ligated at the outer coordination sphere together with more water molecules. This metal binding site is found in all organisms except metazoa, fungi and a few bacteria. The PBGSs that have a Zn2+ binding site, but no Mg2+ site (metazoa and fungi), has a second Zn2+ bound in proximate position to the first. This ion is however not crucial for the reaction.37 There are different proposed mechanisms of PBGS. The main differences Figure 1.5 Schematic illustration of the active site of PBGS with the two 5-ALA lay in how many Schiff bases that are formed in the active site, and in substrate molecules covalently bound at the A- (red) and P-site (blue) via Schiff- which order the intersubstrate bonds are formed. Other differences are 34 base linkages based on the yeast PBGS crystal structures PDB ID: 1H7O and which roles the zinc ion and the basic residues play in the active site.32,44 By 35 1OHL . the proofs of the crystal structures, a consensus has now been built up that there are two Schiff bases formed in the active site; one to each 5-ALA There are at least two different sequences for metal binding in PBGS, one molecule.44 A majority of recently studies also suggest that the C–C inter- primary for zinc ions, and one for magnesium ions. substrate bond is formed before the C–N bond.28,34,35,44-46 The first one is located in the active site. In archaea, some bacteria, metazoa (multicellular animals) and yeast organisms the sequence is very 1.2.1.2 Uroporphyrinogen III decarboxylase cysteine rich with the general sequence DXCXCX(Y/F)X3G(H/Q)CG, where the underlined cysteines coordinate a Zn2+ ion (shown in Figure 1.5). Uroporphyrinogen III (URO-III) is the first cyclic compound in the heme In other organisms this sequence is instead aspartate rich biosynthesis. The enzyme uroporphyrinogen III decarboxylase (UROD; EC 2+ 4.1.1.37) catalyses the decarboxylation of the acetyl chains of URO-III to (DXALDX(Y/F)X3G(H/Q)DG), which could bind either Mg or monova- lent ions such as K+ or Na+.36,37 form coproporphyrinogen III (CP-III): The Zn2+ ion in human and yeast PBGS coordinates to the thiolates of the three cysteines in the sequence above. Since zinc can coordinate four or

22 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 23 in for example humans and yeast two arginine residues (Arg220 and five ligands, it is possibly involved in the reaction mechanism, forming 34 35 Arg232) of the lid, that form hydrogen bonds to the carboxylate of the A- bonds to either H2O, or the substrates/product. Experimental pH, site 5-ALA (see Figure 1.5).34,35 mutagenesis and kinetic studies have suggested that the zinc ion plays an important role in substrate binding at the A-site and in stabilizing intermediates and transition structures during the catalytic mechanism, but however not the binding of the first (P-site) 5-ALA molecule.38-41 This has been further supported by experimental NMR studies42 on the enzyme-bound product complex and crystal structures obtained from human and yeast PBGS with an ‘almost product’ intermediate bound within their active sites (PDB ID: 1E5143 and 1OHL35). In both these crystal structures, the terminal amino group corresponding to the A-site bound 5-ALA was found to be neutral and coordinated to the Zn2+ ion. It has also been proposed that P-site the carbonyl of A-site 5-ALA also coordinates the zinc ion; however this has not been observed in any crystal structures.44 The second sequence for metal binding includes a glutamine, two invari-

22 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 23

make a 180º flip after the decarboxylation of the ring D acetyl, followed by 90º rotations after each the remaining decarboxylations. Since a 180º flip seems to be very unlikely to happen, it has been proposed that after the D- acetyl is decarboxylated, the substrate flips over to the other monomer’s (1.4) active site, and the rest of the decarboxylations take place there.60 How- ever, there are many results that indicate that a single unique catalysing site is used throughout the reaction, a mechanism which is now generally ac- cepted.47,50,52,57,58,61-63 where P specifies the propionate side chains. Under physiological substrate concentrations the enzyme starts the decarboxylation of the acetyl chain of ring D, followed by the acetyls of the A, B, and C rings, whereas at higher concentration the order is random.47,48 The overall structure revealed from X-ray structures of human, tobacco and Bacillus subtilis shows that UROD has a homodimer quaternary structure. The active site of UROD can generally be divided into three regions; one negative, one positive/polar, and one non-charged region.49 The non- charged region is believed to bind the relatively non-polar core of the tetrapyrrole. The negative region is consistent only by the invariant residue Asp86 (human UROD numbering), which coordinates to the –NH–groups in the tetrapyrrole, and is proposed to be deprotonated to stabilize the positively charged nitrogen atom in the intermediate structures. Indeed, through mutagenesis investigations it was found that a D86G mutation essentially killed the enzyme.50 In the positive/polar region there are four invariant residues; two arginines (Arg37 and 41), a histidine (His339) and a tyrosine (Tyr164). The arginines are found to interact with the carboxyl groups of the product.50,51 Several mutagenesis studies have suggested that His339 and Tyr164 are not in fact essential for catalysis.50,52 interestingly, mutation of His339 was found to have little or no effect on the rate at Figure 1.6 Proposed general acid/base mechanism for decarboxylation of the pyr- which the initial decarboxylation, that of ring D, occurred. However, it role acetates in URO-III. HA and HB represent the general acids.56-58 resulted in accumulation of the first mechanistic intermediate; ring D of URO-III decarboxylated while the A, B and C rings remained unaltered.52 1.2.2 5-ALA-PDT In contrast, it has been found that one or all arginine residues within the As already mentioned, 5-ALA-PDT is an elegant method of using the proc- active site (human UROD: Arg37, 41 and 50) are essential for esses inside the cell to produce the photosensitizer in situ. catalysis.49,51,53-55 However, the exact role of the residues remains unclear. The formation of 5-ALA is the rate-determining step in the heme synthe- Barnard and Akhtar have proposed the general mechanism for UROD sis and it is regulated by heme with feedback inhibition. Addition of extra- enzymes given in Figure 1.6.56-58 There are several proposed mechanisms, corporeal 5-ALA will bypass the feedback inhibition, and ferrochelatase and research groups currently contend about which residues are involved (FC) instead becomes the rate-determining enzyme. PpIX, the substrate of in the catalysis.50,53,59,60 In addition, the existing proposed mechanisms ar- FC, is then accumulated. To even more enhance the concentration of PpIX, gue the placement of URO–III within the catalytic site. Finally, it has been a FC inhibitor or an iron chelator can be added. Fortunately, the activity of debated that the mechanism involves more than a single active site. It has FC is found to be lower in tumour cells compared with other cells.64,65 been suggested that if catalysis occurs in a single site, the substrate needs to Thereby the excess of 5-ALA will lead to PpIX accumulation especially in

24 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 25

24 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 25 cancer cells. This is one of the advantages with 5-ALA-PDT compared to The second limitation relates to the two step functionality of 5-ALA- other methods. Another advantage is that PpIX is primary located in the PDT; the uptake of 5-ALA and the PpIX formation, are not easily con- mitochondria during illumination, which causes a higher degree of apop- trolled compared to other photosensitizers, where only the uptake of the tosis than other photosensitizers.66 Furthermore, the deliverance of 5-ALA drug is needed to be taken into account. is very flexible, since it is the only PDT-drug that can be administered both These limitations challenge researchers to find out workarounds to get topically and systemically. an increased effect of the 5-ALA-PDT modality. To solve the penetration problem, different derivatives of 5-ALA have been developed, such as the 1.2.2.1 Fluorescence methyl- and hexyl- esters of 5-ALA, and 5-ALA containing dendrimers.67 PpIX is, like most of the other PDT photosensitizers fluorescent, which has Another possibility is to make more persisting modifications of the 5-ALA, been found to be very useful combined with its high accumulation in can- to change the chemical structure of the photosensitizer, and hence red-shift cerous cells. The fluorescent light shows where the tumour is located, the absorption band. This however can be problematic, since the enzymes which is not always consistent with the shape of the lesion (e.g. on the in the biosynthesis of PpIX need to recognize them as substrates. skin). 5-ALA and other photosensitizers are approved for use in fluores- PDT has been started to be used in combination with other methods for cence photodiagnosis of cancer in dermatology, and the hexyl ester of 5- cancer treatment. There are two approaches of combining other methods ALA is approved for use in fluorescent diagnosis of bladder cancer.66 with PDT. i) A prior treatment which will increase the tumour cells suscep- tibility to the PDT treatment. ii) A second treatment after PDT, which is 68 1.2.2.2 Photobleaching dependent on the prosurvival molecular responses to the PDT modality. PpIX is a very photolabile photosensitizer and undergoes photobleaching 1.3 Tautomerism much easier compared with other synthetically made photosensitizes. PpIX reacts with singlet oxygen formed during the illumination and is converted Isomerism is a quite wide concept, including both structural isomers and into a form that is not photoactive anymore. However, photobleaching stereoisomers. Tautomerism is a type of structural isomerism, where the may actually have some important clinical effects.66 chemical composition is the same, but the position of a double bond and a Upon treatment of large tumours, PpIX in the outer regions of the tu- hydrogen atom differs. Unlike other structure isomers, tautomers intercon- mour is filtering the radiation, so that the inner parts are not reached by vert spontaneously into each other, and there is always an equilibrium the light, but when PpIX in the outer parts is photobleached the light can reaction between the tautomers. In most cases the equilibrium position is penetrate deeper. far to one side of the reaction. The most common types of tautomeric reac- A relatively small amount of PpIX is accumulated into healthy tissue tions are keto-enol-, amide-imidic acid-, amine-imine- and lactam-lactim compared to cancerous tissue. The PpIX molecules in normal tissue are tautomerism. then also more rapidly photobleached, before they can cause any serious Keto-enol tautomerization is the equilibrium reaction between a ketone damage. Because of this phenomenon light overdosage does not cause any and one or two enol forms: problem. Increasing the light dosage will only increase the effect on the OH O OH tumour, since the PpIX in normal tissue is already deactivated. C C C C C C H H H2 2 H2 H2 H (1.5) 1.2.2.3 Limitations The ketone is in most cases the most stable, and the equilibrium position There are two major limitations of 5-ALA-PDT. The first limitation is the lies far to the ketone. The enolic forms are stabilized if there is a stabilizing penetration of both light and drug into the tissue. 5-ALA has an ability to group next to the enolic hydroxyl group, as for example 2,4-pentanedione: stay superficial and not penetrate deep enough, due to its polarity. The H light penetrates, as already mentioned, deeper into human tissue at longer O O O O wavelengths and since PpIX has its long-wavelength absorption peak at C C C C 635 nm, the light penetration is not optimal. H3C C CH3 H3C C CH3 H2 H (1.6)

26 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 27 cancer cells. This is one of the advantages with 5-ALA-PDT compared to The second limitation relates to the two step functionality of 5-ALA- other methods. Another advantage is that PpIX is primary located in the PDT; the uptake of 5-ALA and the PpIX formation, are not easily con- mitochondria during illumination, which causes a higher degree of apop- trolled compared to other photosensitizers, where only the uptake of the tosis than other photosensitizers.66 Furthermore, the deliverance of 5-ALA drug is needed to be taken into account. is very flexible, since it is the only PDT-drug that can be administered both These limitations challenge researchers to find out workarounds to get topically and systemically. an increased effect of the 5-ALA-PDT modality. To solve the penetration problem, different derivatives of 5-ALA have been developed, such as the 1.2.2.1 Fluorescence methyl- and hexyl- esters of 5-ALA, and 5-ALA containing dendrimers.67 PpIX is, like most of the other PDT photosensitizers fluorescent, which has Another possibility is to make more persisting modifications of the 5-ALA, been found to be very useful combined with its high accumulation in can- to change the chemical structure of the photosensitizer, and hence red-shift cerous cells. The fluorescent light shows where the tumour is located, the absorption band. This however can be problematic, since the enzymes which is not always consistent with the shape of the lesion (e.g. on the in the biosynthesis of PpIX need to recognize them as substrates. skin). 5-ALA and other photosensitizers are approved for use in fluores- PDT has been started to be used in combination with other methods for cence photodiagnosis of cancer in dermatology, and the hexyl ester of 5- cancer treatment. There are two approaches of combining other methods ALA is approved for use in fluorescent diagnosis of bladder cancer.66 with PDT. i) A prior treatment which will increase the tumour cells suscep- tibility to the PDT treatment. ii) A second treatment after PDT, which is 68 1.2.2.2 Photobleaching dependent on the prosurvival molecular responses to the PDT modality. PpIX is a very photolabile photosensitizer and undergoes photobleaching 1.3 Tautomerism much easier compared with other synthetically made photosensitizes. PpIX reacts with singlet oxygen formed during the illumination and is converted Isomerism is a quite wide concept, including both structural isomers and into a form that is not photoactive anymore. However, photobleaching stereoisomers. Tautomerism is a type of structural isomerism, where the may actually have some important clinical effects.66 chemical composition is the same, but the position of a double bond and a Upon treatment of large tumours, PpIX in the outer regions of the tu- hydrogen atom differs. Unlike other structure isomers, tautomers intercon- mour is filtering the radiation, so that the inner parts are not reached by vert spontaneously into each other, and there is always an equilibrium the light, but when PpIX in the outer parts is photobleached the light can reaction between the tautomers. In most cases the equilibrium position is penetrate deeper. far to one side of the reaction. The most common types of tautomeric reac- A relatively small amount of PpIX is accumulated into healthy tissue tions are keto-enol-, amide-imidic acid-, amine-imine- and lactam-lactim compared to cancerous tissue. The PpIX molecules in normal tissue are tautomerism. then also more rapidly photobleached, before they can cause any serious Keto-enol tautomerization is the equilibrium reaction between a ketone damage. Because of this phenomenon light overdosage does not cause any and one or two enol forms: problem. Increasing the light dosage will only increase the effect on the OH O OH tumour, since the PpIX in normal tissue is already deactivated. C C C C C C H H H2 2 H2 H2 H (1.5) 1.2.2.3 Limitations The ketone is in most cases the most stable, and the equilibrium position There are two major limitations of 5-ALA-PDT. The first limitation is the lies far to the ketone. The enolic forms are stabilized if there is a stabilizing penetration of both light and drug into the tissue. 5-ALA has an ability to group next to the enolic hydroxyl group, as for example 2,4-pentanedione: stay superficial and not penetrate deep enough, due to its polarity. The H light penetrates, as already mentioned, deeper into human tissue at longer O O O O wavelengths and since PpIX has its long-wavelength absorption peak at C C C C 635 nm, the light penetration is not optimal. H3C C CH3 H3C C CH3 H2 H (1.6)

26 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 27

Here actually the enolic form is in excess at equilibrium (~78% of the reac- CHAPTER 2 tion mixture are in the enol form).69 The enol form is stabilized by hydro- 2 Computational Methods gen bonding to the second carbonyl group, which also can be explained by resonance stabilization: All work presented in this thesis is performed by computational methods to investigate the chemical and physical properties of 5-ALA and its derivatives. The three following calculation methods have been applied: H - H + O O O O C C C C 1. Density Functional Theory (DFT), which is derived from Quantum H3C C CH3 H3C C CH3 H H (1.7) Mechanics (QM) have been performed for the optimization of the structures and calculation of the energetic properties of small sys- 5-ALA has a more stable keto form than its two enolic forms (3enol and tems (Paper I, II, IV and V). 4enol, Figure 1.7). Jaffe et al. studied 5-ALA with 13C- and 1H NMR and 2. Molecular Mechanics (MM) and Molecular Dynamics (MD) meth- found the enolic forms not detectable, i.e. less than 0.3% of these were ods have been used to elucidate the movements of 5-ALA and its formed. Deuterium exchange rates indicate that the tautomerization is 4 derivatives in a lipid bilayer (Paper III), and in parts of the UROD times faster for the 4enol than for the 3enol in phosphate buffer at pH 6.8. Jaffe et al. also compared 5-ALA with the similar compounds 5- study (Paper VI). chlorolevulinic acid (5-CLA) and Levulinic acid (LA)70. The rate of 5-CLA 3. The third one is a combined method of QM and MM simply called is three times slower than 5-ALA for 3enol but in the same order for the QM/MM, which is used to elucidate the enzyme mechanism of 4enol, whereas LA has more than 100 times slower rates than 5-ALA. The UROD (Paper VI). use of phosphate buffer resulted in the fastest rates, which indicates that In this chapter, these methods will be explained and discussed briefly. phosphate is probably involved in the mechanism. O 2.1 Quantum Mechanics HO NH2 Quantum chemistry, which is the application of QM in chemistry, is based O 5-ALA on the Schrödinger equation.71 The most commonly used version of the Schrödinger equation is the time-independent, non-relativistic Schrödinger equation, which is generally expressed: OH OH (2.1) H2O HO NH2 HO NH2 where is the Hamiltonian operator,� Ψ the wave function, and E the en- 5-ALA-4enol 5-ALA-3enol �Ψ � �Ψ O O ergy of the system. Mathematically, Eqn. (2.1) is an eigenvalue problem, � which � means that when the Hamiltonian operator acts upon Ψ, the out- H2O H2O come is the energy of the system (the scalar E) times the same wave func- OH tion (Ψ). Ψ is hence an eigenfunction to . The Hamiltonian operator is a HO NH 2 sum of the kinetic and potential energy operators. OH O �� 5-ALA-hyd (2.2) �� Figure 1.7 The isomerisation of 5-ALA into its various tautomeric (5-ALA-3enol, If Ψ is known, the probability�� of� �finding� � �� particles at given positions in 5-ALA-4enol) and hydrated forms (5-ALA-hyd). space, is given by the integral of the square of the wave function . An arbitrary property A of the system can be calculated as an expectation� value , by the use of the property-specific operator : �|Ψ| dτ

��

28 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 29

��

28 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 29

(2.3) ordinates. By using a single determinant, the electron correlation is � �� |��|�� neglected, and the electron-electron interaction is only taken into � The denominator represents the� �normalization� � �� |�� of the wave function, and is account as an average effect.74 equal to 1 if Ψ is already normalized. • MO-LCAO approximation. For molecules, the one-electron molecular orbitals (MOs) are approximated to be a linear combination 2.1.1 Hartree-Fock (LC) of atomic orbitals (AOs): The Schrödinger equation can be solved analytically for systems with one electron and one nucleus, for example the hydrogen atom. However, ap- (2.6) proximations have to be made to solve larger systems. The Hartree-Fock � � � where c are coefficients � and � ∑ � the� AOs. The atomic orbitals are (HF) method contains a number of approximations of the many-electron i expressed as a linear combination of functions; a basis set (which is wave function (Ψ), from which many other computational methods are �� derived. The major approximations in HF are: discussed in next section). • Self consistent field. The variational principle states, that the energy • The Born-Oppenheimer approximation is the decoupling of the mo- calculated from an approximate wave function is higher than, or tions of the electrons and the nuclei. Since the mass of the smallest equal to the energy of the exact wave function. The equality holds nucleus (a proton) is 1836 times heavier than an electron, the elec- only if it is the exact wave function. Consequently, the wave func- tron movement is much faster than the movement of the nucleus. tion could be determined by an iterative manner.74 To solve a HF Hence, in relation to the electrons, the nucleus is approximately problem, the wave function is guessed, followed by the minimiza- fixed in space. The Born-Oppenheimer approximation implies that tion of the energy, to get a new and better wave function as input the electronic wave function is only dependent on the positions of for the energy minimization. This iterative procedure is called self the nuclei, but independent of their momenta.72 The nucleus-nucleus consistent field (SCF). interaction is now a constant, and could be added to the electronic Even though approximations are made, HF (and HF derived) methods are 73 energy (Eelectronic) when it has been solved. defined as ab initio (Latin for from the beginning) calculations. Ab initio indicates, in contrast to semi-empirical methods, that the calculations are performed without any input of experimental data. The ab initio methods ��Ψ�� Ψ�� (2.4) are derived from theoretical principles and the only parameters included �� are the initial coordinates of the nuclei and universal physical constants. A B �� Z Z There are variations in the HF method depending on the electron con- � � � � � � AB A�� B R figuration of the system. The restricted Hartree-Fock (RHF) method is • In the HF method, the electronic wave function is expressed in a designed for closed-shell systems, i.e. systems that only have paired elec- single Slater determinant: trons, while the Restricted Open-shell Hartree-Fock (ROHF) and Unre- stricted Hartree-Fock (UHF) allow calculations of open-shell systems.

2.1.2 Basis sets ��1� ��1� � ��1� (2.5) � � � The functions which are combined to form an AO are called a basis set. A SD 1 � �2� � �2� � � �2� Φ � � �� complete basis set is when an infinite number of functions are used. How- √�� ever, since the use of an infinite number of functions is impossible; finite where each column represents an electrons spin orbital; a product basis sets always imply an approximation of the AOs. There are mainly of a spatial single-electron wave function and a spin function two function types used for this purpose, the Slater type orbitals (STO) and ( ). The rows of the determinant represent the electron co- the Gaussian type orbitals (GTO).

�� σ�

30 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 31

�� σ�

30 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 31

The STOs are functions that mimic an atomic orbital, and have the fol- versions of s- and p-orbitals have a higher probability to find the electrons lowing mathematical form: further away from the nuclei. The 6-31+G(d,p) basis set is the 6-31G(d,p) basis set with diffuse functions on the heavy atoms, and if another + is (2.7) added, diffuse functions are included at the hydrogen atoms, as well. �� where N is a normalization��,�,ℓ,��, �,constant, �� ,��, �� the �angular part (spherical There are several more extensions and improvements (like the correla- harmonic functions), r the radii, (zeta) the orbital exponent, and n, and tion consistent (cc) basis sets), and basis sets designed for special molecules �,� m are quantum numbers. Integration � of ��, these �� functions are however not (for example LANL2DZ, which is designed for large nuclei), which are not very practical. Therefore, GTOs �have been proposed to be used to modelℓ considered in the current thesis.76 the orbitals with the following mathematical form: 2.1.3 Density Functional Theory (2.8) � ℓ� ℓ� ℓ� �� Density Functional Theory (DFT) is basically a very accurate method, �,ℓ�,ℓ�,ℓ� The main difference between� � �, STO �, �� and� �� GTO� � lies� in that the GTO is ex- which is derived from the Schrödinger equation of quantum chemistry. To pressed in cartesian instead of polar coordinates, and that the r is squared make it more practical however, some parameters are added to the DFT in the power of the exponential function. Thus GTOs are easier to inte- computations. Therefore DFT methods are usually not termed ab initio 74 grate, and the product of two GTOs gives another GTO. methods, but first principle methods. A minimal basis set is the lowest number of basis functions that can be The DFT method is based on the electron density ρ instead of the wave used for a system, where only enough functions are used to contain all the function Ψ. Compared to HF, the number of variables is reduced from 4N electrons in a neutral system. One example is the STO-3G basis set, which (three spatial and one spin variable per electron) to three spatial variables is a STO function, approximated by a linear combination (contraction) of in DFT, where N is the number of electrons74. This means that DFT calcu- three primitive GTOs per atomic orbital. lations are much faster than ab initio methods, and much larger systems A double zeta basis set is the first improvement of the minimal basis set. can be handled. The energy of the system is in DFT a functional of the For each STO in the minimal basis, another STO is included in the basis electron density E[ρ]. The definition of a functional is a function of a func- set. Similarly, a triple or quadruple zeta basis set contains three or four tion. The energy (E) is a function of the electron density ρ(r), which is a times as many functions as a minimal basis, respectively. Among double function of the positions (r). and triple zeta basis sets, the split valence basis sets are very popular. These Kohn and Sham introduced a formalism, which provides a practical way basis sets are not pure double/triple zeta, since minimal basis are used on of calculating the electronic energy in an iterative SCF manner similar to the core orbitals. For example the 6-31G basis set has the minimal basis in the HF method (though more computation-expensive with 3N variables).77 the core orbitals, each a contraction of six primitive GTOs. The valence The Kohn-Sham orbitals can be computed numerically, or be built up by a orbitals however, are each described by two contractions; the first with set of basis functions. There are special basis sets designed for DFT meth- three primitive GTOs and the second with only one primitive GTO. ods, however in comparison with conventional HF basis sets, very small or To extend the basis set further, a set basis functions with angular mo- no improvements of the results were found. It is also found that further mentum higher than the valence shell are added (e.g. d-orbitals are added enlargement of the basis sets does not enhance the results either. Therefore, to heavy atoms and p-orbitals to hydrogen atoms). These orbitals tend to conventional basis sets, for example 6-31g(d) are the most widely used in characterize the polarity of the molecules, and are therefore called polar- DFT studies.75 ized basis sets. For example, the 6-31G(d) basis set contains d-orbital func- The functional of the electron density consists of four terms; the kinetic tions on the heavy atoms and the 6-31G(d,p) basis set contains both d- energy of the non-interacting electrons (T[ρ]), electron-nucleus attraction orbitals on the heavy atoms and p-orbital functions to the hydrogen at- (Ene[ρ]), Coulomb interactions (J[ρ]) and the exchange-correlation energy oms.75 (EXC[ρ]). To describe systems where the electrons could be further away from a nucleus (for example: anions, systems with lone pairs and excited states), (2.9) diffuse functions are added to basis set. Technically, these more diffuse ܧሾߩሿ ൌ ܶሾߩሿ ൅ ܧ௡௘ሾߩሿ ൅ ܬሾߩሿ ൅ ܧ௑஼ሾߩሿ

32 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 33

The first three terms can be obtained analytically. However, the last term, (2.10) which involves non-classical contributions, needs to be calculated ap- � To make a good mathematical�� ℓ representation� � �� of a bond stretch, the proximately. Therefore, a number of different functionals in DFT have � �� Morse potential is the most accurate. However, since it is quite computa- been developed to approximate the exchange-correlation term. The elec- tional expensive, the approximate harmonic potential function is used in tron correlation referrers to that the dynamic movement of one electron is most force fields: dependent of all other electrons in the molecule. As a consequence of the Pauli principle, there is also an exchange interaction between the electrons (2.11) that imply that the electrons of the same spin are avoiding each others. �ℓ � Usually the exchange-correlation functional is divided into an exchange �ℓ � � �ℓ � ℓ�� where is the spring constant,�� the2 bond length and the reference and a correlation functional term.78(p. 316-320) bond length value. ℓ � The �angular energy is also calculatedℓ by a harmonic potentialℓ similar to 2.1.4 Hybrid methods the bond stretch (Eqn. 2.11), while the torsion angles are calculated by one A method to improve the exchange-correlation functional is to take parts or several cosine functions: of it from HF. The HF approach makes the approximation that each electron moves in a mean field created by the rest of the electrons. This ap- (2.12) proximation neglects correlation energy, but gives accurately calculated � � � exchange energy. In DFT calculations however, which are based on the � � � � �� where controls the barrier�� height,�� 2 n is the multiplicity (i.e. how many total electron density, the correlation energy is included from the begin- minima there are in 360º), the torsion angle and the phase factor (i.e. ning. It is often more accurate to use a hybrid HF-DFT method, where part � the location� of the minima). Out-of-plane bendings can be modelled as of the exchange energy is taken from HF method and the electron correla- distances to the plane, angles� to the plane or by improper� torsions, where tion energy is calculated with DFT methodology. the latter is the most common. The B3LYP functional79-81 is a hybrid method with three constants fitted The non-bonded interactions are divided into the long range electrostatic to optimize the proton affinities, atomization energies and ionization energies of a number of compounds. B3LYP has shown very good accuracy and interactions between charged particles ( ) and the van der Waals interac- stability; the errors of B3LYP lie within 2 kcal/mol.82 Therefore, the QM tions between neutral particles ( ): �� calculations in this work were performed using the B3LYP functional. �� (2.13) 2.2 Molecular Mechanics & Molecular Dynamics where the electrostatic interactions�� are �� modelled� �� by the Coulomb’s law: Molecular mechanics (MM) is a less accurate method compared to QM methods, since it is based on classical physical laws. However, in studies of NA NB (2.14) � � larger systems like lipid membranes and enzymes, this methodology is very �� useful. In MM the nuclear motions are modelled and the motion of the and the van der Waals interactions�� are�� in4πε mostr force fields modelled with electrons is ignored. It is usually not possible to model any reactions with the Lenard-Jones potential: MM when the bonds between the atoms are defined in parameters. These parameters are based on experimental data, and are collected in a force N N (2.15) 83 �� field. �� N �� The potential energy in MM (Etot(r )) – a function of the positions (r) of � � � � 4ε �� r r N particles, is usually calculated as a sum of four terms related to: bond In some force fields other so called cross terms are added, to model the lengths ( ), bond angles ( ), torsion angles ( ) and non-bonded interac- couplings between the terms in Eqn. (2.10). tions ( ), where the first three are considered as bonded interactions. Molecular dynamics (MD) is a method to simulate a system over time. �ℓ �� Newtonian mechanics and statistical physics are employed to calculate the ��

34 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 35

34 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 35 trajectory of all atoms (i.e. the movement of all atoms in the system under van der Waals interactions between the subsystems are treated at the MM a period of time) and physical properties, such as radial distribution func- level.85 tions and diffusion coefficients. An ensemble is chosen for the calculation to decide which parameters to be conserved during the simulation. The 2.4 Computational methods in the current studies most common are the NVT and NPT ensembles. In both cases the number of particle (N) and the temperature (T) are conserved. In NVT the volume 2.4.1 Paper I and II is constant, while in NPT the pressure is conserved. All calculations in of the work presented in paper I and II are performed in The forces acting on each atom and the energy of the system are calcu- the GAUSSIAN03 program package.86 As mentioned above, the B3LYP79-81 lated using a MM force field, and the positions of next time step are calcu- functional has been used throughout in the DFT calculations. The basis set 84 lated with an appropriate finite difference method. The time step in the 6-31+G(d,p) was chosen, with diffuse functions (+), since anions were stud- –15 time difference method is chosen in the magnitude of fs (10 s), which is ied, and with polarized functions (d,p) for both hydrogen and heavier ele- much shorter than the time between collisions and vibrations in the system. ments, since the protonation and hydrogen bonding was important. MM methods are used to perform molecular docking. Docking predicts The optimizations were carried out in gas phase and in bulk solvation the orientation of one molecule to another molecule. It is widely used to using the integral equation formalism of the polarizable continuum model suggest the binding of substrates or inhibitors in the enzymatic active site. (IEFPCM).87,88 In these calculations, water was used as a solvent, with the The enzyme is usually rigid during the docking process, and the molecule value 78.4 of the dielectric constant (ε). To investigate how the different to be docked could either be rigid or flexible. The structures are ranked molecules behave in a highly non-polar environment such as lipid mem- using a scoring function, and there are several scoring functions available, branes, single point calculations with IEFPCM were also carried out with taking various parameters into account, such as non-bonded interactions, ε = 4.0, in both studies. Harmonic vibrational frequency calculations were solvation effects and the torsional angles degrees of freedom. performed in aqueous and lipid environments, to obtain zero-point vibrational energies (ZPE), thermal corrections to the Gibbs’ free energies and 2.3 QM/MM method the enthalpy at the temperatures 298.15 K (P I & P II) and 310.15 K (P I)). When pure QM calculations are too heavy, but reactions with bond form- Proton affinities were determined as the difference between the ZPE- ing and bond breaking needs to be considered (which is usually the case in corrected internal energies of the protonated and non-protonated forms. studies of enzyme mechanisms), at least two different methods can be used. For proton affinity calculations, in reactions involving solvated protons, First of all the system can be pruned to a size that is possible to handle the estimated solvation energy of H+ of 267.68 kcal/mol previously estab- with QM methods. Secondly a hybrid method can be utilized. In the lished in our group,89 was employed to obtain reaction free energies. QM/MM approach, the system is divided into (at least) two parts – the The accuracy of gradient corrected DFT for the calculations of proton

QM part and the MM part. The Hamiltonian (Htot) is then divided into affinities is well studied, and generally lies within 1–7 kcal/mol in compari- three parts: son with experimental data.90,91 H = H + H + H (2.16) tot QM MM QM/MM 2.4.2 Paper III where HQM is the Hamiltonian of the QM part, HMM of the MM part and In this study MD has been used to calculate the behaviour of the solutes in the HQM/MM is the Hamiltonian of the interactions between the QM and a membrane. For this, the GROMACS program package92 and its built-in MM parts. GROMACS standard force field has been used. The membrane model used The interactions can be described in different levels in the QM/MM was downloaded from the webpage of P. Tieleman and his group.93,94 The methods. The lowest is the mechanical embedding, where all the QM–MM model consists of 64 dipalmitoylphosphatidylcholine (DPPC) molecules interactions are treated at the MM level. In the electrostatic embedding and 3846 water molecules forming a 4.9 × 4.4 × 9.4 nm periodic box. A method the electrostatic interactions of the MM part are taken into ac- short equilibration run of 1 ns was performed of the pure membrane for count in the QM calculation. The bonded interactions and the non-bonded verification.

36 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 37 trajectory of all atoms (i.e. the movement of all atoms in the system under van der Waals interactions between the subsystems are treated at the MM a period of time) and physical properties, such as radial distribution func- level.85 tions and diffusion coefficients. An ensemble is chosen for the calculation to decide which parameters to be conserved during the simulation. The 2.4 Computational methods in the current studies most common are the NVT and NPT ensembles. In both cases the number of particle (N) and the temperature (T) are conserved. In NVT the volume 2.4.1 Paper I and II is constant, while in NPT the pressure is conserved. All calculations in of the work presented in paper I and II are performed in The forces acting on each atom and the energy of the system are calcu- the GAUSSIAN03 program package.86 As mentioned above, the B3LYP79-81 lated using a MM force field, and the positions of next time step are calcu- functional has been used throughout in the DFT calculations. The basis set 84 lated with an appropriate finite difference method. The time step in the 6-31+G(d,p) was chosen, with diffuse functions (+), since anions were stud- –15 time difference method is chosen in the magnitude of fs (10 s), which is ied, and with polarized functions (d,p) for both hydrogen and heavier ele- much shorter than the time between collisions and vibrations in the system. ments, since the protonation and hydrogen bonding was important. MM methods are used to perform molecular docking. Docking predicts The optimizations were carried out in gas phase and in bulk solvation the orientation of one molecule to another molecule. It is widely used to using the integral equation formalism of the polarizable continuum model suggest the binding of substrates or inhibitors in the enzymatic active site. (IEFPCM).87,88 In these calculations, water was used as a solvent, with the The enzyme is usually rigid during the docking process, and the molecule value 78.4 of the dielectric constant (ε). To investigate how the different to be docked could either be rigid or flexible. The structures are ranked molecules behave in a highly non-polar environment such as lipid mem- using a scoring function, and there are several scoring functions available, branes, single point calculations with IEFPCM were also carried out with taking various parameters into account, such as non-bonded interactions, ε = 4.0, in both studies. Harmonic vibrational frequency calculations were solvation effects and the torsional angles degrees of freedom. performed in aqueous and lipid environments, to obtain zero-point vibrational energies (ZPE), thermal corrections to the Gibbs’ free energies and 2.3 QM/MM method the enthalpy at the temperatures 298.15 K (P I & P II) and 310.15 K (P I)). When pure QM calculations are too heavy, but reactions with bond form- Proton affinities were determined as the difference between the ZPE- ing and bond breaking needs to be considered (which is usually the case in corrected internal energies of the protonated and non-protonated forms. studies of enzyme mechanisms), at least two different methods can be used. For proton affinity calculations, in reactions involving solvated protons, First of all the system can be pruned to a size that is possible to handle the estimated solvation energy of H+ of 267.68 kcal/mol previously estab- with QM methods. Secondly a hybrid method can be utilized. In the lished in our group,89 was employed to obtain reaction free energies. QM/MM approach, the system is divided into (at least) two parts – the The accuracy of gradient corrected DFT for the calculations of proton

36 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 37

ture coupling and Parrinello-Rahman semi-isotropic pressure coupling has (2.19) been used during the simulations. The time step was set to 2 fs, and all 1 � � � � bond lengths were constrained during the simulation. � �� where the limits were approximated�� to�� –45 � and 45� Å, �� respectively. Simulations were performed on systems including one, two and four �� solute molecules; inserted in the middle of the bilayer, where the lipid den- 2.4.3 Paper IV and V sity is low. After equilibration of 2-7 ns a production run of 20 ns was performed. The densities of the solutes were calculated based on these pro- In both studies the system of the enzyme PBGS was pruned to a reduced duction simulations. part of the enzyme; containing only the substrate and parts of the two co- For each of the solutes; free energy profiles across the membrane, local valent bonding residues of the active site. diffusion constants and total permeability constants were calculated. How- The calculations were as in the studies presented in P I and P II, per- ever, the latter two could not be performed for the hexyl ester, due to formed in GAUSSIAN03,86 with B3LYP level of theory.79-81 In these studies problems in the calculations. the 6-31G(d) basis set was used. In comparison with P I and P II, a smaller The free energy, diffusion and permeability calculations in this study is basis set was chosen since the systems are slightly larger, and the diffuse based on the methodology first applied by Marrink and Berendsen.97,98 functions do not make any difference since no anions were studied. Snapshots from the equilibration and production simulations were used as Optimized geometries were obtained in gas phase, and their correspond- input structures for a number of 1 ns MD-simulations. The snapshots were ing harmonic vibrational frequencies were calculated in gas phase and in chosen in order that the distance between the solute centre of mass and the water bulk (IEFPCM and ε = 78.4). The free energy of the systems in aque- bilayer centre of mass possessed discrete steps from 0 to 45 Å (i.e. 46 snap- ous continuum was calculated by adding the thermal correction value to shots per solute). During the 1 ns simulation, this solute–bilayer distance Gibbs’ free energy obtained from frequency calculations of the system in was constrained along the bilayer normal (the z-direction). The force re- the gas phase. To validate the transition state optimizations, IRC (intrinsic quired to maintain the constraint of the solute was obtained in every MD reaction coordinate) calculations were performed.99,100 time step. The free energy difference (ΔG(z)) of solute transfer between the bulk 2.4.4 Paper VI water and a specific distance (z) to the bilayer is calculated by integration In this study the enzyme UROD was pruned to the URO-III substrate and of the average over time of force acting on the solute ( ): the surrounding active site (i.e. the first shell residues and water molecules ᇱ ௧ of the active site), and applied to the QM/MM approach. Only parts of the substrate and the side chains of the residues suggested to be involved in the ۄሺݖ ሻܨۃ ௭ (2.17) ᇱ catalytic reaction was involved in the QM part. -௧݀ݖԢ To generate input structures for the QM/MM calculations MM dockۄሺݖ ሻܨۃ ሺݖሻ ൌ െ නܩȟ ௪௔௧௘௥ The lower limit of the integral was௕௨௟௞ set to 45 Å, as a reference point for ings, minimizations and MD simulations were performed in the Molecular bulk water, which corresponds to the furthest distance in the study. Operating Environment (MOE) program.101 CHARMM22 force field was The local 1-dimention diffusion constants (D(z)) along the bilayer nor- used, since it is well parameterized for heme-like compounds.102 mal were calculated from: For the QM/MM calculations the GAUSSIAN03 program86 was used in conjunction with the ONIOM code with mechanical embedding.103-109 The LYP 79-81 ଶ (2.18) level of theory in the QM part was B3 and 6-31G(d) basis set, and ሺܴܶሻ in the MM part: AMBER94 force field.110 ሺݖሻ ൌ ஶܦ ሺݐሻ݀ݐܥ ׬଴

38 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 39

The simulations were performed with periodic boundary conditions with where R is the gas constant, T the temperature and C(t) is the autocorrela- the NPT ensemble at 1 atm and 323 K, sufficiently above the DPPC melt- tion function of the constrained force. ing point (mp = 305 K computationally95 and 315 K experimentally96). The permeability constant P, was then obtained according to: The Particle-Mesh Ewald (PME) electrostatics, Nose-Hoover temperature coupling and Parrinello-Rahman semi-isotropic pressure coupling has (2.19) been used during the simulations. The time step was set to 2 fs, and all 1 � � � � bond lengths were constrained during the simulation. � �� where the limits were approximated�� to�� –45 � and 45� Å, �� respectively. Simulations were performed on systems including one, two and four �� solute molecules; inserted in the middle of the bilayer, where the lipid den- 2.4.3 Paper IV and V sity is low. After equilibration of 2-7 ns a production run of 20 ns was performed. The densities of the solutes were calculated based on these pro- In both studies the system of the enzyme PBGS was pruned to a reduced duction simulations. part of the enzyme; containing only the substrate and parts of the two co- For each of the solutes; free energy profiles across the membrane, local valent bonding residues of the active site. diffusion constants and total permeability constants were calculated. How- The calculations were as in the studies presented in P I and P II, per- ever, the latter two could not be performed for the hexyl ester, due to formed in GAUSSIAN03,86 with B3LYP level of theory.79-81 In these studies problems in the calculations. the 6-31G(d) basis set was used. In comparison with P I and P II, a smaller The free energy, diffusion and permeability calculations in this study is basis set was chosen since the systems are slightly larger, and the diffuse based on the methodology first applied by Marrink and Berendsen.97,98 functions do not make any difference since no anions were studied. Snapshots from the equilibration and production simulations were used as Optimized geometries were obtained in gas phase, and their correspond- input structures for a number of 1 ns MD-simulations. The snapshots were ing harmonic vibrational frequencies were calculated in gas phase and in chosen in order that the distance between the solute centre of mass and the water bulk (IEFPCM and ε = 78.4). The free energy of the systems in aque- bilayer centre of mass possessed discrete steps from 0 to 45 Å (i.e. 46 snap- ous continuum was calculated by adding the thermal correction value to shots per solute). During the 1 ns simulation, this solute–bilayer distance Gibbs’ free energy obtained from frequency calculations of the system in was constrained along the bilayer normal (the z-direction). The force re- the gas phase. To validate the transition state optimizations, IRC (intrinsic quired to maintain the constraint of the solute was obtained in every MD reaction coordinate) calculations were performed.99,100 time step. The free energy difference (ΔG(z)) of solute transfer between the bulk 2.4.4 Paper VI water and a specific distance (z) to the bilayer is calculated by integration In this study the enzyme UROD was pruned to the URO-III substrate and of the average over time of force acting on the solute ( ): the surrounding active site (i.e. the first shell residues and water molecules ᇱ ௧ of the active site), and applied to the QM/MM approach. Only parts of the substrate and the side chains of the residues suggested to be involved in the ۄሺݖ ሻܨۃ ௭ (2.17) ᇱ catalytic reaction was involved in the QM part. -௧݀ݖԢ To generate input structures for the QM/MM calculations MM dockۄሺݖ ሻܨۃ ሺݖሻ ൌ െ නܩȟ ௪௔௧௘௥ The lower limit of the integral was௕௨௟௞ set to 45 Å, as a reference point for ings, minimizations and MD simulations were performed in the Molecular bulk water, which corresponds to the furthest distance in the study. Operating Environment (MOE) program.101 CHARMM22 force field was The local 1-dimention diffusion constants (D(z)) along the bilayer nor- used, since it is well parameterized for heme-like compounds.102 mal were calculated from: For the QM/MM calculations the GAUSSIAN03 program86 was used in conjunction with the ONIOM code with mechanical embedding.103-109 The LYP 79-81 ଶ (2.18) level of theory in the QM part was B3 and 6-31G(d) basis set, and ሺܴܶሻ in the MM part: AMBER94 force field.110 ሺݖሻ ൌ ஶܦ ሺݐሻ݀ݐܥ ׬଴

38 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 39

Furthermore, calculations of the proton affinities were performed of CHAPTER 3 small models of the substrate and involved residues in order to choose 3 Summary of results which reactant complex to start with. The B3LYP/6-31G(d) level of theory was used in these calculations and the IEFPCM87,88 solvating scheme was Figure 3.1 shows an overview of the studies of 5-ALA presented the current thesis. The first two papers present studies of the properties of 5-ALA used in order to compare the proton affinities at three different dielectric in solution (1). Further, 5-ALA has been studied in a cell membrane; how it constant values (ε = 4, 10 and 78.39). The latter constant is that of water, and the two former have earlier been used to model the polarity within and its esters behave and how high the barriers are to enter the cell (P III) (2). Step number (3) in the figure represents the metabolism of 5-ALA in enzymes.111 the heme biosynthesis, which has been studied in part (P IV-VI). 2.5 Computational facilities In the following sections a summary of the results described in the papers will be presented and discussed. Several computer systems have been used for these studies; in house facilities in Örebro, resources from Swedish National Infrastructure for Com- puting (SNIC) at National Supercomputer Centre (NSC) in Linköping and clusters in Canada (see Table 2.1). 1 Table 2.1 The computational facilities used in the current studies. System # of # of CPUs / node RAM/node Location nodes (GiB) 2 CELL Hydra 30 1 x s Intel P IV 2.8 GHz 0.5 Örebro Mitochondrion

Albatross 13 2 x d AMD 1 GHz 16 3 Örebro 23 2 x q AMD 2.3 GHz

Phoenix 66 1 x q Intel Xeon 2.33 GHz 4 Karlskoga

Monolith 198 2 x s Intel Xeon 2.2 GHz 2 Linköping

Neolith 6440 2 x q Intel Xeon E5345 2.3 GHz 16/32 Linköping

Sharcnet 168-384 2-4 x s/d/q 4-32 (several systems) AMD Opteron/ Intel Xeon Figure 3.1 A schematically overview of the outlines of 5-ALA-PDT and the studies Canada 2.2-2.5 GHz presented in the current thesis. 1) Investigations of 5-ALA and its derivatives in aqueous solution. 2) The penetration of 5-ALA into the cell. 3) The metabolism of s = single, d = double and q = quadruple core. 5-ALA.

3.1 QM calculations of 5-ALA and its derivatives (P I & P II) The aim of the study presented in P I was to theoretically investigate the properties of 5-ALA and three of its esters; the methyl-, ethyl- and the hexyl ester. Even if a lot of experimental and clinical studies have been

40 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 41

Albatross 13 2 x d AMD 1 GHz 16 3 Örebro 23 2 x q AMD 2.3 GHz

Phoenix 66 1 x q Intel Xeon 2.33 GHz 4 Karlskoga

Monolith 198 2 x s Intel Xeon 2.2 GHz 2 Linköping

Neolith 6440 2 x q Intel Xeon E5345 2.3 GHz 16/32 Linköping

40 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 41 performed on 5-ALA and its esters, no theoretical study of the drugs has the longer hexyl ester cation is slightly twisted. The neutral esters however been performed before. (Figure 3.3 a, c and e), have a 90 degrees twisted torsion angle between the In the second paper the reactivity of 5-ALA is described. 5-ALA is a ke- carbonyl and the carboxylic acid group. tone that can undergo keto-enol tautomerization to its enolic forms. There are four different enolic tautomers of 5-ALA, as already mentioned in chapter 1.3. The double bond can be formed either between C3 and C4 or between C4 and C5 in 5-ALA (Figure 1.3). Furthermore, for each of them the two Z and E symmetries can be formed. In addition, a hydrated form of 5-ALA is formed in solution, with a mol fraction of ~0.5%.70 P II de- scribes the theoretical study of these derivatives of 5-ALA and its intercon- vertional reaction mechanisms (Figure 1.7).

3.1.1 Structural properties The optimized structures of 5-ALA’s different protonation states are shown in Figure 3.2. The figure displays that all of them except the zwitterion have an all-trans orientation. The zwitterion (Figure 3.2c) instead has a ring structure with a hydrogen bond (1.497 Å) between the carboxylic acid and the amine moieties of the molecule.

Figure 3.3 Optimized structures of the neutral (a, c, e) and cationic forms (b, d, f) of a-b) Hexyl-, c-d) Ethyl- and e-f) Methyl 5-ALA esters.

3.1.2 Free energies According to calculated free energies, the order of the solutes’ stability in water (IEFPCM solvation) is (cf. Figure 3.4a):

� � � � – � where the5‐ALA energy �used 5‐ALA for the � � proton� 5‐ALA is specified� � �in 5‐ALAmethod� section. �� In lipid environment however, the neutral form is the most stable (Figure 3.4b). The esters are most stable in their protonated form in aqueous solution, Figure 3.2 Optimized structures of a) 5-ALA, b) 5-ALA–, c) 5-ALA+– and d) 5- whereas in lipid environment deprotonation is spontaneous by 6.1 − 7.4 ALA+. kcal/mol. The optimized structures of the neutral and cationic forms of the three 5-ALA esters studied are displayed in Figure 3.3. The methyl- and ethyl- cationic esters (Figure 3.3 d and f) also have an all-trans geometry, while

42 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 43 performed on 5-ALA and its esters, no theoretical study of the drugs has the longer hexyl ester cation is slightly twisted. The neutral esters however been performed before. (Figure 3.3 a, c and e), have a 90 degrees twisted torsion angle between the In the second paper the reactivity of 5-ALA is described. 5-ALA is a ke- carbonyl and the carboxylic acid group. tone that can undergo keto-enol tautomerization to its enolic forms. There are four different enolic tautomers of 5-ALA, as already mentioned in chapter 1.3. The double bond can be formed either between C3 and C4 or between C4 and C5 in 5-ALA (Figure 1.3). Furthermore, for each of them the two Z and E symmetries can be formed. In addition, a hydrated form of 5-ALA is formed in solution, with a mol fraction of ~0.5%.70 P II de- scribes the theoretical study of these derivatives of 5-ALA and its intercon- vertional reaction mechanisms (Figure 1.7).

Figure 3.3 Optimized structures of the neutral (a, c, e) and cationic forms (b, d, f) of a-b) Hexyl-, c-d) Ethyl- and e-f) Methyl 5-ALA esters.

3.1.2 Free energies According to calculated free energies, the order of the solutes’ stability in water (IEFPCM solvation) is (cf. Figure 3.4a):

42 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 43

The hydrolysation free energy of the non-charged esters according to fol- 50 lowing reaction: A 45 5ALA-hyd 4-enol E 2 40 4-enol Z range from thermoneutralX‐�‐�� H O for� �‐�� the methyl � X‐OH ester, X� �� to be�� H� spontaneous by 3-enol E 3-enol Z 5.5 kcal/mol for the hexyl ester. Hence, there is a clear relation between 35 5ALA ester chain length and hydrolysation free energy. 30 As shown in Figure 3.4, all the 5-ALA enolic forms are thermodynamically less stable than the keto form in both water and lipid environment. 25 The stability of the protonation state of the enols follows the same pattern as 5-ALA, with two exceptions; the zwitterion of the Z 3enol isomer is G (kcal/mol) G 20 Δ slightly more stable than the neutral form in aqueous solution and the 15 cation of 5-ALA-hyd is more stable than its neutral form in lipid environment. 10 The enols are on the same energy level, differing with only a few 5 kcal/mol, not following any clear pattern. The hydrated forms are less thermodynamically stable than all the enols in both lipid and aqueous envi- 0 ronment, except the cation in aqueous solution, which is slightly more cation neutral zwitterion anion stable than the most stable enol (Figure 3.4a).

65 3.1.3 Proton affinities B 60 5ALA-hyd The calculated proton affinities (PA) in vacuum are in very well agreement 4-enol E with theoretical and experimental data of the similar compounds: alanine, 55 4-enol Z 3-enol E glycine and levulinate as shown in Table 3.1. The first column of data lists 3-enol Z 50 5ALA the protonation of the nitrogen atom, and the second data column, the protonation of the carboxylic oxygen atom. The difference between the PA 45 of oxygen and nitrogen in gas phase is large, but in solution they are much 40 closer (Table 3.2). This is expected since ions are much more stable in solution than in gas phase. The 5-ALA esters are stronger proton bases than 5-

G (kcal/mol) G 35

Δ ALA itself in gas phase but weaker bases in aqueous and lipid environment 30 (Table 3.2). As seen in Table 3.2 and Figure 3.5, all solutes except 5-ALA-hyd+− are 25 more likely to be protonated to their cationic form in water than in lipid, 20 while the anions take up a proton in lipid much easier than in water. This is a predicted result since ionized compounds are not stable in non-polar 15 fluids. It is also found that the neutral and zwitterionic forms of 5-ALA- cation neutral zwitterion anion hyd are more likely protonated than any of their corresponding 5-ALA or Figure 3.4 Mass balanced relative free energies (kcal/mol) in A) water and B) lipid 5-ALA enol compounds. Finally, the enols (in particular their anions) with + environment. The energy of 5-ALA (aq) + H2O (aq) is set to zero. E symmetry have slightly higher PAs than their corresponding Z enols, with a few exceptions. Thus, the E symmetry enols are generally stronger bases than the Z enols.

44 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 45

G (kcal/mol) G 35

44 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 45

Table 3.1 Proton affinities (kcal/mol) of 5-ALA in comparison with similar compounds in vacuum. Table 3.2 Calculated proton affinities (kcal/mol) of 5-ALA and Compound Method Nitrogen Oxygen its esters in gas phase, aqueous and lipid environment. Atom Vacuum Water Lipid ‡ 5-ALA Theoretical 213.8 330.3 5-ALA N 213.8 279.4 265.1 Glycine Experimental 211.6 a 342.4 e 5-ALA+− O – 280.9 270.5 b 212 5-ALA− O 330.3 279.6 292.5 c 208.2 N – 277.8 286.0 d 211.9 Me-5-ALA N 217.8 276.8 262.0 Theoretical 211 f Et-5-ALA N 217.1 277.5 262.7 g 211.5 He-5-ALA N 225.6 276.5 261.0

Alanine Experimental 214.8 a 340.7 e 212.3 b 212.2 c 3.1.4 Tautomerization mechanism 215.5 d The tautomerization process of the Z isomers of 5-ALA-3enol and 5-ALA- 4enol to the ketone form (5-ALA) has been studied, as well as the hydra- Theoretical 215.5 g tion of 5-ALA to 5-ALA-hyd. The results show that the tautomerization of Levulinate Experimental 340.7 h 5-ALA is a hydrogen transfer reaction; due to the fact that the Mulliken charge on the hydrogen atom hardly change during the process. ‡ 112 113 114 This work; also displayed in Table 3.2. References: a: , b: , c: , d: Three mechanisms were tested for both the 3enol and the 4enol; direct 115, e: 116, f: 117, g: 91, h: 118. transfer, transfer via a bridging water molecule, and a self-catalysed

320 mechanism. The direct transfer was the least favourable mechanism with an activation energy of 68.2 and 57.6 kcal/mol, respectively. The bridging neutral (N) water neutral (N) lipid water molecule did almost halve the barriers to 36.5 and 30.9 kcal/mol, 310 zwitterion (O) water zwitterion (O) lipid respectively. Anyway, the most probable mechanism was the self-catalysed anion (O) water anion (O) lipid one (shown in Figure 3.6), with the rate-determining activation energy of 300 anion (N) water 10.8 and 15.1 kcal/mol for the 3enol and the 4enol, respectively (Figure anion (N) lipid 3.7). The carboxylic acid oxygen picks up the enol hydrogen (TS1) and H (kcal/mol) H

Δ 290 moves it like a ‘crane’ (TS2) to the carbon atom in question (C3 or C5). The hydration process of 5-ALA has a high activation energy of 35 kcal/mol even with an explicit water molecule included in the model. This 280 is not in agreement with experimental data that show that 5-ALA-hyd is more common than the enolic forms.70 An explanation could be the pH 270 Proton affinty, - affinty, Proton dependence of the reactions or the influence of the phosphate buffer used in the experiments. 260

5ALA 3enol (E) 3enol (Z) 4enol (E) 4enol (Z) 5ALAhyd Figure 3.5 Proton affinities of the solutes in water and in lipid environment.

46 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 47

5ALA 3enol (E) 3enol (Z) 4enol (E) 4enol (Z) 5ALAhyd Figure 3.5 Proton affinities of the solutes in water and in lipid environment.

46 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 47

1.286 a b 1.388 1.499 70 1.072 1.032 68.2 A Self-catalyzed 1.339 60 Water bridge 1.362 1.348 Direct transfer 1.347 50 5-ALA-4enol+− 5-ALA-3enol+− 40 36.5 0.976 0.979 30 2.432 2.404 1.294 20 2.301 2.696 14.8 10.8 10.3 10 1.303 1.375 0.0 1.368 0 TS1self-cat 4enol -6.4 self-cat 3enol Freeenergy (kcal/mol) TS1 -10 -8.8 0.991 1.275 1.010 -20 2.077 1.894 +- 5ALAenol TS1 Intermed TS2 5ALA+-

1.294 60 1.399 57.6 1.375 B Self-catalyzed 50 Water bridge IM1self-cat 4enol IM1self-cat 3enol Direct transfer 40 1.229 1.178 30.9 1.421 1.461 1.251 30

1.259 20 15.1 1.417 1.440 10.0 9.9 self-cat 3enol 10 TS2 TS2self-cat 4enol 0 0.0

1.496 -8.9 1.496 -10 1.518 Freeenergy (kcal/mol) -19.5 1.518 1.222 -20 1.222 -30 +- +- 5ALAenol TS1 Intermed TS2 5ALA

+− 5-ALA 5-ALA+− Reaction coordinate Figure 3.7 Relative reaction free energy potential energy surface of the tautomeriza- +− +− Figure 3.6 The tautomerization mechanism a) 5-ALA-3enol+− to 5-ALA+− b) 5- tion of A) 5-ALA-3enol , and B) 5-ALA-4enol (the energies are given in ALA-4enol+− to 5-ALA+−. (Distances are shown in Å). kcal/mol).

48 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 49

1.294 60 1.399 57.6 1.375 B Self-catalyzed 50 Water bridge IM1self-cat 4enol IM1self-cat 3enol Direct transfer 40 1.229 1.178 30.9 1.421 1.461 1.251 30

1.259 20 15.1 1.417 1.440 10.0 9.9 self-cat 3enol 10 TS2 TS2self-cat 4enol 0 0.0

1.496 -8.9 1.496 -10 1.518 Freeenergy (kcal/mol) -19.5 1.518 1.222 -20 1.222 -30 +- +- 5ALAenol TS1 Intermed TS2 5ALA

48 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 49

3.2 MD simulations of 5-ALA and its esters in membrane (P III) To elucidate the permeability of 5-ALA and its methyl-, ethyl- and hexyl esters through cell membranes, MD simulations have been performed of them within a model membrane consistent of a bilayer of DPPC lipid molecules (Figure 3.8). It has previously been proved that various transfer processes are involved in the transport of 5-ALA,119-123 but passive diffusion may play a role in the transport of the 5-ALA esters.119 The current study focuses on the passive diffusion of the solutes over the membrane bilayer. A B

Water

Head groups

Lipid tails

Figure 3.9 Free energy profiles in kJ/mol of zwitterionic and neutral 5-ALA, and Head groups methyl-, ethyl- and hexyl 5-ALA esters.

The free energy profile of the membrane penetration was calculated for the Water systems and is plotted in Figure 3.9 against the distance to the bilayer centre of mass. All solutes have similar curve shapes; a well at 1–2 nm from

the bilayer centre of mass and a local energy maximum in the middle. This

means that all solutes are attracted by the polar lipid head groups (with Figure 3.8 A unit cell of the equilibrated membrane with A) one neutral 5-ALA highest density at 1–2 nm), and have varying difficulties of passing the molecule, and B) one neutral He-5-ALA molecule. most apolar, but less dense, middle of the membrane. The zwitterionic 5-ALA molecule has a much higher maximum in the After equilibration, the molecules are located at different depths into the middle than the neutral 5-ALA (66 and 33 kJ/mol, respectively). There is lipid bilayer, where they stay during the production simulation. The neu- no apparent connection between ester length and barrier height, since Me- tral 5-ALA is most of the time found in the interface, between the lipid-tail 5-ALA has the lowest barrier of all compounds (26 kJ/mol) and Et-5-ALA and the polar head-group regions (Figure 3.8a). The more hydrophilic has a higher barrier than He-5-ALA (49 and 39 kJ/mol, respectively). zwitterion is predominately located in the head-group region, which is The permeability constants of 5-ALA, the methyl- and the ethyl ester expected since the zwitterion is doubly charged. The esters are more dis- were calculated (Table 3.3). The zwitterion has the lowest permeability tributed in the membrane, between the head-group and the lipid-tail re- constant; six orders of magnitude lower than neutral 5-ALA. The perme- gions. The two esters with the longest chains have their polar part pre- ability of 5-ALA and Me-5-ALA are in the same magnitude, with a slightly dominantly in contact with the polar head groups and its apolar part di- higher permeability of the ester. The ethyl ester is not penetrating as fast as rected to the middle of the membrane (cf. Figure 3.8b). the methyl ester, since it has both a broader well in the head-group region, and a higher maximum in the middle of the membrane.

50 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 51

Water

Head groups

Lipid tails

Figure 3.9 Free energy profiles in kJ/mol of zwitterionic and neutral 5-ALA, and Head groups methyl-, ethyl- and hexyl 5-ALA esters.

the bilayer centre of mass and a local energy maximum in the middle. This

50 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 51

the Schiff base formation step, some rearrangements have to be done, but Table 3.3 Calculated permeability coefficients (P). their barriers are not high enough to affect the rate of the reaction.

Solute P / cm s−1 5-ALA+− 6.44 × 10–5 5-M A2MWH+ 5-ALA 1 HO 1.89 × 10 5-MH+ Me-5-ALA 5.28 × 101 AMWH+ O –3 5-2MH+ Et-5-ALA 7.45 × 10 AMW H O 2 To fully understand how the ester chain length affects the permeability, AM further investigations have to be performed with a wider spectrum of es- NH O 2 ters. On the basis of present results, it is not clear where the trend is broken. According to experimental studies, both the methyl- and the hexyl + H N LysP-site ester diverge from the linearity of increased ester chain length vs. penetra- NH 2 A2MWH+ 3 tion of the molecule. Cells in vitro produce less PpIX from the methyl ester 5-M than from 5-ALA, although the picture changes to show an increase from A2MH+ 5-MH+ ethyl to pentyl ester in PpIX levels. For hexyl and longer ester chains the 5-2MH+ LysA-site produced PpIX levels once again becomes lower in level.124,125 5-2MH+ Aa2MH+ 3.3 Enzymatic reactions (P IV-VI) Figure 3.10 A schematic picture of how the system is pruned in the study of the The aim of the third part of the current work is to study the metabolism of Schiff base formation. The atoms within the white box are included in all systems, 5-ALA. The story does not end when the drug molecules finally have found whereas the atoms in the light grey boxes are included in the listed systems beside their way to the target cells and penetrated their cell membranes. To be- each box and the atoms in the dark grey areas were not included in any system. come photoactive, eight 5-ALA molecules are combined together to protoporphyrin IX (PpIX) within the cell. This process follows the heme bio- The calculations were performed on small model systems, where the lysine synthesis and is catalysed by in total six enzymes in the cytosol and mito- residues were pruned to a methylamine [CH3NH2], and the substrate mole- chondria (cf. Figure 1.4). Two of these enzymes were studied; Porphobili- cule (5-ALA) was either included as a whole or pruned to aminoacetone nogen synthase (PBGS) and Uroporphyrinogen-III decarboxylase (UROD). [NH2CH2C(=O)CH3] or acetone [CH3C(=O)CH3]. The content of each system is schematically shown in Figure 3.10. Both neutral and protonated 3.3.1 The mechanism of PBGS (PIV-V) systems were studied as well as the effect of an explicit water molecule. The names of the systems in Figure 3.10 are based on their content; 5-ALA The PBGS enzyme catalyses the asymmetrical condensation of two 5-ALA (5-), aminoacetone (Aa), acetone (A), methylamine (M), water (W) and molecules. The active site contains, as already mentioned in chapter 1, two protonated (H+). catalytic lysine residues. The two 5-ALA substrates bind to each of the two The results show that an explicit water molecule catalyses the carbi- lysines by Schiff base linkages. The study of the formation of a Schiff base nolamine formation step of the neutral system and halves the barrier from is described in P IV and the mechanism of the cyclization reaction of PBGS 36.6 to 18.1 kcal/mol (Figure 3.11). is described in P V. The energies are given as free energies in aqueous solu- In the protonated systems, the carbinolamine formation barrier is low- tion at 298.15 K and 1 atm ( ) unless otherwise noted. ered to 22.3–26.3 kcal/mol when the amino group of the substrate is acting ଶଽ଼ ௔௤ 3.3.1.1 The Schiff base formation߂ܩ as a proton transporter (5-MH+, 5-2MH+, Aa2MH+). However, when a second lysine residue transfer the proton from the nucleophilic amine to The Schiff base formation is divided into two steps; the carbinolamine the carbonyl oxygen atom (A2MH+ and A2MW+), the barrier is further formation and the Schiff base formation. To orient the carbinolamine for reduced to 16.2–20.1 kcal/mol (Figure 3.12).

52 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 53

the Schiff base formation step, some rearrangements have to be done, but Table 3.3 Calculated permeability coefficients (P). their barriers are not high enough to affect the rate of the reaction.

52 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 53

TS1AMW Ea = 18.1 TS35-M Ea = 14.1 The Schiff base formation barrier in the neutral systems is halved by a bridging water molecule (from 48.7 kcal/mol of the AM system to 26.2 kcal/mol of the AMW system), but even more reduced by the catalysis of the carboxylic acid moiety of 5-ALA (5-M system) to 14.1 kcal/mol (Figure 3.11). Water catalyse the Schiff base formation of the protonated system (AMWH+) with a barrier of 21.3 kcal/mol, while 5-ALA amine catalysis (5-MH+) reduces the barrier to 9.0 kcal/mol. However, the addition of a second lysine (5-2MH+, A2MWH+, Aa2MH+ and A2MH+) reduces the Figure 3.11 The neutral systems. The lowest barriers for the various mechanisms of barrier even further to 6.0–7.5 kcal/mol (Figure 3.12). (left) the carbinolamine formation (TS1) and (right) the Schiff base formation step Hence, we can conclude that the second lysine is the best catalyst for (TS3). The free energy activation energies are given in kcal/mol. both the carbinolamine formation and the Schiff base formation steps.

3.3.1.2 Schiff base transfer

TS1A2MH+ Ea = 16.2 TS15-MH+ Ea = 22.3 It has been proposed that the 5-ALA substrate is first bound to the A-site lysine, followed by a transfer of the Schiff base to the P-site lysine, so that a second 5-ALA molecule can bind to the A-site (cf. Figure 1.5).45 We therefore studied this transfer process of a neutral and a protonated model system. The system was pruned in the same way as in the Schiff base formation study (P IV). The input structures include 5-ALA bound with a Schiff base to a methylamine representing the A-site lysine, and a second methylamine as a model for the free P-site lysine. In the neutral system the nucleophilic attack of the P-site lysine and the proton transfer from P-site amine to the deprotonated A-site imine is found to be a one-step reaction. This process is catalysed by the carboxylic acid of 5-ALA, with a barrier of 13.0 kcal/mol. From the now formed aminal TS3AMWH+ Ea = 21.3 TS35-MH+ Ea = 9.0 TS35-2MH+ Ea = 6.0 (two amines bonded to the same carbon), the cleavage of the A-site C–NH bond is attained by the reversed reaction, as shown in Figure 3.13a as a mirror of TS1, with a barrier of 7.2 kcal/mol. 25 12 A B gas phase 10 IM2 water 20 TS1 TS2 "TS2" "TS1" 8

15 TS1 "TS1" (kcal/mol)

6 (kcal/mol)

"IM1' "

energy 4 IM1' 10 IM1 IM1

energy "IM1"

2 Relative RC PC 5 Relative gas phase Figure 3.12 The protonated systems. The lowest barriers for the various mecha- 0 water nisms of (upper row) the carbinolamine formation (TS1) and (bottom row) the AC PC 0 ‐2 Schiff base formation (TS3). The free energy activation energies are given in Figure 3.13 Relative potential energy surfaces in gas phase (ΔEel in red) and in kcal/mol. water ( in green) given in kcal/mol for the neutral (A) and protonated (B) system of theଶଽ଼ Schiff base transfer. ߂ܩ௔௤

54 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 55

3.3.1.2 Schiff base transfer

15 TS1 "TS1" (kcal/mol)

6 (kcal/mol)

"IM1' "

energy 4 IM1' 10 IM1 IM1

energy "IM1"

54 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 55

With a protonated A-site Schiff base the P-site lysine can attack the A- 3.3.1.3 Cyclization reaction mechanism site without the need of a proton transfer. This process has a barrier of 7.4 When the two substrates are bound to the active site, the cyclization reac- kcal/mol. However, a proton needs to be transferred between the amines in tion is catalysed. Three different mechanisms were compared in the study. order to break the A-site C–NH bond. This proton transfer can be cata- The differences in these mechanisms lie in which order the intersubstrate lysed by the amine of 5-ALA. The proton is transferred from A-site nitro- bonds are formed. According to the first path, the C–N bond is formed gen to 5-ALA amine by a cost of 6.9 kcal/mol. The cleavage of A-site C– prior to the C–C bond (Figure 3.14). The second and third paths suggest NH bond is as in the neutral system attained by the mirrored reaction that the C–N bond is formed after the C–C bond. In the second path the (Figure 3.13b) which has the corresponding barriers 1.2 and 3.3 kcal/mol, C–N bond is formed subsequent to the C–C bond formation (Figure 3.15), respectively. whereas in the third path the enzyme substrate linkage is broken prior in between the two intersubstrate bond formations (Figure 3.16). OH HO O O O O OH H OH H N N + CH3 H N CH N 2 3 + H N H H N NH 2 2 H2N CH H C 3 3 1A 1B 1C

HO O O HO O O 1A 2B 2C

H OH H OH + H N NH

+ CH CH N 3 N 3 H H2N H N H H N 2 H N 2 2 CH CH 3 3

1C 1D 1E

2C 2D 1E

Figure 3.15 The Path 2 mechanism of PBGS, where the C–C inter-substrate bond is formed first.35,36,44-46 The last steps (1E-1H) follow the same mechanism as Path 1 in Figure 3.14.

1E 1F 1G The potential energy surfaces of the minimized local minima of the three mechanisms are plotted in Figure 3.17. The transition state calculations show that the highest barrier of path 1 is the first hydrogen transfer step (1C–1D) with 20.1 kcal/mol. The highest barrier in path 2 is the formation of the C–C bond (2B-2C), which is the first intersubstrate bond formed. The activation energy of this step is 19.4 kcal/mol. However, some of the proton transfer barriers are not investigated, and therefore we can not tell 1G 1H for sure if these are the highest barriers. Anyway, the calculated activation

energies are in good agreement with experimental activation energy inves- Figure 3.14 The Path 1 mechanism of PBGS, where the C–N inter-substrate bond tigations, which found a barrier of 18.4 kcal/mol.127 is formed first.40,126

56 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 57

HO O O HO O O 1A 2B 2C

H OH H OH + H N NH

+ CH CH N 3 N 3 H H2N H N H H N 2 H N 2 2 CH CH 3 3

1C 1D 1E

2C 2D 1E

Figure 3.15 The Path 2 mechanism of PBGS, where the C–C inter-substrate bond is formed first.35,36,44-46 The last steps (1E-1H) follow the same mechanism as Path 1 in Figure 3.14.

56 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 57

2C 3D 3E

O OH HO

O 1 Path 2 Path Z 3 Path E 3 Path NH + H 2 N H N H N CH 2 2 H 3 CH 3 3E 3F 1G

Figure 3.16 The Path 3 mechanism of PBGS, where the C–C inter-substrate bond is formed first, but the P-site enzyme-substrate bond is broken prior to the cyclization step.32,34,38 The first steps (1A–2C) follow the same mechanism as Path 2 (Figure 3.15) and the last step (1G–1H) the same as Path 1 (Figure 3.14).

The third path follows the same steps as path 2 with the C–C bond formation barrier mentioned above. The unique steps of the third path have very low calculated barriers. However, the activation energy of the 2C-3D step was not calculated, which is known to be a very endergonic step (10.9 and 17.0 kcal/mol for the formation of 3D with E and Z symmetry, respectively). In addition, the proton transfer of the 2C–3D step proceeds via a Reaction coordinate Reaction four-membered ring transition structure if there is not another residue that can catalyse the proton transfer. Maybe this would be possible with a catalysing active site water molecule present. Since the highest barrier of path 1 is a hydrogen transfer, it is probable that this step can be catalysed by a base (e.g. the A-site lysine), which can reduce this barrier. Furthermore, many of the steps are less endergonic in A B B' C D D' E F F' G H path 1, compared to the other two. Therefore, we believe that this path is the most probable. However, the effect of the surrounding active site cavity and the zinc ion is not considered in this study, which can influence the 0 50 40 30 20 10

barriers and the geometries. -10 -20 (kcal/mol) energy Free

298 Figure 3.17 Potential energy surfaces in aqueous solution (ΔGaq ) of Path 1 (―black), Path 2 (- - -red) and Path 3 Z (– – blue) and E symmetry (- – - green).

58 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 59

2C 3D 3E

O OH OH

O 1 Path 2 Path Z 3 Path E 3 Path NH + H 2 N NH NH CH 2 2 H 3 CH 3 3E 3F 1G

barriers and the geometries. -10 -20 (kcal/mol) energy Free

298 Figure 3.17 Potential energy surfaces in aqueous solution (ΔGaq ) of Path 1 (―black), Path 2 (- - -red) and Path 3 Z (– – blue) and E symmetry (- – - green).

58 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 59

3.3.2 The mechanism of UROD (PVI) higher proton affinity than the DC2 before the decarboxylation in water (ε = The enzyme UROD catalyses the decarboxylation of the acetyl side chains 78.39), but lower values when the dielectric constant is reduced. If the of URO-III to CP-III. The mechanism of the first decarboxylation of the pyrrole amine is coordinated by hydrogen bond to Asp86 the difference in ring D acetyl has been investigated in the current study. PA is further increased in the calculations with lower dielectric constants. The initial docking study generated 30 structures of the enzyme…URO- This means that the DC2 is able to take the proton from Arg37 in the active III systems. The ones of these that were mechanistically relevant can be site of UROD. In the last step in Figure 1.6, since the PA of DC2 of the divided into three groups based on the acidic residue coordinated to the product is lower than the guanidinium at all dielectric constants DC2 is also acetate on ring D. The three residues are: i) Arg50, ii) Tyr164 and iii) able to donate the proton back to Arg37. His339. The top scored structures in each group was chosen for further The PA was calculated for the structure that corresponds to the protona- investigations as conformer I, II and III, respectively (Figure 3.18). tion of DC3' after the decarboxylation step (third structure in Figure 1.6). In It was found that conformer I had the strongest interaction energy to the comparison with the PA of arginine, it is found that that Arg50 is able to enzyme (−246.3 kcal/mol), while conformer II and III had weaker interac- act as the acid HB and protonate DC3'; independent of the dielectric con- tions by 14.5 and 22.8 kcal/mol, respectively. stant when the pyrrole ring is coordinated to the acetate. Large system QM/MM calculations were performed on conformer I in order to calculate the energy barriers for the reaction. Conformer I was chosen on the following basis: a) conformer I is proven to have the strongest interaction energy, b) conformer II and III are questioned since experimental studies have found that Tyr164 and His339 are not essential for the enzymatic mechanism,49,50 c) the PA discussion above show that Arg37 and Arg50 are good candidates to be the catalysing residues.

Figure 3.18 Schematic illustration of the three active site-bound substrate confor- 20 mations: I, II and III. A and P denotes the substrates acetate and propionate 0.0 13.7 0.9 ‐5.3 groups, respectively and substrate…enzyme/solvent interactions are indicated by 0 dashed lines. ‐20 (kcal/mol)

proton of Arg37 and DC2 during the MD simulations were calculated to Relative ‐81.5 2.41, 2.40 and 2.97 Å for conformer I, II and III, respectively. This means ‐100 that Arg37 is more likely to catalyse this step than water, and at least the former two distances are in the range of a hydrogen bond. RC TS1 I1 TS2 I2 TS3 P To examine if Arg37 is a good candidate to act as the acid HA in Figure Figure 3.19 Relative potential energy surface for the enzymatic reaction of 1.6, the active site environment effect on the proton affinities (PAs) of UROD. ଶଽ଼ ௔௤ Arg37 and DC2 was investigated. The PAs were calculated for models rep- ߂ܩ resenting Arg37 and the pyrrole moiety, in a bulk solvation with the dielec- The barriers of the reaction are shown in Figure 3.19. The TS1, which is tric constants set to 78.39, 10.0 and 4.0. The results show that arginine has found to be the rate-determining step with a barrier of 13.7 kcal/mol, re-

60 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 61

It has earlier been suggested that the solvent is acting as the acid in the first ‐40 step of the mechanism (Figure 1.6).50 However, the distance between C2 of energy pyrrole D (DC2) and the closest water molecule is in the three conformers ‐60 between 5.82 and 7.31 Å. Therefore, we conclude that there must be an- ‐78.4 other catalysing acid. The average distances between the a guanidinium ‐80 ‐91.5 proton of Arg37 and DC2 during the MD simulations were calculated to Relative ‐81.5 2.41, 2.40 and 2.97 Å for conformer I, II and III, respectively. This means ‐100 that Arg37 is more likely to catalyse this step than water, and at least the former two distances are in the range of a hydrogen bond. RC TS1 I1 TS2 I2 TS3 P To examine if Arg37 is a good candidate to act as the acid HA in Figure Figure 3.19 Relative potential energy surface for the enzymatic reaction of 1.6, the active site environment effect on the proton affinities (PAs) of UROD. ଶଽ଼ ௔௤ Arg37 and DC2 was investigated. The PAs were calculated for models rep- ߂ܩ resenting Arg37 and the pyrrole moiety, in a bulk solvation with the dielec- The barriers of the reaction are shown in Figure 3.19. The TS1, which is tric constants set to 78.39, 10.0 and 4.0. The results show that arginine has found to be the rate-determining step with a barrier of 13.7 kcal/mol, re-

60 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 61 fers to the proton transfer from Arg37 to DC2. TS2 refers to the decarboxy- CHAPTER 4 lation with the concomitant proton transfer from Arg50 to DC3', which is 4 Conclusions and future perspectives actually found to be spontaneous. TS3 refers to the proton transfer of the 5-ALA and Me-5-ALA are now widely used in photodynamic therapy as initially transferred proton back from DC2 to Arg37. The rate-determining prodrugs in treatment of actinic keratosis and basal cell carcinoma8. The barrier agrees well with experimental results of various species, which are found in the range of 2.0–12.3 kcal/mol.51,128 aim of the studies presented in this thesis was to find out more about the properties of 5-ALA and its derivatives by the use of various computational techniques such as DFT, MM, MD and QM/MM. This research has led to more and deeper knowledge on how 5-ALA is acts in aqueous solution and in a lipid membrane, and parts of its biosynthesis has been elucidated. The main results are:

• The studies of the stability of the various protonation states of 5- ALA show that the protonated form is the most stable in aqueous solution (P I). • The tautomerization reaction of 5-ALA follows a self-catalysed mechanism with a barrier of 15 kcal/mol (P II). • The free energy profiles and permeability constants show that 5- ALA and its methyl ester are diffusing fastest trough a lipid bilayer (P III). • The reaction mechanisms of PBGS and UROD, where various pathways were compared. The activation energies were found to be in good agreement with experimental data (P IV-VI). Of course, much more research can be done in this field to find new derivatives with enhanced effectiveness in their drug delivery. More specifically, the studies of the transport over lipid bilayers can be enhanced to involve a wider spectrum of molecules. Furthermore, transport mechanisms with for example transmembrane proteins can be studied. Computational methods are powerful tools to elucidate the mechanisms of enzymes, and the results are found to be in good agreement with experiments. To evaluate the effects of the active site cavity and the zinc ion, larger models of PBGS with QM/MM methodology can be used. Further- more, the following three decarboxylation steps in UROD are under further studies. Finally, there are still several other heme enzymes for which the enzymatic reaction is not fully understood. Their mechanisms could also be further investigated by computational methods.

62 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 63 fers to the proton transfer from Arg37 to DC2. TS2 refers to the decarboxy- CHAPTER 4 lation with the concomitant proton transfer from Arg50 to DC3', which is 4 Conclusions and future perspectives actually found to be spontaneous. TS3 refers to the proton transfer of the 5-ALA and Me-5-ALA are now widely used in photodynamic therapy as initially transferred proton back from DC2 to Arg37. The rate-determining prodrugs in treatment of actinic keratosis and basal cell carcinoma8. The barrier agrees well with experimental results of various species, which are found in the range of 2.0–12.3 kcal/mol.51,128 aim of the studies presented in this thesis was to find out more about the properties of 5-ALA and its derivatives by the use of various computational techniques such as DFT, MM, MD and QM/MM. This research has led to more and deeper knowledge on how 5-ALA is acts in aqueous solution and in a lipid membrane, and parts of its biosynthesis has been elucidated. The main results are:

62 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 63

Acknowledgements First of all I would like to thank Professor Leif Eriksson, who has been my supervisor during these four years of PhD studies. Thank you for all your help and support and thanks for helping me to focus on the right things. Thank you Dr James Gauld, for that you have generously shared your knowledge in the field of enzyme catalysis and have given me so much of your time and effort in supervision. Thanks also for your (and the rest of the group) hospitality and all your support during my stay at your research group in Windsor, (ON) Canada. A special thanks to Eric Bushnell and Dr Jorge Llano for introducing me to calculations with QM/MM. Thank you MD Lennart Löfgren for your knowledge on the clinical field and for interesting and improving discussions about 5-ALA. A special thanks to Dr Daniel dos Santos for supervision in the lipid membrane studies. Thanks for being available on e-mail and instant mes- saging, not hesitating to answer my questions. I am very thankful to my workmates in the research group! I wish to give a special thank to my friends in PhD and MSc projects: Klefah, Emma, Li, Viarja, Magnus, Ann-Louise, Min, Boxue, Samuel and Ismael. Thank you postdocs and seniors for help and discussions about work: Yaoquan, Dragan, Patricia, Oles, Salama, David, Sofia and Rubo. Thanks all others in the department and especially you within the Life Science research centre and Modelling and Simulation centre. A special thanks to docent Jana Jass, the head of the Life science research school. Thanks Lisa, Ann, Klefah, Emma, Yaoquan, Eric and Dragan for your help in proof reading the thesis! I would also like to acknowledge the faculty of Science and Technology of Örebro University, the Modelling and Simulation Centre (MoS), the Swedish Science Research Council (VR) and the Swedish Chemical Society for financial support. I also acknowledge the National Supercomputer Centre (NSC) in Linköping and SHARCNET for generous grants of computing time. Jag skulle även vilja tacka min fru Lisa för din kärlek och att du stått ut med min arbetsbörda speciellt under de senaste månaderna. Tack mamma och pappa för all kärlek och uppskattning. Tack alla mina vänner här i Örebro, och ett speciellt tack till er i Brickebergskyrkan – tack för den goda gemenskapen och era böner! Slutligen, tack Gud för din nåd.

64 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 65

References 1. Nobel Lectures, Physiology or Medicine 1901-1921, Amsterdam: Elsevier Publishing Company, 1967. 2. Allison, RR, Mota, HC and Sibata, CH, Clinical PD/PDT in North America: An historical review, Photodiagnosis and Photodynamic Therapy, 2004 1(4):263-77. 3. Mitton, D and Ackroyd, R, A brief overview of photodynamic therapy in Europe, Photodiagnosis and Photodynamic Therapy, 2008 5(2):103-11. 4. Dougherty, TJ, Gomer, CJ, Henderson, BW, Jori, G, Kessel, D, et al, Photodynamic therapy, Journal of Natlional Cancer Institute, 1998 90(12):889-905. 5. Bown, SG, Rogowska, AZ, Whitelaw, DE, Lees, WR, Lovat, LB, et al, Photodynamic therapy for cancer of the pancreas, Gut, 2002 50(4):549-57. 6. Moore, CM, Pendse, D and Emberton, M, Photodynamic therapy for prostate cancer - a review of current status and future promise, Nat Clin Pract Urol, 2009 6(1):18-30. 7. Moor, ACE, Ortel, B and Hasan, T, Mechanisms of photodynamic therapy In: Patrice, T. (editor) Photodynamic Therapy, RSC Pubishing, 2003; p. 19-57. 8. O'Connor, AE, Gallagher, WM and Byrne, AT, Porphyrin and Nonporphyrin Photosensitizers in Oncology: Preclinical and Clinical Advances in Photodynamic Therapy, Photochem. Photobiol., 2009 85(5):1053-74. 9. Chang, R, Physical Chemistry for the Chemical and Biological Sciences, University Science Books, 2000. 10. Patrice, T, Photodynamic Therapy, RSC Pubishing, 2003 Comprehensive Series in Photochemical & Photobiological Sciences. 11. Robertson, CA, Evans, DH and Abrahamse, H, Photodynamic therapy (PDT): A short review on cellular mechanisms and cancer research applications for PDT, J. Photochem. Photobiol. B: Biol., 2009 96(1):1-8. 12. Sharman, WM, Allen, CM and van Lier, JE, Photodynamic therapeutics: basic principles and clinical applications, Drug Discov. Today, 1999 4(11):507-17. 13. Josefsen, LB and Boyle, RW, Photodynamic therapy: novel third- generation photosensitizers one step closer?, Br. J. Pharmacol., 2008 154(1):1-3. 14. Chen, J, Stefflova, K, Niedre, MJ, Wilson, BC, Chance, B, et al, Protease-Triggered Photosensitizing Beacon Based on Singlet

66 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 67

Oxygen Quenching and Activation, J. Am. Chem. Soc., 2004 29. Akagi, R, Yasui, Y, Harper, P and Sassa, S, A novel mutation of 126(37):11450-51. delta-aminolaevulinate dehydratase in a healthy child with 12% 15. Castano, AP, Demidova, TN and Hamblin, MR, Mechanisms in erythrocyte enzyme activity, Br. J. Haematol., 1999 106(4):931- photodynamic therapy: part one--photosensitizers, 37. photochemistry and cellular localization, Photodiagnosis and 30. Erskine, PT, Newbold, R, Brindley, AA, Wood, SP, Shoolingin- Photodynamic Therapy, 2004 1(4):279-93. Jordan, PM, et al, The X-ray structure of yeast 5-aminolaevulinic 16. Svaasand, LO and Eixingsen, R, Optical Penetration in Human acid dehydratase complexed with substrate and three inhibitors, Intracranial Tumors, Photochem. Photobiol., 1985 41(1):73-76. J. Mol. Biol., 2001 312(1):133-41. 17. Ronan, P, Electromagnetic spectrum, Filename: EM_spectrum.svg, 31. Shoolingin-Jordan, PM, Spencer, P, Sarwar, M, Erskine, PE, Acessed from: http://en.wikipedia.org/wiki/File:EM_spectrum.svg Cheung, KM, et al, 5-Aminolaevulinic acid dehydratase: metals, 18. Moan, J and Sommer, S, Oxygen Dependence of the mutants and mechanism, Biochem. Soc. Trans., 2002 30:584-90. Photosensitizing Effect of Hematoporphyrin Derivative in NHIK 32. Neier, R, Chemical synthesis of Porphobilinogen and studies of its 3025 Cells, Cancer Res., 1985 45(4):1608-10. biosynthesis In: Moody, C.J. (editor) Advances in Nitrogen 19. Moan, J and Wold, E, Detection of singlet oxygen production by Heterocycles London, England: JAI Press, 1996; p. 35-146. ESR, Nature, 1979 279(5712):450-51. 33. Erskine, PT, Senior, N, Awan, S, Lambert, R, Lewis, G, et al, X- 20. Smith, G, McGimpsey, WG, Lynch, MC, Kochevar, IE and ray structure of 5-aminolaevulinate dehydratase, a hybrid Redmond, RW, An Efficient Oxygen Independent Two-Photon aldolase, Nat Struct Mol Biol, 1997 4(12):1025-31. Photosensitization Mechanism, Photochem. Photobiol., 1994 34. Erskine, PT, Coates, L, Newbold, R, Brindley, AA, Stauffer, F, et 59(2):135-39. al, The X-ray structure of yeast 5-aminolaevulinic acid 21. DeRosa, MC and Crutchley, RJ, Photosensitized singlet oxygen dehydratase complexed with two diacid inhibitors, FEBS Lett., and its applications, Coord. Chem. Rev., 2002 233-234:351-71. 2001 503(2-3):196-200. 22. Ochsner, M, Photophysical and photobiological processes in the 35. Erskine, PT, Coates, L, Butler, D, Youell, JH, Brindley, AA, et al, photodynamic therapy of tumours, J. Photochem. Photobiol. B: X-ray structure of a putative reaction intermediate of 5- Biol., 1997 39(1):1-18. aminolaevulinic acid dehydratase, Biochem. J., 2003 373:733-38. 23. Foote, CS, Definition of Type I and Type II Photosensitized 36. Frere, F, Schubert, WD, Stauffer, F, Frankenberg, N, Neier, R, et Oxidation, Photochem. Photobiol., 1991 54(5):659-59. al, Structure of porphobilinogen synthase from Pseudomonas 24. Llano, J, Raber, J and Eriksson, LA, Theoretical study of aeruginosa in complex with 5-fluorolevulinic acid suggests a phototoxic reactions of psoralens, J. Photochem. Photobiol. A: double Schiff base mechanism, J. Mol. Biol., 2002 320(2):237-47. Chem., 2003 154(2-3):235-43. 37. Jaffe, EK, An Unusual Phylogenetic Variation in the Metal Ion 25. Castano, AP, Demidova, TN and Hamblin, MR, Mechanisms in Binding Sites of Porphobilinogen Synthase, Chem. Biol., 2003 photodynamic therapy: part two--cellular signaling, cell 10(1):25-34. metabolism and modes of cell death, Photodiagnosis and 38. Jordan, PM, The biosynthesis of 5-aminolaevulinic acid and its Photodynamic Therapy, 2005 2(1):1-23. transformation into uroporphyrinogen III In: Jordan, P.M. (editor) 26. Labbe-Bois, R and Beaumont, C, Haem Biosynthesis In: Biosynthesis of Tetrapyrroles, Amsterdam: Elsevier, 1991. Encyclopedia of Life Sciences, John Wiley & Sons, Ltd: Chichester 39. Jaffe, EK, Martins, J, Li, J, Kervinen, J and Dunbrack Jr., RL, The http://www.els.net [doi:10.1038/npg.els.0000556], 2001. Molecular Mechanism of Lead Inhibition of Human 27. Astner, I, Schulze, JO, van den Heuvel, J, Jahn, D, Schubert, WD, Porphobilinogen Synthase, The Journal of Biological Chemistry, et al, Crystal structure of 5-aminolevulinate synthase, the first 2001 276(2):1531-37. enzyme of heme biosynthesis, and its link to XLSA in humans, 40. Jarret, C, Stauffer, F, Henz, ME, Marty, M, Luond, RM, et al, EMBO J., 2005 24(18):3166-77. Inhibition of Escherichia coli porphobilinogen synthase using 28. Breinig, S, Kervinen, J, Stith, L, Wasson, AS, Fairman, R, et al, analogs of postulated intermediates, Chem. Biol., 2000 7(3):185- Control of tetrapyrrole biosynthesis by alternate quaternary forms 96. of porphobilinogen synthase, Nat. Struct. Biol., 2003 10(9):757- 63.

68 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 69

41. Jaffe, EK and Hanes, D, Dissection of the early steps in the 54. Decatabbi, SB, Demolina, M. D. R., Deviale, Lcsm, Studies on the porphobilinogen synthase catalyzed reaction. Requirements for Active-Center of Rat-Liver Porphyrinogen Carboxylase Invivo Schiff's base formation, J. Biol. Chem., 1986 261(20):9348-53. Effect of Hexachlorobenzene, Int. J. Biochem., 1991 23(7-8):675- 42. Jaffe, EK, Markham, GD and Rajagopalan, JS, Nitrogen-15 and 79. carbon-13 NMR studies of ligands bound to the 280 000-dalton 55. Jones, RM, Jordan, P. M., Purification and Properties of the protein porphobilinogen synthase elucidate the structures of Uroporphyrinogen Decarboxylase from Rhodobacter-Sphaeroides, enzyme-bound product and a Schiff base intermediate, Biochem. J., 1993 293:703-12. Biochemistry (Mosc). 1990 29(36):8345-50. 56. Akhtar, M, The modification of acetate and propionate side 43. Mills-Davies, NL, Thompson, D, Cooper, JB, Wood, SP and chains during the biosynthesis of haem and chlorophylls: Shoolingin-Jordan, PM, PDB code: 1E51; Crystal Structure of mechanistic and stereochemical studies, Ciba Found. Symp., 1994 Native Human Erythrocyte 5-Aminolaevulinic Acid Dehydratase, 180:131-52. 2001. 57. Barnard, GF, Akhtar, M., Stereochemistry of Porphyrinogen 44. Jaffe, EK, The porphobilinogen synthase catalyzed reaction Carboxy-lyase Reaction in Heme Biosynthesis, J. Chem. Soc., mechanism, Bioorg. Chem., 2004 32(5):316-25. Chem. Commun., 1975 13:494-96. 45. Goodwin, CE and Leeper, FJ, Stereochemistry and mechanism of 58. Barnard, GF, Akhtar, M., Stereochemical and Mechanistic studies the conversion of 5-aminolaevulinic acid into porphobilinogen on the decarboxylation of uroporphyrinoen-III in heme catalysed by porphobilinogen synthase, Organic & Biomolecular biosynthesis, J. Chem. Soc., Perkin Trans. 1, 1979(10):2354-60. Chemistry, 2003 1(9):1443-46. 59. Silva, PJ, Ramos, M. J., Density-Functional Study of Mechanisms 46. Frere, F, Nentwich, M, Gacond, S, Heinz, DW, Neier, R, et al, for the Cofactor-Free Decarboxylation Performed by Probing the active site of Pseudomonas aeruginosa Uroporphyrinogen III Decarboxylase, J. Phys. Chem. B, 2005 porphobilinogen synthase using newly developed inhibitors, 109:18195-200. Biochemistry (Mosc). 2006 45(27):8243-53. 60. Martins, BM, Grimm, B, Mock, H-P, Huber, R and 47. Heinemann, IU, Jahn, M and Jahn, D, The biochemistry of heme Messerschmidt, A, Crystal Structure and Substrate Binding biosynthesis, Arch. Biochem. Biophys., 2008 474(2):238-51. Modeling of the Uroporphyrinogen-III Decarboxylase from 48. Luo, J and Lim, CK, Order of uroporphyrinogen III Nicotiana tabacum, J. Biol. Chem., 2001 276(47):44108-16. decarboxylation on incubation of porphobilinogen and 61. Akhtar, M, New Comprehensive Biochemistry: Biosynthesis of uroporphyrinogen III with erythrocyte uroporphyrinogen Tetrapyrroles, London: Elsevier, 1991, p. 67-76. decarboxylase, Biochem. J., 1993 289(2):529-32. 62. Straka, J, Kushner, J.P., Purification and characterization of 49. Fan, J, Liu, Q, Hao, Q, Teng, M and Niu, L, Crystal Structure of bovine hepatic uroporphyrinogen decarboxylase, Biochemistry Uroporphyrinogen Decarboxylase from Bacillus subtilis, J. (Mosc). 1983 22(20):4664-72. Bacteriol., 2007 189(9):3573-80. 63. Phillips, JD, Warby, CA, Whitby, FG, Kushner, JP and Hill, CP, 50. Phillips, JD, Whitby, FG, Kushner, JP and Hill, CP, Structural Substrate Shuttling between Active Sites of Uroporphyrinogen basis for tetrapyrrole coordination by uroporphyrinogen Decarboxylase Is Not Required to Generate Coproporphyrinogen, decarboxylase, EMBO J., 2003 22(23):6225-33. J. Mol. Biol., 2009 389(2):306-14. 51. Chaufan, G, de Molina, MDR and de Viale, LCS, How does 64. Navone, NM, Frisardi, AL, Resnik, ER, Batlle, AMdC and Polo, hexachlorobenzene treatment affect liver uroporphyrinogen CF, Porphyrin biosynthesis in human breast cancer. Preliminary decarboxylase?, Int J Biochem Cell Biol, 2001 33(6):621-30. mimetic in vitro studies, Med. Sci. Res., 1988 16(2):61-2. 52. Wyckoff, EE, Phillips, J.D., Sowa, A.M., Franklin, M.R., Kushner, 65. van Hillegersberg, R, van den Berg, JWO, Kort, WJ, Terpstra, OT J.P., Mutational analysis of human uroporphyrinogen and Wilson, JHP, Selective accumulation of endogenously decarboxylase, Biochim. Biophys. Acta, Protein Struct. Mol. produced porphyrins in a liver metastasis model in rats, Enzymol., 1996 1298(2):294-304. Gastroenterology, 1992 103(2):647-51. 53. Lewis, CA, Wolfenden, R., Uroporphyrinogen decarboxylation as 66. Krammer, B, Malik, Z, Pottier, R and Stepp, H, Basic Principles a benchmark for the catalytic proficiency of enzymes, PNAS, In: Pottier, C., Krammer, B., Stepp, H., Baumgartner, R. (editors.), 2008 105(45):17328-33.

70 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 71

Photodynamic Therapy with ALA, RSC Pubishing, 2006; p. 15- 82. Koch, W and Holthausen, MC, A Chemist’s Guide to Density 77. Functional Theory. 2nd Ed., Weinheim, Germany: Wiley-VCH 67. Casas, A and Batlle, A, Aminolevulinic acid derivatives and Verlag GmbH, 2001. liposome delivery as strategies for improving 5-aminolevulinic 83. Grant, GH and Richards, WG, Computational chemistry, Oxford: acid-mediated photodynamic therapy, Curr. Med. Chem., 2006 Oxford University Press, 1995 Oxford chemistry primers, No. 29. 13(10):1157-68 and references therein. 84. Allen, MP and Tildesley, DJ, Computer simulations of lipids, 68. Verma, S, Watt, GM, Mai, Z and Hasan, T, Strategies for Oxford, UK: Oxford University Press, 1987. Enhanced Photodynamic Therapy Effects, Photochem. Photobiol., 85. Lin, H and Truhlar, D, QM/MM: what have we learned, where 2007 83(5):996-1005. are we, and where do we go from here?, Theoretical Chemistry 69. Allred, AL and Thompson, DW, Keto-enol equilibriums in 2,4- Accounts: Theory, Computation, and Modeling (Theoretica pentanedione and 3,3-dideuterio-2,4-pentanedione, J. Phys. Chimica Acta), 2007 117(2):185-99. Chem., 1971 75(3):433-35. 86. Gaussian 03, Frisch, MJ, Trucks, GW, Schlegel, HB, Scuseria, GE, 70. Jaffe, EK and Rajagopalan, JS, Nuclear-Magnetic-Resonance Robb, MA, et al, 2003, Gaussian, Inc., Pittsburgh PA Studies of 5-Aminolevulinate Demonstrate Multiple Forms in 87. Cancès, E and Mennucci, B, Comment on "Reaction field Aqueous-Solution, Bioorg. Chem., 1990 18(4):381-94. treatment of charge penetration" [J. Chem. Phys. 112, 5558 71. Schrödinger, E, An Undulatory Theory of the Mechanics of Atoms (2000)], J. Chem. Phys., 2001 114(10):4744-45. and Molecules, Physical Review, 1926 28(6):1049. 88. Cancès, E, Mennucci, B and Tomasi, J, A new integral equation 72. Born, M and Oppenheimer, R, Zur Quantentheorie der Molekeln, formalism for the polarizable continuum model: theoretical Annalen der Physik, 1927 84:457-84. background and applications to isotropic and anisotropic 73. Mueller, M, Fundamentals of Quantum Chemistry Molecular dielectrics, J. Chem. Phys., 1997 107(8):3032-41. Spectroscopy and Modern Electronic Structure Computations, 89. Llano, J and Eriksson, LA, First principles electrochemistry: New York: Kluwer Academic Publishers, 2001. Electrons and protons reacting as independent ions, J. Chem. 74. Jensen, F, Introduction to Computational Chemistry 2nd edition, Phys., 2002 117(22):10193-206. John Wiley & Sons Ltd, 2007. 90. Fitzgerald, G and Andzelm, J, Chemical Applications of Density 75. Young, DC, Computational Chemistry: A Practical Guide for Functional Theory - Comparison to Experiment, Hartree-Fock, Applying Techniques to Real-World Problems., New York: John and Perturbation-Theory, J. Phys. Chem., 1991 95(26):10531-34. Wiley & Sons, Inc., 2001. 91. Topol, IA, Burt, SK, Toscano, M and Russo, N, Protonation of 76. Foresman, JB and Frisch, Æ, Exploring Chemistry with Electronic glycine and alanine: proton affinities, intrinsic basicities and Structure Methods Pittsburgh (PA) Gaussian Inc., 1996. proton transfer path, Journal of Molecular Structure: 77. Kohn, W and Sham, LJ, Self-Consistent Equations Including THEOCHEM, 1998 430(1-3):41-49. Exchange and Correlation Effects, Physical Review, 1965 92. GROMACS v. 3.3.1 / 3.3.2, Spoel, Dvd, Lindahl, E and Hess, B, 140:A1133. 2006, University of Groningen, The Netherlands 78. Atkins, PW and Friedman, RS, Molecular Quantum Mechanics 93. Marrink, S-J, Berger, O, Tieleman, P and Jahnig, F, Adhesion 4th ed., Oxford University Press, 2004. Forces of Lipids in a Phospholipid Membrane Studied by 79. Becke, AD, Density-Functional Thermochemistry .3. the Role of Molecular Dynamics Simulations, Biophys. J., 1998 74(2):931-43. Exact Exchange, J. Chem. Phys., 1993 98(7):5648-52. 94. Structures and Topologies, Biocomputing at University of Calgary, 80. Lee, CT, Yang, WT and Parr, RG, Development of the Colle- 2010, Acessed from: Salvetti Correlation-Energy Formula into a Functional of the http://moose.bio.ucalgary.ca/index.php?page=Structures_and_Top Electron-Density, Physical Review B, 1988 37(2):785-89. ologies 81. Stephens, PJ, Devlin, FJ, Chabalowski, CF and Frisch, MJ, Ab- 95. Leekumjorn, S and Sum, AK, Molecular studies of the gel to Initio Calculation of Vibrational Absorption and Circular- liquid-crystalline phase transition for fully hydrated DPPC and Dichroism Spectra Using Density-Functional Force-Fields, J. DPPE bilayers, Biochimica et Biophysica Acta (BBA) - Phys. Chem., 1994 98(45):11623-27. Biomembranes, 2007 1768(2):354-65.

72 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 73

96. Janiak, MJ, Small, DM and Shipley, GG, Nature of the thermal Addition, The Journal of Physical Chemistry, 1996 pretransition of synthetic phospholipids: dimyristoyl- and 100(50):19357-63. dipalmitoyllecithin, Biochemistry (Mosc). 1976 15(21):4575-80. 108. Svensson, M, Humbel, S and Morokuma, K, Energetics using the 97. Marrink, SJ and Berendsen, HJC, Simulation of Water Transport single point IMOMO (integrated molecular orbital+molecular through a Lipid-Membrane, J. Phys. Chem., 1994 98(15):4155- orbital) calculations: Choices of computational levels and model 68. system, The Journal of Chemical Physics, 1996 105(9):3654-61. 98. Marrink, SJ and Berendsen, HJC, Permeation process of small 109. Vreven, T and Morokuma, K, On the application of the IMOMO molecules across lipid membranes studied by molecular dynamics (integrated molecular orbital + molecular orbital) method, J. simulations, J. Phys. Chem., 1996 100(41):16729-38. Comput. Chem., 2000 21(16):1419-32. 99. Gonzalez, C and Schlegel, HB, An improved algorithm for 110. Cornell, WD, Cieplak, P, Bayly, CI, Gould, IR, Merz, KM, et al, A reaction path following, The Journal of Chemical Physics, 1989 Second Generation Force Field for the Simulation of Proteins, 90(4):2154-61. Nucleic Acids, and Organic Molecules, J. Am. Chem. Soc., 1995 100. Gonzalez, C and Schlegel, HB, Reaction path following in mass- 117(19):5179-97. weighted internal coordinates, The Journal of Physical Chemistry, 111. Himo, F, Quantum chemical modeling of enzyme active sites and 1990 94(14):5523-27. reaction mechanisms, Theoretical Chemistry Accounts: Theory, 101. Molecular Operating Environment (MOE) (2007.09), Chemical Computation, and Modeling (Theoretica Chimica Acta), 2006 Computing Group, Montreal, Canada 116(1):232-40. 102. MacKerell, AD, Bashford, D, Bellott, M, Dunbrack, RL, Evanseck, 112. Lias, SG, Liebman, JF and Levin, RD, Evaluated Gas Phase JD, et al, All-atom empirical potential for molecular modeling Basicities and Proton Affinities of Molecules; Heats of Formation and dynamics studies of proteins, Jorunal of Physical Chemistry of Protonated Molecules, J. Phys. Chem. Ref. Data, 1984 B, 1998 102(18):3586-616. 13(3):695-808. 103. Dapprich, S, Komáromi, I, Byun, KS, Morokuma, K and Frisch, 113. Afonso, C, Modeste, F, Breton, P, Fournier, F and Tabet, JC, MJ, A new ONIOM implementation in Gaussian98. Part I. The Proton affinities of the commonly occurring L-amino acids by calculation of energies, gradients, vibrational frequencies and using electrospray ionization-ion trap mass spectrometry, Eur. J. electric field derivatives, Journal of Molecular Structure: Mass Spectrom., 2000 6(5):443-49. THEOCHEM, 1999 461-462:1-21. 114. Meot-Ner, M, Hunter, EP and Field, FH, Ion thermochemistry of 104. Humbel, S, Sieber, S and Morokuma, K, The IMOMO method: low-volatility compounds in the gas phase. 1. Intrinsic basicities Integration of different levels of molecular orbital approximations of a-amino acids, J. Am. Chem. Soc., 1979 101(3):686-89. for geometry optimization of large systems: Test for n-butane 115. Hunter, EPL and Lias, SG, Evaluated Gas Phase Basicities and conformation and S[sub N]2 reaction: RCl+Cl[sup - ], The Proton Affinities of Molecules: An Update, J. Phys. Chem. Ref. Journal of Chemical Physics, 1996 105(5):1959-67. Data, 1998 27(3):413-656. 105. Maseras, F and Morokuma, K, IMOMM: A new integrated ab 116. Locke, MJ and McIver Jr, RT, Effect of solvation on the acid/base initio + molecular mechanics geometry optimization scheme properties of glycine J. Am. Chem. Soc., 1983 105(13):4226-32. of equilibrium structures and transition states, J. Comput. Chem., 117. Pepe, C, Rochut, S, Paumard, JP and Tabet, JC, Ab initio 1995 16(9):1170-79. calculations of proton affinities of glycine, proline, cysteine and 106. Matsubara, T, Sieber, S and Morokuma, K, A test of the new phenylalanine: comparison with the experimental values obtained ldquointegrated MO + MMrdquo (IMOMM) method for the using an electrospray ionisation ion trap mass spectrometer, conformational energy of ethane and n-butane, Int. J. Rapid Commun. Mass Spectrom., 2004 18(3):307-12. Quantum Chem, 1996 60(6):1101-09. 118. Graul, ST, Schnute, ME and Squires, RR, Gas-Phase Acidities of 107. Svensson, M, Humbel, S, Froese, RDJ, Matsubara, T, Sieber, S, et Carboxylic-Acids and Alcohols from Collision-Induced al, ONIOM: A Multilayered Integrated MO + MM Method for Dissociation of Dimer Cluster Ions, Int. J. Mass Spectrom. Ion Geometry Optimizations and Single Point Energy Predictions. A Processes, 1990 96(2):181-98. Test for Diels−Alder Reactions and Pt(P(t-Bu)3)2 + H2 Oxidative 119. Rud, E, Gederaas, O, Hogset, A and Berg, K, 5-Aminolevulinic Acid, but not 5-Aminolevulinic Acid Esters, is Transported into

74 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof I 75

Adenocarcinoma Cells by System BETA Transporters, Photochem. Photobiol., 2000 71(5):640-47. 120. Moretti, MB, Garcia, SC, Perotti, C, Batlle, A and Casas, A, delta- aminolevulinic acid transport in murine mammary adenocarcinoma cells is mediated by beta transporters, Br. J. Cancer, 2002 87(4):471-74. 121. Döring, F, Walter, J, Will, J, Föcking, M, Boll, M, et al, Delta- aminolevulinic Acid Transport by Intestinal and Renal Peptide Transporters and Its Physiological and Clinical Implications, The Journal of Clincal Investigation, 1998 101(12):2761-67. 122. Whitaker, CJ, Battah, SH, Forsyth, MJ, Edwards, C, Boyle, RW, et al, Photosensitization of pancreatic tumour cells by delta- aminolaevulinic acid esters, Anti-Cancer Drug Des., 2000 15(3):161-70. 123. Gederaas, OA, Holroyd, A, Brown, SB, Vernon, D, Moan, J, et al, 5-aminolaevulinic acid methyl ester transport on amino acid carriers in a human colon adenocarcinoma cell line, Photochem. Photobiol., 2001 73(2):164-69. 124. Kaliszewski, M, Juzeniene, A, Juzenas, P, Kwasny, M, Kaminski, J, et al, Formation of protoporphyrin IX from carboxylic- and amino-derivatives of 5-aminolevulinic acid, Photodiagnosis and Photodynamic Therapy, 2005 2(2):129-34. 125. Kloek, J, Akkermans, W and Beijersbergen van Henegouwen, GMJ, Derivatives of 5-Aminolevulinic Acid for Photodynamic Therapy: Enzymatic Conversion into Protoporphyrin, Photochem. Photobiol., 1998 67(1):150-54. 126. Jordan, PM and Seehra, JS, Mechanism of action of 5- aminolevulinic acid dehydratase: stepwise order of addition of the two molecules of 5-aminolevulinic acid in the enzymic synthesis of porphobilinogen, J. Chem. Soc., Chem. Commun., 1980(5):240-42. 127. Schlösser, M and Beyersmann, D, Zinc and Cadmium 5- Aminolevulinate Dehydratase. Metal-Dependent pH Profiles, Biol. Chem. Hoppe-Seyler, 1987 368(2):1469-78. 128. Juaréz, AB, Aldonatti, C., Vigna, M. S., Ríos de Molina, M. C., Studies on uroporphyrinogen decarboxylase from Chlorella kessleri, Can. J. Microbiol., 2007 53:303-12.

76 I EDVIN ERDTMAN 5-Aminolevulinic acid and derivatives thereof