Optimizing Cellular to Improve Chronic Skin Wound Healing

James J. Slade Honors Thesis

Luis Felipe Ramirez Biomedical Engineering Rutgers University, New Brunswick

Under the direction of Dr. Francois Berthiaume and Dr. Gabriel Yarmush

Abstract—Despite significant advances, chronic skin wounds Wound healing is a complex process with four identifiable remain a large problem both in terms of morbidity and cost. It stages: hemostasis, inflammation, proliferation, and remodel- is estimated that in the United States, this problem afflicts 6.5 ing [6]. Hemostasis involves the formation of a blood clot that million people a year and costs more than 30 billion dollars for di- abetic foot ulcers alone [4,11]. Currently approved treatments are stops the loss of blood at the site of injury [6]. Growth factors often ineffective. This thesis seeks to leverage the large amount of released by activated platelets during the hemostasis stage information that has accumulated about metabolism in the hu- recruit immune cells such as neutrophils and macrophages [6]. man body, and to mine that information with computational mod- The infiltration of the injured tissue with these immune cells eling. It seeks to uncover whether metabolites commonly available leads to the inflammatory phase [6]. The role played by inflam- in the human body can be used to bolster the metabolism of wounded cells such as keratinocytes (the cells that form the mation is classified as both positive and negative [5]. On the epidermis), and therefore improve the natural process of wound one hand the inflammatory response clears the wound site of healing. This would offer treatment options with fewer side effects pathogens and dead cells, on the other hand the inflammatory than what is currently offered to improve wound healing. The response is thought to prevent the wound from healing and author first reduced the global set of reactions and metabolites to be responsible for scarring [5]. In the proliferation stage, in a publicly available database of cellular metabolism (the recon Database), to 357 reactions and 339 metabolites that are most keratinocytes repopulate the injury and angiogenesis leads to applicable to keratinocytes. Then, metabolic flux through a key revascularization of the tissue [6]. Keratinocytes migrate from set of 25 reactions was defined for mass balance analysis. Lastly, the edge of the injury and from hair follicles towards the the impact of variations in metabolic inputs on the metabolic flux deeper portions of the injury in part due to stimulation from distribution was investigated. By analyzing the variability of the transforming growth factor (TGF)-β [6]. Hypoxic conditions flux through each reaction in Monte Carlo simulations, reactions that are likely to be amenable to manipulation were identified. within the injury site lead to release of vascular endothelial This analysis points to the citric acid cycle as the pathway growth factor, which results in angiogenesis [6]. Fibroblasts most amenable to manipulation. Other key reaction categories are also important in the proliferation stage. In particular, they identified were glycolysis, the pentose phosphate pathway, amino regenerate the extracellular matrix by releasing proteins such acid metabolism, transport reactions, and energy production and as collagen, elastin, and matrix metalloproteinase [6]. The homeostasis. remodeling stage lasts the longest, and involves remodeling of the extracellular matrix [6]. Of particular importance is I.INTRODUCTION the conversion of type III collagen to type I collagen, which strengthens the injured tissue [6]. A. The Medical Problem Chronic wounds are characterized by a sustained inflam- Chronic skin wounds can be defined as wounds that do not matory response, increased presence of microbes, and fewer heal within three months, and include three types of wounds: and functionally deficient fibroblasts [5]. The sustained in- vascular ulcers, diabetic ulcers, and pressure ulcers [5, 2]. In flammatory response presents a problem for wound healing the United States alone, this problem is said to afflict 6.5 as it taxes oxygen and nutrient availability. The immune cells million patients [11]. The burden of diabetic foot ulcers alone that infiltrate chronic wounds also appear to have diminished on the health system in the United States is estimated to be phagocytic ability, leading to a buildup of necrotic debris [5]. $30 billion annually [4]. The prevalence of diabetic foot ulcers A common problem with chronic wounds is also the presence alone was estimated to be 68.4/1000 person-years [3]. All of of bacteria and fungi that invade the wound [5]. The interplay this information points to a set of conditions that cause a severe between the sustained inflammatory response and the presence burden on individuals and the health system, and that merits of microbes is unclear [5]. Fibroblasts found in chronic sustained research efforts. wounds have also been found to lack a normal response to growth factor and to lack normal migratory capacities [5]. TABLE I Chronic wound fluid has also been found to have an inhibitory SUMMARY OF THE METABOLISM OF WOUNDS effect on the proliferation of fibroblasts and keratinocytes, indicating problems with the factors present in chronic wounds Type of Condition Condition Sub-type Positive/ [6]. Matrix Metalloproteinases have also been found to be Negative Effect overactive in chronic wounds, which leads to a degradation Acute Hypoxia- Has positive of the extracellular matrix [6]. leads to platelet effects aggregation and the release of B. Metabolism of Wounds Hypoxia cytokines Chronic Hypoxia- Has large Many processes of wound healing demand large amounts leads to impaired negative energy production, effects of energy [13]. An especially important factor in the ability a prolonged of cells to obtain energy is the oxygen availability. Shortly inflammatory phase, after the onset of the wound, hypoxic conditions occur in and too much ROS the wound area as a result of vessel constriction and platelet accumulation Xanthine - Negative effect aggregation [8]. Acute hypoxic conditions can have benefi- Purine Catabolism - cial impacts on wound healing, whereas chronic hypoxia is Excessive ROS believed to have large deleterious effects on wound healing production L-Arginine Negative effect [10]. Chronic hypoxia impairs cellular function by limiting Deaminase- the production of ATP through aerobic glycolysis, the citric Arginine Metabolism ROS produced acid cycle, and fatty acid oxidation [10]. This ATP production Arginase - No Positive effect ROS produced is critical for sustaining all of the steps of wound healing, Decreased flux Negative effect including cell proliferation, bacterial defense, and collagen through glycolysis, synthesis [10]. Furthermore, the oxygen dependent NADPH the citric acid cycle, nucleotide synthesis, produces reactive oxygen species (ROS) that are and fatty acid crucial for bacterial control, cytokine release, cell proliferation, catabolism- leads and angiogenesis [10]. As mentioned above, acute hypoxia is to reduced energy in the cells available also believed to play a beneficial role in wound healing [10]. for wound healing The initial hypoxia, for example, leads to platelet aggregation Impaired Glucose Uptake Increased catabolism Negative effect and the stimulation of cytokine release through the production of amino acids and carboxylic acids- of ROS [10]. Moreover, it is believed that leukocytes use the leads to breakdown of ROS gradient to infiltrate the wound tissue [10]. proteins and a limited ability to construct new Chronic wounds are characterized by local hypoxia that proteins needed for is caused by alterations of the blood vessels and reduced wound healing vascular perfusion [10]. These hypoxic conditions lead to Reduced flux through Negative effect the Pentose Phosphate reduced energy for cells, as well as to signaling cascades Pathway- less that lead to inflammatory responses [10]. In addition, hypoxia reducing equivalents leads to increased production of cytokines that promote the produced movement of neutrophils and macrophages into the wound, and the subsequent production of pro-inflammatory molecules [10]. Prolonged hypoxic conditions also prevent the production of nitric oxide through the [10]. produces reactive oxygen species [1]. Conversely, arginine Nitric oxide normally prevents the accumulation of large quan- metabolized by arginase does not generate ROS [1]. tities of ROS, and its absence can lead to harmful accumulation of ROS [10]. Due to increased energy demand, central carbon metabolism Another key metabolic pathway involved in wound healing is very important in wound healing. Defective glucose uptake, is purine metabolism. In particular, it has been found that for example, has been found to negatively impact wound due to increased energy demand, cells engage in substantial healing [14]. The researchers found that the decline in glucose purine catabolism during wound healing [13]. Weinstein et al. uptake led to significant reductions in flux through glycol- found that xanthine oxidoreductase leads to excessive ROS ysis, the citric acid cycle, nucleotide synthesis, and fatty formation in diabetic patients, thus contributing to impaired acid catabolism [14]. The cells compensated by increasing wound healing [13]. A similar pathway involving arginine catabolism of amino acids and carboxylic acids [14]. Also of metabolism has also been studied in wound healing [1]. interest, the cells showed increased oxidative stress due to their Arginine can be metabolized through two different , inability to produce reducing equivalents through the pentose one is l-arginine deaminase while the other is arginase [1]. phosphate pathway [14]. If arginine is metabolized through l-arginine deaminase, it The information in this section is summarized in table 1. C. Metabolic Pathways they are transferred to oxygen [7]. The energy released by the transfer of electrons down the transport chain is used to Glycolysis is an almost universal pathway, that is present in pump protons into the mitochondrial intermembrane space [7]. all skin cell types [7]. Glycolysis uses glucose as its starting The chemical energy stored in this proton gradient is then material, and breaks it down into two three carbon sugars used to produce ATP [7]. It is important to note that both known as pyruvate [7]. The pathway is composed of 10 steps the citric acid cycle and the electron transport chain require that occur in the cytosol [7]. Nearly all other polysaccharides oxygen, meaning that they are significantly reduced in hypoxic and monosaccharides can serve as substrates for glycolysis [7]. conditions. Moreover, because fatty acid catabolism leads to Under hypoxic conditions, cells will engage in fermentative these pathways, it will also be significantly diminished in pathways that regenerate NADH for glycolysis [7]. Under hypoxic conditions. aerobic conditions, the pyruvate produced by glycolysis will Anabolic pathways, which require the utilization of energy, undergo further degradation to acetyl CoA that can enter are also necessary for the proper functioning of skin cells. the citric acid cycle [7]). Gluconeogenesis is used to create Amino acids are synthetized from a number of intermediates glucose when the concentration of glucose is low [7]. This of glycolysis and the citric acid cycle [7]. Histidine, isoleucine, pathway is mostly specific to the liver and does not occur in , lysine, methionine, phenylalanine, threonine, trypto- skin cells [7]. Glucose can also serve as the starting point of phan, and valine are known as essential amino acids and the pentose phosphate pathway, which produce pentoses that cannot be synthesized by humans [7]. Purine biosynthesis can be used to synthetize nucleotides, coenzymes, DNA, and occurs in 11 common steps that then branch for the formation RNA [7]. Equally important, the pentose phosphate pathway of either adenylate or guanylate [7]. Pyrimidine synthesis, on produces NADPH which can be used in reductive pathways the other hand, occurs in seven steps that use aspartate as a or to mitigate the impact of ROS [7]. Although glucose can precursor [7]. be channeled into either glycolysis or the pentose phosphate Fatty acid biosynthesis occurs in the cytosol of cells [7]. pathway, normally the predominant pathway is glycolysis [7]. The process occurs by the repetition of four steps until a 16 Lastly, glucose can enter an anabolic pathway in which glucose carbon chain fatty acid is produced [7]. The process utilizes monomers are joined to make glycogen [7]. The ability to form acetyl-CoA, ATP, and NADPH as its substrates [7]. The 16 glycogen, however, is not present in skin cells [7]. carbon fatty acid can be elongated in the smooth endoplasmic Fatty acid oxidation mainly occurs via a process called β- reticulum by what is known as the fatty acid elongation system oxidation that takes place in the mitochondria [7]. β-oxidation [7]. Triacylglycerols are synthetized by first forming glycerol involves the repetition of four steps in which two carbon and the tail groups, and then joining them [7]. Cholesterol is fragments of the fatty acid are broken of into acetyl-CoA [7]. synthesized from acetate [7]. This reaction takes place mostly Unsaturated fatty acids require an additional two steps [7]. in the liver [7]. Another minor pathway for catabolism of fatty acids called ω oxidation only occurs in the liver and kidney [7]. Ketone D. Utilizing the Recon Database body formation, which comprises an alternative fate for acetyl- This project will make use of the Recon database, which CoA and is an important source of energy when glucose is not strives to recompile all of the metabolic pathways in the available, is primarily liver-specific and does not occur in skin human body [12]. Recon 2.2 is one of the latest iterations of cells [7]. the database and contains 5324 metabolites, 7785 reactions, Amino acid catabolism is another major source of energy in and 1675 associated genes [12]. A paper written in 2014 humans [7]. Although there are many pathways through which outlined the process for curating the recon database for a amino acids are degraded, they converge on a few common specific cell type [9]. The researchers manage to do this by products that can enter the citric acid cycle in skin cells ([7]]. limiting the inputs and outputs to the model, limiting the Ammonia produced from the breakdown of amino acids is generation and utilization of ATP, getting rid of reactions channeled to the urea cycle, which does not occur in skin cells that become redundant, and eliminating reactions that are [7]. Nucleotide catabolism can yield either uric acid in the case thermodynamically infeasible [9]. This paper will follow this of purines, or urea in the case of pyrimidines [7]. There are proven method of reducing the entire set of reactions to a additional pathways known as salvage pathways that recycle subset that is more applicable to skin cells. This method will purines and pyrimidines [7]. produce a reduced set of 357 reactions and 339 metabolites. The acetyl-CoA that is produced through all of the previ- From there a key set of 20 reactions will be defined and used ously mentioned reactions enters the citric acid cycle for fur- to solve the flux through all of the remaining reactions using ther oxidation [7]. This cycle takes place in the mitochondria metabolic flux analysis. [7]. The citric acid cycle directly produces one molecule of The overarching goal of this project is to develop better ATP, and also produces reduced coenzymes that will serve as methods for improving healing in chronic wounds. As alluded the inputs to the electron transport chain [7]. In the electron to earlier, metabolic limitations provide a great impediment to transport chain, the electrons from the reduced coenzymes chronic wound healing. A natural question to ask is whether produced in glycolysis and the citric acid cycle are transferred one can leverage the vast amount of accumulated information sequentially to carriers with increasing reducing potential until about metabolic pathways in the human body to improve chronic wound healing. This paper tries to look at the most B. Metabolic Flux Analysis important set of reactions concurrently, making use of systems The output stoichiometric matrix from the reduction step biology. Thus, rather than artificially separating metabolic was then used as the starting point for metabolic flux analysis. pathways, this project will try to take into account the entire To carry out this task, the flux through a subset of 25 key metabolism of a cell. It will make use of sensitivity analysis reactions was set. The 25 reactions are the exchange reactions of the derived metabolic flux model to find the metabolites for every amino acid except cysteine, the biomass reaction, that lead to the largest improvement in cellular metabolism. the exchange reaction for glucose, the exchange reaction for No other piece of literature utilizes this approach to improve lactate, the exchange reaction for oxygen, and the exchange chronic wound healing, and so it is hoped that this project will reaction for ammonium. The idea here is to use the principle make a meaningful contribution to the area. of mass balance as well as a subset of measured fluxes, to E. Monte Carlo Simulations determine the flux through reactions in the metabolic network.. The last aspect of this project will involve Monte Carlo Here one is trying to solve for m in the equation S · m = O, simulations. This will involve varying the flux through the var- where S is the stoichiometric matrix, m is the list of reactions, ious reactions to ascertain their effect on global metabolism. and O is the flux for each reaction. O can be determined by It is hoped that the use of these simulations will allow for first finding the inverse of the stoichiometric function with- the identification of the metabolites that lead to the largest out the reactions corresponding to the measured fluxes, and improvement in metabolism. This will then allow for testing multiplying it by the stoichiometric function corresponding to see whether these metabolites will lead to improvements in to the measured fluxes. This is equivalent to saying there wound healing. is an equation W · s = I, where W is the stoichiometric matrix without including the reactions that were measured, II.METHODS s is the set of reactions that relates the two, and I is the A. Reducing the number of reactions stoichiometric matrix for the measured reactions. Here one is solving for s. After this is done, the output is multiplied by In 2014, Quek et al. outlined a method for reducing the the measured fluxes. The solution to this last multiplication is entirety of the Recon database to a set of 357 reactions and the flux through every reaction in the system. 339 metabolites that were more appropriate for cells in culture In addition to calculating the flux, a standard deviation is [9]. The author used the methods outlined in that paper to also calculated. This is done by finding the variance associated achieve a reduced set of reactions. The reduction process with the matrix that relates the two stoichiometric matrices, involves four steps: reserved functions, futile ATP, redundant, and adding it to the variance associated with the measured and loopless [9]. In the first step, certain types of reactions fluxes. were constrained. This was done by manually identifying all of the reactions involved in key processes such as ATP and All of the calculations involved in this step were carried out NADPH production, the transport of compounds across the on MATLAB. cell membrane, and the fixation of [9]. Each category C. Monte Carlo Simulations of reactions was constrained such that only the most important contributing reactions were included. Moreover, reactions that For the last part of this thesis, Monte Carlo simulations are not relevant to cells in culture were set to 0 [9]. of every reaction in the system were carried out. A random In the second step of the reduction process, reactions number generator was used to vary the value of different fluxes that consume ATP without a clear metabolic purpose were in the output of the model using the inputs from the Quek identified [9]. These sets of reactions were then grouped into paper. The random number generator followed the standard a category termed ATP maintenance [9]. Some of the reactions normal distribution. A total of 183 simulations were carried that consume ATP in a cyclic fashion were set to zero after out. The effects of those variations on the other reactions performing flux variability analysis to assess their importance were quantified. Excel was used to calculate the average and in the presence of a biomass objective function [9]. standard deviation of every flux output for each simulation. To In the third step reactions that are redundant as a result of determine the reactions that had a high variability a cut-off of compartmentalization or different use were identified 30 mol/gDW/h was used. [9]. The set of reactions was decomposed into a minimum All of the simulations for this step were also carried out in number of sub-models according to a flux constraint [9]. Matlab. Within each sub-model only one reaction was retained. In the final step, thermodynamically infeasible reactions were III.RESULTS identified using flux variability analysis [9]. These reactions A. Reduction process may have been infeasible at the outset, or they may have become infeasible as a result of the previous reduction steps By carrying out the reduction process described in the [9]. Quek paper, a matrix of 357 reactions X 339 metabolites was This process of reduction was entirely carried out using obtained. The summary of the reactions grouped by function is MATLAB. presented in table 2; the full list of reactions is in the appendix. B. Metabolic Flux Analysis In this step, an output was obtained for all 357 reactions in TABLE II the system. A figure illustrating the main metabolic pathways SUMMARY OF REACTIONS GROUPEDBY FUNCTION is included in figure 1. The calculated fluxes from using the Reaction Type Reaction Number measured fluxes in the supplement are in figure 2. Biomass 50 The set of reactions in the Quek paper was used to check Glycolysis 96, 134, 147, 175, 196, 260, 262, 263 the agreement of this paper’s model with that of the Quek Citric Acid Cycle 21, 32, 69, 141, 185, 207, 256, Paper’s model. The results are contained in table 3. 335, 336, In total, the Quek paper lists 12 reactions [9]. Of those 12 Oxidative Phosphorylation 48, 71, 72, 97, 98 reactions, 10 of the calculated fluxes were within the range Pentose Phosphate Pathway 145, 165, 227, 264, 324, 325 337, 341, 342 listed in the Quek paper. Glycogenesis 146 Glycogenolysis 266 C. Monte Carlo Simulations Gluconeogenesis 258 A total of 68 reactions had a standard deviation above 30 Cholesterol Synthesis 80, 86, 87, 89, 181, 182, 191, 200, 215, 273, 333, 334 mol/gDW/h. These reactions are grouped by category in table Synthesis 2, 51, 54, 92, 201, 300, 301 4. 307 Fatty Acid Synthesis 16, 17, 52, 93, 94, 129, 130, IV. DISCUSSION 131, 132, 133 Fatty Acid Catabolism 12, 18, 95, 173, 174, 219 The process outlined in the Quek paper was effectively 220, 277 recreated as a part of this thesis. The 357 reactions from Phospholipid Metabolism 8, 27, 40, 41, 56, 57, 58, 59, table 1, include the most important set of reactions for cells 60, 64, 88, 143, 167, 217, 218, 259, 267, 268, 274, 275, 276, in culture. They also got rid of redundant and infeasible 284, 285, 288, 326, 331, 332 reactions. Of particular importance to wound healing, the Amino Acid Synthesis 28, 42, 74, 75, 81, 152, 157, reactions for biomass creation, for nucleotide synthesis, and 180, 213, 226, 261, 270, 285, 287, 295, 296, 327, 329, 340 for lipid synthesis are included. Amino Acid Catabolism 4, 5, 6, 7, 11, 35, 36, 45, 46, In the second part of the thesis, metabolic flux analysis was 47, 84, 136, 139, 140, 149, used to calculate the flux through all 357 reactions. A set 154, 155, 156, 160, 166, 176, of 25 key metabolites were found and the flux through their 177, 178, 187, 192, 194, 198, 204, 206, 216, 221, exchange reaction was defined. This set of 25 reactions was 222, 223, 244, 245, 251, effectively used to solve for all of the remaining fluxes. Figure 252, 270, 284, 291, 292, 1 lists all of the solved fluxes. The fluxes calculated in this 293, 294, 297, 298, 330, 346 thesis were also compared to the fluxes calculated in the Quek Urea Cycle 38, 144 paper (table 3). The discrepancies are likely not due to errors Purine Synthesis 24, 25, 26, 30, 31, 158, 189, in calculations, but rather due to differences in the reactions 283, 299 Purine Catabolism 135 defined. The Quek paper does not list the set of reactions they Pyrimidine Synthesis 43, 55, 82, 83 defined. Rather, they list flux for metabolites. Several of those Transport 1, 6, 29, 33, 37, 44, 49, 68, 77, 78, 79, 91, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115,116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 137, 138, 142, 151, 153, 159, 161, 162, 163, 170, 171, 172, 179, 183, 184, 186, 188, 195, 197, 199, 202, 203,205, 214, 228, 229, 230, 239, 240,241, 242, 253, 257, 269, 271, 272,281, 282, 286, 290, 302, 303, 304,308, 309, 310, 311, 312, 313, 314,315, 316, 317, 318, 328, 339, 347,348, 352, 353, 355 Nucleoside Phosphate Maintenance 22, 23, 70, 76, 85, 90, 150, 164, 190, 231, 232, 233, 234, 235, 236, 237, 238, 250, 275, 320, 321, 322, 323, 343, 345, 350 Carboxylic Acid Dissociation 169

Fig. 1. Picture of Metabolic Pathways [7] TABLE III SUMMARY OF REACTIONS GROUPEDBY FUNCTION TABLE IV REACTIONS WITH HIGH FLUX ARRANGED BY REACTION CATEGORY reaction This paper’s This paper’s Quek paper calculated calculated (umol/gDW/h) Reaction Category Reaction average standard Glycolysis Glucose-6-Phosphate , (umol/gDW/h) deviation Pyruvate Kinase, Lactate (umol/gDW/h) Dehydrogenase Hexokinase 623.3 11.9 599-645 Gluconeogenesis Pyruvate Carboxylase Pyruvate 979.5 33.0 787-2623 Citric Acid Cycle Pyruvate Dehydrogenase, Kinase 2-Oxoglutarate Dehydrogenase, Pyruvate -104.2 53.9 33-241 Citrate Synthase, Fumarase, Dehydrogenase Malate Dehydrogenase, Pyruvate Lactate -1207.3 24.2 (-)1257- (-)1163 Dehydrogenase, Succinate Dehydrogenase Dehydrogenase, Succinate Coa G6P 145.7 29.5 0-605 Lygase Dehydrogenase Oxidative Phosphorylation ATP Synthase, Cytochrome Citrate 8.9 51.8 0-204 C oxidase, Dehydrogenase Ubiquinol-6 Cytochrome C 2-Oxoglutarate 75.1 19.6 0-211 Reductase, Electron Transfer Dehydrogenase , Electron Transfer Succinate 108.0 19.7 10-222 Flavoprotein-Ubiquinone Dehydrogenase Oxidoreductase, NADH ATP 1937.7 124.3 1822-4014 Dehydrogenase, Synthase Pentose Phosphate Pathway Glucose 6-Phosphate ATP 390.2 28.3 0-3560 Dehydrogenase, Phosphogluconate Maintenance Dehydrogenase, Alanine 236.6 58.5 161-240 6-Phosphogluconolactonase Transport Energy Metabolism Adenylate Kinase, Adenosine (in) Kinase, Aspartate-Glutamate Glutmamine -525.3 54.5 -15-50 Mitochondrial Shuttle Transport ADP/ATP Transport Mitochondrial, (in) Citrate Transport, Malate Transport, Malic Enzyme, Pyruvate Mitochondrial Transport, Lipid Synthesis Acetyl-CoA Carboxylase, ATP-Citrate , metabolites have several reactions for their transport across Inorganic Diphosphatase the plasma membrane. As a result, it is difficult to determine Amino Acid Metabolism Aspartate Exchange of Glutamine, exactly how they apportioned the fluxes through the various Glutamine Synthetase, transport reactions for each metabolite. This author merely L-glutamate secretion via defined one reaction for each metabolite, namely the exchange Secretory vesicle (ATP Driven) Glycine Passive Transport reactions. It is therefore very likely that differences in defining to Mitochondria, Glycine the fluxes for each measured metabolite are what is leading to N-Methyltransferase, Methionine Synthase, Sarcosine Transport, S Adenosyl Methionine Adenosylhomocysteinase, Metabolism Methionine Adenosyltransferase Tetrahydrofolate Metabolism 10-Formyltetrahydrofolate Mitochondrial Transport via Diffusion, Methenyltetrahydrofolate Cyclohydrolase, Methylenetetrahydrofolate Dehydrogenase, 5,10-Methylenetetrahydro folatereductase, 5,6,7,8-Tetrahydrofolate Transport, Diffusion Transport CO2 Transporter via Diffusion, CO2 Exchange, H2O Exchange, Exchange Reaction for Oxygen, H2O Transport via Diffusion, H2O Transport, L-lactate Reversible Transport via Proton Symport, O2 Transport Diffusion, Phosphate Trasporter Fig. 2. calculated fluxes for each reaction in the order they appear in table 2. discrepancies. Given these unknowns, the level of agreement [8] N. Ojha, et al., “Assessment of Wound-Site Redox Environment and the is very satisfactory. Significance of Rac2 in Cutaneous Healing,”, Free Radical Biology and Medicine, vol. 44,no. 4, pp. 172–178, 2008. A high variability of flux through a reaction in Monte Carlo [9] L. Quek, et al., “Reducing Recon 2 for Steady-State Flux Analysis of simulations helps pinpoint reactions that are likely to have HEK Cell Culture,”, Journal of Biotechnology, vol. 184,, pp. 682–691, an impact on global metabolism. In this project, 68 reactions 2014. [10] S. Schreml, et al., “Oxygen in Acute and Chronic Wound Healing,”, with a flux of 30 mol/gDW/h or above were identified. Many British Journal of Dermatology, vol. 163,no. 2, pp. 257–268, 2010. of these reactions are involved in key metabolic pathways. Of these the citric acid cycle was almost entirely represented [11] S. Sen, et al., “Human Skin Wounds: A Major and Snowballing Threat to Public Health and the Economy,”, Wound Repair and Regeneration, within this subset of reactions. This likely indicates that the vol. 17,no. 6, pp. 763–771, 2009 citric acid cycle is a good target for manipulation. Encourag- [12] N. Swainston, et al. “HRecon 2.2: from Reconstruction to Model of ingly, many reactions that are key for energy production such Human Metabolism,”, Metabolomics, vol. 12,no. 7, 2016 [13] a. Weinstein, et al. “Normalizing Dysfunctional Purine Metabolism as ATP synthase and those of glycolysis were also represented. Accelerates Diabetic Wound Healing,”, Wound Repair and Regeneration, Likewise important, the reactions of the pentose phosphate vol. 23,no. 1, pp. 14–21, 2015 pathway were represented. This deserves to be highlighted, [14] Z. Zhang, et al. “725 Epidermal Deletion of Glut1 Highlights Essential Roles for Glucose Metabolism in Wound Healing and the Response to because excessive production of reactive oxygen species can UVB Irradiation,”, Journal of Investigative Dermatology, vol. 137,no. 5, occur in chronic wounds. Amino acid metabolism and lipid 2017 synthesis reactions were represented. These are also important reactions in mediating a correct response to a wound. Lastly, a number of transport reactions that can be easily manipulated were present within this subset.

V. CONCLUSION This paper sought out to use a systems biology approach to make a novel contribution to the area of wound healing. It was hoped that natural metabolites could be identified, and that these metabolites could lead to an improvement in chronic skin wound healing. The Recon database was reduced to a set of 357 reactions and 339 metabolites. A set of 25 reactions were then used to solve for the remaining reactions in the system. Lastly, Monte Carlo simulations were used to identify metabolites that have the largest impact on global metabolism. In the end, several reaction categories were identified as likely targets of manipulation. These reactions were present in key pathways such as the citric acid cycle, glycolysis, the pentose phosphate pathway, amino acid metabolism, and energy production and homeostasis. These reactions will likely point to the metabolites that can help ameliorate the global metabolism of chronic wounds.

VI.REFERENCES REFERENCES [1] M. Albina, “Temporal Expression of Different Pathways of 1-Arginine Metabolism in Healing Wounds,” The Journal of Immunology, vol. 144,no. 10, pp. 3877–80, May 1990. [2] R. Frykberg, J. Banks, “Challenges in the Treatment of Chronic Wounds,”, Advances in Wound Care, vol. 4,no. 9, pp. 560–582, Sept. 2015. [3] Z. Graves, “The Prevalence and Incidence of Chronic Wounds: A Literature Review,” Wound Practice Research, vol. 22, no. 1, pp. 4–19, Mar. 2014. [4] M. Kathawala, “Healing of Chronic Wounds: An Update of Recent Developments and Future Possibilities,”, Tissue Engineering Part B: Reviews, vol. 25,no. 5, pp. 429–444, 2019. [5] P. Martin, R. Nunan, “Cellular and Molecular Mechanisms of Repair in Acute and Chronic Wound Healing,”, British Journal of Dermatology, vol. 173,no. 2, pp. 370–378, 2015. [6] L. Morton, T. Phillips, “Wound Healing and Treating Wounds: Differ- ential Diagnosis and Evaluation of Chronic Wounds,”, Journal of the American Academy of Dermatology, vol. 74,no. 4, pp. 589-605, 2016. [7] D. Nelson, M. Cox, Lehninger Principles of Biochemistry, 7th ed., W.H. Freeman, 2017. VII.APPENDIX

TABLE V: List of Reactions by Reaction Number

Reaction Number reaction name 1 Mitochondrial carrier 2 hydroxysteroid (17-beta) dehydrogenase 7 3 10-Formyltetrahydrofolate mitochondrial transport via 4 2-Aminoacrylate hydrolysis 5 2-Oxoadipate:lipoamde 2-oxidoreductase(decarboxylating acceptor-succinylating) (mitochondria) 6 2-oxoadipate shuttle (cytosol/mitochondria) 7 4-Hydroxyphenylpyruvate:oxygen oxidoreductase 8 3-Dehydrosphinganine reductase 9 3-hydroxyanthranilate 3,4-dioxygenase 10 3-hydroxyisobutyryl-CoA , mitochondrial 11 L-aminoadipate-semialdehyde dehydrogenase (NADH), mitochondrial 12 acetyl-CoA C-acetyltransferase, mitochondrial 13 acetyl-CoA C-acetyltransferase 14 acetyl-CoA C-acetyltransferase, mitochondrial 15 Acetoacetate mitochondrial transport via H+ symport 16 acetyl-CoA carboxylase 17 ATP-Citrate lyase 18 acyl-CoA dehydrogenase (2-methylbutanoyl-CoA), mitochondrial 19 acyl-CoA dehydrogenase (isovaleryl-CoA), mitochondrial 20 acyl-CoA dehydrogenase (isobutyryl-CoA), mitochondrial 21 Aconitate hydratase 22 adenylate kinase 23 adenosine kinase 24 adenylosuccinate lyase 25 adenylosuccinate lyase 26 adenylosuccinate synthase 27 1-acylglycerol-3-phosphate O-acyltransferase 1 28 adenosylhomocysteinase 29 diffusion of S-Adenosyl-L-homocysteine 30 phosphoribosylaminoimidazolecarboxamide formyltransferase 31 phosphoribosylaminoimidazole carboxylase 32 2-oxoglutarate dehydrogenase 33 alpha-ketoglutarate/malate transporter 34 L-alanine reversible transport via proton symport 35 L- 36 2-aminomuconate reductase 37 diffusion of S-Adenosyl-L-methionine 38 arginase 39 L-arginine transport via diffusion 40 R group artificial flux 41 R group to palmitate conversion 42 asparagine synthase (glutamine-hydrolysing) 43 aspartate carbamoyltransferase 44 aspartate-glutamate mitochondrial shuttle 45 46 aspartate transaminase 47 aspartate transaminase 48 ATP synthase (four protons for one ATP) 49 ADP/ATP transporter, mitochondrial 50 Generic human biomass reaction 51 C-3 sterol dehydrogenase 52 production of butyrylcarnitine 53 transport of butyryl carnitine in the mitochondrial matrix for final hydrolysis 54 C-4 sterol methyl oxidase 55 carbamoyl-phosphate synthase 56 phosphatidylinositol synthase 57 phosphatidate cytidylyltransferase 58 choline phosphotransferase 59 choline phosphate cytididyltransferase 60 Choline kinase 61 transport of cholesterol into the cytosol 62 citrulline mitochondrial transport via proton anti 63 citrate transport, mitochondrial 64 cardiolipin synthase 65 CO2 transporter via diffusion 66 CO2 endoplasmic reticular transport via diffusion 67 CO2 transport (diffusion), mitochondrial 68 CoA transporter 69 citrate synthase – cit ac cyc 70 CTP synthase (glutamine) 71 cytochrome c oxidase, mitochondrial Complex IV 72 ubiquinol-6 cytochrome c reductase, Complex III 73 L-cysteine transport via diffusion – extra to cyt 74 cystathionine g-lyase 75 cystathionine beta-synthase 76 cytidylate kinase 77 dATP diffusion in nucleus 78 dCTP diffusion in nucleus 79 dGTP diffusion in nucleus 80 7-dehydrocholesterol reductase 81 dihydrofolate reductase 82 dihydoorotic acid dehydrogenase 83 dihydroorotase 84 6,7-dihydropteridine reductase 85 DM atp(c) 86 dimethylallyltranstransferase 87 diphosphomevalonate decarboxylase 88 dihydrosphingosine N-acyltransferase 89 Desmosterol reductase 90 dTMP kinase 91 dTTP diffusion in nucleus 92 3-beta-hydroxysteroid-delta(8),delta(7)-isomerase 93 3-hydroxyacyl-CoA dehydratase (3-hydroxyisobutyryl-CoA) (mitochondria) 94 3-hydroxyacyl-CoA dehydratase (3-hydroxybutanoyl-CoA) (mitochondria 95 2-Methylprop-2-enoyl-CoA (2-Methylbut-2-enoyl-CoA), mitochondrial 96 enolase 97 electron transfer flavoprotein 98 Electron transfer flavoprotein 99 exchange reaction for L-alanine 100 L-Arginine exchange 101 exchange reaction for L-asparagine 102 L-Aspartate exchange 103 CO2 exchange 104 exchange reaction for L-cysteine 105 D-Glucose exchange 106 exchange reaction for L-glutamine 107 L-Glutamate exchange 108 exchange reaction for Glycine 109 exchange reaction for proton 110 H2O exchange 111 Exchange of hydrosulfide 112 exchange reaction for L-histidine 113 L-Isoleucine exchange 114 L-Lactate exchange 115 L-Leucine exchange 116 L-Lysine exchange 117 L-Methionine exchange 118 Ammonia exchange 119 exchange reaction for oxygen 120 exchange reaction for L-phenylalanine 121 Phosphate exchange 122 L-Proline exchange 123 exchange reaction for L-serine 124 L-Threonine exchange 125 L-Tryptophan exchange 126 L-Tyrosine exchange 127 Urea exchange 128 L-Valine exchange 129 fatty acyl-CoA synthase (n-C10:0CoA) 130 fatty-acyl-CoA synthase 131 fatty-acyl-CoA synthase 132 fatty-acyl-CoA synthase 133 fatty acyl-CoA synthase 134 fructose-bisphosphate aldolase 135 formate dehydrogenase 136 N-Formyl-L-kynurenine amidohydrolase 137 FOR transporter, endoplasmic reticulum 138 transport of Farnesyl diphosphate into the endoplasmic reticulum 139 formimidoyltransferase cyclodeaminase 140 fumarylacetoacetase 141 fumarase, mitochondrial 142 fumarate transport, mitochondrial 143 glycerol-3-phosphate dehydrogenase 144 L-glutamate 5-semialdehyde dehydratase 145 glucose 6-phosphate dehydrogenase 146 UTP-glucose-1-phosphate uridylyltransferase 147 glyceraldehyde-3-phosphate dehydrogenase 148 phosphoribosylglycinamide formyltransferase 149 glycine hydroxymethyltransferase 150 guanylate kinase - 151 glucose transport (uniport) 152 glutamine synthetase 153 L-glutamine transport via electroneutral transporter 154 glutamate dehydrogenase 155 glutamate dehydrogenase 156 Glutamate formimidoyltransferase 157 glutaminase (mitochondrial) 158 glutamine phosphoribosyldiphosphate amidotransferase 159 L-glutamate reversible transport via proton symport, mitochondrial 160 glutaryl-CoA dehydrogenase 161 L-glutamate secretion via secretory vesicle (ATP driven) 162 glycine reversible transport via proton symport 163 glycine passive transport to mitochondria 164 GMP synthase 165 phosphogluconate dehydrogenase 166 glycine N-methyltransferase 167 glycerol-3-phosphate acyltransferase 168 geranyltranstransferase 169 carboxylic acid dissociation 170 H2O transport via diffusion 171 H2O endoplasmic reticulum transport 172 H2O transport, mitochondrial 173 3-hydroxyacyl-CoA dehydrogenase 174 3-hydroxyacyl-CoA dehydrogenase 175 hexokinase 176 Homogentisate:oxygen 1,2-oxidoreductase 177 3-hydroxyisobutyrate dehydrogenase 178 histidase 179 L-histidine transport via diffusion and Transport of L-Histidine by y+ transporter (HISCAT1) 180 3-Hydroxy-L-kynurenine hydrolase 181 Hydroxymethylglutaryl CoA reductase 182 Hydroxymethylglutaryl CoA synthase 183 Hydroxymethylglutaryl-CoA reversible mitochondrial transport 184 H transporter, endoplasmic reticulum 185 Isocitrate dehydrogenase 186 Isoleucine mitochondrial transport 187 isoleucine transaminase 188 L-isoleucine transport via diffusion 189 IMP cyclohydrolase 190 IMP dehydrogenase 191 isopentenyl-diphosphate D-isomerase 192 Imidazolonepropionase 193 KHte 194 kynurenine 3-monooxygenase 195 L-lactate reversible transport via proton symport 196 L-lactate dehydrogenase 197 leucine mitochondrial transport 198 Leucine transaminase 199 L-leucine transport via diffusion 200 lanosterol synthase 201 Lathosterol oxidase 202 L-lysine transport via diffusion 203 Lysine mitochondrial transport via ornithine carrier 204 maleylacetoacetate isomerase 205 malate transport, mitochondrial 206 methylcrotonoyl-CoA carboxylase 207 malate dehydrogenase 208 malate dehydrogenase, mit 209 malic enzyme 210 malic enzyme (NADP) 211 malic enzyme (NADP), mitochondri 212 methionine adenosyltransferase 213 methionine synthase 214 L-methionine transport via diffusion (extracellular to cytosol) 215 mevalonate kinase 216 methylglutaconyl-CoA hydratase 217 myo-inositol 1-phosphatase 218 myo-Inositol-1-phosphate synthase 219 methylmalonyl-CoA epimerase 220 methylmalonyl-CoA mutase 221 methylmalonate-semialdehyde dehydrogenase 222 methenyltetrahydrofolate cyclohydrolase 223 methenyltetrahydrifikate cyclohydrolase, mitochondrial 224 methylenetetrahydrofolate dehydrogenase 225 methylenetetrahydrofolate dehydrogenase (NAD), mitochondrial 226 5,10-methylenetetrahydrofolatereductase 227 NADH dehydrogenase, mitochondrial 228 NADPH transporter, endoplasmic reticulum 229 NADP transporter, endoplasmic reticulum 230 Na+/K+ exchanging ATPase 231 nucleoside-diphosphatase 232 nucleoside-diphosphate kinase (ATP:GDP) 233 nucleoside-diphosphate kinase (ATP:UDP) 234 nucleoside-diphosphate kinase (ATP:CDP) 235 nucleoside-diphosphate kinase (ATP:dTDP) 236 Nucleoside-diphosphate kinase (ATP:dGDP) 237 nucleoside-diphosphate kinase (ATP:dCDP) 238 Nucleoside-diphosphate kinase (ATP:dADP) 239 ammonia transport via proton antiport 240 o2 transport (diffusion) 241 O2 transport, endoplasmic reticulum 242 O2 transport (diffusion) 243 2-Oxobutanoate dehydrogenase, cytosolic 244 3-oxoacid CoA- 245 2-oxoisovalerate dehydrogenase 246 2-oxoisovalerate dehydrogenase 247 2-oxoisovalerate dehydrogenase 248 orotidine-5’-phosphate decarboxylase 249 proline anab 250 orotate phosphoribosyltransferase 251 1-pyrroline-5-carboxylate dehydrogenase 252 pyrroline-5-carboxylate 253 phosphatidylcholine flippase 254 picolinic acid decarboxylase 255 pyruvate carboxylase 256 pyruvate dehydrogenase 257 phosphatidylethanolamine scramblase 258 Phosphoenolpyruvate carboxykinase 259 phosphatidylethanolamine N-methyltransferase 260 phosphofructokinase 261 phosphoglycerate dehydrogenase 262 glucose-6-phosphate isomerase 263 phosphoglycerate kinase 264 6-phosphogluconolactonase 265 phosphoglycerate mutase 266 phosphoglucomutase 267 Phosphatidylglycerol phosphate phosphatase 268 phosphatidyl-CMP: glycerophosphate phosphatidyltransferase 269 L-phenylalanine transport via diffusion (extracellular to cytosol) 270 L-Phenylalanine,:oxygen oxidoreductase (4-hydroxylating) 271 phosphate transporter, mitochondrial 272 phosphate transport, endoplasmic reticulum 273 phosphomevalonate kinase 274 inorganic diphosphatase 275 inorganic diphosphatase, endoplasmic reticulum 276 phosphatidic acid phosphatase 277 Propionyl-CoA carboxylase, mitochondrial 278 phosphoribosylglycinamide synthase 279 phosphoribosylaminoimidazolesuccinocarboxamide synthase 280 phosphoribosylformylglycinamidine synthase 281 L-proline reversible transport via proton symport 282 L-proline transport, mitochondrial 283 phosphoribosylpyrophosphate synthetase 284 Phosphatidylserine decarboxylase 285 phosphoserine transaminase 286 phosphatidylserine flippase 287 phosphoserine phosphatase 288 Phosphatidylserine synthase 289 pyruvate kinase 290 pyruvate mitochondrial transport via proton symport 291 L-Glutamate 5-semialdehyde:NAD+ oxidoreductase 292 L-Alanine:2-oxoglutarate aminotransferase 293 L-Asparagine amidohydrolase 294 L-Cysteine L-homocysteine-lyase 295 glycine synthase Nitrogen metabolism 296 L-2-Aminoadipate:2-oxoglutarate aminotransferase 297 N6-(L-1,3-Dicarboxypropyl)-L-lysine:NAD+oxidoreductase 298 2-Aminomuconate semialdehyde:NAD+ 6-oxidoreductase 299 2-(Formamido)-N1-(5-phosphoribosyl)acetamidine cyclo- (ADP-forming) 300 4,4-dimethyl-5a-cholesta-8,24-dien-3b-ol:NADP+ D14-oxidoreductase 301 Lanosterol,NADPH:oxygen oxidoreductase (14-methyl cleaving) 302 Free diffusion 303 Free diffusion 304 Free diffusion 305 Mitochondrial Carrier 306 Major Facilitator 307 Biosynthesis of Enzyme catalyzed 308 Active transport 309 Facilitated diffusion 310 Transport reaction 311 Transport reaction 312 Utilized transport 313 Major Facilitator 314 Mitochondrial Carrier 315 Major Facilitator 316 Major Facilitator 317 Major Facilitator 318 Major Facilitator 319 RE2675 - ? 320 ribonucleoside-diphosphate reductase 321 ribonucleoside-diphosphate reductase (GDP) 322 ribonucleoside-diphosphate reductase (CDP) 323 ribonucleoside-diphosphate reductase (UDP) 324 ribulose 5-phosphate 3-epimerase 325 ribose-5-phosphate isomerase 326 R total flux 327 saccharopine dehydrogenase 328 Sarcosine transport (mitochondrial) 329 Sarcosine dehydrogenase - 330 L-Serine hydro-lyase 331 serine C-palmitoyltransferase 332 Sphingomyelin synthase 333 Squalene epoxidase 334 Squalene synthase 335 succinate dehydrogenase 336 Succinate–CoA ligase 337 transaldolase 338 Tetrahydrobiopterin-4a-carbinolamine dehydratase 6,7-dihydropteridine reductase 339 5,6,7,8-Tetrahydrofolate transport, diffusion 340 L-threonine deaminase – branched chain 341 transketolase 342 transketolase 343 thymidylate synthase 344 triose-phosphate isomerase 345 thioredoxin reductase 346 L-Tryptophan:oxygen 2,3-oxidoreductase (decyclizing) 347 L-tryptophan transport 348 L-tyrosine transport 349 tyrosine transaminase 350 UMP kinase 351 URCN 352 Urea transport via facilitate diffusion 353 Valine reversible mitochondrial transport 354 valine transaminase, mitochondrial 355 L-valine transport via diffusion (extracellular to cytosol) 356 Exchange of GlutaMAX 357 Hydrolysis of GlutaMAX