1
Characterization of the Carnitine Transporter, OCTN2: Functional Impact of Mutations and Its
Role in COVID-19 Treatment Related Drug- Drug-Interactions
Author: Mattias Rödin
Degree Project in Pharmacokinetics, 30 hp, Spring Semester 2020
Supervisors: Kathleen M. Giacomini & Megan Koleske
Examiners: Maria Kjellson & Margareta Hammarlund-Udenaes
Giacomini Laboratory
Membrane Transporter Functions and Genomics Laboratory Department of Bioengineering and Therapeutic Sciences University of California, San Francisco
Division for Pharmacokinetics
Department of Pharmaceutical Life Sciences Faculty of Pharmacy
Uppsala University
2
Abstract
Organic Cation Transporter Novel 2 (OCTN2) belongs to the Solute Carrier 22 (SLC22) family and is transcribed from the SLC22A5 gene. OCTN2 mainly transports carnitine, vital for the individual’s fatty acid metabolism. Deleterious mutations in SLC22A5 may result in low carnitine levels and lead to Carnitine Transporter Deficiency (CTD). The symptoms can be as severe as hypoketotic hypoglycemia, liver failure and arrythmia.
OCTN2 may also be inhibited by drugs and lead to imbalance in carnitine homeostasis.
The goal with this study was to quantify the impact of 28 mostly novel point mutations in SLC22A5. Of these, 10 has been observed in CTD-patients, and 15 was found in GnomAD dataset v2.1.1 and predicted to be some of the most deleterious by Combined Annotation Dependent Depletion (CADD) and Polyphen-2. Additionally, 25 drugs currently clinically tested for Covid-19 was screened for potential inhibition of OCTN2.
Human Embryonic Kidney-293 (HEK-293) cells were transfected with EV, WT-SLC22A5 or mutants and C
14-carnitine uptake studies was conducted. For the Covid-19 screen the investigational drugs were also added and instead only WT-SLC22A5 was used.
Interestingly, and quite expected, all mutations found in CTD-patients proved to drastically impair OCTN2. G152D, Y211H, P266L, S362L, G411V and R471C completely abolished its function. N91S, D139N, L202P, V527L, W275C and R227S drastically reduced OCTN2 uptake to 42-60 % of WT-OCTN2 function. Furthermore, most deleterious mutations were found in transmembrane regions of the transporter. Ritonavir and Tofacitinib fully inhibited OCTN2 at 100
µM while Umifenovir reduced carnitine uptake to close to EV at the same concentration.This study concludes that the SLC22A5-mutations found in CTD-patients are more deleterious than the ones found purely in GnomAD. Mutations in transmembrane domains tend to be more deleterious than mutations in aqueous parts of the protein.
CADD and Polyphen-2 are quite poor predictors of mutational impact on SLC22A5,
expressing the need for improvement of computational predictors. Ritonavir may inhibit
OCTN2, reduce carnitine concentration and lead to Cardiovascular Diseases (CVD) and
might be a cause for concern whether or not it is approved for Covid-19 as it is already
a life-long treatment for HIV-patients (74,75).
3
Populärvetenskapligt Abstrakt
Cellerna i kroppen kräver många olika molekyler såsom näringsämnen, signalsubstanser och aminosyror för att fungera. Proteinerna som möjliggör cellernas upptag, och även utflöde, av dessa molekyler kallas transportörer. Dessa befinner sig på cellernas membran och agerar som en länk mellan cellen och dess omgivning. Karnitintransportproteinet OCTN2 är ett exempel på en transportör som tar upp karnitin till cellen. Karnitin möjliggör nedbrytningen av långa fettsyror, en viktig komponent för cellens möjlighet att utvinna energi. OCTN2 finns i de flesta av kroppens celler, särskilt i njurarna där transportören motverkar att alltför mycket karnitin utsöndras i urinen. Skadliga mutationer i SLC22A5, genen som kodar för OCTN2, minskar dess effektivitet vilket reducerar karnitinkoncentrationen i kroppen och kan leda till Primär Karnitinbrist (CTD). Symtom för denna sjukdom är bland annat låga koncentrationer ketoner och glukos, leversvikt och dödliga arrytmier. Fortsättningsvis kan OCTN2, precis som alla andra proteiner i kroppen, bli inhiberade av olika läkemedel vilket även det kan riskera låg karnitinkoncentration. Målet med den här studien var att avgöra hur 28 olika punktmutationer i SLC22A5 påverkar funktionen hos OCTN2. Av dessa har 10 observerats i CTD-patienter. 15 av mutationerna har ej observerats i CTD-patienter men har, med hjälp av Combined Annotation Dependent Depletion (CADD) och Polyphen-2, predikterats att vara skadliga. Det andra målet var att tillsätta 25 läkemedel vilka för tillfället testas som behandling av Covid-19, såsom Ritonavir och Tofacitinib, och avgöra om de inhiberar OCTN2:s normala funktion.
Humana njurceller (HEK-293) stimulerades att tillverka OCTN2, muterat OCTN2 eller ingetdera.
Karnitinupptaget bestämdes sedan genom att tillsätta karnitin märkt med radioaktivt kol. Detta repeterades men då vid samtidig tillsats av de läkemedel som testas för Covid-19.
Studien visade att de mutationer som observerats hos CTD-patienter var skadligare än de mutationerna i individer som ej har diagnosticerats. 6 av mutationerna som setts i CTD-patienter inhiberade OCTN2 totalt och de övriga 4 minskade funktionen drastiskt. Av de resterande mutationerna minskade endast 2 funktionen påtagligt, medan en mutation ökade funktionen med ca 22 %. 100 µM Ritonavir och Tofacitinib inhiberade funktionen hos OCTN2 till fullo.
Slutsatsen är att mutationerna som hittats i CTD-patienter med säkerhet är orsak till diagnosen. Det finns utrymme för förbättring hos CADD och Polyphen-2 och deras förmåga att förutse en mutations inverkan på OCTN2. Slutligen, Ritonavir kan vara skadlig då den reducerar funktionen hos OCTN2 kraftigt, har påvisat minskad karnitinkoncentration och redan används som en livslång behandling av HIV.
4
Table of contents
1. Introduction ... 5
1.1 Transporters ... 5
1.2 Organic Cation Transporter Novel 2 (OCTN2) ... 5
1.3 Carnitine ... 5
1.4 Carnitine Transporter Deficiency (CTD) ... 6
1.5 Mutations & characterization of OCTN2 ... 7
1.6 Covid-19... 7
1.7 Covid-19 clinical trials ... 8
1.8 Drug-screening against human transporters ... 8
1.9 End-goal: Mutation characterization ... 9
1.10 End-goal: Covid-19 drug candidate screen ... 10
1.11 Aim ... 10
2.0 Methods ... 10
2.1 Choice of mutations and CADD scores ... 10
2.2 Site directed mutagenesis (SDM) ... 12
2.3 Miniprep and sequence validation ... 12
2.4 Growth and splitting of HEK-293 cells ... 13
2.5 Seeding of 96 well-plate ... 13
2.6 Transfection of HEK-293 cells ... 13
2.7 Carnitine uptake assay ... 13
2.8 Covid-19 drug screen on solute carrier OCTN2 ... 14
2.9 Data Analysis ... 14
3.0 Results ... 15
4.0 Discussion ... 20
4.1 Discussion of the uptake results ... 20
4.1.1 Deleterious mutations ... 20
4.1.2 Mutations not deleterious to OCTN2 function ... 20
4.1.3 Mutations not observed in gnomAD ... 20
4.1.4 Comparison between results and available studies on the mutations ... 21
4.2 Assay Limitations ... 21
4.2.1 Low/negative protein counts ... 21
4.2.2 Edge-effect and its impact on the results ... 21
4.2.3 Exclusion of wells and its impact on the results ... 21
4.3 CADD and Polyphen-2 predictions ... 22
4.4 Amino acid substitutions: Location and impact ... 22
4.6 Covid-19 screen results... 23
5.0 Conclusions ... 24
5.1 OCTN2 mutations ... 24
5.2 Covid-19 drug screen conclusions ... 24
5.3 Conclusion summary ... 25
6.0 Special thanks ... 25
7.0 References ... 26
5
1. Introduction
1.1 Transporters
Transporters are the gates of cells, allowing them to survive and communicate with surrounding cells and their close environment. Some transporters help with the influx of small molecules into a cell, others in efflux, while some transport molecules in both directions. There are a vast number of different substrates transported throughout the body such as drugs, nutrients, metabolites, cell debris, signaling molecules, etc. Some transporters have a narrow spectrum of substrates while others have a broad spectrum. Two superfamilies of transporters are the ATP-binding Cassette (ABC) and the Solute Carrier (SLC) transporters. In the human genome, the ABC superfamily consists of 48 closely related transporters divided into several subfamilies (1). Though new families and members are constantly being added, the current literature reports 55 SLC families, one of which is the SLC22 family (2,3). 23 transporters can be found in the SLC22A family, divided into at least 6 subfamilies, as follows; Organic Anion Transporter (OAT), OAT-like, OAT-related, Organic Cation Transporter (OCT), Organic Cation/Carnitine Transporter (OCTN) and OCT/OCTN-related (4). ABC-transporters and SLC- transporters are overlapping in substrate specificity and transport a vast number of different drugs, nutrients and other molecules. Potential drug-drug interactions (DDI) due to inhibition, induction and modulation of these transporters are a possibility (5). This dissertation research will focus on the Organic Cation Transporter Novel 2 (OCTN2).
1.2 Organic Cation Transporter Novel 2 (OCTN2)
OCTN2 belongs to the SLC22 transporter family and is encoded by the SLC22A5 gene with a full- length transcript of 3237 base pairs. Within the transcript there are 10 coding exons with the total number of 1674 base pairs (6). This translates to a protein with 557 amino acids resulting in a transporter with 12 predicted transmembrane domains, 6 extracellular loops and 7 intracellular loops. A long extracellular loop of 100 amino acids can be found between the first and the second transmembrane section, and this loop seems to be conserved among SLC22-transporters and is thought to have a key role in function, see Figure 1. It has in previous studies been concluded that this protein transports cations such as tetraethylammonium and the zwitterions carnitine and acylcarnitines (7). Carnitine transport is sodium-dependent while cation transport is sodium-independent. The primary substrate for OCTN2 is carnitine with a Km of 3-5 µM. The transporter is abundantly expressed in most tissues, with high levels of expression in muscle, kidney and in the cerebellum. (8,9).
1.3 Carnitine
Carnitine is an amino acid derivative, which includes a hydrophilic quaternary amine. Its main function is to conjugate to long-chain fatty acids and transport them from the cytosol into the mitochondria for
Figure 1. Predicted 2-D transmembrane structure of OCTN2 in the plasma membrane (9). The highlighted amino acid substitutions are not connected with the following experiment and should be ignored.
6 subsequent beta-oxidation. If the carnitine level for an individual is too low, fatty acid metabolism may be drastically reduced, and the body will no longer be able to rely on long-chain fatty acids as a major energy source. Glucose and small fatty acids will instead be the major source of energy. As glycogen storages are depleted after a few hours the individual will suffer from hypoglycemia. Furthermore, as fatty acid oxidation is reduced, there will be no or minimal ketones in the plasma which will lead to hypoketosis. Additionally, excessive fat will be stored throughout the body since fatty acid oxidation is not taking place, including cardiac and skeletal muscle as well as the liver. This aberrant lipid storage could lead to complications such as myopathies in skeletal and cardiac muscles as well as liver failure.
Furthermore, the excessive fat stored in the cardiac cells has the potential to interfere with the electrical conduction in the heart and can cause severe disruptions such as arrythmias (9).
Humans can synthesize carnitine to a small extent, but mostly obtain it from the intake of animal-based food sources. Carnitine is transported by OCTN2 which is highly expressed in the intestine for absorption and the kidney proximal tubule for carnitine reabsorption. Under healthy conditions, carnitine reabsorption is inversely proportional to intake. Additionally, carnitine binds acyl residues which helps containing/removing abnormal organic acids which could potentially be harmful (9). See Figure 2 for an overview of carnitine’s cycle throughout the human body (10).
1.4 Carnitine Transporter Deficiency (CTD)
A major determinant of carnitine levels is OCTN2-mediated carnitine absorption in the intestine and reabsorption in the kidneys (11). The normal plasma concentration of carnitine in a healthy individual ranges between 25-50 µM. Lower concentrations raise concerns for dysfunction and a concentration lower than 8 µM is a sign that the individual can suffer from Carnitine Transporter Deficiency (CTD) and might need supplemental carnitine (9). CTD, also called Systemic Carnitine Deficiency (SCD), is a rare recessive disease with a incidence of 1:142 000 in the U.S ranging to as high as 1:300 in the Faroe Islands (12). The estimated prevalence of CTD in newborns in Europe is 1:20 000 – 1:70 000 (13). This disease results in skeletal and cardiac myopathy, hypoketotic hypoglycemia, encephalopathy and even acute liver failure. If left untreated, this disease can result in death caused by metabolic crisis or
7 arrythmias due to the altering of electrical activity in the cardiac cells by fatty acid accumulation. CTD is caused by a dysfunction in the carnitine transporter as a result of different mutations in the gene SLC22A5 coding for this protein (9).
1.5 Mutations & characterization of OCTN2
The SLC22A5 gene is not yet fully understood. Although the OCTN2 protein has no available crystal structure, predictive methods have assigned its amino acids to various protein domains, see Figure 3 (14). Though, increased knowledge of the transporter’s 3D-structure, its folding in the cellular membrane as well as what parts of the protein have more or less impact on the transporter’s function is needed.
1.6 Covid-19
Recently, a new virus surfaced and has caused a pandemic affecting all countries over the globe. In the beginning of May 2020 3,7 million confirmed cases had been reported whereas 258 000 had died due to the resulting disease, designated COVID-19 (15). Affected individuals’ health is compromised, and the health care system is strained along with the global economy. Countries are under lockdown and the beginning of 2020 has been plagued with a huge challenge, causing major effects at both the individual level as well as industrial and government levels. The virus is called SARS-CoV-2 (Severe Acute Respiratory Syndrome Coronavirus 2) and belongs to the Coronaviridae family, which is divided into four genera; Alpha-, Beta-, Gamma- and Deltacoronavirus. SARS-CoV-2, along with SARS-CoV (Severe Acute Respiratory Syndrome Coronavirus) and MERS-CoV (Middle East Respiratory Syndrome-related Coronavirus) all belong to the Betacoronavirus group. In total there are seven human coronavirus, of which 5 belong to the Betacoronavirus and two belong to the Alphacoronavirus group (16,17). SARS-CoV emerged in 2003 while MERS-CoV followed up in 2012, both without any present vaccine. The mortality rate of SARS-CoV was 10-15 % while MERS-CoV was as high as 37 % (18–
20). WHO estimated the mortality rate of SARS-CoV-2 to 3,4 % globally (21,22) but in the beginning of May almost 7 % of the people infected had died (15).
Figure 3. An overview of the OCTN2 transporter’s different protein domains. TM = Transmembrane, EL = Extracellular loop and IL = Intracellular loop. The numbers represent the amino acid sequence (15).
8
1.7 Covid-19 clinical trials
The discovery, development, and approval of a novel compound showing antiviral efficacy towards SARS-Cov-2 as well as a satisfying safety profile could take years. The average time to approval for a novel compound in the drug discovery pipeline is about 12 years (23). Instead, to cure this virus or at least minimize its spread at the present time scientists all over the globe are conducting research on alternative medicines. Currently a wide variety of clinical trials are recruiting and beginning their studies on several approved drugs such as, but not limited to, Baricitinib, Hydroxycholoroquine and chloroquine, with the goal of repurposing an already approved drug for treatment of COVID-19.
Additionally, a drug not yet approved called Remdesivir, is also showing potential effect on SARS- CoV-2 (20,24,25).
1.8 Drug-screening against human transporters
As drugs are in clinical trials for COVID-19 treatment, there are many factors that will influence a successful candidate. The drugs tested for Covid-19 have a risk, as do all other drugs, to result in unwanted side effects. These can be mild, moderate or very severe. One reason for such adverse events is the unwanted impact on the innate transporters in the human body. Drug transporters are the site of many clinically relevant drug-drug interactions (DDIs) (26). Medicines can inhibit or induce transporters either directly or indirectly, which can result in an imbalance for the cells in the body. For example, if OCTN2 is inhibited reabsorption of carnitine could be drastically reduced resulting in lower plasma carnitine concentrations. With less carnitine, intake of long fatty-acids into the mitochondria could also be reduced and the metabolism in the body will be imbalanced, as mentioned earlier (9,27). Of particular concern in the case of COVID-19 is the fact that patients with co-morbidities such as diabetes and hypertension are more at risk and have more severe cases. These patients are frequently on a multitude of drugs to treat their co-morbidities and are at increased risk for DDIs, particularly if they were to be given an experimental drug treatment for COVID-19. There is a vast landscape of transporters and proteins with the potential to be inhibited or induced by experimental therapeutics and all cannot be imagined fitting into one or even several papers.
9
Table 1. Overview of 25 drugs currently tested for covid-19, with the exception of Triazavirin which could be tested in the future. Each drug’s predicted mechanism of action for covid-19 treatment and how many clinical trials currently enrolled are specified.
Drug Current clinical trials found in Clinicaltrials.gov
Reference
Chloroquine 59 (28)
Hydroxychloroquine 185 (29)
Azithromycin 78 (30)
Remdesivir 22 (31)
Umifenovir (Arbidol) 8 (32)
Favipiravir 14 (33)
Lopinavir 58 (34)
Ritonavir 60 (35)
Darunavir 5 (36)
Baricitinib 10 (37)
Camostat 5 (38)
Fingolimod 1 (39)
Thalidomide 3 (40)
Colchicine 10 (41)
Losartan 10 (42)
Oseltamivir 17 (43)
Sildenafil 1 (44)
Tetrandrine 1 (45)
Piclidenoson 1 (46)
Ribavirin 7 (47)
Tofacitinib 1 (48)
Leflunomide 1 (49)
Prazosin 1 (50)
Ruxolitinib 14 (51)
Triazavirin 0
1.9 End-goal: Mutation characterization
The goal of this dissertation research is to functionally characterize genetic mutations in OCTN2 for the purposes of (a) understanding structure function relationships in the transporter, and (b) informing computational methods for predicting function of missense variants in the transporter. This latter goal is tied mainly to informing newborn screening, which amongst other diseases tries to identify CTD.
Unfortunately, current predictive methods are inadequate giving false positives as well as negatives subsequently resulting in failure to treat infants with CTD or treating healthy individuals unnecessarily.
To achieve our goal, novel point mutations observed in humans that mostly have not yet been studied were characterized. 25 point mutations in the SLC22A5 gene were chosen, of which 15 are the most frequent in 5 different ancestry groups (Europeans, Latino, African, South Asian and East Asian) according to the Genome Aggregation Database (GnomAD) while the remaining 10 mutations have been observed in patients suffering from CTD. Out of these only a few has ever been characterized before. The mutations were introduced in the SLC22A5 gene and the plasmids was transfected into Human Embryonic Kidney 293 (HEK-293) cells.
10
1.10 End-goal: Covid-19 drug candidate screen
A second goal of this dissertation research is to screen 25 drugs currently in clinical trials for Covid-19 (see Table 1) against OCTN2 to identify drugs that may exacerbate carnitine deficiency in susceptible individuals. Those individuals may include those with malnutrition, individuals with CTD or heterozygotes with low carnitine levels. These data are then to be used to improve the health care of patients.
1.11 Aim
To reveal how missense mutations can affect this transporter’s function, increase the knowledge of OCTN2’s structure for future computational predictions and determine the potential for COVID-19 treatment inhibition of OCTN2.
2.0 Methods
2.1 Choice of mutations and CADD scores
25 mutations in the SLC22A5 gene were chosen in total for experimentation, see Table 3. Of these, 10 mutations were chosen because they have been found in CTD patients but have not yet been extensively studied (11,12,52–59). The remaining 15 were found in the GnomAD v2.1.1 dataset with the SLC22A5- 001 transcript as a reference. Missense mutations were the subject of this study as they are the most common type of mutation observed in CTD patients and are easily assayed (6,60). The mutations were filtered by frequency and the most frequent mutations were chosen. To enrich for underrepresented populations, 3 mutations were studied from each of five ancestry groups. Mutations occurring in 2 or more ancestry groups were added to a “shared” group and excluded from the ancestry groups of which they were found. Therefore, the mutations in each ancestry group’s is exclusively found in that specific ancestry group. In total 5 ancestry groups were included; African, Latino, European, South Asian and East Asian. The shared group was not studied due to lack of time.
The Combined Annotation Dependent Depletion (CADD) tool was in turn used on each single nucleotide mutation and CADD PHRED-scores annotated. These scores indicate the likelihood that a specific mutation will be observed or not. A higher PHRED-score suggests that a mutation is less likely to be observed and therefore is probably more deleterious (61). In each ancestry group the three mutations with the highest PHRED-scores were chosen. All 25 mutations were also run through the Polyphen-2 database to predict mutations impact with yet another computational model (62). The 25 mutations were introduced individually in the SLC22A5_CGFP_017destination plasmid from Giacomini lab in UCSF, San Francisco, California, USA. See Figure 3 below (63). The 25 plasmids were created by site-directed mutagenesis by Genscript in Piscataway, New Jersey, USA (64).
The frequency of SLC22A5 mutations found in the gnomAD dataset was spread evenly across the exons but two protein domains (extracellular loop 2 and 5) had no observed mutations and transmembrane domain 5 had a very low frequency of mutations, see Table 2 and Figure 4 below. These domains are small and consist of 4, 3 and 21 amino acids, respectively, which may explain why there are less mutations occurring there. Another explanation could be that these domains are highly conserved and therefore very important to the transporter’s function. To test this, three extra mutations were introduced, one in each domain. The mutations were created through Site Directed Mutagenesis (SDM) in the lab.
M195L, I237M and T429A were the chosen mutations in these three domains.
11
Table 2. The 199 SLC22A5 mutations with the highest allele frequency from gnomAD and their corresponding protein domains. TM = Transmembrane, EL = Extracellular loop, IL = Intracellular loop
Figure 4. Plasmid used for OCTN2 expression in HEK-293 cells. The plasmid has the SLC22A5 gene conjugated to a fluorescent GFP-tag as well as resistance genes for kanamycin, penicillin and streptomycin. The empty vector lacks the SLC22A5 gene, the wild type has the SLC22A5 gene and the plasmids with mutations have a point mutation in the SLC22A5 gene. Plasmid is from Giacomini lab in UCSF, San Francisco, California, USA.
12
Table 3. 25 chosen mutations selected for study based on the method described above in [1.6]. Nucleotide change is the point mutation made in the SLC22A5 gene at that specific position, amino acid substitution is the resulting change in the protein.
The mutation abbreviation is used in this paper to simplify in diagrams. Ancestry group from which the mutation was identified in gnomAD is also specified. The CADD score and Polyphen-2 score are the scores found for each mutation by running the mutation through the respective computational model. The highlight colors are polyphen-2 predictions where green means
“benign”, orange means “possibly damaging” and red means “probably damaging”
Nucleotide change
Amino acid substitution
Group CADD
PHRED score
Polyphen-2 score
c.272A>G N91S Disease 24,4 0,943
c.415G>A D139N Disease 23,4 0,048
c.455G>A G152D Disease 26,8 1,000
c.605T>C L202P Disease 28 0,977
c.631T>C Y211H Disease 27,8 1,000
c.797C>T P266L Disease 25,9 0,933
c.1085C>T S362L Disease 31 1,000
c.1232G>T G411V Disease 28,3 0,999
c.1411C>T R471C Disease 31 1,000
c.1579G>C V527L Disease 22,5 0,000
c.825G>T W275C African 35 0,931
c.1586G>T G529V African 34 0,871
c.158C>G P53R African 30 1,000
c.338G>A C113Y East Asian 32 1,000
c.679C>A R227S East Asian 29,9 1,000
c.700G>C G234R East Asian 29,5 1,000
c.391G>A E131K European 32 0,999
c.1262C>T P421L European 30 0,965
c.1475A>G Y492C European 29,3 1,000
c.496A>G R166G Latino 33 0,879
c.1138G>A A380T Latino 29 0,911
c.341T>C L114P Latino 28,3 0,972
c.304G>A D102N South Asian 32 1,000
c.43G>T G15W South Asian 32 1,000
c.1310T>C F437S South Asian 29,9 0,988
2.2 Site directed mutagenesis (SDM)
6 different primers (3 forward and 3 corresponding reverse) were ordered from IDT to synthesize the following mutations: M195L, I237M and T429A. The primers melting temperatures were calculated with IDT’s Oligoanalyzer Tool to determine the temperature for the subsequent thermocycling (65) and the high-fidelity PCR instructions from IDT were used to conduct the experiment.
2.3 Miniprep and sequence validation
To validate the sequences 5 isolated colonies from each plate were picked with a pipet tip under sterile conditions and placed in separate culture tubes with 4 ml kanamycin-containing LB broth and grown overnight for 17 hours in 37 oC and then stored at 4 oC. Subsequently, to isolate the DNA and prepare the plasmids for sequence validation the Zyppy™ Plasmid Miniprep Kit was used (66). When the miniprep was done the samples was sent to McLab sequencing for validation.
13
2.4 Growth and splitting of HEK-293 cells
Nutrition media was prepared containing Dulbecco’s Modified Eagle Medium (DMEM), 10% Fetal Bovine Serum (FBS) and 1% Penicillin-Streptomycin (PS). HEK-293 cells were grown in 15 mL nutrition mixture on a 100 mm plate at 37oC with 5% CO2. To keep the cells healthy, they were split at 90-100 % confluency. This was done by removing the supernatant from the plate, washing the plate with 5 ml 37oC Phosphate-Buffered Saline (PBS) and the addition of 2 ml 37oC trypsin-EDTA. The plate was carefully shaken to detach the cells after which 10 ml 37oC nutrition media was added. The mixture, containing the cells, was transferred to a 15 ml falcon tube and centrifuged for 5 minutes at 3000 rpm.
The supernatant was discarded, and the pellet was reconstituted in 5 ml 37oC nutrition media. 1-2 ml was transferred to the 100 mm plate and 15 ml nutrition was added before placing the plate in the incubator at 37oC with 5% CO2. DMEM, FBS, PS, PBS and trypsin-EDTA were acquired from Gibco© in Carlsbad, California, USA.
2.5 Seeding of 96 well-plate
10 µL of the suspension with the dissolved pellet from step [2.2] was mixed with 10 µL of Tryphan Blue (Gibco© in Carlsbad, California, USA). 10 µL of the mixture was analyzed in a TC20™ Automated Cell Counter (Bio-Rad in Hercules, California, USA) and the cells counted. 30 000 cells in 200 µL nutrition media were added to each well in a 96-well plate using a multi-channel pipet. The well-plate was incubated for 24 hours at 37oC.
For assays done in a 24 well-plate, 180 000 cells and 1 mL nutrition media were added to each well, otherwise the procedure stated above was followed.
2.6 Transfection of HEK-293 cells
Microscopy was used to visualize confluency of the cells in the well-plate, which was about 70%. The HEK-293 cells were transfected with 100 ng DNA using Opti-MEM (Gibco© in Carlsbad, California, USA) and Lipofectamine LTX (Invitrogen© in Carlsbad, California, USA)
For the 24 well-plate uptake assay the same transfection procedure was used but instead with 500 ng DNA/well and 1 mL nutrition media/well.
2.7 Carnitine uptake assay
Hank’s Balanced Salt Solution (HBSS) from Gibco™ in Carlsbad, California, USA was heated to 37oC.
The 96 well-plate with the transfected HEK-293 cells was grown to about 100% confluency, the supernatant was removed and 300 µL HBSS was added to each well before incubation for 8 minutes at 37oC. The HBSS was discarded and 75 µL uptake buffer containing 0,89 µM C14-Carnitine in HBSS was added to each well before the plate was incubated for 10 minutes at 37oC for the uptake assay. The uptake buffer was removed, and the plate put on ice. Each well was washed with 300 µL ice-cold HBSS.
To break up the cells 250 µL lysis buffer was added to each well and the well plate was shaken at 300 rpm for 90 minutes at room temperature. 200 µL cell lysates from each well was added to scintillation tubes and liquid scintillation fluid and a scintillation counter was used to analyze the amount of radioactive carnitine in each well.
14 For the assay made in a 24 well-plate the same uptake procedure was used, instead with 300 µL uptake buffer and 750 µL lysis buffer of which 690 µL was subsequently transferred to the scintillation tubes.
To correct for differences in each well’s cell density a BCA assay was done using The Pierce™ BCA Protein Assay Kit from Fisher Scientific in Carlsbad, California, USA. 25 µL from each well was transferred to a 96 well-plate and a standard solution of albumin with 12 different concentrations ranging from 2 mg/ml to 0,98 µg/ml was made. Then the Pierce™ BCA Protein Assay Kit’s protocol was followed step by step. A standard curve was created with the absorbance of the albumin solutions and their known concentrations. The equation of the curve was then used to calculate the protein amount in each well.
2.8 Covid-19 drug screen on solute carrier OCTN2
This method followed the same protocol as the carnitine uptake assay previously described, with a different layout of the 96 well-plate. 1 µM carnitine was also added in addition to the 0,89 µM radiolabeled carnitine. 12 wells were transfected with an empty vector while the remaining 84 wells were transfected with SLC22A5 WT (OCTN2) plasmid, see Figure 3. The 25 drugs were tested in triplicates and 100 µM Verapamil was used as the control inhibitor. All drugs were tested as inhibitors at a fixed concentration and added to the uptake buffer for their respective wells. All 25 drugs were screened at 100 µM except for Azithromycin, Baricitinib and Tetrandrine which due to solubility issues were tested at the concentrations 50 µM, 50 µM and 10 µM, respectively. The EV wells and four OCTN2 wells were screened without any inhibitors present and instead only DMSO and carnitine.
2.9 Data Analysis
The disintegration per minute (DPM) for each well was divided by the average protein count in the triplicate for that mutation. All uptake was normalized to WT function. All data in the bar diagrams are presented in % uptake function related to WT. Outlier tests was made on the wells with low protein values and one WT-well, two EV-wells and several mutation-wells was excluded, see Figure 6.
15
3.0 Results
As a control, we examined the function of mutations not found in gnomAD. See Figure 5 for the uptake results of M195L, I237M and T429A. Though T429A had about 50% function, a Student’s t-test did not show a significant difference between this or other mutations and WT-OCTN2.
Results analyzing the function of 25 mutations found using method [2.1] are shown in Figure 7.
In the functional analysis, results from the mutants D102N, G15W and F437S was excluded because two of each of their triplicate determinations had extremely low or negative protein counts caused by edge-effects (see Figure 6). Because both bottom rows of the plate were affected by edge-effects, A380T and L114P were also excluded. Thus, A380T, L114P, D102N, G15W and F437S are excluded in all further analyses.
Carnitine uptake by OCNT2 containing the missense mutations G529V, P53R, C113Y, G234R, E131K, P421L and Y492C was not significantly different from uptake by WT-OCTN2. For the mutations G152D, Y211H, P266L, S362L, G411V and R471C, carnitine uptake was 16 %, 21 %, 28 %, 25 %, 29
% and 17 %, respectively, compared to WT-OCTN2. All mutations found in CTD-patients significantly impaired the function of OCTN2 (Figure 7).
Some mutations partially decreased the OCTN2-mediated carnitine transport to 45-60 % of normal function. N91S, D139N, L202P, V527L, W275C and R227S had the following carnitine uptakes compared to WT-OCTN2: 45 %, 55 %, 44 %, 58 %, 61 % and 60 %, respectively. Conversely, R166G significantly increased the transporting function of OCTN2 to 123 % of WT-OCTN2.
Figure 5. Function of mutations with respect to carnitine transport compared with SCL22A5 WT.
Uptake was made following the procedure in [2.7] in a 24 well-plate.
0,00
1,00 1,08
0,84
0,54
-0,20 0,00 0,20 0,40 0,60 0,80 1,00 1,20
EV WT M195L I237M T429A
% WT
Avg %WT SLC22A5 carnitine uptake
16 The results were analyzed further to identify trends between different parameters. A significant difference (p < 0,0001) was observed between the function of the mutations found in CTD-patients compared to mutations found in different ancestry groups from the gnomAD dataset (Figure 8). Average carnitine uptake for mutations found in CTD-patients is ~35 % of WT whereas average uptake is ~95 % of WT for the other mutations.
Figure 6. Plate layout of the 96 well-plate for the carnitine uptake assay. All wells highlighted in yellow have been excluded due to negative/low protein count or suspected edge-effect.
Figure 7. The DPM has been normalized to the average protein count for that plasmid and then divided by the uptake of WT-OCTN2. Black dots = values for one well in that mutation type, red dots = average values for that triplicate, red line = standard deviation. The highlighted wells shown in Figure 6 have been excluded. All data has been multiplied by 100 to get a % comparison with the function of WT.
17 A weak correlation was observed between carnitine uptake and allele frequency of each of the mutations, see Figure 9. The correlation suggests that an OCTN2 mutation occurring more frequently has a higher uptake than one with a lower allele frequency and is consistent with purifying selection. However, the p-value = 0,30 and is not significant.
Figure 9. The uptake for each mutation compared to WT is plotted on the y-axis compared to the mutation’s allele frequency in GnomAD. 17 mutations were included. All mutations observed in CTD-patients was included, except for S362L and G411V which were excluded as no allele data for those two mutations could be found in the GnomAD dataset. V527L was not included in this analysis because of a much higher allele frequency compared to the other mutations. A380T, L114P, D102N, G15W and F437S were excluded in this analysis as for all other analyzes.
A boxplot was made to compare the uptake of mutations found in each protein domain. The mutations were grouped into extracellular, intracellular or transmembrane group depending on their location in the protein’s predicted secondary structure. The three groups were compared and can be seen in Figure 10.
Though not significant, the uptake of carnitine in mutations occurring within the transmembrane domain was slightly lower than the other domains suggesting that the parts of the protein stretching through the cellular membrane may impact the function of OCTN2 more than other mutations.
Figure 8. Boxplot with the WT% values for mutations found in CTD-patients (blue box) compared with the remaining 10 mutations from other ancestry groups (gray box).
y = 332305x + 45,628 R² = 0,1915
0 20 40 60 80 100 120 140
0 0,00005 0,0001 0,00015 0,0002
Uptake (%WT)
Allele frequency
Uptake (%WT) compared to allele
frequency
18 The functional data from the experiments were compared with the scores from the two computational predictors CADD and Polyphen-2. No correlation was seen between the CADD PHRED-score and the allele frequency for each mutation. No significant difference could be seen between mutations predicted to be benign, possibly damaging or probably damaging by the Polyphen-2 predictor. The uptake for each mutation compared to CADD PHRED-score showed a weak correlation that an OCTN2 mutant with a higher CADD PHRED-score has a higher uptake compared with mutants with lower CADD PHRED- scores (Figure 11). The Polyphen-2 score for each mutation plotted against the allele frequency for the mutations is shown in Figure 12. A significant correlation (p-value < 0,0055) can be seen between Polyphen-2 score and allele frequency suggesting that a mutation occurring more frequently has a lower Polyphen-2 score and is therefore predicted to be less deleterious, as expected.
Figure 10. The uptake compared to WT for each mutation where the mutations has been grouped to what type of protein domain they are found in. 5 mutations found in extracellular loops, 4 in intracellular loops and 11 in transmembrane domains.
Figure 11. % uptake for each mutation compared to WT on the y-axis, CADD PHRED-score on the x- axis.
y = 4,4871x - 65,741 R² = 0,1602
0 50 100 150
0 5 10 15 20 25 30 35 40
Uptake (%WT)
CADD PHRED-score
Uptake (%WT) compared to predicted CADD
PHRED-score
19 The results of the Covid-19 drug-screen against OCTN2 are shown in Figure 13. Verapamil, a control inhibitor, fully inhibited the transport of carnitine by OCTN2 at 100 µM. Of the interventional drugs Tofacitinib and Ritonavir both fully inhibited OCTN2, reducing uptake to lower than EV, while Umifenovir reduced the carnitine uptake to 14,4 % compared to WT-OCTN2.
Ruxolitinib, Lopinavir and Remdesivir reduced the carnitine uptake to 30,5 %, 29,0 % and 37,1 %, respectively, while Chloroquine reduced the uptake to 47,9 %. The drugs with the least inhibition of OCTN2 were Triazavirin, Ribavirin and Tetrandrine, lowering the uptake to 94,9 %, 88,6 % and 88,4
%, respectively. The remaining 15 drugs reduced the carnitine uptake to 59,6 % – 78,7 %.
Figure 13. Shows the average uptake of carnitine compared to WT-OCTN2, expressed in % of WT uptake. Each drug was tested in triplicate at the concentration 100 µM, except for Tetrandrine, Baricitinib and Azithromycin which were tested at 10 µM, 50 µM and 50 µM, respectively. Verapamil is the positive control used at 100 µM. * = p-value < 0,05, **= p-value < 0,01,
*** = p-value < 0,001.
Figure 12. Polyphen-2 scores for each mutation on the y-axis and the allele frequency for each mutation on the x-axis. S362L and G411V is not included as no available data on allele frequency was found.
y = -2785,1x + 1,1008 R² = 0,3911
0 0,2 0,4 0,6 0,8 1 1,2
0 0,00005 0,0001 0,00015 0,0002 0,00025 0,0003
Polyphen-2 score
Allele frequency
Polyphen-2 score compared to mutations
allele frequency
20
4.0 Discussion
4.1 Discussion of the uptake results
4.1.1 Deleterious mutationsCTD is a monogenic, autosomal recessive disease in which individuals harboring two mutant alleles of OCTN2 (SLC22A5) are affected. The severity of the disease is variable ranging from mild almost asymptomatic disease to cardiomyopathy and sudden death. Fortunately, the disease can be treated with early intervention by administering large doses of carnitine. Early detection of the disease in all newborns is increasingly occurring and transitioning from analyte detection in blood spots with low carnitine levels being associated with the disease to NextGen sequencing. The latter may be more reliable, but current methods predicting whether or not a missense mutation is deleterious are problematic. To evaluate the reliability of current prediction algorithms and to better understand the underlying structure function relationships between missense mutations and function, this dissertation examined the function of disease-causing mutations in OCTN2 and mutations found randomly in volunteers from different ancestral backgrounds, who do not have the disease. The first finding was that all mutations found in CTD-patients showed a drastic deleterious impact on the OCTN2 transporter function consistent with these mutations being causal for the development of CTD in individuals.
Interestingly, variation in function even within the disease-causing mutations was observed (Figure 7).
Previous studies (67) have shown that functional effects of mutations in transporters may associate with disease severity, e.g., for Fanconi Bickel syndrome, a monogenic disorder caused by mutations in SLC2A2. Though we did not perform such associations, it will be interesting to do so in future studies.
This dissertation’s second observation was that a number of random mutations, not associated with CTD also exhibited deleterious effects on the OCTN2 transporter and significantly reduced the carnitine uptake, see Figure 7. This result suggests that many mutations within OCTN2 exist in the population which may have the potential to cause disease. Such mutations should be cataloged in a CTD database, and if identified in newborn screening should be noted as potentially disease causing. We are not aware of a functional catalog of mutations in any transporter that may potentially be informative for newborn screening.
4.1.2 Mutations not deleterious to OCTN2 function
Not surprisingly, a number of mutations that did not significantly impact the transporting function of OCTN2 (i.e., G529V, P53R, C113Y, G234R, E131K, P421L and Y492C, Figure 7) was identified.
Mutations that have no effects on function are termed “neutral” and should also be identified and cataloged. R166G showed a significant 23 % increase of carnitine transport compared to WT-OCTN2.
This is unlikely to significantly affect an individual’s health as carnitine concentrations tends to differ a lot between healthy individuals (~25-50 µM) (9).
4.1.3 Mutations not observed in gnomAD
The mutations M195L, I237M and T429A have not been observed but instead simulated. The results show that mutations in these regions of the protein is not deleterious for OCTN2’s transporting function, see Figure 5. This suggests that these protein domains are in fact not particularly evolutionary conserved.
If they were more vital to the protein’s function compared to other regions a significant impact on the transporting function should be seen. Instead, why no mutations have been observed in these regions is probably due to small numbers of amino acids rather than these regions being more evolutionary conserved compared to other regions of the protein.
21 4.1.4 Comparison between results and available studies on the mutations
Most of the mutations used in this study are novel and no previous data has been generated for comparison. G152R has been found in CTD-patients and shown drastically reduced carnitine levels. As G152D shown a total impairment of the OCTN2 function G152R might also be the reason for some individuals developing CTD. This suggests that glycine is important in this position. G411V has also been previously tested and seen in CTD-patients. This mutation has shown reduced carnitine levels which is confirmed by the results in this study, where G411V resulted in ~0 % OCTN2 function (68)
4.2 Assay Limitations
4.2.1 Low/negative protein counts
Due to low protein counts, some wells in the 96 well-plate during the carnitine uptake assay were excluded. The very low protein values could not be used to normalize the DPM for those wells because the DPM divided with protein would result in an uptake as high as 105 times the function of WT, which is unrealistic. Technical reasons can perhaps partly explain these results. The small volumes with HEK- 293 cells added to each well are susceptible to the concentration of cells. Sedimentation of the cells may result in a non-homogenous solution leading to different numbers of cells in each well. This might be part of the reason why the rows seeded last has lower protein counts compared to others. Additionally, the poly-D coating of the well-plate might be lower in some wells reducing the possibility for the cells to attach and instead being washed off. Repetition of the experiment with a well-plate from another batch can rule out or confirm if that has an impact.
4.2.2 Edge-effect and its impact on the results
Out of 91 wells transfected, the results from 67 are presented, see Figure 6-7. Most of the two bottom rows were excluded due to low or negative protein counts. This suggests that the cells in the bottom edge of the well, closest to the door in the incubator, seem to be less healthy than cells elsewhere on the plate. This so-called edge-effect can be the issue. How this effect truly works is still discussed and theories include that more fluid evaporates in the outer edges and the wells are heated more quickly making the cells less comfortable and causing cell death (69). The remaining 7 excluded wells are instead spread across other parts of the plate and are not likely due to any edge-effect, but rather assay variability that remains to be optimized. The assay would need to be repeated to validate if it is due to coincidence or if there is a pattern. Due to lack of time the assay could not be repeated at this time.
4.2.3 Exclusion of wells and its impact on the results
Two of the excluded wells were transfected with the EV-plasmid, one with the WT-plasmid and the remaining 21 with any of the mutated plasmids. As there were 8 wells with EV and WT the remaining 6 and 7 wells, respectively, used in the results is enough to give a confident result that can be expected to reflect the reality. As all mutations are standardized with the WT the mutations can still be reliably compared. A380T, L114P, D102N, G15W and F437S all were subjects for the edge-effect and excluded, making conclusions for these mutations impossible. D139N, G152D, R471C, G529V, P421L and Y492C all had one of their wells excluded. As they still have duplicates these results are more reliable.
Of the remaining mutations not yet discussed, V527L and P53R have the highest spread between the different wells. The rest of the mutations are mostly clustered and give similar results throughout the triplicate.
22
4.3 CADD and Polyphen-2 predictions
CADD scoring can be used as a tool to compare between mutations, but is not applicable for determining a mutation’s grade of impact on a protein (70). CADD predicts the impact of nucleotide changes over the whole human genome and sees the whole human DNA as one entity, giving a score from 1 to 99 with higher scores likely to be more damaging (70). Polyphen-2 considers the amino acid substitution within a specific protein rather than the whole human genome. This gives a score from 0 to 1 and will calculate the impact on only that protein (71). Naturally, both predictive methods will have different predictions for the same mutation, which were seen in my studies. D139N and V527L both have the lowest CADD scores and are predicted “benign” by Polyphen-2; however, both show a 42-45 % reduction of the OCTN2 carnitine transporter function (Figure 7). That reduction suggests that those mutations are potentially harmful and shows that the predictive models are not quantitative and may not be useful when identifying individuals who harbor disease. Despite not being fully accurate many of the mutations predicted damaging by both CADD and Polyphen-2 were confirmed by the results in this study. Out of the mutations N91S, D139N, G152D, L202P, Y211H, P266L, S362L, G411V, R471C, V527L, W275C and R227S which all exhibit deleterious effects on OCTN2 carnitine uptake, all but D139N and V527L were predicted damaging by both CADD and Polyphen-2. Thus, in this limited screen, the two methods have a false negative rate of about 17%. This rate would be unacceptable in newborn screens as a substantial percentage of infants would go undetected and untreated.
No correlation between allele frequency of a mutation and its CADD PHRED-score could be seen.
Polyphen-2 has a weak correlation showing this pattern, but as the allele frequencies were very low this correlation can only be concluded on mutations occurring in ~0,027 % of individuals or less, see Figure 12. Interestingly, CADD PHRED-scores show a weak correlation that mutations not impairing the OCTN2 function have higher CADD PHRED-scores than mutations that impair its function, see Figure 11. This contradicts the hypothesis that higher CADD PHRED-scores translates to a more deleterious mutation, underscoring the need for new methods. Whether or not it would predict better if more mutations were to be studied, this shows that CADD is not robust enough to be completely trusted.
Further studies and data need to be collected to improve the accuracy in computational predictors.
4.4 Amino acid substitutions: Location and impact
Membrane transporters have domains that are within the plasma membrane and domains that are found in either the cytosol or extracellular fluid. Interestingly, all mutations fully impairing the OCTN2 carnitine transporting activity were found in the transmembrane domains of the OCTN2 transporter (see Figure 3 and 7). In Figure 10 a pattern, even though not significant, can be seen that mutations found in transmembrane domains impair the OCTN2 function more than mutations in amino acids found in aqueous compartments. Many of these mutations have substitutions making the position more hydrophilic, which is mostly not beneficial in the hydrophobic cellular membrane. For example, G152D could be deleterious due to the small and nonpolar nature of glycine, whereas aspartic acid is bigger, more hydrophilic and carries a negative charge. Furthermore, in the Y211H mutation, the stability of the aromatic ring in tyrosine which histidine is lacking makes histidine more hydrophilic. A substitution from alanine to threonine in position 380 will make this position more hydrophilic due to an addition of a hydroxyl. P266L is a slight change but the carbon chain of leucine might be enough to sterically hinder
23 the folding of the protein or the transporting function itself. Ser362Leu shows that even though leucine is less hydrophilic, which should be favorable in the transmembrane domain, its longer chain disrupts the function of the protein and somehow impairs OCTN2’s transporting function. Lastly, a substitution of arginine to cysteine in position 471 results in a major functional difference in the protein. Arginine is both large and has a charged end. Despite this, the results show that this substitution is deleterious. A potential explanation is that arginine’s charged end stabilizes the protein or that cysteine results in sulfhydryl bonding within the protein, which may be deleterious.
The two mutations reducing the OCTN2 transporting activity to lower than 10 %, L114P and G15W, are instead found in the N-terminal tail and extracellular loop 1. Glycine substituted to tryptophan is a major difference. Tryptophan is a large amino acid while glycine is the smallest. Also, the N-terminal tail in OCTN2 has in earlier studies shown to be vital for carnitine uptake (72). Such a major change in the N-terminus tested in this study is not surprisingly affecting the function of the protein drastically, which is confirmed by the results.
Of the remaining 7 deleterious mutations only 2 are in transmembrane domains. As these mutations did not decrease the transport function nearly as much as the other deleterious mutations this may point to a fact that extracellular and intracellular domains are generally not as important for the transporter’s function.
4.6 Covid-19 screen results
The drugs showing the highest inhibition of OCTN2 were Ritonavir, Tofacitinib and Umifenovir. Of the 25 screened drugs, Ritonavir has the third-most ongoing Covid-19 clinical trials, following Azithromycin and Hydroxychloroquine (Table 1).
An interesting aspect of the results from this experiment is the drastic inhibition of OCTN2 by Ritonavir and the lesser inhibition by Lopinavir. As these are already used for treatment of HIV which is a life- long disease, individuals treated with these drugs might suffer from low carnitine levels without knowing it (73). Interestingly, studies have actually shown that patients treated for HIV have lower carnitine levels and also an increased risk for cardiovascular diseases (CVD) due to low carnitine levels (74,75). Our results suggest the possibility that anti-viral drugs such as ritonavir might affect OCTN2 mediated carnitine absorption and reabsorption, thus lowering carnitine levels. However, our results are based on a single concentration. Further studies have to be conducted in order to determine the IC50
towards OCTN2 and compare that to plasma and tissue concentrations after treatment of HIV, and also Covid-19, if these drugs would be approved for that indication.
Two other highly interesting candidates are Chloroquine (CQ) and Hydroxychloroquine (HCQ). They are closely related and are currently used for treatment of Malaria. They have also shown promise as a treatment for Covid-19 infection. Though, CQ shows higher toxicity for the patient compared to HCQ and currently there are more trials on HCQ in humans compared to CQ (see Table 1) (76). Additionally, the results from the Covid-19 screen show a higher inhibition of OCTN2 by CQ compared to HCQ. CQ reduces the carnitine uptake to 47,9 % while HCQ reduces it to 63,6 % following the trend that CQ is more toxic than HCQ in humans. However, to our knowledge levels of carnitine in patients with malaria have not been studied.
Remdesivir is also a candidate showing promising results for Covid-19 which inhibits carnitine uptake to 37,1 %. Though it is not yet approved for any indication, transporter interaction data will be useful to predict and prevent adverse events or DDIs. Lopinavir and Ruxolitinib reduce the uptake to 29 % and 30,5 %, respectively, which is important to keep in mind as these are already approved and might soon reach the market with Covid-19 as indication.
24
5.0 Conclusions
5.1 OCTN2 mutations
A380T, L114P, D102N, G15W and F437S were excluded and cannot be concluded upon.
The first conclusion is that the mutations found in CTD-patients have proven to be deleterious and likely explains the disease in those patients. G152D, Y211H, P266L, S362L, G411V and R471C have all been found in CTD-patients and completely abolish OCTN2 activity. Additionally, four of the mutations found in CTD-patients drastically decrease the activity. G152D, Y211H, P266L, S362L, G411V and R471C all occur in transmembrane domains. These data point to the pattern that amino position within the protein structure is vital to the transporter’s function, although the experiment should be repeated to validate that hypothesis.
Of the three “invented” mutations none significantly impacted OCTN2 function. Despite that, T429A shows a trend toward reduced function and should be studied further. M195L is located in an extracellular loop and does not significantly impair OCTN2, adding to the theory that mutations in transmembrane domains are more deleterious than one in e.g. extracellular loops. Though, I237M does not significantly impair OCTN2 either and is located in a transmembrane domain. As said, more studies are warranted in order to validate the hypothesis that mutations in transmembrane domains are actually more deleterious compared to other mutations.
There are too few data to draw a certain conclusion about whether or not one ancestry group suffers from more deleterious mutations or not. Only three mutations were significantly affecting OCTN2 out of all mutations identified in different ancestry groups.
The computational predictors CADD and Polyphen-2 are not robust enough to predict a mutation’s impact on the protein. From these data, they poorly predicted whether or not a mutation will be deleterious, illustrating the need for improvement and further studies similar to the one conducted in this paper. The goal is that the predictors should be able to reliably pinpoint an isolated SNP’s effect on the protein it changes, which could not be seen in this study.
5.2 Covid-19 drug screen conclusions
Ritonavir is already approved and has numerous ongoing Covid-19 clinical trials. As it completely inhibits OCTN2-mediated carnitine transport, this drug might prove to result in severe adverse events such as CVD. The finding that patients with HIV, many of whom are taking ritonavir have lower carnitine levels may potentially be explained by the drug rather than the disease. The concentrations of ritonavir needed to inhibit should be studied further to establish its IC50 towards OCTN2. Additionally, Tofacitinib and Umifenovir also pose a risk for carnitine-related adverse events. To draw this conclusion their inhibitory concentrations have to be compared with clinically relevant plasma concentrations used
25 for treating patients. Although less extensively tested in clinical trials, further information about their inhibition of OCTN2 is valuable. As Ruxolitinib, Lopinavir and Remdesivir also show drastic inhibition, further studies should be conducted to determine their IC50. Though, as they have been screened at very high concentrations, which might be higher than clinical concentrations used for potentially treating Covid-19 it should be kept in mind they might not pose any problems at clinically relevant concentrations. Furthermore, the impact of the Covid-19 disease on the patient has not been taken into account in this study. No data is available to show whether or not the disease itself has a functional impact on OCTN2 to any extent which might be a point of interest in future studies.
Also, as Ritonavir, Lopinavir and Tofacitinib are already approved and being used for life-long diseases such as HIV and RA, their interactions with carnitine transport at OCTN2 should be studied further. If they inhibit OCTN2 at clinically relevant concentrations, many individuals suffering from these diseases may have a drastically reduced carnitine concentration. This could pose severe health complications and in the worst case scenario increase the morbidity and mortality rate of arrythmias and liver failures for these patients, which should not be taken lightly.
5.3 Conclusion summary
These studies show that SLC22A5-mutations, especially mutations located in transmembrane domains, can lead to drastic impairment of OCTN2. This study has characterized novel mutations in SLC22A5 strengthening the fact that these mutations may cause CTD. In this study, CADD and Polyphen-2 shown poor predictions of a mutation’s impact on OCTN2 and these predictors need to be tested further. Antiviral treatment such as Ritonavir may inhibit OCTN2 and lead to low carnitine levels which has shown correlation with increased risk for CVD in HIV-patients.
6.0 Special thanks
I would like to thank everyone in the whole Giacomini laboratory in University of California, San Francisco (UCSF) for giving me this opportunity and welcoming me with open arms. I would specifically like to thank Megan Koleske for mentoring me throughout my time in San Francisco and also back here in Sweden, without her and her guidance I would have certainly been lost. I would also like to thank Kathy Giacomini for accepting me to her lab and together with Sook-Wah Yee making every measure needed in order to make it possible for me to continue with my project despite the Covid-19 situation, thank you both. Lastly, I want to thank Kathy Giacomini and Megan Koleske again for giving feedback and points of improvement of my dissertation.
26
7.0 References
1. Xiao Q, Zhou Y, Lauschke VM. Ethnogeographic and inter-individual variability of human ABC transporters. Hum Genet. maj 2020;139(5):623–46.
2. SLC superfamily of solute carriers | Transporters | IUPHAR/BPS Guide to PHARMACOLOGY [Internet]. [citerad 22 april 2020]. Tillgänglig vid:
https://www.guidetopharmacology.org/GRAC/FamilyDisplayForward?familyId=863 3. He L, Vasiliou K, Nebert DW. Analysis and update of the human solute carrier (SLC) gene
superfamily. Hum Genomics. 01 januari 2009;3(2):195–205.
4. Nigam SK. The SLC22 Transporter Family: A Paradigm for the Impact of Drug Transporters on Metabolic Pathways, Signaling, and Disease. Annu Rev Pharmacol Toxicol. 06 2018;58:663–87.
5. Nigam SK. What do drug transporters really do? Nat Rev Drug Discov. januari 2015;14(1):29–44.
6. Transcript: SLC22A5-201 (ENST00000245407.7) - Summary - Homo sapiens - Ensembl genome browser 99 [Internet]. [citerad 29 april 2020]. Tillgänglig vid:
https://uswest.ensembl.org/Homo_sapiens/Transcript/Summary?db=core;g=ENSG00000197375
;r=5:132369752-132395614;t=ENST00000245407
7. Biotechnology S. OCTN2 - Transporters - Solvo Biotechnology [Internet]. [citerad 22 maj 2020].
Tillgänglig vid: https://www.solvobiotech.com/transporters/octn2
8. Koepsell H. Organic Cation Transporters in Health and Disease. Daws LC, redaktör. Pharmacol Rev. januari 2020;72(1):253–319.
9. Longo N, Frigeni M, Pasquali M. CARNITINE TRANSPORT AND FATTY ACID OXIDATION. Biochim Biophys Acta. oktober 2016;1863(10):2422–35.
10. Liepinsh E, Makrecka M, Kuka J, Cirule H, Makarova E, Sevostjanovs E, m.fl. Selective inhibition of OCTN2 is more effective than inhibition of gamma-butyrobetaine dioxygenase to decrease the availability of l-carnitine and to reduce myocardial infarct size. Pharmacol Res. juli 2014;85:33–8.
11. Urban TJ, Gallagher RC, Brown C, Castro RA, Lagpacan LL, Brett CM, m.fl. Functional Genetic Diversity in the High-Affinity Carnitine Transporter OCTN2 (SLC22A5). Mol Pharmacol. 01 november 2006;70(5):1602–11.
12. Frigeni M, Balakrishnan B, Yin X, Calderon FRO, Mao R, Pasquali M, m.fl. Functional and molecular studies in primary carnitine deficiency. Hum Mutat. 2017;38(12):1684–99.
13. Orphanet: Systemic primary carnitine deficiency [Internet]. [citerad 19 maj 2020]. Tillgänglig vid:
https://www.orpha.net/consor/cgi-
bin/Disease_Search.php?lng=EN&data_id=3316&Disease_Disease_Search_diseaseGroup=Carniti ne&Disease_Disease_Search_diseaseType=Pat&Disease(s)/group%20of%20diseases=Systemic- primary-carnitine-
