Improving safety and establishing episomal maintenance of Adeno-associated viral vectors

Adeno-associated viral (AAV) vectors are among the most widely used gene transfer systems in basic and pre-clinical research and have been employed in more than 160 clinical trials. AAV vectors are commonly produced in producer cell lines like HEK293 by co-transfection with a so-called vector plasmid and one (in this work) or two so-called helper plasmids. The vector plasmid contains the transgene cassette of interest (TEC) flanked by AAV’s inverted terminal repeats (ITRs) which serve as packaging signals, whereas the helper plasmid provides the required AAV and helper virus functions in trans. A pivotal aspect of AAV vectorology is the manufacturing of AAV vectors free from impurities arising during the production process. These impurities include AAV vector preparations that contain capsids containing prokaryotic sequences, e.g. antibiotic resistance genes originating from the producer plasmids. In the first part of the thesis we aimed at improving the safety of AAV vectors. As we found that encapsidated prokaryotic sequences (using the ampicillin resistance gene as indicator) cannot be re-moved by standard purification methods we investigated whether the producer plasmids could be replaced by Minicircles (MCs). MCs are circular DNA constructs which contain no functional or coding prokaryotic sequences; they only consist of the TEC and a short sequence required for production and purification. MC counterparts of a vector plasmid encoding for enhanced green fluorescent (eGFP) protein and a helper plasmid encoding for AAV serotype 2 (AAV2) and helper Adenovirus (Ad) genes were designed and produced by PlasmidFactory (Bielefeld, Germany). Using all four possible combinations of plasmid and MCs, single-stranded AAV2 vectors (ssAAV) and self-complementary AAV vectors (scAAV) were produced and characterized for vector quantity, quality and functionality. The analyses showed that plasmids can be replaced by MCs without decreasing the efficiency of vector production and vector quality. MC-derived scAAV vector preparations even exceeded plasmid-derived preparations, as they displayed up to 30-fold improved transduction efficiencies. Using MCs as tools, we found that the vector plasmid is the main source of encapsidated prokaryotic sequences. Remarkably, we found that plasmid-derived scAAV vector preparations contained a much higher relative amount of prokaryotic sequences (up to 26.1 %, relative to TEC) compared to ssAAV vector preparations (up to 2.9 %). By replacing both plasmids by MCs the amount of functional prokaryotic sequences could be decreased to below the limit of quantification. Additional analyses for DNA impurities other than prokaryotic sequences showed that scAAV vectors generally contained a higher amount of non-vector DNA (e.g. adenoviral sequences) than ssAAV vectors. For both, ssAAV and scAAV vector preparations, MC-derived vectors tended to contain lower amounts of foreign DNA. None of the vectors tested could be shown to induce immunogenicity. In summary we could demonstrate that the quality of AAV vector preparations could be significantly improved by replacing producer plasmids by MCs. Upon transduction of a target tissue, AAV vector genomes predominantly remain in an episomal state, as duplex DNA circles or concatemers. These episomal forms mediate long-term transgene expression in terminally differentiated cells, but are lost in proliferating cells due to cell division. Therefore, in the second part of the thesis, in cooperation with Claudia Hagedorn and Hans J. Lipps (University Witten/Herdecke) an AAV vector genome was equipped with an autonomous replication element (Scaffold/matrix attachment region (S/MAR)). AAV-S/MAR encoding for eGFP and a blasticidin resistance gene and a control vector with the same TEC but lacking the S/MAR element (AAV-ΔS/MAR) were produced and transduced into highly proliferative HeLa cells. Antibiotic pressure was employed to select for cells stably maintaining the vector genome. AAV-S/MAR transduced cells yielded a higher number of colonies than AAV-ΔS/MAR-transduced cells. Colonies derived from each vector transduction were picked and cultured further. They remained eGFP-positive (up to 70 days, maximum cultivation period) even in the absence of antibiotic selection pressure. Interestingly, the mitotic stability of both AAV-S/MAR and control vector AAV-ΔS/MAR was found to be a result of episomal maintenance of the vector genome. This finding indicates that, under specific conditions such as the mild selection pressure we employed, “common” AAV vectors persist episomally. Thus, the S/MAR element increases the establishment frequency of stable episomes, but is not a prerequisite.

MicroRNA modulation of aldosterone production in the adrenal gland

Hypertension is the major risk factor for coronary disease worldwide. Primary hypertension is idiopathic in origin but is thought to arise from multiple risk factors including genetic, lifestyle and environmental influences. Secondary hypertension has a more definite aetiology; its major single cause is primary aldosteronism (PA), the greatest proportion of which is caused by aldosteroneproducing adenoma (APA), where aldosterone is synthesized at high levels by an adenoma of the adrenal gland. There is strong evidence to show that high aldosterone levels cause adverse effects on cardiovascular, cerebrovascular, renal and other systems. Extensive studies have been conducted to analyse the role that regulation of CYP11B2, the gene encoding the aldosterone synthase enzyme plays in determining aldosterone production and the development of hypertension. One significant regulatory factor that has only recently emerged is microRNA (miRNA). miRNAs are small non-coding RNAs, synthesized by a series of enzymatic processes, that negatively regulate gene expression at the posttranscriptional level. Detection and manipulation of miRNA is now known to be a viable method in the treatment, prevention and prognosis of certain diseases. The aim of the present study was to identify miRNAs likely to have a role in the regulation of corticosteroid biosynthesis. To achieve this, the miRNA profile of APA and normal human adrenal tissue was compared, as was the H295R adrenocortical cell line model of adrenocortical function, under both basal conditions and following stimulation of aldosterone production. Key differentially-expressed miRNAs were then identified and bioinformatic tools used to identify likely mRNA targets and pathways for these miRNAs, several of which were investigated and validated using in vitro methods. The background to this study is set out in Chapter 1 of this thesis, followed by a description of the major technical methods employed in Chapter 2. Chapter 3 presents the first of the study results, analysing differences in miRNA profile between APA and normal human adrenal tissue. Microarray was implemented to detect the expression of miRNAs in these two tissue types and several miRNAs were found to vary significantly and consistently between them. Furthermore, members of several miRNA clusters exhibited similar changes in expression pattern between the two tissues e.g. members of cluster miR-29b-1 (miR-29a-3p and miR-29b-3p) and of cluster miR-29b-2 (miR-29b-3p and miR-29c- 3p) are downregulated in APA, while members of cluster let-7a-1 (let-7a-5p and let-7d-5p), cluster let-7a-3 (let-7a-5p and let-7b-5p) and cluster miR-134 (miR- 134 and miR-382) are upregulated. Further bioinformatic analysis explored the possible biological function of these miRNAs using Ingenuity® Systems Pathway Analysis software. This led to the identification of validated mRNAs already known to be targeted by these miRNAs, as well as the prediction of other mRNAs that are likely targets and which are involved in processes relevant to APA pathology including cholesterol synthesis (HMGCR) and corticosteroidogenesis (CYP11B2). It was therefore hypothesised that increases in miR-125a-5p or miR- 335-5p would reduce HMGCR and CYP11B2 expression. Chapter 4 describes the characterisation of H295R cells of different strains and sources (H295R Strain 1, 2, 3 and HAC 15). Expression of CYP11B2 was assessed following application of 3 different stimulants: Angio II, dbcAMP and KCl. The most responsive strain to stimulation was Strain 1 at lower passage numbers. Furthermore, H295R proliferation increased following Angio II stimulation. In Chapter 5, the hypothesis that increases in miR-125a-5p or miR-335-5p reduces HMGCR and CYP11B2 expression was tested using realtime quantitative RT-PCR and transfection of miRNA mimics and inhibitors into the H295R cell line model of adrenocortical function. In this way, miR-125a-5p and miR-335-5p were shown to downregulate CYP11B2 and HMGCR expression, thereby validating certain of the bioinformatic predictions generated in Chapter 3. The study of miRNA profile in the H295R cell lines was conducted in Chapter 6, analysing how it changes under conditions that increase aldosterone secretion, including stimulation Angiotensin II, potassium chloride or dibutyryl cAMP (as a substitute for adrenocorticotropic hormone). miRNA profiling identified 7 miRNAs that are consistently downregulated by all three stimuli relative to basal cells: miR-106a-5p, miR-154-3p, miR-17-5p, miR-196b-5p, miR-19a-3p, miR-20b- 5p and miR-766-3p. These miRNAs include those derived from cluster miR-106a- 5p/miR-20b-5p and cluster miR-17-5p/miR-19a-3p, each producing a single polycistronic transcript. IPA bioinformatic analysis was again applied to identify experimentally validated and predicted mRNA targets of these miRNAs and the key biological pathways likely to be affected. This predicted several interactions between miRNAs derived from cluster miR-17-5p/miR-19a-3p and important mRNAs involved in cholesterol biosynthesis: LDLR and ABCA1. These predictions were investigated by in vitro experiment. miR-17-5p/miR-106a-p and miR-20b-5p were found to be consistently downregulated by stimulation of aldosterone biosynthesis. Moreover, miR-766-3p was upregulation throughout. Furthermore, I was able to validate the downregulation of LDLR by miR-17 transfection, as predicted by IPA. In summary, this study identified key miRNAs that are differentially-expressed in vivo in cases of APA or in vitro following stimulation of aldosterone biosynthesis. The many possible biological actions these miRNAs could have were filtered by bioinformatic analysis and selected interactions validated in vitro. While direct actions of these miRNAs on steroidogenic enzymes were identified, cholesterol handling also emerged as an important target and may represent a useful point of intervention in future therapies designed to modulate aldosterone biosynthesis and reduce its harmful effects.

Regulation of prostate cancer cell function by activators of AMP-activated protein kinase

AMP-activated protein kinase (AMPK) is a key regulator of cell energy homeostasis. More recently, it has become apparent that AMPK regulates cell proliferation, migration and inflammation. Previous evidence has suggested that AMPK may influence proliferation and invasion by regulating the pro-proliferative mitogen-activated protein kinases (MAPKs). However, the mechanisms underlying this crosstalk between AMPK and MAPK signalling are not fully understood. As AMPK activation has been reported to have anti-proliferative effects, there has been increasing interest in AMPK activation as a therapeutic target for tumourigenesis. The aim of this study was to investigate whether AMPK activation influenced prostate cancer (PC) cell line proliferation, migration and signalling. Therefore, different PC cell lines were incubated with two structurally-unrelated molecules that activate AMPK by different mechanisms, AICAR and A769662. Both chemicals activated AMPK in a concentration- and time-dependent manner in PC3, DU145 and LNCaP cell lines. AMPK activity as assessed by AMPK activating phosphorylation as well as phosphorylation of the AMPK substrate ACC increased along with tumour severity in PC biopsies. Furthermore, both activators of AMPK decreased cell proliferation and migration in the androgen-independent PC cell lines PC3 and DU145. Inhibition of proliferation by A769662 was attenuated in AMPK α1-/- AMPK α2-/- knockout (KO) mouse embryonic fibroblasts (MEFs) compared to wild type (WT) MEFs, and the inhibitory effect on migration of AICAR lost significance in PC3 cells infected with adenoviruses expressing a dominant negative AMPK α mutant, indicating these effects are partially mediated by AMPK. Furthermore, long-term activation of AMPK was associated with inhibition of both the phosphatidylinositol 3’-kinase/protein kinase B (PI3K/Akt) signalling pathway in addition to the extracellular signal-regulated kinase 1/2 (ERK1/2) signalling pathway. Indeed, the actions of AMPK activators on PC cell line viability were mimicked by selective inhibitors of Akt and ERK1/2 pathways. In contrast to the effects of prolonged incubation with AMPK activators, short-term incubation with AMPK activators had no effect on epidermal growth factor (EGF)-stimulated ERK1/2 phosphorylation in PC cell lines. In addition, AMPK activation did not influence phosphorylation of the other MAPK family members p38 and JNK. Interestingly, both AICAR and A769662 decreased EGF-stimulated ERK5 phosphorylation in PC3, DU145 and LNCaP cells as assessed with an anti-phospho-ERK5 antibody. Further characterisation of this effect indicated that prior stimulation with the AMPK activators had no effect on ERK5 phosphorylation stimulated by transient transfection with a constitutively active ERK5 kinase (MEK5DD), which represents the only known canonical kinase for ERK5. Intriguingly, the pattern of EGF-stimulated ERK5 phosphorylation was distinct from that mediated by MEK5DD activation of ERK5. This finding indicates that AMPK activation inhibits EGF-stimulated ERK5 phosphorylation at a point at or above the level of MEK5, although why EGF and constitutively active MEK5 stimulate markedly different immunoreactive species recognised by the anti-phospho-ERK5 antibody requires further study. A769662 had a tendency to reduce EGF-stimulated ERK5 phosphorylation in WT MEFs, yet was without effect in MEFs lacking AMPK. These data indicate that AMPK may underlie the effect of A769662 to reduce EGF-stimulated ERK5 phosphorylation. Prolonged stimulation of PC cell lines with AICAR or A769662 inhibited EGF-stimulated Akt Ser473 phosphorylation, whereas only incubation with A769662 rapidly inhibited Akt phosphorylation. This difference in the actions of the different AMPK activators may suggest an AMPK-independent effect of A769662. Furthermore, AICAR increased phosphorylation of Akt in WT MEFs, an effect that was absent in MEFs lacking AMPK, indicating that this effect of AICAR may be AMPK-dependent. Taken together, the data presented in this study suggest that AMPK activators markedly inhibit proliferation and migration of PC cell lines, reduce EGF-stimulated ERK1/2 and Akt phosphorylation after prolonged incubation and rapidly inhibit ERK5 phosphorylation. Both AMPK activators exhibit a number of effects that are likely to be independent of AMPK in PC cell lines, although inhibition of ERK1/2, ERK5 and Akt may underlie the effects of AMPK activators on proliferation, viability and migration. Further studies are required to understand the crosstalk between those signalling pathways and their underlying significance in PC progression.