Recombinant protein production in Escherichia coli provides a cheap and efficient way of producing medically and industrially relevant proteins. Sequence features of individual genes and especially their 5’ terminal coding sequences act on the efficiency of gene expression by complex regulatory mechanisms which are still not fully understood. This study aimed to investigate the features of the 5’ coding region of recombinant mRNAs, and to optimize them for increased expression in E. coli. A previous study had found that a synonymous change of the bla reporter gene 2nd codon leads to an increased expression, and accordingly a synonymous library in the 5’ bla coding sequence was created by a directed evolution approach building on this feature. Variants conferring up to three-fold increases in active enzyme amounts were identified, and the increased expression was shown to stem from increased transcriptional efficiency. The effect of changing the 2nd codon synonymously was further investigated by synonymous substitutions of the 2nd codons of the bla and two other reporter genes, phoA and celB. These experiments showed that the effect of 2nd codon changes on the gene expression is determined by the sequence context, as changes in expression levels appeared to be gene specific. All the coding sequences of the study were also analysed in silico, and an application for calculating the tRNA adaptation index was programmed in Python and made freely available online.As the synonymous codon changes did not lead to a great improvement in protein amounts and any sequence features affecting the expression were hard to pinpoint, an alternative strategy involving 5’ terminal gene fusions was investigated. Combinatorial mutagenesis coupled to an effective screening technique was applied to further optimize a 5’ terminal fusion partner, previously shown to improve expression of several eukaryotic genes. The application of the best identified fusion partner candidate yielded a 3.8-fold improvement in IFN-α2b protein amounts over the original fusion, and showed twice as high protein amounts than a pelB-IFN-α2b fusion previously proven to give industrial expression amounts. The developed peptide fusion is thus an eligible candidate for further development for use in heterologous protein production.

Oral delivery is considered a convenient route for administration of pharmaceuticals. Great effort has been made to optimize oral delivery vehicles to increase the bioavailability of the pharmaceutical, and enhance patient compliance to ease swallowing. Emulsion-based gelled matrices have shown promising features as delivery systems. They are soft chewable matrices that are easy to swallow, and have the ability to entrap the pharmaceutical, providing prolonged, and controlled release to avoid fast dissolution in the stomach. The purpose of this study was divided into two main objectives. The first aim was to study emulsion-based gelatin matrices intended for oral drug delivery by investigation of the influence of gelatin type and/or oil content on the matrix properties. The second aim was to investigate the properties of multiple emulsions, regarding their potential as oral drug delivery systems for water-soluble pharmaceuticals, and controlled drug delivery. Emulsion-based matrices stabilized by 160g Bloom gelatins, either type A or B, and containing various amounts of corn oil (0, 10, 30 and 50 wt %) were subjected to rheological characterization and in vitro dissolution studies at simulated gastric conditions. The results showed an increase in viscosity, storage modulus and gelling and melting temperatures in line with increased oil content in the matrix, and to a larger extend for matrices with type A gelatin compared to type B. Longitudinal deformations of the gelled matrices did show a trend of slightly increasing Young’s modulus when oil was added to the matrix, but no clear trend was observed for force and strain at break. A correlation between rheological matrix properties and dissolution time was observed: An increase in dissolution time with higher fractions of oil and prolonged dissolution time for matrices with type A gelatin compared to type B. Overall the results showed that different oil contents and gelatin types changed the physical properties of the matrices, providing a possibility to tailor matrices to obtain suitable delivery systems for various pharmaceuticals. Water-in-oil-in-water double emulsions, stabilized by either 226g Bloom gelatin type B or tween80 was examined by long time stability studies, and by in vitro lipolysis studies simulating small intestinal conditions. The water-soluble marker tartrazine, was entrapped in the inner water phase of the emulsions. The release of tartrazine was measured during a period of 78 days and both double emulsions with gelatin and tween80 were found to possess long-term stability at room temperature. In vitro lipolysis of gastric stable double emulsions stabilized by gelatin was conducted in a dissolution medium containing bile extracts, with or without lipases. A complete release of tartrazine was obtained; both in the presence and absence of lipases, while a faster release was observed when lipases were present. Although the release mechanism was not completely determined, the results indicate that release of drugs can occur in the small intestines due to lipolysis. The double emulsions thus offer great potential in delivery of gastric unstable pharmaceuticals.

Breast cancer is a complex disease comprising subtypes with varying clinical behavior, biological features and treatment response. Breast cancer heterogeneity characterized by different subtypes, is responsible for the high mortality among breast cancer patients, since patients with identical diagnosis can have very different prognosis. Metabolite profiles and biomarkers can hopefully be used for the improved diagnosis and optimal therapeutic treatment.The scope of this study was split in two parts. The first aim was to develop an optimal method for the complete extraction of polar and non-polar metabolites, from invasive ductal carcinoma xenografts. The optimization experiments were performed with tissue samples of a basal-like xenograft model (MAS98.12), and a Precellys 24 homogenizer equipped with a cooling unit. Polar metabolites were detected by absolute quantification analysis by gas-chromatography triple quadrupole mass spectrometry (GC-QqQ-MS) after methyl chloroformate (MCF) derivatization. Non-polar metabolites were detected as fatty acid methyl esters (FAMEs) by quantitative analysis by GC-Q-MS (single quadrupole GC-MS). TMS derivatization was also evaluated, but MCF derivatization was concluded to be a more sensitive method for the polar extracts than TMS derivatization. The complete extraction was achieved after three homogenization rounds with methanol and chloroform, respectively.The second aim was to use the optimal method in metabolite profiling experiments of luminal-like (MAS98.06) and basal-like (MAS98.12) xenografts, in addition to metabolite profiling of basal-like xenografts treated with a cancer drug called MK-2206. Polar metabolite profiles were obtained by absolute quantitative MCF GQ-QqQ-MS and compared by principle component analysis (PCA) and Student`s t-tests. The statistical analyses showed that the MAS98.12 xenograft has significant higher concentrations of lactate and glycine compared to the MAS98.06 xenograft, while the MAS98.06 xenograft has significant higher concentrations of O-acetyl-L-serine and aspartate. The untreated MAS98.12 xenograft has significant higher concentration of lactate than the MK-2206 treated MAS98.12 xenograft. Classification of breast cancer subtypes can therefore be made based on the polar metabolite profiles. Non-polar metabolite profiles were not found after flow injection (FIA) MS of a non-polar extract, much due to a limited research time. Both the extraction method and the metabolite profiles need further validation in the search for biomarkers, that can serve as prognostic tools from the development of breast cancer and determination of prognosis to the establishment of personalized treatment.

Experiment-based study: The 5’-untranslated region (5’-UTR), and the DNA region corresponding to it, have been shown to have a significant influence on the expression of genes both at transcriptional and translational level. Since transcription and translation are two independent mechanisms a 5’-UTR sequence will probably not be optimal for both. The suggested solution presented in this study is to design a long 5’-UTR composed of one transcription stimulating and one translation stimulating region. The stimulating regions consist of 5’-UTR variants independently identified as transcription or translation stimulating by screening for the desired trait, as well as translation stimulating 5’-UTR variants designed using a bioinformatics tool.The results indicated that 5’-UTR fusions tend to introduce limiting factors yielding a reduced gene expression. However, some 5’-UTR fusions successfully resulted in high gene expression and one variant surpassed both of its components showing a possible additive effect of stimulating both transcription and translation in the form of 5’-UTR fusions. This indicates that testing a relatively small number of different sequences gives a good chance of success. The method also proved viable to increase the expression of low expressive 5’-UTR variants while maintaining low uninduced expression. In addition 5’-UTR fusions containing in silico designed translation stimulating regions have the potential of reaching expression levels on par with the levels reached by fusions containing 5’-UTR variants identified through screening. Literature-based study: The nucleotide sequence at the gene 5’-end has a great influence on the expression level of genes, being the location of central mechanisms like transcription and translation initiation. Because of this the 5’-end sequence is an important target when designing genes for recombinant expression. This review will focus on recent research trends, covering the traits of the 5’-end that influence gene expression, as well as on approaches and tools targeting this region that have been utilized or show potential to be used to achieve desired recombinant expression levels in E. coli. In recent years it have become evident that the entire 5’-untranslated region as well as the initial coding sequence has great influence on gene expression, showing that there is more to designing genes for recombinant expression then picking a strong promoter and an optimal SD sequence.

In industry today standard oxidative quality parameters are based on measurements of primary and secondary oxidation products, measured by PV and AV respectively. These methods are all prone to limitations and weaknesses, and their suitability for application on marine oils is not well documented. An increase in fish oil products with added flavor, color compounds, antioxidants and vitamins has entered the market in recent years. However, no documentation on the effect of these additives on the oxidation parameters has been found. The aim of this thesis was therefore to study the effect of variations in procedures in an attempt to highlight weaknesses and further to establish the most suitable procedures for each method when performing measurements on marine oils. In addition, the effect of antioxidants and additives on the oxidation parameters in cod liver oil has been evaluated.In this thesis, PV measurements by iodometric titration and by the ferric thiocyanate method were used to measure primary oxidation products, and AV and TBARS measurements were used to measure secondary oxidation products. Uncertainty of the methods was determined by performing n -measurements at different stages of the oxidation process. Measurement by the iodometric titration method was found to have a lower detection limit of PV >2.0 mEq peroxide kg-1 oil, with an uncertainty of ± 2%. Measurement by the ferric thiocyanate method was found to have a lower detection limit of PV ≥3.6 mEq peroxide kg-1 oil, with an uncertainty of ±10%. Measurement by the AV method was found to have a lower detection limit of AV≥ 1.3, with an uncertainty of ±5%. Measurements by the TBARS method was found to have a lower detection limit of 0.7 μM TBARS /g sample, with an uncertainty of ±12%. The published method of the International Dairy Federation (IDF) for PV determination was evaluated by comparison with a modified version of the method. Factors such as type of solvent used, deaeration of reagents, premixing of reagents and addition of antioxidant were differences between the methods. It was observed significant difference in absorbance in the two methods, and it was therefore concluded that the varied factor had an influence on the method. It is necessary to perform further experiments to determine which of the varied factors that cause variations in the absorbance measurements.PV measurements by the iodometric titration method were found to be influenced by the stirring method, reagent reaction time and oxygen removal. Stirring by magnetic stirring was found to give a higher PV compared to gentle stirring. The importance of the 1 minute reagent reaction time was strengthened as the PV was found to rapidly increased at prolonged reagent reaction times. It was demonstrated that this to a higher degree is important for marine oils compared to vegetable oils, as new hydroperoxides are formed more rapidly in the unstable marine oils. A significant influence of oxygen removal in reagents was detected in cod liver oil. The findings in this thesis suggests that stirring by magnetic stirring, 1 minute reagent reaction time and deaeration of all reagents should be standard procedure when PV is determined in marine oils by the iodometric titration method.Among eight investigated antioxidants and additives, Q10, tocopherol, vitamin K1, lemon – and peppermint extract was found to significantly elevate the PV measured by iodometric titration. For PV determination by the ferric thiocyanate method, lemon extract was found to significantly elevate the PV. Rosemary extract was found to significantly lower the AV measurement, while lemon extract to a very high degree elevated the AV measurement. In measurements by the TBARS method only lemon extract was found to significantly interfere with the method, leading to an elevated TBARS value. Both methods for PV detection were influenced by several of the investigated antioxidants and additives. Clearly there is need for reevaluation of the methods is use today and development of new methods. New methods for measurements of secondary decomposition products are especially needed for fish oils with added lemon extracts as today’s measurements by the AV and TBARS method give highly unreliable results.

Alginates are a class of natural unbranched linear polysaccharides, and consist of the monomers β-D-mannuronic acid and α-L-guluronic acid. The inherent physical properties, relative ease of modification, wide availability and good biocompatibility of alginates have gained a great deal of attention with regards to therapeutic applications. Heparin is a highly negatively charged linear glycosaminoglycan that is widely used as an anticoagulant. The presence of carboxyl groups and several sulfate groups gives heparin its negative charge, while iduronic acid moieties confer a high degree of flexibility to the polysaccharide by being able to assume different stable conformations. Heparin shows diversity in molecular weight, monomer sequence and modification pattern, resulting in a vast range of biological effects. When administered therapeutically, this can in cause an unpredictable dose response and potentially severe adverse effects in certain patients. The main objective of this study was to create a structural analog of heparin exhibiting a more regular structure and distribution, through chemical sulfation of alginate using chlorosulfonic acid. Other important aims were to characterize the analog in terms of structure, distribution and sulfation degree, and assess protein binding and anticoagulating properties of the sulfated alginates in comparison with heparin and the unmodified alginate templates. Sulfation was performed using chlorosulfonic acid in formamide on a polymannuronic acid (poly-M) and a polyalternating alginate with a guluronic acid fraction of FG = 0.46 (poly-MG), introduced through enzymatic epimerization. FTIR, elemental analysis with HR-ICP-MS and carbon NMR were employed to detect the attached sulfate groups on the alginate. The average molecular weights and the mass distributions of the alginate samples were studied using SEC-MALLS. Elemental analysis was used to estimate the sulfation degrees of the alginates, and 13C NMR was employed to study substitution patterns, provide additional DS estimates and assess sample purity. The protein binding properties of the sulfated alginates were evaluated by studying their ability to release hepatocyte growth factor and osteoprotegerin bound to myeloma cells. Anticoagulating properties were studied by measuring prolongation of plasma coagulation time as a result of sulfated alginate supplementation. The alginates were successfully sulfated and exhibited different degrees of sulfation obtained by varying the chlorosulfonic acid concentration used (1 - 10 %), as estimated by elemental analysis. The poly-MG alginate showed increased solubility during the sulfation reaction, resulting in a higher estimated DS at lower chlorosulfonic acid concentrations compared with poly-M. No apparent degradation of the alginates as a result of the sulfation was observed, although preliminary acid hydrolysis resulted in a molecular weight disparity between poly-M and poly-MG samples. Analysis of carbon NMR spectra allowed characterization of novel peaks and secondary DS estimations for the sulfated poly-M samples, while the complexity of the sulfated poly-MG spectra prevented confident characterization of the structures. Sulfation resulted in a profound improvement of the protein binding properties of the alginates, and showed prolongation of the plasma coagulation time at high treatment concentrations.

Alginate hydrogel made from alginate and crosslinking divalent ions is a natural biomaterial that is biocompatible, has low toxicity, is relatively cheap and has mild gelation chemistry. It is a porous material that allows diffusion of small molecules. Alginate hydrogel is a polymeric network that contains 95-99% water and it does in many ways resemble the natural extracellular matrix (ECM) that surrounds cells in the body. It is also hydrophilic, which reduces friction in body fluids and minimizes protein adsorption and it is easily stored and sterilized.Alginate is produced by both algae and bacteria, and it is initially synthesized as mannuronan (M) with 100% M-residues. Guluronic acid residues (G) are introduced in a post-polymerization step by enzymes called mannuronan C-5 epimerases that catalyze conversion of M into G without breaking the glycosidic bond. Seven different mannuronan C-5 epimerases have been sequenced, cloned and produced recombinantly, and these enzymes introduce MG-blocks, G- blocks or both in the alginate chains. With the use of these mannuronan C-5 epimerases it is now possible to engineer alginate with desired and known structure. It is also possible to covalently modify alginates with coupling of cell specific adhesion molecules to the carboxylic group in the monomers. An example is the RGD peptide (arginine-glycine-aspartic acid) that is commonly found in collagen and fibronectin in the ECM. The RGD peptide is the smallest sequence that integrin receptors can recognize and bind to.Central nervous system (CNS) damage is still one of the major causes of both death and disability, despite intense research efforts to achieve neurogenesis and restore functional synaptic connection of CNS neurons. None of he current therapy strategies promote regeneration or regrowth of neural cells or axons. In vitro and in vivo studies has shown that CNS axons can regenerate when located in a permissive environment and it is known that on-going neurogenesis occurs in certain areas of the adult brain, such as the olfactory bulb. Olfactory ensheathing cells (OECs) are found in the olfactory mucosa and olfactory bulb and secrete neurotrophins, provide necessary ECM molecules and substrates for axon elongation and myelination. They do not activate and induce inhibitory molecules or hypertrophy in astrocytes, and are therefore believed to be a promising candidate for cell-mediated repair of the CNS.The major aims of this study was to investigate whether encapsulation of OECs in different types of alginate matrices would improve cell viability over time and induce change of cell morphology, as a future goal is to transplant OECs into the CNS. Viability of OECs up to 14 days in 1.8% UP-LVG capsules have been reported by Kristin Karstensen (Karstensen, 2010), and similar results were achieved in an experiment in this project. Indications of cell concentration dependency on viability were observed in this experiment, with higher viability in capsules with low cell concentration (1.5 mil cells/mL alginate, 3.0 and 5.0 mil/mL). It was decided to conduct an encapsulation of high and low OEC concentration (4.0 mil/mL and 1.0 mil/mL) in 1.0% UP-LVG Ca2+/Ba2+ alginate, with the aim of examining whether reduced alginate concentration would improve cell viability. The results were promising, with a live cell percentage of 50% in the low cell concentration batch after 51 days. The high cell concentration batch was discarded after 22 days with estimated 30% live cells. This result strengthened the hypothesis that lower cell concentration enhanced cell viability, and confirmed that lower alginate concentration improved cell viability notably. These indications were supported by the results of a second encapsulation with similar settings. High and low concentrations (1.5 mil/mL and 5.0 mil/mL) of OECs were encapsulated in 1.0% epimerized Ca2+ alginate with and without 0.2 % RGD peptide graft. The experiment did not show an effect of the RGD peptide on cell viability or morphology. The viability of the cells was extended with one week and viable cells could be observed for 22 days, but in this experiment increased viability as a result of lower cell concentration was less pronounced. This experiment was therefore inconclusive in terms of improved viability connected to cell concentration, but indicated that a lower alginate concentration had a beneficial impact on cell viability. Star shaped channels were observed inside all capsules in this experiment, and a large fraction of dead cells were found to be located inside these channels. This experiment was later repeated with another source of epimerized alginate grafted with ≈ 0.4% RGD peptide with comparable results in terms of cell viability and morphology.Two encapsulations of low cell concentration in 1.0% UP-LVG Ca2+/Ba2+ alginate mixed with three different concentrations of gelatin (0.5%, 1.0% and 2.0%) were carried out, with the aim of observing capsule stability and cell viability. In first experiment the capsule stability appeared to be inversely proportional with gelatin concentration. This was not confirmed when the experiment was repeated, as the batch with the middle gelatin concentration was perceived as most stable. The cell viability was overall high for both encapsulations. Finally, four batches of 1.5 mil/mL OECs were encapsulated in 0.9% UP-LVG Ca2+/Ba2+ alginate gel with one type of ECM molecule mixed with the alginate per batch to yield a concentration of 1.0 mg/mL. Sulphated MG alginate was mixed with 0.9% UP-LVG Ca2+/Ba2+ alginate to a final concentration of 1.0 mg/mL, and included in the experiment. The experiment was terminated at day 28, with varying cell viabilities in the different batches. Common for all was overall lower cell viability compared with the viability observed for cells with similar concentration encapsulated in pure 1.0% UP-LVG, but the capsules proved to be relatively stable. In conclusion, reducing the alginate concentration from 1.8% to 1.0% had notable positive effect on cell viability. High cell concentration in the alginate capsules also proved to have a negative impact on cell viability, but this effect was most evident in the UP-LVG alginate gels. The negative effect on cell viability related to high cell concentration was not as profound in the epimerized alginate gels.RGD peptide grafted onto alginate did not show any unambiguous effect on cell viability and no effect on cell morphology, regardless of 0.2 % peptide graft or ≈ 0.4% peptide graft. The gelatin-1.0% UP-LVG alginate mixes also failed to induce morphology change in the OECs, and neither did any of the ECM molecule-1.0% UP-LVG alginate mixes or the sulphated alginate-1.0% UP-LVG alginate mix. The cells encapsulated in gelatin-alginate mix capsules displayed an overall high viability, while the cells encapsulated in ECM molecule- alginate mix and sulphated alginate- alginate mix displayed lower viability than cells encapsulated in pure UP-LVG alginate. All capsule varieties displayed generally good stability in culture, with the exception of the gelatin-alginate mix capsules that progressively dissolved in culture.

The mucus layers on the internal surfaces of the human body serve as an important barrier against foreign material, but it also create restrictions regarding drug delivery. Discovering methods to overcome this barrier would lead us closer to an efficient delivery of larger drugs and nanoparticles. Recent studies have shown that alginate G-block polymers can modify the physical properties of mucus, and that the G-blocks make the mucin network more open and increase particle transport through mucus. This shows that G-blocks are an interesting candidate, for use in drug delivery across mucosal barriers, but further work to understand the mechanisms by which the G-blocks interact with mucus is still desirable. Studies have shown that neutral particles diffuse more easily through mucus than charged particles. In this thesis the interactions between nanoparticles with different surface structures and mucus components were compared by dynamic light scattering and the effect of G-block on these interactions was established. The diffusions of all particle types were compared in pig gastric mucin (PGM) from Sigma, by the use of confocal microscopy and multiple particle tracking (MPT). Then alginate G-blocks were added and the diffusion of the particle types was compared. The results showed that G-blocks can reduce the amount of mucin components accumulating on to positively charged surfaces but not to negatively charged particles. The MPT showed that the surface charge of the nanoparticles is the primary determining factor when it comes to diffusion through Sigma mucin, and that the effect on the diffusions caused by G-blocks is relatively small, most probably due to the matrix of Sigma mucin, lacking the large networking polymers on which G-blocks previously have shown their effect. It was found that G-blocks make the distribution of trajectories more homogenous, but that this did not affect the mean displacements.

Several studies have suggested that the host genetics may influence the composition of gut microbiota, but few genes involved in host control have been proposed (Sekirov et al., 2010). Lipocalin 2 (Lcn2) prevents growth of bacteria that rely on catechol type siderophores for iron acquisition (Goetz, et al., 2003; Flo, et al., 2004). It was hypothesized that Lcn2 may impart a selection pressure on establishment of the gut microbiota and thus influence the commensal diversity. The aim of this study was to find out whether the antimicrobial protein, Lipocalin 2, has a determining effect on the colonization of the gut microbiota in mice. Two factors were investigated: genotype (Wt, Ht and Lcn2 KO) and habitation (single-housing and co-housing). The naturally developing gut microbiota of wild type mice (Wt), heterozygote mice (Ht) and lipocalin 2 deficient mice (Lcn2 KO) were studied, as well as re-established microbiota after antibiotic perturbation, by collecting stool samples. Microbial community profiles were generated by the use of PCR and denaturing gradient gel electrophoresis (DGGE). The gels were analysed by the software program gel2K (Norland, 2002) and band intensity profiles were compared by statistical analysis. The study of mice gut microbiota revealed differences in the microbial profiles between Wt-, Ht- and Lcn 2 KO-mice. The result showed that both the factor of genotype and habitation were significant factors for the observed differences. For the single-housed mice (mice of same genotype), a significant difference of gut microbiota was found between Wt/Ht-mice and Lcn 2 KO-mice, indicating that the genotype was the main factor for the observed differences. Lcn 2 thus seems to influence the natural colonization of the mice gut, as well as the re-establishment of the microbiota after a perturbation with antibiotics treatment.For the co-housed mice (mice of mixed genotypes) both the effect of genotype and maternity seemed to influence the composition of the microbiota, although the factor of maternity was not taken into account in the analysis. The experimental set-up was designed with the intention of comparing littermates in order to minimize other effects than the knockout of Lcn2. Unfortunately this design meant that the effect of genotype and maternity could not be differentiated.

All animals live in symbiosis with complex microbial communities. The gastrointestinal system in vertebrates is a natural environment for microbes, and this leads to a complex and numerous microbiota. The gastrointestinal (GI) microbiota has several functions of importance to the host, and the development of molecular biological methods for investigation of microbial communities has lead to a new understanding of this environment. The hypothesis of this thesis was that growth rate in larval fish is partly explained by the composition of the GI microbiota. This was tested by comparing the GI microbiota of slow and fast growing Atlantic cod (Gadus morhua) and mangrove killifish (Kryptolebias marmoratus) of the same age. The GI microbiota was characterized by PCR/DGGE (denaturing gradient gel electrophoresis) and sequencing of bands from the DGGE gels. There was a significant difference between the GI microbiota in fast and slow growing individuals from cod and the mangrove killifish strain DAN. The mangrove killifish strain PAN-RS also showed differences, but these findings were only marginally significant. This can partially be explained by the low number of samples analyzed. The GI microbiota of the PAN-RS juveniles had similarities with the microbial composition of both the feed and water, and showed that the GI microbiota is affected by both. In an experimental test it was attempted to examine if exposure to the culturable microbiota from either slow or fast growing individuals could reproduce size differences. However, the cultured bacteria from fast and slow growing mangrove killifish PAN-RS larvae were not significantly different in composition. Thus it was not expected to find any size difference between the fish larvae supplied with the different cultured bacteria. This was confirmed analytically, but the fish larvae supplied with the bacteria had a larger variation in size than the control group.The results in this thesis indicate a difference in the composition of the GI microbiota between fast and slow growing fish in the early stages of development. Further studies are required to verify if this is a causal relationship where differences in the GI microbiota of individuals results in differences in somatic growth.