Introduction

Using synthetic biology for pathway engineering will require enzymes to be used in-vivo rather than isolated or purified. This is particularly true for complex enzymes such as the alkane mono-oxygenase (AlkB) of Pseudomonas putida GPo1 which is an integral membrane protein, requiring two coenzymes (AlkG and AlkT) as well as two cofactors (NADH and FAD). Screening the substrate specificity of such enzymes in this context will rely on the assumption that transport of the substrate into the cell occurs at an appropriate rate. The alkane degradation pathway native to Pseudomonas putida GPo1 has been the focus of extensive research over the past decades and much is understood about the components of the pathway, the substrate range1,2, enzyme mechanism3, electron transport coenzymes4,5,6,7, regulatory system8,9,10, effect on host cell physiology11,12,13,14, performance in recombinant hosts15,16,17,18 and potential applications19. It has previously been hypothesised20,21,22 from the presence of alkL in the alk operon and its position in the outer membrane23 that it plays an important role in transport of alkanes into the cell. Despite this, the role of the alkL component in alkane transport has remained unproven with no apparent loss of function observed when alkL is knocked out23. The authors of the work speculated that the reason no loss of activity was observed is because alkL may only be necessary for uptake when growing in conditions where the alkane concentration available is extremely low and therefore diffusion across the membrane alone will not be sufficient. Redundancy of outer membrane transporters with the same function has also been speculated as a possible explanation21.

The scientific literature indicates that whilst both the purified alkane hydroxylase complex and the wild-type containing the alk genes were able to utilize aliphatic alkanes in the range of C6 –C12, the recombinant P.putida S81 pGEc41, which contains all alk genes with the exception of alkJ (an alcohol dehydrogenase), alkK (an acyl CoA synthetase) and alkL, is not capable of oxidizing n-dodecane1,16. Use of the E.coli pGEc47ΔJ strain, which contains all alk genes except for alkJ, has been shown to be capable of n-dodecane bio-oxidation24. Although other explanations are available, such as increased metabolism of the product or other intrinsic host properties in the case of the P.putida host, the presence of alkL on the plasmid is a notable difference between the two strains and therefore in this study its role in uptake of medium chain alkanes is investigated. Recent work by others has also identified that AlkL can enhance oxidation of nonane, methyl dodecanoate25 and limonene26 in resting cells which further supports the hypothesis for the role of alkL in the uptake of medium-chain alkanes.

Results

Factorial experimental design for systematically identifying the function of alkL

To test the hypothesis that alkL is necessary for uptake of C12 alkanes, E.coli expressing the alkane hydroxylase complex of P.putida GPo1 was used as a test system to determine if uptake was occurring. The alkL gene was expressed under control of the pALK promoter on a plasmid containing the alkB,F,G,T genes, which code for the alkane-1-monooxygenase (AlkB), rubredoxin (AlkG) and rubredoxin reductase (AlkT) which form the alkane hydroxylase complex responsible for the oxidation of n-alkanes to 1-alkanols (Figure 1). The regulatory protein AlkS was also present to control expression of the alk genes. An equivalent plasmid without the alkL gene was used as a negative control. Previous attempts to oxidise dodecane to 1-dodecanol have identified overoxidation and metabolism of 1-dodecanol as a bottleneck with the pGEc47ΔJ plasmid24 and so the plasmid for the new biocatalyst was also designed to minimize metabolism by removal [from the pGEc47ΔJ plasmid] of the aldehyde dehydrogenase (alkH) and acyl CoA synthetase (alkK) which will contribute to the overoxidation of 1-dodecanol to dodecanoic acid and further into the β-oxidation pathway.

Figure 1
figure 1

Schematic of proposed mechanism for alkane uptake and bioconversion via the AlkB monooxygenase complex.

Full size image

The Clontech in-fusion system27 was used for generating these new plasmids containing the minimal set of genes for alkane oxidation in a one-step multi-fragment recombination reaction.

The plasmids were tested using a factorial design in DH1 and HB101 strains of E.coli according to Table S1. The results show (Figure 2) that from 16 biological replicates of n-dodecane oxidation using the E.coli pSTBFG cells (without the alkL gene product), only two showed detectable levels of oxidised product, with yields of 0.004 and 0.006 goxidised products/gdcw after 20 hours, the other 14 showed no detectable product. The host cells containing AlkL showed activity in all 16 replicates on n-dodecane with a maximum yield of 0.73 goxidised products/gdcw (average 0.18 goxidised products/gdcw). Oxidised products of n-octane were detected in all host cells containing either pSTBFG or pSTBFGL. This demonstrated that an effective alkane hydroxylase complex was being expressed and was functional in the hosts lacking alkL and therefore demonstrating that the presence of alkL is necessary for activity on n-dodecane.

Figure 2
figure 2

Specific yields (grams of product per gram of dry cell mass) of the alcohol, aldehyde and fatty acid products resulting from bio-oxidation of n-octane and n-dodecane using the pGEC47ΔJ, pSTBFG and pSTBFGL plasmids transformed into E.coli DH1 and HB101 according to the factorial design in table S1.

The graphs show the average of microwell bioconversions from 8 separate colonies under each condition.

Full size image

The factorial screening experiment revealed that activity of the STBFGL plasmid was achieved over all conditions tested upon n-dodecane. Analysis of variance (ANOVA) was performed on the data and confirmed that the cooperative effect between the plasmid in combination with the substrate (C8 or C12 alkane) was validated over all conditions and was the most significant of the factors varied (p < 0.001).

SDS-PAGE analysis was performed to confirm alkB expression. AlkL levels were too low to detect over the background using SDS-PAGE but its expression was demonstrated by Western blot (Figure S1) and correlated with whole-cell bioxidation activity on dodecane. The work also demonstrates that alkL is not strictly necessary for activity on n-octane in E.coli although the work of others has shown alkL to improve initial rates of n-octane oxyfunctionalisation by 4-fold25. In this case, the less hydrophobic n-octane substrate may be able to diffuse across the membrane without a transporter or may be able to be transported into the cell by native transporters in E.coli, in particular FadL or OmpW. It seems that effective transport of the longer chain alkanes (C12-C16) does not occur through these proteins. This may be due to the selectivity of the transporters or that: if these transporters do enable longer chain alkane transport they may do so at insufficient levels for detectable activity with the alkB complex, or alternatively these native transporters may not be adequately expressed. E.coli strains expressing alkL were also tested for a direct measurement of n-dodecane uptake via resuspending the cells to a constant cell density, incubation with the alkane substrate for 15 minutes, centrifugation and washing of the E.coli pellet twice, extracting the intracellular alkanes into ethyl acetate overnight and quantifying by gas chromatography. The results of this showed that the control cells not expressing alkL bound up to 20% of the alkane compared to 40% in the cells expressing alkL (figure S4).

Removal of alkH and alkK to reduce overoxidation of 1-alkanols to fatty aldehydes and fatty acids

We have previously reported that overoxidation of the desired 1-dodecanol product to the 1-dodecanal and dodecanoic acid products is a major issue when using the pGEc47ΔJ strain24. Since the phenomenon of excessive acid build-up in the absence of the alcohol dehydrogenase alkJ was witnessed it was considered that removal of the downstream aldehyde dehydrogenase, alkH, would provide information on whether alkH was primarily responsible for the conversion from alkanol to aldehyde to acid.

Removal of alkK and alkH from the plasmid resulted in producing a greater average proportion of alkanol and aldehyde products and reducing overoxidation to the carboxylic acid product in the case of n-dodecane (Figure 3). Aldehyde proportion increases from 2% of oxidized products to 10% in the ΔalkH strain. An increase in aldehyde yield was expected, however, dodecanoic acid is still the majority product (77% of total oxidized products). This hints that AlkH was not the primary cause of carboxylic acid formation in the pGEc47ΔJ strain. It also provides evidence that AlkB may be responsible for carboxylic acid formation directly. This has been shown to occur in other oxygenase systems such as P450ALK of Candida tropicalis which is capable of producing dicarboxylic acids directly from alkane substrates28. The isolated alkB enzyme is already known to overoxidise to the aldehyde2,29 but it has yet to be proven that it is capable of carboxylic acid formation. Although there is evidence linking it to a component under control of the alk promoter in pGEc47ΔJ24 and the work reported in this study further supports the hypothesis that AlkB may be capable of oxidation to the carboxylic acid product. However, endogenous E. coli enzymes or non-enzymatic oxidation of the unstable aldehyde could also explain this overoxidation or contribute to it.

Figure 3
figure 3

Proportion of 1-dodecanol, 1-dodecanal and dodecanoic acid products as a percentage of the total quantity of oxidized products for bio-oxidation of n-dodecane using the pGEC47ΔJ and pSTBFGL plasmids transformed into E.coli DH1 and HB101 according to the factorial design in table S1.

The graphs show the average of microwell bioconversions from 16 separate colonies with each plasmid.

Full size image

Effect of alkB and alkL expression on plasmid stability

The specific activity demonstrated by the new pSTBFG and pSTBFGL constructs was lower than that of the pGEc47ΔJ plasmid for both octane and dodecane substrates. The cell growth was also lower, particularly in the cells containing the pSTBFGL construct (Figure S2), indicating a particularly detrimental effect on cell viability of expressing the alkL protein. There is much evidence in literature of poor stability of alk gene expression13,17,30 with instability in expressing the alkane hydroxylase enzyme (AlkB) observed in both the P.putida wild-type host13 and in E.coli strains W3110 and DH117. SDS-PAGE demonstrated that the AlkB component was being expressed to levels at least as high as the pGEc47ΔJ cells (Figure S3). This indicates that the inferior performance of the new constructs may be caused by negative consequences of overexpression from the high copy number plasmid. This may be as a result of excessive quantities of the electron transport coenzymes AlkG and AlkT (although not seen on the SDS PAGE gel over the background) inclusion body formation of one or more of the Alk components, inadequate translocation of the AlkB protein to the inner membrane, inappropriate ratios of the AlkB, AlkF and AlkG components, compromising the integrity of the membrane or some other limitation. A previous study into this enzyme complex also indicated that overexpression of the complex in W3110 leads to compromised stability and does not result in higher specific activities6.

Controlled expression of the alkL protein to overcome toxicity and control intracellular enzyme activity via transport protein expression level

To overcome the issue of plasmid stability caused by alkL, the alkL protein was cloned into a rhamnose inducible plasmid which allowed tighter control of expression. The plasmid was then co-transformed alongside the pGEC41 plasmid (which lacks alkL) and using this configuration yields of up to 8 g/L of total oxidation products were observed after 24 hours (1.7 g/g) albeit with a greater proportion of dodecanoic acid compared to 1-dodecanol (Figure 4).

Figure 4
figure 4

Effect of alkL expression level on volumetric (A) and specific (B) product yields and biomass concentration (C).

Graphs A&B show the change in oxidation products of the n-dodecane bio-oxidation using alkB, alkG and alkT from the pGEc41 plasmid by co-expressing alkL using either a low-range or high range expression plasmid under different inducer concentrations. The Y-axis shows the yields of products measured after 24 hours compared to the negative control with only the pGEc41 plasmid. The numbers on the x-axis represent inducer concentrations in μM of Rhamnose.

Full size image

The alkL protein was expressed from two different plasmids with different levels of expression. Using the low-range (weak rbs) expression vector showed that n-dodecane oxidation activity could be directly correlated with expression level and that the maximum activities achieved by the un-induced high-range (strong rbs) expression vector could be matched. By changing the ribosome binding site (RBS) to contain an increased spacer region between the RBS and the start codon it was possible to introduce this fine level of rheostatic control of expression by modulating the concentration of rhamnose inducer. This demonstrates a powerful approach which can be applied to optimizing the substrate feeding rate into the cell by modulating the transport characteristics of the cell. Furthermore the overall product yields of dodecanoic acid were twice that of the construct with the original RBS. The specific yield of product (g/g) was equivalent for the two variants but higher cell densities were able to be reached with the modified RBS thus indicating that this strategy seems to be effective to overcome the toxic burden of overexpression of the transport protein as well as finely controlling the the expression level of the transport protein (Figure 4).

Alkane substrate range

Now that the role of AlkL in alkane transport has been established, the substrate range of compounds upon which AlkL could act was investigated. It was found that the AlkL protein also enabled activity on tetradecane and hexadecane. The strain containing the pSTBFG plasmid showed only activity on n-octane and no activity on C12-C16 alkanes, further confirming that AlkL is the necessary component for activity on these longer chain alkanes in E.coli (Figure 5).

Figure 5
figure 5

Average specific yield of products per gram of dry cell weight resulting from bio-oxidation of linear alkanes of carbon chain length C8- C16 (n = 4) cultured according to the conditions stated in the material and methods section.

Full size image

This supports previous reports that C12-C16 aliphatic alkanes are viable substrates for this enzyme complex1,2 and explains the previously observed discrepancy between in-vivo activity using P. putida Gpo1 pGEc41 (ΔalkL) and in-vitro activity of the purified enzyme complex1.

Using the rhamnose inducible vector co-transformed with pGEc41 also demonstrated initial rate improvements of up to 10-fold with n-octane, up to 120 μmol/min/g and 1 g/L/h (Figure S6). Improved activities on heptane were also observed with rates improving by 75% from 125 mg/L/h to 228 mg/L/h for production of 1-heptanol. With hexane, alkL expression did not improve activity significantly under the conditions tested (110 mg/L/h). Heptadecane oxidation was also attempted which resulted in no detectable C17 oxidation products. Since the AlkB enzyme has never before been shown to work on alkanes >C16 it remains unclear whether the AlkL protein can act on alkanes longer than hexadecane.

Activation of the AlkS biosensor

As an alternative method to demonstrate that AlkL facilitates uptake of alkanes, the AlkL protein was co-expressed alongside an alkane biosensor based on the transcription factor AlkS and tested for the rate of uptake of C10, C11 and C14 n-alkanes. Figure 6 demonstrates that as induction of the AlkL protein is increased the rate of specific activation by both C10 and C11 n-alkanes is increased above the control strain and the strain where alkL is not induced. In the case of decane, an 8-fold increase in specific fluorescence activation was observed when alkL was induced compared to a 3-fold activation in the control cells and uninduced AlkL strains. In the case of undecane no activation was observed without AlkL expression. When AlkL was expressed a 4.5-fold activation was observed with undecane. As expected, no activation was observed with tetradecane as previous studies have shown that AlkS is only activated by alkanes of chain length C5 – C1143. A repeat of the experiment using n-dodecane in place of tetradecane confirmed that dodecane also failed to activate AlkS (data not shown).

Figure 6
figure 6

Average specific fluorescence (GFP emission/OD) of E.coli containing pSB50C7 (alkS-GFP) by different alkane effector molecules (C10-C14) with and without expression of the alkL protein.

Full size image

Structural study

A homology model of AlkL (Figure 7) was generated by reference to the structures of the E.coli OmpW protein (PDB 2f1t). This was the closest template for which an x-ray crystal structure was available (29% sequence identity). The model structure exhibited the same 8 stranded β barrel with a hydrophobic core and the same lateral opening, exiting into the outer membrane. This lateral opening is formed by 2 proline molecules at equivalent positions as found in OmpW; the structural model is therefore consistent with the lateral diffusion hypothesis which is suggested to be the primary mechanism of hydrophobic transport across the cell membrane. The proline molecules surround an opening from the core of the channel into the membrane. The Van den Berg group also proposed there to be a narrowing of the OmpW channel residues which blocks passage through the channel and directs molecules to exit through the lateral opening. In the case of OmpW, side chain interactions between leucine and tryptophan residues on opposite sides of the protein channel seem to act to close the channel beneath the lateral opening (Figure 7). In the AlkL homology model, equivalent side chain interactions are predicted at the same position below the lateral opening but with phenylalanine and isoleucine. The model also indicates that the AlkL protein contains a hydrophobic channel as in OmpW. Docking studies indicated that the predicted region of highest binding energy using n-dodecane as a ligand was observed in an extracellular hydrophobic binding pocket. The equivalent position in the native E.coli OmpW protein also bound n-dodecane but the region of predicted binding was narrower (Figure 7D). Due to the seemingly low sequence identity between the OmpW template and AlkL, supplementary homology models were generated using the 2f1tb template and the next closest homologue with an x ray crystal structure, OprG (Pdb 2 × 27) using the Swissmodel, Modweb and Phyre servers. The alignment of the alternative homology models (Figure S5) show some deviation in the extracellular and periplasmic loop regions but the hydrophobic channel and lateral opening are common between the Swissmodel models generated from both the 2f1t template and 2 × 27 templates as well as when the model was generated using the modweb server. Whilst the docking and homology modeling data offer plausible functional explanations for the mechanism of transport through AlkL, this remains speculative at present.

Figure 7
figure 7

Structural comparison between the homology model of AlkL and the x-ray crystal structure for E.coli OmpW (A) Surface model indicating hydrophobic side chains for AlkL (blue) and OmpW (green) (B) Cartoon structure of AlkL model with cavities shown as a mesh. (C) Close up of the proline residues forming the lateral opening into the outer membrane and the side chain interactions closing the channel beneath the lateral opening. (D) Cartoon structure of AlkL and ompW showing 100 positions of n-dodecane docked using autodock 4.

Full size image

Discussion

This study provides clear evidence of the role of AlkL in the uptake of medium chain alkanes in the range of C12-C16. Our findings are complemented by a recent report showing that coexpression of AlkL significantly increased oxygenation of nonane and also dodecanoic acid methyl esters. This supports the evidence of a role in transport hypothesized previously due to the position of the AlkL protein in the alkane degradation operon23 and the additional structural parallels between the AlkL protein's hydrophobic core and lateral opening with the fatty acid transporter FadL31. The current report now shows that, whilst AlkL is not required for n-octane uptake and subsequent conversion it is necessary for C12-C16 uptake and subsequent oxidation by the alkane monooxygenase complex. It is therefore likely to play a crucial role in enabling growth of alkane assimilating bacteria on medium-chain alkanes in the wild type P.putida GPo1 strain. Further work to mutate the lateral opening residues to block transport would be a useful follow up to demonstrate conclusively that transport does indeed occur directly through this protein rather than as a result of other less likely possibilities such as weakening of the cell membrane as a result of overexpression of the protein. Such lateral opening mutants of the transporter FadL have been used successfully to demonstrate that transport of fatty acids does indeed occur via the lateral opening20.

As reported in work by the Van den Berg group, AlkL and OmpW family proteins are widespread in Gram-negative bacteria and yet little is known about the majority of these proteins. The characteristic feature of this particular protein family is the hydrophobic core which is unusual for outer membrane channels which usually contain a hydrophilic core filled with water21. The demonstration of the requirement of AlkL in the uptake of C12-C16 alkanes shown in this paper combined with the hydrophobic channel shared by members of the family hints that uptake of hydrophobic compounds could be the primary function of this protein family. The presence of homologues in operons for the degradation of other hydrophobic molecules such as dibenzothiopene and naphthalene support this hypothesis. The removal of the OmpW outer membrane protein in Pseudomonas fluorescens prevented uptake of naphthalene32; this is the only other OmpW protein for which the function has so far been proven.

The work also provides further evidence that supports the theory that the outer membrane and likely the hydrophilic lipopolysaccharide layer, is the principle barrier to uptake of long-chain alkanes in E.coli. This has been shown to be the case in the past with lipophilic drugs41,42,43. In the case of methyl dodecanoate, which is structurally similar to dodecane, the addition of the metal chelator EDTA, which is known to disrupt the formation of lipopolysaccharide, has also been shown to improve the permeability of E.coli by 2.8-fold25.

In conclusion, the work demonstrates that supplementing a whole-cell biocatalyst with an importer active on the substrate of interest can have an enormous impact on the observed activity and should be an important consideration in the engineering design of such systems. The work also demonstrates that when studying the rate of import through a candidate uptake protein, particularly with the intention of building a highly productive whole-cell biocatalyst, the use of a highly active intracellular enzyme can be a particularly effective way to measure the true uptake rate since other assay set-ups such as growth based assays may not reveal the true rate of uptake if uptake is not the rate limiting step.

Methods

Material and methods

Medium composition

The composition of the aqueous phase in all experiments was either luria broth or as described by Wubbolts et al30. KH2PO4, 4 g/L; K2HPO4 (3H2O), 15.9 g/L; Na2HPO4 (12H2O), 7 g/L; (NH4)2SO4, 1.2 g/L; NH4Cl, 0.2 g/L (all from Sigma Aldrich); yeast extract (Oxoid), 5 g/L; L-leucine, 0.6 g/L; L-proline, 0.6 g/L; thiamine, 5 mg/L (All from Alfa Aesar). After autoclaving, MgSO4 (7H2O) (BDH), 1 g/L (BDH); 1 mL of trace minerals (composition below); 1 mL of 4% (w/v) CaCl2(2H2O) (Alfa Aesar) and 10 g/L D-glucose (Sigma Aldrich) were added having all been heat sterilized separately. 1 ml of 10 mg/ml filter sterilised tetracycline (Sigma Aldrich) was also added.

The solution of trace minerals contained per liter of 5 M HCL: FeSO4 (7H2O) (Sigma Aldrich), 40 g; MnSO4 (H2O) (Sigma Aldrich), 10 g; CoCl2 (6H2O) (Fluka), 4.75 g; ZnSO4 (7H2O) (VWR), 2 g; MoO4Na2 (2H2O) (Sigma Aldrich), 2 g; CuCl2, (2H2O) (Riedel-de Haen), 1 g; H3BO3 (Sigma Aldrich), 0.50 g.

Construction of new plasmids

The E.coli GEC137 pGEc47ΔJ strain33 contains all components of the OCT plasmid with the exception of the knocked out alcohol dehydrogenase alkJ. This was used as a template for PCR amplification of the alk genes used to construct the new plasmids.

From the purified pGEc47ΔJ plasmid the alkST, alkBFG and alkL regions were amplified by PCR using the Phusion Hotstart DNA polymerase according to the manufacturers recommended protocol. The primers were designed with a 15 bp overlap which was designed to be complimentary to a 15 bp region at the end of the neighbouring fragment in the desired final construct. In-fusion cloning27 was then used to join the fragments together using a 3′ to 5′ exonuclease which reveals the 15 bp overlapping regions which anneal and construct the alkST, alkBFG and pUC19 fragments into the plasmid in the desired order and orientation.

The alkL DNA fragment was introduced into the plasmid by ligating the amplified alkL fragment into the TOPO vector and then using restriction digestion to clone from the TOPO-alkL vector into the pSTBFG vector. A synthetic palkB promoter was then ligated upstream of the alkL to ensure expression under control of the alkS regulator.

The Rhamnose inducible plasmid was constructed from the pRHA67K plasmid44 and the alkL gene synthesised by DNA 2.0.

The arabinose inducible plasmid was constructed from the pBTB2 plasmid45.

pSB50C7 (alkS-GFP) was constructed by amplifying the alkS gene and pALKB promoter into pSB50C3 plasmid46.

The pSTBFG and pSTBFGL plasmids were purified by the Qiagen miniprep protocol and pGEc47ΔJ was purified by the Qiagen maxiprep protocol. All DNA concentrations were verified by nanodrop and electrophoresis gel. The composition of all plasmids was verified by sequencing.

Whole-cell bio-oxidation

Microwell biotransformation experiments

Carried out in polypropylene 24 deep square well plates with between 1.2 and 2 ml of two-liquid phase medium agitated at 250 rpm with 25 mm throw diameter at 37°C as described previously44. The aqueous phase was inoculated with 5% v/v seed culture with 17% v/v organic phase. The medium used is as described under media composition; the organic phase consisted of the respective alkane substrate with 0.05% dicyclopropylketone (DCPK) as inducer of the alk system.

Cell density

Cell density measurements were taken by sacrificing the well contents and centrifuging the two-phase samples at 13000 rpm (19000 g) for 15 minutes, marking the aqueous volume on the side of the graduated Eppendorf tube; washing the pellets with tris-HCL pH7.4 and drying in an 80°C oven until a constant mass was reached (24–96 hours).

Gas chromatography

Determination of oxidation products in microwells was done via sacrificial sampling and extraction of both phases into ethyl acetate directly. 800 μl of ethyl acetate was added to the two-phase supernatant following centrifugation at 13000 rpm for 5 minutes in a microfuge. The samples were vortexed for 3 × 20 seconds before further centrifugation for 1minute and removal of 100 μl of the organic phase to be analysed by GC-FID as previously mentioned. The proportion of organic phase was determined by cross-referencing the n-alkane peak size to a calibration curve of known standards.

The column used was FID-GC using a SGE BPX5 (30 m long; 0.53 mm internal diameter, 1 μm film) capillary column with helium as carrier gas under a constant pressure of 4PSI. For n-dodecane, n-tetradecane, n-hexadecane oxidation product determination, the samples were eluted at an initial temperature of 150°C for 2minutes, followed by a linear increase of °C minute−1 to reach a final temperature of 240°C. Injector and detector temperatures were both 280°C. The concentrations were determined by cross-referencing to a set of the corresponding alkane, 1-alkanol, 1-alkanal and alkanoic acid standards analysed in the same run. For n-octane oxidation product determination the samples were eluted at 70°C followed by a linear increase of 5°C minute−1 to reach a final temperature of 145°C. All standards were purchased from Alfa Aesar at the highest purity available (>98%).

Design of experiments

Statistical experimental design and interpretation of the results was aided by use of Design Expert 7 from Statease. All experimental designs reported in this paper were general factorial designs.

The comparison of pSTBFG, pSTBFGL and pGEc47ΔJ experiment varied seed culture media (LB or an optimized media described above +10 g/L glucose), antibiotic concentration (100 mg/L or 500 mg/L ampicillin; 4 mg/L or 20 mg/L tetracycline), harvest time (8 hours and 24 hours), substrate (n-octane or n-dodecane), host strain (DH1 or HB101) and plasmid (pSTBFG, pSTBFGL and pGEc47ΔJ) (Table S1). Single colonies transformed with the corresponding plasmid the previous day were used to inoculate 3 ml of seed culture according to the experimental design and 100 μl subcultured after 11 hours into 900 μl fresh media containing the antibiotic concentration set out in the experimental design and 200 μl of the alkane substrate, 4% v/v Dimethyl sulphoxide (DMSO), 0.1% v/v triton-x-100 and 0.05% v/v dicyclopropylketone (DCPK).

The substrate range experiment varied plasmid (pSTBFG and pSTBFGL), substrate (n-octane, n-dodecane, n-tetradecane and n-hexadecane) and harvest time (8 hours and 20 hours). The experimental protocol was the same as above but with the following fixed factors: the host strain was DH1; the seed medium (see composition above) contained 10 g/L glucose; the antibiotic concentration was 500 mg/L ampicillin.

Assays using the pGEc41 plasmid cotransformed with the pRha67K alkL vector were performed in HB101 and used the protocol as described above without DMSO addition and using 10 g/L glycerol in place of glucose as the carbon source so that glucose repression of the RhaR RhaS system did not inhibit AlkL expression. The dodecane oxidation study was harvested after 24 hours and the substrate screening experiment pre-induced the AlkB and AlkL expression for 4 hours prior to substrate addition and the reaction was quenched and harvested after 1 hour by addition of ice-cold ethyl acetate.

Structural modeling

Structural alignment and viewing the structure of AlkL and OmpW was performed using Pymol. The models were generated using Swiss Model34,35 using the 2f1t x-ray crystal structure as a template. The membrane position was predicted using the orientations of proteins in membranes database36

SDS-PAGE AlkB expression determination

Lysate samples were prepared by resuspending cell pellets in distilled water and sonicating for 10 × 10 s pulses with 10 s between each pulse. The Bradford assay37 was then used to determine protein concentration before loading standardised protein concentrations of 5 μg per lane. SDS-PAGE38 was then carried out according to the Invitrogen Nupage method39. Densitometry was used to compare band intensity as a % of the whole lane after background intensity was subtracted using the rolling disk method.

Western blot

To confirm expression, a C-terminal 6HIS alkL construct was synthesised which expressed alkL from a second plasmid under control of the rhamnose promoter and under kanamycin resistance. This was used to transform and complement the pGEC41 (ΔalkL) construct to restore the dodecane oxidation phenotype. HB101 Cells were cultured overnight and induced with 10 mM Rhamnose and harvested after 24 hours. SDS-PAGE was performed as described above loading 10 μg of protein per lane. Semi-dry transfer was performed after thoroughly washing the gel in phosphate-bufffered saline. Blocking was performed using Amresco's rabidblock™ solution followed by incubation overnight at 4°C in the anti-HIS rabbit antibody with near infrared emitting dye conjugate. The blot was then washed thoroughly in PBS + 0.01% Tween and analysed using the Licor Odyssey40.

Alkane biosensor assay

E.coli DH5αZ1 and E.coli HB101 were transformed with an alkS-Superfolder GFP biosensor and cotransformed with either (i) a control expression vector pBTB2 containing no gene or (ii) pBTB2 containing alkL. The strains were grown up overnight from single colonies in parallel in 24 Deep square well plates (1 ml per well) at 37°C and 250 rpm (25 mm throw diameter) in Luria broth containing 5 g/L glucose and then sub cultured (4% incoculum) into the media described above containing 10 g/L glycerol, grown for 2 hours, induced using either 0, 0.3 mM or 3 mM L-arabinose as appropriate. After 3 hours growth, 1% v/v alkane substrates were added and grown for 3 hours before analysing fluorescence (excitation 485 nm; emission 517 nm) and OD using the BMG Clariostar plate reader.