ARTICLES

Expression of Vitreoscilla Hemoglobin in Escherichia coli EnhancesRibosome and tRNA Levels: A Flow Field-Flow Fractionation Study

Mikael Nilsson,† Pauli T. Kallio,‡ James E. Bailey,‡ Leif Bu1 low,§ andKarl-Gustav Wahlund*,†

Technical Analytical Chemistry, Center for Chemistry and Chemical Engineering, Lund University, P.O. Box124, SE-221 00 Lund, Sweden, Institute of Biotechnology, ETH-Zurich, CH-8093 Zurich, Switzerland, and Pureand Applied Biochemistry, Center for Chemistry and Chemical Engineering, Lund University, P.O. Box 124,SE-221 00 Lund, Sweden

Asymmetrical flow field-flow fractionation (FFF) was used to separate and quantitate70S ribosomes, the 30S and 50S subunits, and tRNA in one single analytical procedure.The method was applied to an investigation of the effect of Vitreoscilla hemoglobin(VHb) on the translational machinery of the recombinant Escherichia coli cells. Thenumber of active 70S ribosomes per cell increased dramatically, more than 2-fold, asdid also the tRNA levels for the VHb-expressing strain relative to VHb-negative controlat the end of a 30-h fed-batch cultivation. This was accompanied by a corresponding61% increase of a cloned marker enzyme activity. The results clearly indicate thatVHb promotes the level of translational components. There should be many other casesin bioengineering where it is important to relate the protein production level in abioreactor to the ribosome and tRNA levels.

IntroductionExpression of VHb in oxygen-limited Escherichia coli

cultures has been shown to increase final cell density (1)and also to improve expression levels of cloned enzymes,with studies of â-galactosidase, chloramphenicol acetyl-transferase (CAT), and R-amylase so far reported (2, 3).Higher cell density at harvest implies that the specificgrowth rate of the VHb-expressing cells must be greaterthan that of wild-type cells during some portion of thecultivation (1).

Perturbations in metabolism correlated with the pres-ence of VHb have been previously documented which areconsistent with higher growth rates of VHb-containingcells. A sequence of experimental studies on synthesis ofATP, content of cytochrome complexes, NAD(P)H utiliza-tion, and oxygen metabolism of microaerobic E. coliproducing VHb indicated involvement of VHb in enhanc-ing the activity and efficiency of the electron transportchain under oxygen-scarce conditions (4-8). However, noprior study has investigated the influence of VHb on 70Sribosome content of the cells. Such information can givenew insight into the action of VHb as it would enableone to study if there is a correlation between the growthrate and the ribosome content of VHb-expressing cells.

Ribosome organization, i.e., how ribosome componentsare partitioned between unassociated 30S and 50S

subunits and active 70S ribosomes, is known to beimportant, since translational activity depends on thecontingent of assembled 70S ribosomes in the cells. Thisis known to be affected by primary metabolism, specifi-cally via cellular content of GTP. A short review of thismechanism is provided in ref 9. Since VHb affects severalkey parameters of primary metabolism, it may also alterribosome organization.

In addition, it should be interesting to estimate thetRNA level to study its relationship to the translationalcapacity of the bacterial cells.

A method, asymmetrical flow field-flow fractionation(FFF) (10-12), is now available for rapid simultaneousmeasurement of ribosome content, organization, andtRNA levels (13). Separation between the ribosomalspecies and tRNA is achieved on the basis of differencesin their diffusion coefficients which in turn depend onthe hydrodynamic size. The separation takes place in athin channel in which an aqueous carrier liquid (buffer)is pumped and creates an axial parabolic flow profile inthe channel (Figure 1). The differently sized particles aresubjected to a flow of the carrier in a direction perpen-dicular to the axial channel flow. The particles willsample axial flow velocities reflecting the balance be-tween the cross-flow-induced transport and the oppositelydirected diffusional transport caused by Brownian mo-tion. In this way elution times can be converted intodiffusion coefficients and these in turn to hydrodynamicdiameters. Separations are completed in 10 min, whichmakes the method a fast alternative to ultracentrifuga-tion, which often is burdened by run times up to severalhours.

† Technical Analytical Chemistry, Lund University.‡ ETH-Zurich.§ Pure and Applied Biochemistry, Lund University.* Corresponding author. E-mail: karl-gustav.wahlund@

teknlk.lth.se. Telephone: + 46 46 222 83 16. Fax: + 46 46 222 4525.

158 Biotechnol. Prog. 1999, 15, 158−163

10.1021/bp9900155 CCC: $18.00 © 1999 American Chemical Society and American Institute of Chemical EngineersPublished on Web 03/05/1999

The objective of this work was to investigate the effectof VHb expression on the translational machinery of E.coli by quantitating the amounts of ribosomal particles(inactive subunits, 30S and 50S, as well as active 70Sribosomes) and tRNA using asymmetrical flow FFF.Samples were withdrawn at various time points ofmicroaerobic fed-batch bioreactor cultivations. Both thenumber of ribosomes per cell and the mass fraction ofactive 70S ribosomes were calculated for VHb-expressingE. coli cells and compared with VHb-negative controlcells. In addition, an estimate of total tRNA level for aVHb-producing strain and for the wild-type control wasextracted from the FFF data. The results were correlatedto the cellular activity of â-lactamase, a cloned markerenzyme.

Materials and Methods

Strains and Plasmids. Experiments were conductedusing E. coli MG1655 (λ-, F-; Cold Spring HarborLaboratory) carrying either the control plasmid pKT1 orthe IPTG inducible VHb-expression plasmid pKTV1.Constructions of these plasmids have been describedpreviously (7).

Media and Bioreactor Cultivations for RibosomeAssay. Preculture bioreactor inoculations of eitherMG1655::pKT1 or MG1655::pKTV1 were grown at 37 °Cand 250 rpm for 14 h in 100-mL shake flasks containing20 mL of Luria-Bertani medium (1% tryptone, 0.5%yeast extract, and 1% NaCl). Fed-batch bioreactor culti-vations were performed in a glucose batch medium thatcontained 4 g/L glucose, 0.4 g/L (NH4)2SO4, 4.35 g/L K2-HPO4, 1.5 g/L KH2PO4, 1 mL/L trace metal mix (8.3 mMNa2MoO4, 7.6 mM CuSO4, 8 mM H3BO3), 1 mL/L vitaminmix (0.042% riboflavin, 0.54% pantothenic acid, 0.6%niacin, 0.14% pyridoxine, 0.006% biotin, 0.004% folicacid), 1 mM MgSO4, 0.05 mM CaCl2, 0.2 mM FeCl3, and100 mg/L ampicillin and was supplemented with 150mg/L casamino acid (Difco) and 30 mg/L yeast extract(Difco).

Fed-batch cultivations, in triplicate, were performedin a Sixfors bioreactor (Infors, Switzerland) containing300 mL of glucose batch medium as described above. Tostart Sixfors bioreactor cultivations, 3.0 mL of overnightseeding culture was added and process parameters weremaintained at 37 °C, pH 7 ( 0.2 (adjusted with either 2M NaOH or 2 M H3PO4), 300 rpm, and 0.4 vvm of airsupply. The feed medium consisted of 250 g/L glucose,110 g/L (NH4)2SO4, 8 g/L MgSO4, 1 mL/L vitamin mix, 1mL/L trace metal mix, 0.05 mM CaCl2, 0.2 mM FeCl3,

30 mg/L casamino acid, 15 mg/L yeast extract, and 100mg/L ampicillin. Expression of VHb in E. coli MG1655::pKTV1 was induced by IPTG addition (final concentra-tion of 1 mM) to the culture when the absorbance at 600nm (A600) reached approximately 1.0, which took place 5h after inoculation of the Sixfors reactor. Fed-batch modewas commenced with 1 mL/h of feed medium when theculture reached an A600 of 2.5 and increased to 2 mL/hwhen A600 reached approximately 5 and then was main-tained constant at this level until the end of the 30-hcultivation.

Samples for FFF ribosome assays and enzymaticanalysis were withdrawn at four different time pointsduring the cultivations (9, 24, 27, and 30 h after inocula-tion). Dissolved oxygen (DO) concentration was moni-tored with a pO2 electrode (Ingold, Inc.), and exhaustgases (CO2 and O2) from the Sixfors bioreactors weremonitored using an emission monitor (Bruel & Kjaer,Emissions Monitor Type 3427).

Cell Harvest and Preparation of Ribosomes.Samples (15 mL) from the different time points werepoured on 20-25 g of crushed ice supplemented with 3.25mL of 1% NaN3 to stop bacterial growth. The suspensionswere then centrifuged for 5 min at 5900g at 4 °C. Thepellets were resuspended in 1 mL of buffer A (25 mMTris with 10 mM MgSO4 and 30 mM K2SO4, pH 7.4) andcentrifuged 3 min at 13 000 rpm in an Eppendorffcentrifuge and then resuspended in 1 mL of sucrosebuffer (buffer A with 18% (w/v) sucrose). The resuspen-sions were frozen in liquid nitrogen in 40-µL aliquots andstored at -20 °C. Ribosomes were prepared from the 40-µL aliquots after thawing by adding 10 µL of sucrosebuffer containing lysozyme (5 mg/mL) and incubating 10min on ice. The suspensions were then exposed to afreeze-thaw cycle (liquid nitrogen-water at 20 °C) fourtimes. Then 400 µL of buffer A, 50 µL of 5% (w/v) Brij 58in buffer A, and 3 µL of RNAse-free DNAse were added.The suspensions were incubated 10 min on ice andcentrifuged at 13 000 rpm for 3 min in an Eppendorffcentrifuge. The supernatants were transferred to newtubes and stored at 4 °C awaiting FFF analysis.

Assays. The soluble fraction of cells was prepared byresuspending harvested cells in a sonication buffer (100mM Tris-HCl (pH 8), 50 mM NaCl, 1 mM EDTA) anddisrupting cells using a French press (SLM-Aminco).Globin activities were assayed using CO-reduced minusreduced spectroscopy as described previously (14). Thesoluble protein content was determined enzymatically

Figure 1. Asymmetrical flow FFF channel (top) and separation principle (bottom). The channel shape is defined by the spacer. Aparabolic flow is established by the axial laminar flow. The position of sample particles in the transversal direction is governed bya balance between the cross-flow-induced drift and oppositely directed diffusion. Larger particles (larger dots) are less diffusive andwill therefore on average sample lower flow velocities than smaller particles (smaller dots). Larger particles are thus eluted laterthan smaller particles, and size separation is achieved.

Biotechnol. Prog., 1999, Vol. 15, No. 2 159

with a Beckman SYNCHRON CX5CE autoanalyzer usingthe M-TP assay (Beckman).

â-Lactamase activities encoded by the plasmids pKT1and pKTV1 were determined using the chromogeniccephalosporin nitrocefin (Becton & Dickinson) as asubstrate (15). Briefly, 5 mg of nitrocefin was dissolvedin 0.5 mL of DMSO and 9.5 mL of 50 mM potassiumphosphate buffer (pH 7) was added; the solution wasmixed, and the stock solution was stored at -20 °C. Thecells were resuspended at A600 ) 10 in 50 mM potassiumphosphate (pH 7), 0.4 mM Tris-HCl (pH 7), and 40 µMEDTA buffer containing 0.4 mg/mL lysozyme. The sus-pension was incubated 15 min at 37 °C, and cells werelysed by sonication. Cell debris was removed by centrifu-gation at 15 000 rpm for 15 min at 20 °C. Soluble cellularfraction (1-100 µL) was mixed with 1 mL of the nitro-cefin stock solution, and the change of absorbance at 482nm was recorded at room temperature. The activity ofâ-lactamase is described in units (U) per milligram ofsoluble protein.

Asymmetrical Flow FFF System and Operation.The separation system was similar to that used in aprevious study (13). It consisted of a channel constructedby covering a porous ceramic frit in a Lucite block withan ultrafiltration membrane (molecular weight cutoff10 000), type Nadir UF-C 10 (Hoechst AG, Wiesbaden,Germany). A spacer (thickness 190 µm) with the channelshape cut out was placed on top of the membrane. Theupper wall was a glass plate with inlet and outlet holesdrilled 28.5 cm apart and an injection inlet hole situated2.0 cm from the inlet end. The channel was clampedtogether with six beams across the glass plate using 12bolts through holes in the Lucite block and beams.Channel geometry (see Figure 1) was trapezoidal with atip-to-tip length of 28.5 cm, the breadth at the inlet was2.0 cm, and the breadth at the outlet was 0.50 cm. Theends were tapered with lengths of 2.0 and 0.5 cm at theinlet and outlet, respectively. The area reduction of theaccumulation wall due to the tapered inlet end was 2.1cm. The outlet flow from the channel was monitored at260 nm with a spectrophotometric detector, UVD 340S(Gynkotek, Germering, Germany). A Kontron HPLCpump 422 (Kontron, Zurich, Switzerland) was used todeliver the inlet channel flow of the carrier liquid (bufferA), and a CMA/100 syringe-type microinjection pump(CMA Microdialysis, Stockholm, Sweden) introduced thesample to the channel. To switch the direction of theflows, one two-way and one four-way motor driven valveswere used, both purchased from Valco Europe (Schenkon,Switzerland). The injector was a Rheodyne 9125 syringeinjector (Cotati, CA), and sample volumes used were 60µL by loading the 20-µL sample loop from a 100-µLsyringe. In all cases samples were run in triplicate. Theoutlet and cross-flow flow rates were monitored by twoPhaseSep liquid flow meters (Phase Separations, Queens-ferry, U.K.).

A typical FFF experiment started with a rinsing phasein which the channel was back-flushed for 1 min at 6mL/min with carrier liquid. The flow rate was thenlowered to 2 mL/min, and the flow was divided into twostreams: one entering from the inlet end and the otherone from the outlet end. The two streams met at thesample focusing point 0.5 cm downstream from theinjection inlet point. After 6 s the sample was injectedby starting the microinjection pump which delivered 60µL of carrier liquid at 0.12 mL/min through the sampleloop. After the 30 s of injection the sample was focusedfor yet another 30 s. The flow direction through thechannel was then changed to go from inlet to outlet, and

the desired inlet channel flow rate (7.5 mL/min) was seton the HPLC pump. Control of the pumps and valves aswell as data acquisition were carried out with theGynkosoft software (Gynkotek, Germering, Germany).The channel was kept at ambient temperature (23-25°C).

Ribosomes, Mass Fraction of 70S, and tRNALevel. The number of ribosomes was determined fromthe area under the 70S peak in the fractograms using4.2 × 107 M-1 cm-1 as extinction coefficient for ribosomesat 260 nm. The mass fraction of 70S was determined bydividing the peak area under the 70S peak by the sumof areas under all three ribosomal peaks (70S, 50S, and30S as validated in ref 13). An estimate of the amount oftRNA per A600 unit was obtained by dividing the areaunder the tRNA peak [given as milli-absorbance units(mAU) × minutes) by the A600.

ResultsEffect of VHb Expression on Growth of E. coli

under Microaerobic Cultivations. IPTG-induced,highly controlled, and glucose fed-batch bioreactor cul-tivations of both VHb-negative control cells (MG1655::pKT1) and VHb-expressing cells (MG1655::pKTV1) wereperformed in triplicate to study the effect of VHb expres-sion on activity and amount of ribosomal components.Both the air flow rate (0.4 vvm) and stirrer speed (300rpm) were set low in order to reach oxygen-limitedconditions (dissolved oxygen below 2% of air saturation)before IPTG induction. VHb-expressing cells reached onthe average 30% higher final culture densities relativeto control cells during the 30-h fed-batch cultivations; twotypical growth curves are shown in Figure 2. Samplesfor both ribosomal FFF measurements and â-lactamaseassays were withdrawn from cultures in the logarithmicgrowth phase (after 9 h) and during the late fed-batchphase (after 24, 27, and 30 h) when the VHb-expressingE. coli cells grew better than the unmodified VHb-negative control cells under hypoxic conditions (Figure2).

Fractograms of the Ribosome Preparations. Theribosomes from each sample were prepared and thenanalyzed with asymmetrical flow FFF. The fractograms

Figure 2. Growth trajectories of VHb-expressing E. coli(MG1655::pKTV1, closed symbols) and VHb-negative controlcells (MG1655::pKT1, open circles) grown under hypoxic condi-tions in a Sixfors bioreactor. The fed-batch cultivations werecarried out as described in the Materials and Methods. TheIPTG induction of VHb expression was initiated 5 h afterinoculation of a bioreactor. The samples (1-4) for FFF measure-ments (indicated by arrows) were withdrawn 9, 24, 27, and 30h after inoculation.

160 Biotechnol. Prog., 1999, Vol. 15, No. 2

obtained (Figure 3) all displayed the three peaks char-acteristic of small and large subunit and the completeribosome (at elution times 4.5, 5.5, and 6.5 min). Fur-thermore there was one peak representing tRNA atelution time 1.3 min and broad protein peaks were foundat elution times between 3 and 4 min. These retentiontimes have been proven (13) to correctly identify thecorresponding cellular components.

Number of Ribosomes per A600 Unit. During thefirst 11 h of cultivation the VHb-expressing and the VHb-negative control strain were growing at the same rate(see Figure 2). Thus, it was not surprising that thenumber of active 70S ribosomes per A600 unit was similarwhen this parameter was analyzed for the first time pointafter 9 h of growth (see Table 1 for details).

VHb-expressing cells were growing better relative tocontrol cells during the late fed-batch phase. The FFFresults show that the number of active 70S ribosomesafter 24 and 27 h were approximately 42% and 46%,respectively, higher in VHb-producing cells relative tocontrol cells. The amount of active 70S ribosomes per A600unit was 2.2-fold higher in the VHb-expressing strainrelative to control strain at the end of the 30-h cultivation(Table 1). These results clearly show that the expressionof VHb enables E. coli cells to maintain their 70Sribosomes at a higher level in an oxygen-limited bio-reactor.

There was no difference between the mass fractions ofribosomes for the VHb-expressing cells and the controlcells for any of the time points (data not shown). Thisindicates that the expression of VHb had no effect on thedistribution of ribosomal subunits.

Total Amount of tRNA. Increased translationalcapacity was also reflected in our measurements of tRNAper optical density at 600 nm as displayed in Table 2.For the first time point after 9 h, no difference betweencontrol strain and VHb-expressing strain was observed,but after 24-27 h of growth, the amount of tRNA per

A600 unit was approximately 20-30% higher for the VHb-producing cells relative to the control cells. Furthermore,VHb-expressing cells had over 100% more tRNA per A600unit than the control cells at the end of the 30-h fed-batchcultivation.

â-Lactamase Activities. An immediate objectivebased on our FFF measurements showing that VHb-expressing cells contained more active ribosomes relativeto control cells was to study the formation of a clonedmarker enzyme, â-lactamase. E. coli MG1655::pKT1 andMG1655::pKTV1 were grown as described in the Materi-als and Methods, and samples for â-lactamase assayswere withdrawn at 9, 24, 27, and 30 h after inoculationfrom a bioreactor supplemented with an equal amountof ampicillin (100 µg/mL). The results in Table 3 clearlyindicate that VHb-expressing cells showed higher â-lac-tamase activities relative to VHb-negative control cul-tures by 33-61%. The increase at the end of the 30-hfed-batch cultivation is, however, lower than the corre-sponding increase of the number of active 70S ribosomesin VHb-expressing strain (approximately 2-fold) relativeto the control (Table 1). Cultures of VHb-expressing cellswere also able to reach higher optical densities relativeto control cells during the microaerobic cultivations

Figure 3. Representative fractograms of VHb-expressing E.coli MG1655::pKTV1 (+VHb) and VHb-negative control cellsMG1655::pKT1 (Control). Samples were withdrawn at the fourtime points, 9, 24, 27, and 30 h (designated as 9 h, 24 h, 27 h,and 30 h in the left margin), after inoculation. Fractionationwas performed as described in the Material and Methods witha cross-flow rate of 6.3 mL/min and an outlet flow rate of 1.2mL/min. Separation profiles are based on the absorbance at 260nm. The positions of peaks corresponding to small subunit (30S),large subunit (50S), and ribosome (70S) are indicated by arrows.The transfer RNA peak is also indicated (tRNA).

Table 1. Number of Active 70S Ribosomes per OpticalDensity at 600 nm in E. coli MG1655::pKTV1(VHb-Expressing) and MG1655::pKT1 (Control)a

strain

sampling timeafter inoculation

(h)

MG1655::pKTV1(ribosomes per

A600/10-11)

MG1655::pKT1(ribosomes per

A600/10-11)

relativeincrease

(%)

9 2.8 (0.6) 3.0 (0.2) -724 1.7 (0.02) 1.2 (0.1) 4227 5.4 (0.6) 3.7 (0.3) 4630 1.3 (0.08) 0.4 (0.08) 225

a Samples were withdrawn at four time points during threeseparate fed-batch bioreactor cultivations. The standard deviations(n ) 9) are shown within parentheses. The relative increase wascalculated as [(pKTV1-pKT1)/pKT1] × 100%.

Table 2. tRNA Levels per Optical Density at 600 nm in E.coli MG1655::pKTV1 (VHb-Expressing) andMG1655::pKT1 (Control)a

strainsampling timeafter inoculation

(h)MG1655::pKTV1[(mAU min)/A600]

MG1655::pKT1[(mAU min)/A600]

relativeincrease

(%)

9 0.64 (0.07) 0.69 (0.02) -724 0.17 (0.003) 0.14 (0.006) 2127 0.47 (0.02) 0.37 (0.02) 2730 0.13 (0.005) 0.06 (0.003) 117

a Samples were withdrawn at four time points during threeseparate fed-batch bioreactor cultivations. The standard deviations(n ) 9) are shown within parentheses. The relative increase wascalculated as [(pKTV1-pKT1)/pKT1] × 100%.

Table 3. â-Lactamase Activities Related to Total SolubleProtein in E. coli MG1655::pKTV1 (VHb-Expressing) andMG1655::pKT1 (Control)a

strainsampling timeafter inoculation

(h)MG1655::pKTV1(U/mg of protein)

MG1655::pKT1(U/mg of protein)

relativeincrease

(%)

9 8.9 (1.2) 6.0 (1.4) 4824 3.1 (0.3) 2.0 (0.1) 5527 2.8 (0.2) 2.1 (0.1) 3330 2.9 (0.9) 1.8 (0.1) 61

a Samples were withdrawn at four time points during twoseparate fed-batch bioreactor cultivations. The standard deviations(n ) 2) are shown within parentheses. The relative increase wascalculated as [(pKTV1-pKT1)/pKT1] × 100%.

Biotechnol. Prog., 1999, Vol. 15, No. 2 161

(Figure 2). Part of the translational activity of the larger70S ribosome population is apparently also utilized forenhanced production of cellular components in VHb-expressing strain relative to the VHb-negative controlstrain.

Discussion

The results from the measurements of the number ofactive 70S ribosomes, mass fractions of ribosome, andtRNA levels clearly display the ability of the asym-metrical flow FFF method to monitor the levels ofribosome and tRNA, and thereby the translational capac-ity, during the course of bioreactor cultivations. The shortanalysis time can give more relevant information thanthe alternative method, analytical ultracentrifugation.Sample preparation times are, however, longer than theanalysis time. Thus, our future work will strive towardshortening the sample preparation time in order to beable to evaluate the translational capacity nearly at theon-line level from the bioreactor samples.

In this study we have demonstrated that VHb-express-ing cells contain more tRNA relative to VHb-negativecontrols. ATP is needed for efficient tRNA synthesis inE. coli. It has been shown that VHb-expressing E. colistrains have a larger ATP pool than the VHb-negativecontrol (4, 5). These observations may explain the in-creased amounts of tRNA per VHb-expressing cell rela-tive to the control E. coli cell.

Furthermore, higher ATP levels should also lead toimproved production of GMP and ultimately to elevatedGTP levels in VHb-expressing strains relative to controls,since high ATP concentration favors GMP synthesis andinhibits the synthesis of IMP from GMP in E. coli (16).In addition, GTP enhances the synthesis of AMP fromIMP and GMP (16). The observation that VHb-expressingstrains have higher ATP levels is based on one-dimen-sional 31P nuclear magnetic resonance measurementswhich determines total NTP level. A generalized as-sumption has been made that the increased NTP peakof VHb-expressing cells is due to the enhanced productionof ATP (4, 5). However, it cannot be ruled out that theproduction of either CTP, TTP, or GTP has also increasedin VHb-expressing cells. Thus, VHb-expressing cells mayalso have higher GTP levels. This could explain ourobservation that the number of active 70S ribosomes percell increased for the VHb-expressing strain relative tocontrol because GTP is required for the binding of tRNAto the small ribosomal subunit during the first step oftranslation initiation.

Increased amounts of tRNA per cell may explain theobserved improvement of translational activity in VHb-producing E. coli cells. Measurements of â-lactamaseactivities also support the proposed hypothesis. In addi-tion, both â-lactamase assays and improved cell growthsuggest that one part of the increased translationalactivity is directed toward cellular metabolites (cellgrowth) and another part toward recombinant proteinproduction. However, the translational capacity, as mea-sured by the ribosome and tRNA levels, increases muchmore in the VHb-expressing bacteria than the â-lacta-mase levels. In our future studies on optimizing recom-binant protein production, we therefore intend to identifywhat other factor(s) can limit intracellular translationalefficiency.

Conclusion

This study clearly demonstrates the potential of asym-metrical flow FFF to monitor the level of translational

components during the production of a heterologousprotein in a bioreactor. The novel information gained byfollowing the ribosome pattern should hence be mostvaluable for increasing the yield in various biotechno-logical protein production systems.

AcknowledgmentThis research was supported by the ETH, Lund Insti-

tute of Technology (LTH), Swedish Natural ScienceResearch Council, Astra Hassle AB, and Carl TryggerFoundation.

References and Notes

(1) Khosla, C.; Bailey, J. E. Heterologous expression of abacterial haemoglobin improves the growth properties ofrecombinant Escherichia coli. Nature 1988, 331, 633-635.

(2) Khosla, C.; Curtis, J. E.; DeModena, J.; Rinas, U.; Bailey,J. E. Expression of intracellular hemoglobin improves proteinsynthesis in oxygen-limited Escherichia coli. Bio/Technology1990, 8, 849-853.

(3) Khosravi, M.; Webster, D. A.; Stark, B. C. Presence of thebacterial hemoglobin gene improves R-amylase production ofa recombinant Escherichia coli strain. Plasmid 1990, 24,190-194.

(4) Chen, R.; Bailey, J. E. Energetic effect of Vitreoscillahemoglobin expression in Escherichia coli: an on-line 31PNMR and saturation transfer study. Biotechnol. Prog. 1994,10, 360-364.

(5) Kallio, P. T.; Kim, D. J.; Tsai, P. S.; Bailey, J. E. Intracellularexpression of Vitreoscilla hemoglobin alters Escherichia colienergy metabolism under oxygen-limited conditions. Eur. J.Biochem. 1994, 219, 201-208.

(6) Tsai, P. S.; Rao, G.; Bailey, J. E. Improvement of Escherichiacoli microaerobic oxygen metabolism by Vitreoscilla hemo-globin: new insights from NAD(P)H fluorescence and cultureredox potential. Biotechnol. Bioeng. 1995, 47, 347-354.

(7) Tsai, P. S.; Hatzimanikatis, V.; Bailey, J. E. Effect ofVitreoscilla hemoglobin dosage on microaerobic Escherichiacoli carbon and energy metabolism. Biotechnol. Bioeng. 1996,49, 139-150.

(8) Tsai, P. S.; Nageli, M.; Bailey, J. E. Intracellular expressionof Vitreoscilla hemoglobin modifies microaerobic Escherichiacoli metabolism through elevated concentration and specificactivity of cytochrome o. Biotechnol. Bioeng. 1996, 49, 151-160.

(9) Peretti, S. W.; Bailey J. E. Mechanistically detailed modelof cellular metabolism for glucose-limited growth of Escheri-chia coli B/r-A. Biotechnol Bioeng. 1986, 28, 1672-1689.

(10) Wahlund, K.-G.; Litzen, A. Application of an asymmetricalflow field-flow fractionation channel to the separation andcharacterization of proteins, plasmids, plasmid fragments,polysaccharides and unicellular algae. J. Chromatogr. 1989,461, 73-87.

(11) Litzen, A.; Wahlund, K.-G. Improved separation speed andefficiency for proteins, nucleic acids and viruses in asym-metrical flow field flow fractionation. J. Chromatogr. 1989,476, 413-421.

(12) Litzen, A.; Walter, J. K.; Krischollek, H.; Wahlund, K.-G.Separation and quantitation of monoclonal antibody ag-gregates by asymmetrical flow field-flow fractionation andcomparison to gel permeation chromatography. Anal. Bio-chem. 1993, 212, 469-480.

(13) Nilsson, M.; Bulow, L.; Wahlund, K.-G. Use of flow field-flow fractionation for the rapid quantitation of ribosome andribosomal subunits in Escherichia coli at different proteinproduction conditions. Biotechnol. Bioeng. 1997, 54, 461-467.

(14) Hart, R. A.; Bailey, J. E. Purification and aqueous two-phase partitioning properties of recombinant Vitreoscillahemoglobin. Enzyme Microb. Technol. 1991, 13, 788-795.

(15) O’Callaghan, C. H.; Morris, A.; Kirby, S. M.; Shingler, A.H. Novel method for detection of â-lactamases by using a

162 Biotechnol. Prog., 1999, Vol. 15, No. 2

chromogenic cephalosporin substrate. Antimicrob. AgentsChemother. 1972, 1, 283-288.

(16) Zalkin, H.; Nygaard, P. Biosynthesis of purine nucleotides.In Escherichia coli and Salmonella. Cellular and molecularbiology; Neidhardt, F. C., Curtiss R., III, Ingraham, J. L., Lin,E. C. C., Low, K. B., Magasanik, B., Reznikoff, W. S., Riley,

M., Schaechter, M., Umbarger, H. E., Eds.; ASM Press:Washington, DC, 1996; Vol. 1, pp 561-579.

Accepted February 2, 1999.

BP9900155

Biotechnol. Prog., 1999, Vol. 15, No. 2 163