Analytica Chimica Acta 460 (2002) 257–270
Tuning the parameters for fast respirometry
A. Tzorisa, D. Caneb, P. Maynardc, E.A.H. Halla,∗a Institute of Biotechnology, University of Cambridge, Tennis Court Road, Cambridge CB2 1QT, UK
b Omnifit Ltd., 2 College Park, Coldhams Lane, Cambridge CB1 3HD, UKc Maynard Projects, 72 The Lane, Hauxton, Cambridge CB2 5HP, UK
Received 26 September 2001; received in revised form 6 March 2002; accepted 7 March 2002
Abstract
The aerobic bacterial respiration rate is an indicator of microbial growth and metabolism, essential for monitoring theoxidation process and organic load content of samples in a diverse field of application from influent streams in wastewatertreatment facilities to industrial fermentations. This paper looks at the influence of parameters, such as culture concentrationand volume, sample surface area/volume ratio and headspace volume to achieve optimisation of respirometry measurementand thus design a bench-top respirometric device, based on the monitoring of the pressure changes in a closed chamber wherea bacterial culture is allowed to respire in contact with a sample. Contrary to traditional respirometry, the goal is detection ofbacterial respiration within 5 min in a minimal sample volume. Both qualitative and quantitative data could be derived using asimple equation and fine-tuning of the micro-manometric parameters of the device, with a most important finding being thatminimal headspace volume in combination with elevated bacterial populations maximised absolute pressure change responseand favoured high sensitivity at short response time, even though the conditions indicated oxygen-limitation. Furthermore,in comparison with a commercially available respirometer the typical respiration rate of stationary phaseP. putida M10 gaveoxygen uptake rate (OUR) and specific oxygen uptake rate (SOUR) of 38�mol l−1 min−1 and 5�mol g−1 min−1, respectivelywith the ‘classical’ system, while the�-Warburg device designed here showed a typical response, for the culture with the samedry cell concentration, of 66�mol l−1 min−1 for the OUR and 9�mol g−1 min−1 for the SOUR. The remarkable outcomefrom this data, therefore, is that it appears that the high surface area/volume geometry of the�-Warburg device design hasachieved less respiration limitation, even though the sample is unstirred. This presents important insight regarding futurerespirometer design. © 2002 Elsevier Science B.V. All rights reserved.
Keywords: Respirometry; Oxygen; Manometry; Biosensor; Bacteria; Wastewater
1. Introduction
Microbial respiration is most often associated withenvironmental soil and water research and also withfood science and preservation, insect respiration, tis-sue and skin respiration, plant research and a widerange of other applications.
∗ Corresponding author. Tel.:+44-1223-334160;fax: +44-1223-334162.E-mail address: [emailprotected] (E.A.H. Hall).
Various methods to determine the respiration rate ofbacteria (oxygen uptake rate, OUR) have been used toassess the ability of a bacterial population to removesubstances from wastewater, to determine the effect ofthe substances on the bacteria and also to quantify mi-crobial growth and associate substrate depletion andproduct formation in industrial fermentations. Thesemethods range from the O2 balance method by mon-itoring the inlet and exhaust gases in a bioreactor [1]to the O2 electrode to Warburg manometric methods[2].
0003-2670/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved.PII: S0003-2670(02)00190-3
258 A. Tzoris et al. / Analytica Chimica Acta 460 (2002) 257–270
It has been seen that respirometric methods that relyon the monitoring of the oxygen uptake of a biologicalsystem (micro-organism, plant or a higher organism,such as eukariotic organism, fish, etc.) by observingthe change in volume (constant pressure respirometry)or the change in pressure (constant volume respirom-etry) of the gas phase in contact with the respiringliquid can be used in the place of the dilution bio-logical oxygen demand (BOD) test [3]. The simplic-ity of the method with respect to its operation andinstrumentation also suggests that it might have anadvantage over the current BOD methodology if itcould be developed for faster analysis and on smallersamples.
Respirometric methods have their first report in thelate 19th century, when Adney developed a constantpressure manometric apparatus to observe the rate ofoxygen absorption by polluted water, but later con-cluded that despite the good accuracy of the method,it was not suitable for routine work. Sierp [4] revivedand modified Adney’s direct aeration apparatus toreduce the cumbersomeness of the dilution methodand afford quick, direct and frequent readings. Fromthen onwards, there are many reports on respirom-etry in the literature (for a detailed description see[3,5]) and today there are several such commerciallyavailable instruments, the so-called respirometers.These range from very simple, manually operatedBOD-bottles to fully self-operating instruments thatautomatically perform sampling, calibration and cal-culation of respiration rate. Respirometers are widelyused in wastewater treatment plants to evaluate theorganic load, as part of wastewater treatability, andalso the potential toxicity of the incoming streams[6–8].
All respirometers are based on some technique formeasuring the respiration rate, or OUR, i.e. the rate atwhich biomass takes up dissolved oxygen (DO) fromthe liquid. This can be done directly by measuring DO(chemically and almost uniquely electrochemically) orindirectly by measuring gaseous oxygen (manomet-ric, volumetric and paramagnetic methods). From thework of Spanjers et al. [9], addressing the significanceof respirometry measurements, DO concentration andgaseous oxygen concentration balances are consideredfor a system consisting of a liquid phase, containingbiomass and a gas phase. From these considerations,the DO mass balance over the liquid phase has been
given as follows:
d(VLCL)
dt= FlowinCL, in − FlowoutCL
+ VLKLa(C∗L − CL) − VLr (1)
where CL is the concentration of DO in the liquidphase, with subscript “in” indicating the concentrationin the entering stream and superscript “∗” indicat-ing the saturating concentration.VL is the volume ofthe liquid phase andr is the respiration rate.KLa isthe oxygen mass transfer coefficient. This will be afunction of flow rate or agitation for a given system.As pointed out by Spanjers et al. [9], it can readilybe seen that in quiescent solution the first two termsbecome zero. The third term describes the mass trans-fer of oxygen from the gas phase to liquid phase.Meanwhile, the situation in the gas phase is presentedas follows:
d(VGCG)
dt= FlowinCG, in − FlowoutCG
−VL
VGKLa(C∗
L − CL) (2)
Spanjers et al. have also summarised the measuringprinciples for different flow and quiescent regimes inthe liquid and gas phases. Measurement of decreasein DO in the liquid phase under ‘no-flow’ conditionsgives a measure of respiration, or in the gas phaseCG can be measured viaVG. Traditional designsof respirometers demand a considerable period oftime in order to give an indication of the respira-tory activity of the system under investigation. In thecase of BOD measurement, this can be as long as5 days. Although this is to some extent related tothe specifications of the method followed (for exam-ple, BOD5), the reasons that a faster more versatilemeasurement is not achieved are mainly the slowresponse of the instrument deriving from the low ac-tivity of the micro-organism and also the design ofthe respirometry chamber and its incorporation intothe measurement system.
In this paper, we revisit the principles of respirom-etry with the goal of investigating its potential toachieve a fast method for monitoring the uptake ofoxygen by bacteria using a bench-top simple respiro-metric device. The work presented herein investigatesthe experimental and design parameters, which may
A. Tzoris et al. / Analytica Chimica Acta 460 (2002) 257–270 259
optimise the measurement. Thus, such parametersthat include selection of a model organism, culturevolume, concentration and growth phase and devicegeometry were addressed in this work.
2. Materials and methods
The micro-organisms used wereE. coli JM109(NCIMB 12711),P. putida M10 kindly provided byDr. N. Bruce andP. putida KT2440 (ATCC 47054)kindly provided by Dr. A. Tunnacliffe, both fromthe Institute of Biotechnology, University of Cam-bridge, UK. Dry Baker’s yeast was obtained froma local supermarket. Sodium chloride, glycerol and3,5-dichlorophenol 97% (DCP) were purchased fromAldrich, Poole, Dorset, UK. Bacto agar >99%, Bactoyeast extract and Bacto tryptone were from DifcoLaboratories, Detroit, USA. Soda lime self-indicatinggranules (“carbosorb” 6–16 mesh) was from BDH,Merck Ltd., Poole, Dorset, UK.
Baker’s yeast was suspended in deionised waterprior to use and activated by stirring for 15 min.E. coli andP. putida were both grown in LB medium(Bacto tryptone, Bacto yeast extract, and NaCl; 1,0.5 and 0.5% w/v, respectively in deionised water)in Erlenmeyer flasks at 37 and 30◦C, respectively,under shaking (200 rpm). Bacterial growth was es-timated by turbidity measurements as expressed inunits of absorbance at 600 nm (OD600, 1 ml, 1 cmlight path) after 10-fold dilution of the samples withLB using deionised water as the reference. Cell num-bers (CFUs) were determined by viable counts onLB-agar (LB medium, 1.5% w/v in Bacto agar) platesin triplicates, after serial dilution of the liquid culturewith LB, by plating out 10�l of the 1:107 and 1:108
dilutions. For short-term storage, cultures were grownovernight (17–18 h) at the specified temperature onLB-agar in plastic Petri dishes (9 mm diameter), byspreading 10�l from a diluted (1:105 with LB) freshculture evenly on the agar surface, and then stored at4◦C. For long-term storage, frozen cultures were pre-pared by centrifuging (6000 rpm, 10 m) 1 ml aliquotsof a fresh liquid culture (grown as above until anOD600 value of 4–5), resuspending the cell paste in1 ml LB (20% v/v in sterile glycerol) and storingthe inoculum in 1.5 ml microtubes at−80◦C untilused. For inoculation, the frozen cultures were slowly
thawed (30◦C), the supernatant was discarded aftercentrifugation (6000 rpm, 10 min) and the cell pastewas resuspended in 1 ml LB. For stationary and log-arithmic growth phase the bacteria were grown toOD600 values of 7–8 and 2–3, respectively.
Concentrated bacterial cultures were prepared bycentrifuging freshly grown liquid cultures at 6000 rpmfor 10 min, discarding the supernatant and resuspend-ing the cell paste in a specified volume of LB in orderto reach the desired concentration (mg ml−1) of bacte-ria. The concentrated cultures used in this work were∼71 g wet cell weight l−1, unless otherwise specified,which was equivalent to∼(2–5) × 1010 CFU ml−1
as estimated by viable counts. Dry cell weight wasdetermined after drying the wet cell paste at 60◦Cin pre-weighed micro tubes (1.5 ml) until constantweight.
An electrolytic respirometer, the Merit 20 (Adding-ton Instruments) able to accommodate up to 20samples simultaneously in a controlled-temperatureenvironment was kindly provided by Yorkshire Waterand used according to the following protocol, a vol-ume of sample was placed in the 20 ml “respiration”chamber, which was connected to a second identicalbottle, the “compensation” chamber, which containedan identical volume of deionised water. For this work,the sample was a liquid bacterial culture. The dropin the headspace pressure of the bottle due to oxy-gen consumption was sensed by a flexible membraneseparating the two bottles, which in turn triggered there-generation of oxygen by an electrolytic cell in or-der to equilibrate the pressure drop in the respirationchamber. After pressure re-equilibration, the amountof oxygen produced by the electrolytic cell (whichcorrelates with the amount of oxygen consumed bythe bacteria), in microliters of O2 was the resultingparameter. Test results were downloaded using theAddington Instruments software operating throughMicrosoft Windows. In this work, the Merit was oper-ated at 20◦C. Soda lime, placed in a “tray” suspendedin the headspace of the respiration bottle, was used tocapture the CO2 produced by the metabolic activityof the bacteria, that would otherwise interfere withthe pressure changes in the headspace.
Warburg respirometry was performed using a�-Warburg “open-end” device, constructed “in-house”(Fig. 1a). The respiration chamber consisted of aglass vessel (3.5 cm long, 2.2 cm internal diameter
260 A. Tzoris et al. / Analytica Chimica Acta 460 (2002) 257–270
Fig. 1. The�-Warburg device. (a) Glass capillary/open-end device, (b) device with pressure transducer for direct monitoring of pressurechanges.
(i.d.)) that accommodated the culture/sample and waspush-fit in the body of a Teflon block in such a waythat a small volume of air was allowed above thesample. The volume of this headspace could be reg-ulated by adjusting the position of the glass vessel upor down the Teflon block. A T-shaped narrow portmachined through the body of the Teflon block inter-connected the headspace with a capillary glass tube(0.4 mm i.d.), attached on one side of the Teflon blockand with the atmosphere on the other side of the block.
The respiration chamber could be “closed” or “open”to atmospheric (or external) pressure with the use of atap situated on the top of the Teflon block. The sample(1 ml unless otherwise stated) was introduced into therespiration chamber with a 1 ml syringe via a sampleport also machined through the Teflon block and re-mained in place during the measurement to ensure airtightness. Soda lime for CO2 absorption was placed ina small recess machined on the block above the sam-ple and separated from the sample by a nylon mesh,
A. Tzoris et al. / Analytica Chimica Acta 460 (2002) 257–270 261
held in place by a rubber O-ring. Oxygen depletionin the respiration chamber, due to (aerobic) bacterialrespiration, would cause a local reduction in volume.This would force a liquid bolus situated in the capil-lary glass tube to move inwards (towards the culture)to compensate for the decrease in volume in such an“open-ended” system. Glycerol was used as the liquidin the capillary. Hence the movement of the glycerolin the capillary, captured by a CCD camera could beused as an indication of bacterial respiratory activity.
In another version of the�-Warburg, a SensorTechnics precision differential pressure transducerwith a range of±100 Pa was connected to one sideof the pressure port, replacing the glass capillary(Fig. 1b). Thus, the pressure changes in the headspaceof the respiration chamber due to bacterial activity ofthe sample could be monitored directly. The defaulttotal headspace volume, including the volume of thetube connections, was 2.55 cm3 and the volume ofthe sample used was 1 ml unless otherwise stated. Asmentioned earlier the headspace volume could be ad-justed by altering the positioning of the glass vesselin the Teflon block of the device. The response of thepressure transducer was in Volts with 1 V changecorresponding to pressure of 100 Pa.
3. Results and discussion
3.1. Respirometer design
The goal of this work was to achieve a fast respi-rometry system. Intuitively, higher cell concentrationsshould give higher total respiration rates, but the rateper cell may become limited due to local or bulk oxy-gen depletion, so for a given cell mass, O2 supply toan individual cell may be a function of the geometry(e.g. surface area/depth of the culture medium) as wellas its bacterial cell count. Clearly, therefore, a rapidassay for measurement of bacterial respiration is likelyto require optimal cell design and rapid transductionof the changes in the gas concentration involved insuch activity.
The classical 5 h and >5 h respirometric experimentusing respirometers, such as the Merit often employa culture preparation, e.g. mainlyP. putida absorbedon wood chips. In this instance the cell volume is20 and 10 ml sample is used. Following the standard
Addington Instruments protocol using 10% of woodchips (w/v in deionised water), a typical O2 consump-tion of 1.5�l min−1 (data not shown) is indicatedfor a sample showing neither inhibition nor enhance-ment of activity. At a micro rather than macro level,it is difficult to extract useful quantitative data fromthis, since the wood chip preparation is not hom*oge-neous or otherwise characterised in terms of its totalmicro-organism profile.
A preparation ofE. coli JM109 (∼2 × 108 CFUml−1) showed an O2 consumption of∼5.6�l min−1
during the first hour of the experiment (Fig. 2). In theabsence of CO2 absorber, the apparent respiration ratewas reduced to 4.2�l min−1, indicating a total volumechange due to the respiration ofE. coli during the first5 min of the experiment of 28 and 21�l, respectivelyin the presence and absence of soda lime. This sug-gests an increase in CO2 equivalent to 0.3�mol CO2per 1.2�mol O2 consumed under these conditions.
In contrast, Baker’s yeast (10 ml, 8% w/v culture)showed an O2 consumption of∼12�l min−1 withinthe first hour (Fig. 2). However, assuming that thishas accommodated all of the high CO2 evolution inyeast, it indicated a less active oxygen consumer.From Fig. 2, it can also be seen that at the high res-piration rates, the observed rate becomes independentof culture concentration (compare 4 and 8% values).With an instrument resolution of∼25�l O2, it isapparent that short time respirometry is at the limit ofcapability with these concentrations and conditions sothat clearly some changes are required if short timerespirometry measurements are to be achieved.
As mentioned earlier, intuitively using a high cellcount for the assay might elevate the O2 consump-tion and thus improve the resolution at short times.In Table 1 and Fig. 3, the Merit responses forE.coli JM109 andP. putida M10 are given as a func-tion of the culture population and growth phase forshorter respiration times. A common feature of bothE. coli and P. putida was that the growth phase didnot appear to significantly affect the respiration rateof the micro-organism. For relatively dilute cultures(e.g. stationary phase cultures used without additionalprocessing, after growth to OD600 = 7–8; ∼108 to109 CFU ml−1), the oxygen demand estimated withthe Merit was similar for bothE. coli JM109 andP.putida M10. More concentratedE. coli JM109 cul-tures had no effect on the response, but in contrast
262 A. Tzoris et al. / Analytica Chimica Acta 460 (2002) 257–270
Fig. 2. Oxygen demand for 8% (�) and 4% (�) (w/v) Baker’s yeast aqueous culture, and stationary phase (∼2× 108 CFU ml−1) E. coliJM109 in LB medium with (�) and without (�) soda lime for CO2 absorption, as determined with the Merit 20 respirometer.
Table 1Respiration rates (�l O2 min−1) for stationary cultures ofE. coli JM109 andP. putida M10 as estimated with the Merit 20 respirometer
Dry cellconcentration(g l−1)
Wet cellconcentration(g l−1)
Cell count(CFU ml−1)
E. coli JM109 P. putida M10
Stationary phase Logarithmic phase Stationary phase Logarithmic phase
0.4 3 1 (±0.5) × 109 7.3 ± 0.4 8.9± 1.7 7.6± 0.5 7.6± 0.97.4 ± 0.1 43± 1 1.5 (±0.5) × 1010 6.6 ± 0.9 8.2± 1.5 9.1± 1.3 9.3± 1.2
Fig. 3. Short-term respirometry: typical oxygen demand of stationary phaseP. putida M10: (�) normal (∼108 to 109 CFU ml−1) and (�)concentrated (∼1010 CFU ml−1) culture.
A. Tzoris et al. / Analytica Chimica Acta 460 (2002) 257–270 263
more concentratedP. putida M10 cultures (∼1010
CFU ml−1) showed a 20–22% increase in the res-piratory activity recorded with the Merit (∼10�lO2 min−1; Merit resolution approximately 25�l O2).However, in general, a proportional increase in therespiration activity with concentration was not in-dicated, since further concentration of the bacterialculture did not produce equivalent enhancement. Thismay imply that oxygen demand estimated with theMerit becomes limited by the instrumental resolution,the cell design or a feature of the culture respirationat high concentrations.
This is not altogether encouraging, but the abilityto design highly resolved respiration measurementswithin shorter time periods will be determined by anunderstanding of all the factors that limit it, not justconcentration. In addition to the normal metabolicactivity of a particular organism, these may be con-cerned with gas exchange rates and local oxygenconcentration gradients in the vicinity of the organismso that cell design at a given culture concentrationcan be of importance. Khlebnikov et al. describedsome of the necessary assumptions to be able todescribe the system parameters for respirometry andidentified the need for a zero-order reaction with re-spect to oxygen [10]. In practical terms, design mustalso be a function of desired sample volume or otherlimitations, which might be imposed as a result of aparticular respirometric measurement. Here, we haveinvestigated the idea of deviating from the classicalMerit style design and using a small sample volume(0.5–1.0 ml) and large surface area (3.8 cm2), (surfacearea/volume ratio 7.6:3.8), thus creating the greatestopportunity for fast gas exchange, without employingstirring or shaking. This retains the simple Warburgmanometry principle with the area/volume featuresachieved in the basic design illustrated in Fig. 1.
3.2. Transduction method
The �-Warburg device in Fig. 1a uses a hori-zontal open-ended capillary manometer containing∼5–6 mm of gauge liquid. The position of the liq-uid was monitored optically and is plotted in Fig. 4aand b according to pixel number which correspondsto position (1 mm= 6 pixels). Fig. 4a and b com-pare measurements obtained forE. coli at differentconcentrations. It can be seen that for a culture of
Fig. 4. Respiration monitoring of stationary phaseE. coli JM109cultures with the�-Warburg device. (a) Concentrated culture((2–5) × 1010 CFU ml−1), (�) 0 min; (�) 2 min; (�) 3 min; (�)4 min; (�) 5 min. (b) Normal culture ((2–5) × 108 CFU ml−1),(�) 0 min; (�) 5 min; (�) 10 min; (�) 12 min. The distance cov-ered by the liquid bolus in the capillary as described in Section 2was captured by a CCD camera and is denoted in the graph by thepixel number (x-axis). Low light intensity reflects the liquid bolus(y-axis). (c) Distance travelled by the liquid bolus in the capillaryof the �-Warburg device for normal (�) ((2–5)× 108 CFU ml−1)and concentrated (�) ((2–5) × 1010 CFU ml−1) E. coli JM109culture.
∼(2–5) × 1010 CFU ml−1 an assessment of respira-tion could be obtained within 5 min. However, for aless concentrated culture (approximately(2–5) × 108
CFU ml−1) a much longer period was required for thedevice to record respiration (Fig. 4c). Similar resultswere obtained forP. putida.
To obtain the transduction sensitivity of the capil-lary manometer design, the volume change (πd2h/4)in O2 can be calculated from the movement (h) of the
264 A. Tzoris et al. / Analytica Chimica Acta 460 (2002) 257–270
liquid bolus along the capillary. Thus the sensitivityof the method could be adjusted byd, the i.d. of thecapillary. In this case, a 0.4 mm i.d. was considered agood compromise between dominating capillarity ef-fects and sensitivity inh. The choice of the gaugefluid is also important, not only to minimise capillar-ity effects but also to ensure a well defined meniscusand poor gas exchange across the liquid bolus. How-ever, as can be seen in Fig. 4, the movement of theliquid bolus was not smooth, but suffered from iner-tia, initially slow and later accelerated movement ofthe liquid bolus contradicted the apparently linear re-lationship between oxygen consumption rate and timesuggested from electrolytic respirometry (Fig. 3). Atthe low respiration rates seen in Fig. 4b, the effect ofinertia is accentuated: in this instance, the liquid bolushardly moved within the first 5 min of the experimentonly to accelerate sharply after∼8 min. Furthermore,interpretation of the results was ambiguous since theend-point of the distance travelled by the bolus dif-fered significantly among similar experiments. In prin-ciple, the solution resides in the choice of gauge fluid,but preferred materials like mercury are not suitablefor this open-ended design whereas water, ethanol andliquid paraffin tried as gauge fluids did not give therequired accuracy.
Fig. 5. Typical respiration profiles ofP. putida M10 (�) and E. coli JM109 (�) obtained by monitoring directly the pressure changes inthe respiration chamber of the�-Warburg device. Stationary phase concentrated cultures were used (wet cell concentration 71± 3 g l−1,1 ml). (�) “Blank” sample, i.e. sample containing nutrient medium only (1 ml), no culture.
Fig. 1b shows the same basic design, but the pres-sure transduction is achieved by an in-line solid-statepressure transducer. In this instance, a measurementof p is obtained at constant volume, under quiescentsolution conditions, thus Eq. (2) from Spanjers et al.can be written as follows:
−dCG
dt= piVL
(pi − p)VGKLa(C∗
L − CL) (3)
In Fig. 5, a typical respiration profile ofE. coli andP. putida M10 is shown in comparison with theresponse of a “blank” sample containing onlynutrient medium (LB). The respiration of bothmicro-organisms could be detected in less than 5 minwith P. putida showing a greater pressure drop (ratesfor P. putida and E. coli were estimated 142 and62 Pa min−1, respectively). Referring to Eqs. (1) and(3) above, we note some difficulty in using theseequations directly to obtain the respiration rate,r.First knowledge ofKLa is required and second anabsolute initial value forpi as well asp.
In the case of this design using a solid-state pres-sure transducer (assuming that the drop in pressurein the respiration chamber is due solely to the uptakeof oxygen by the bacteria and that oxygen and nitro-gen behave almost as ideal gases at the temperature
A. Tzoris et al. / Analytica Chimica Acta 460 (2002) 257–270 265
(22–24◦C) and pressure (1 atm) of the experiment) anequation for the OUR of the micro-organisms underthe conditions of the experiment can be derived asfollows.
According to Henry’s law, the fugacity of a solute indilute solution is proportional to its mole fraction. Forlow pressures (up to 1 atm) and ideal gases fugacitycan be equated with vapour pressure and Henry’s lawcan be written as:x = p/k (k is the proportionalityconstant). Because the mole fraction of O2 dissolvedis small
xO2 = nO2
nO2 + nH2O≈ nO2
nH2O(4)
typically in pure water in an air atmosphere (O2 partialpressure: 160 Torr/21.325 kPa/∼1 × 10−5 mol cm−3),this corresponds to
nO2 ≈ xO2nH2O = p
knH2O = 160
3.3 × 107 × 55.5
= 2.7 × 10−4 mol l−1 (5)
Consumption of oxygen by the micro-organism willcause a flux of oxygen from the gas phase to the liquidphase. The respiration in the liquid phase will causean overall reduction in the total oxygen present, result-ing in a lower concentration in both the gas and liq-uid phases, partitioned according to Henry’s constant.Since the concentration of oxygen in the gas phase isapproximately two orders of magnitude greater thanin the liquid phase, then for similarVG andVL:
total O2 = [O2]G + [O2]L ∼ [O2]G (6)
It can therefore be assumed thatp is approximately(1% error results from this approximation in this sys-tem) a measure of the total oxygen uptake by themicro-organism. If the respiration chamber and thetubing have a combined volumeV then after the ad-dition of the sample of volumeVL, the headspacevolumeVG of the chamber will be (cm3):
VG = V − VL (7)
At the beginning of the measurement (chamber opento atmosphere) the pressure in the chamberp1 willbe equal to the atmospheric pressure (assume 1 atm,101.325 kPa). Ifn1 is the mole fraction of oxygenin air (at sea leveln1 ≈ 0.2) and assuming for airapproximately 24.3 l mol−1 at 22–24◦C, then at the
start of the experiment there will benox(1) moles ofoxygen present in the headspace:
nox(1) = VG
24, 300n1 (8)
After a certain time dt before the “opening” of therespiration chamber to atmosphere again,nox molesof oxygen will have been consumed by the bacteria,leavingnox(2) moles of oxygen in the headspace hencethe new mole fraction of oxygen will ben2 and
nox = nox(1) − nox(2) = (n1 − n2)VG
24, 300(9)
wherenox(2) = VGn2/24, 300. This in turn will resultin a drop in pressure dp measured by the pressuresensor, equal to (Pa):
p = p1 − p2 = n1 − n2
n1p0 (10)
wherep1 is the pressure at the beginning of the exper-iment, equal to atmospheric pressure,p2 the pressurein the headspace after time dt at whichnox(2) moles ofoxygen have remained in the headspace of the closedrespiration chamber andp0 is the partial pressure ofoxygen at the start of the experiment (Pa).
Solving Eq. (9) for (n1 − n2) and substituting inEq. (10):
p = p1 − p2 = 24, 300nox
n1VGp0 (11)
but
p = p1 − p2 = 100(U1 − U2) (12)
whereU1 andU2 are the voltage values recorded bythe pressure sensor at the beginning and at the end ofthe measurement, respectively (1 V= 100 Pa), hencecombining Eqs. (5) and (6) the amount of oxygen con-sumed by the bacteria within time dt can be estimated
nox = 100(U1 − U2)VGn1
24, 300p0(13)
with U1 andU2 in Volts, VG in cm3 andp0 in Pa.This can also be expressed in terms of uptake of
oxygen by the bacteria per litre of sample per minute(OUR, �mol O2 l−1 min−1), since the slope of thedecreasing pressure curve with time can be associ-ated with the decrease in oxygen concentration as in
266 A. Tzoris et al. / Analytica Chimica Acta 460 (2002) 257–270
the case of DO measurements [11,12]. For compari-son with estimates from other methods, the data willbe expressed in this form. It should be rememberedhowever, that the sample size here is=1 cm3 and thatobtaining data directly for 1000 cm3 samples requiresindependent design. For the experimental conditionsemployed in this work:
OUR = 40.6(U1 − U2)VG
VLdt(14)
hence for Fig. 5, OURs forP. putida and E. coli,respectively are 144 and 64�mol l−1 min−1.
3.3. Assessment of maximum resolution
In Table 2, the mean OURs forE. coli JM109 andP. putida M10 are given with respect to the cultureconcentration, volume and growth phase for a cellof constant volume,V. P. putida M10 is again seenas the more ‘active’ species with an apparent OURapproximately 1.5–2 times that ofE. coli. Most impor-tantly from this data, it can be seen that logarithmicphase cultures do not give higher apparent respirationrates than stationary ones (even if this might have beenexpected since the microbial specific growth rate ismaximum at logarithmic growth phase) and increase
Table 2OURs (�mol l−1 min−1) as determined by monitoring directly the pressure changes in the respiration chamber of the�-Warburg device
(a) E. coli JM109Wet cell concentration (g l−1) 43 ± 1 71 ± 3Dry cell concentration (g l−1) 7.4 ± 0.1 14.7± 0.2
Stationary phase1.0 ml 68± 17 (n = 8) 69 ± 9 (n = 6)0.5 ml 166± 40 (n = 4) 153± 10 (n = 3)
Logarithmic phase1.0 ml 59± 9 (n = 5) 54 ± 9 (n = 6)0.5 ml 153± 22 (n = 4) 124± 22 (n = 6)
(b) P. putida M10Wet cell concentration (g l−1) 43 ± 1 71 ± 3Dry cell concentration (g l−1) 7.4 ± 0.1 14.7± 0.2
Stationary phase1.0 ml n.d. 155± 12 (n = 12)0.5 ml n.d. 267± 50 (n = 6)
Logarithmic phase1.0 ml 137± 22 (n = 7) 113± 13 (n = 6)0.5 ml 203± 35 (n = 4) 223± 35 (n = 7)
n.d.: not done.
in concentration (population) of the culture may not in-crease OUR. However, if only cells near the surface ofthe culture suspension are predominantly responsiblefor the OURs obtained here, a change in surface arearather than concentration may better describe the eff-ect of cell count on the respiratory activity of thebacteria. The transfer of oxygen from gas to liquid isoften the limiting step in the aerobic microbial process[13,14]. According to the stationary liquid film the-ory a stationary liquid film develops at the interfacesbetween the gas and the liquid phases and betweenthe liquid phase and the biomass in which a con-centration gradient of dissolved gas exists. It followsthat oxygen transfer from gas through a liquid to thebiomass surface will depend, among other factors, onthe interfacial area and the thickness of the stationaryliquid film but also on the distanceL the oxygen has totravel within the liquid before it reaches the biomasssurface. In this instance, using a non-stirred culture,the volume of the sample used, influences the stag-nant liquid layer between the liquid–biomass and thegas–liquid interfaces. It can be seen here that for thesame cell concentration, reducing the sample volumeincreases the apparent OUR (Table 2). The signifi-cance of this increase is highlighted if samples withidentical cellmass are also compared (e.g. compare
A. Tzoris et al. / Analytica Chimica Acta 460 (2002) 257–270 267
14.7 g l−1 cells in 0.5 ml sample with 7.4 g l−1 cellsin 1 ml sample). From Eq. (5), the response of the�-Warburg device can be expected to depend on theheadspace volumeVG. More specifically, for a givennox, i.e. a constant cell count, the magnitude ofp1−p2increases as−VG decreases. This is contrary to theresults in Table 2, however, that seem to suggest thatlower culture volumes give rise to higher responses.Clearly, two differences in the parameters whichdescribe the culture cell are immediately obvioushere,VG is larger for the smaller sample volume, butcoincident with this the distanceL for O2 travel isshorter forV L = 0.5 ml compared with 1.0 ml.
In order to separate these effects, results were ob-tained at a single sample volume and concentrationbut different headspace volumeVG, revealing that thepressure drop rate is attenuated with headspace vol-ume. However, the trend only complies with Eq. (11)whenVG is large. In Table 3, some typical data showsthat the pressure drop deviates from that predicted byEq. (11) for change inVG and relative to data at thelargestVG, theexpected pressure drop for decreasingVG is higher than the actualp measured. This raisesa paradox, since clearly the ‘sensitivity’ (i.e. absolutepressure change measured) of the device could be en-hanced simply by decreasing the headspace volumeof the respiration chamber. However, if the headspacevolume is not large enough for the DO concentration tobe above a critical threshold, the respiration of the bac-teria may become oxygen limited and thus OUR andspecific oxygen uptake rate (SOUR) estimates may re-duce. From the data in Table 3 and other data, there isan indication from the OUR and SOUR that the largestVG gives the greatest OUR and SOUR, in accordancewith similar reports in the literature [15]. This impliesthat an accurate SOUR requires excessVG, but in spe-cific applications where fast responses would be bene-ficial, for example, if ‘toxicity’ assessment is desired,then smallVG and high pressure change sensitivityshould be selected. From Table 3(a) and (b), highercell densities also seem to favour pressure change sen-sitivity and hence response time.
In Fig. 6, the effect of cell concentration on the re-sponse of the device is shown. By increasing the cellconcentration the pressure drop rate in the respirationchamber increases thus allowing for further tuning ofthe sensitivity of the�-Warburg. This is also reflectedin the OUR (Table 3). The fact that the increase in
Fig. 6. Effect of cell concentration on the response (pressure droprate) of the�-Warburg device. Culture: stationary phaseP. putidaKT2440, 1 ml. Headspace volume: 2.55 cm3.
OUR with concentration is not linear and also thatthe SOUR actually decreases with increasing cellconcentration (compare data in Table 3(a–c)) is alsoconsistent with the idea that the system is operatingunder limiting conditions. This can be related to theratio between liquid phase surface area and number ofcells. The effective surface area of a stagnant constantvolume culture decreases with increasing cell con-centration. Therefore, oxygen is available to a lesserproportion of cells. This is reflected in the SOUR cal-culation that assumes a uniform oxygen availabilityand unlimited activity for all cells.
3.4. Comparative performance
Comparing the results obtained from the Merit andthe�-Warburg revealed that the two instruments gaveconsistent qualitative information, but on differenttime-scales and with some inconsistency in the quan-titative information. In view of significant differencesin the geometry of the two systems and in light of thedata obtained here for the�-Warburg and the impor-tance of diffusion pathways and oxygen-limitation, thequantitative differences cannot be seen as exceptional,but it does raise some question about the comparisonof data even between the currently available ‘large’
268A
.T
zoriset
al./Analytica
Chim
icaA
cta460
(2002)257–270
A. Tzoris et al. / Analytica Chimica Acta 460 (2002) 257–270 269
respirometers on the market. Both instruments usedin this study indicated thatP. putida M10 is more‘active’ than E. coli JM109. With the Merit the typ-ical respiration rate of stationary phaseP. putidaM10 (∼1.5× 1010 CFU ml−1, 10 ml culture, Table 1)at 20◦C was estimated at∼9.1�l O2 min−1. Thisvalue is equivalent to∼0.38�mol O2 min−1, hencethe OUR and SOUR, for a dry cell concentration of7.4 g l−1, can be calculated as 38�mol l−1 min−1 and5�mol g−1 min−1, respectively. With the�-Warburgdevice the typical response for 1 ml stationary phaseP. putida M10 culture with the same dry cell con-centration was 66�mol l−1 min−1 for the OUR and9�mol g−1 min−1 for the SOUR at 22–24◦C (av-erage room temperature). The remarkable outcomefrom this data therefore, is that it appears that thehigh surface area/volume geometry of the�-Warburgdevice design has achieved less respiration limitation,even though the sample is unstirred. This presents im-portant insight regarding future respirometer design.
4. Conclusions
Oxygen consumption can be used as a surrogatemeasure for substrate consumption because of the sto-ichiometric link between the two processes for aerobicgrowth. Hence oxygen uptake profiles yield the sameinformation as substrate depletion profiles [16].
Respirometry is an established method for the deter-mination of BOD. However, most assays and meth-ods involve long and quite often labour intensiveprocedures, bulky and complex instruments and incases trained personnel.
In this work, initial data on a rapid and simplerespirometric method is reported. A simple bench-top�-Warburg device follows the respiration of micro-organisms in contact with a sample in a closedrespirometric chamber of defined volume. This isachieved by a very sensitive pressure sensor thatmonitors the pressure change in this chamber causedby the oxygen consumption, due to the activity of thebacteria.
It is shown that the�-Warburg device is capableof measuring minute pressure changes caused by therespiration of bacteria in less than 5 min utilising aminimal (1 ml) volume of sample! The respirationactivity of P. putida KT2440 andE. coli JM109 were
successfully monitored and fast responses were ac-complished by careful tuning of the micro-manometricparameters, such as the culture concentration and vol-ume, sample surface area/volume ratio and headspacevolume. Both qualitative and quantitative data, thelatter in the form of OUR and SOUR, could bederived using a simple equation. More importantly,oxygen-limiting conditions (small headspace volume)suggested that faster responses could be achieved,in terms of higher pressure change sensitivity, evenif accurate SOUR determination seemed to requireexcess headspace volume.
Qualitative data from both a standard respirometerand the�-Warburg device was in agreement however,the latter appeared to have achieved less respirationlimitation even though the sample was unstirred.
It is envisaged that the�-Warburg device could beused in several applications that require rapid oxygenconsumption determination, such as in situ methodsfor monitoring the oxidation process, organic loadcontent and treatability of influent streams in wastew-ater treatment facilities. These methods focus on theidentification of the DO dynamics as recorded by aDO probe. Reliability concerns however, arise dueto DO tendency to fouling and drift. Additionally,in case of systems that focus on input wastewaterswhere the respiration rate caused by the presence ofwastewaters is rather high [17] and either importantdilution [18] or small sample sizes [19] are neces-sary to allow measurement of the respiration rate, thedevice describe here could have a significant advan-tage. Furthermore, the limitations in the experimentsrequired to estimate biokinetic parameters in indus-trial fermentations can be overcome with a simpleand rapid respirometric method, which has all the ad-vantages of a batch technique coupled with the needfor little experimental and analytical effort. Biofilmrespiration, permeability testing, yeast and bacteriacontamination and biotoxicity testing could also besome potential applications of the�-Warburg devicereported here.
In theory, any chemical or biochemical reactioninvolving a pressure change due to gas productionor consumption, for example and enzymatic oxida-tion, could be monitored with the�-Warburg devicedescribed in this work implementing simple and in-expensive in situ instrumentation and possibly fastresponses.
270 A. Tzoris et al. / Analytica Chimica Acta 460 (2002) 257–270
References
[1] J. Zhong, K.B. Konstantinov, T. Yoshida, J. Ferment. Bioeng.77 (1994) 445.
[2] W.W. Umbreit, R.H. Burris, J.F. Stauffer, Manometric andBiochemical Techniques, Burgess, 1972, p. 1.
[3] H.A.C. Montgomery, Water Res. 1 (1967) 631.[4] F. Sierp, Ind. Eng. Chem. 20 (1928) 247.[5] D. Jenkins, Waste Treatment, Pergamon Press, Oxford, 1960,
p. 99.[6] C.P.L. Grady, J.S. Dang, D.M. Harvey, A. Jobbagy, X.-L.
Wang, Water Sci. Technol. 21 (1989) 957.[7] J.S. Cech, J. Chudoba, P. Grau, Water Sci. Technol. 17
(1994) 259.[8] J. Surmacz-Gorska, K. Gernaey, C. Demuynck, P. Vanrol-
leghem, W. Verstraete, Water Res. 30 (1996) 1228.[9] H. Spanjers, P. Vanrolleghem, G. Olsson, P. Dold, Water
Sci. Technol. 34 (1996) 117.[10] A. Khlebnikov, F. Samb, P. Peringer, J. Chem. Technol. Bio-
technol. 73 (1998) 274.
[11] M. de Bel, L. Stokes, J. Upton, J. Watts, Water Sci. Technol.33 (1996) 289.
[12] K. Gernaey, H. Bogaert, P. Vanrolleghem, L. van Vooren, W.Verstraete, Environmental Biomonitoring. The Biotechno-logy Ecotoxicology Interface, Cambridge University Press,Cambridge, 1998, p. 180.
[13] S.J. Pirt, Principles of Microbe and Cell Cultivation, Black-well Scientific Publications, Oxford, 1975, p. 87.
[14] P.F. Stanbury, A. Whitaker, S.J. Hall, Principles of Ferment-ation Technology, Elsevier/Pergamon, Amsterderm/Oxford,1995, p. 247.
[15] Z.-Y. Wen, J.-J. Zhong, Biotechnol. Tech. 9 (1995) 521.[16] R.G. Riefler, D.P. Ahlfeld, B.F. Smets, Biotechnol. Bioeng.
57 (1998) 35.[17] P.A. Vanrolleghem, Z. Kong, G. Rombouts, W. Verstraete, J.
Chem. Technol. Biotechnol. 59 (1994) 321.[18] M. Kohne, Environ. Tech. Lett. 6 (1985) 546.[19] P. Venrimmen, E.R. Henken, J.C. Lamb, J. Water Pollut. Con-
trol Fed. 39 (1967) 1006.
