Embed Size (px)
Citation preview
HAL Id: hal-02143257https://hal.archives-ouvertes.fr/hal-02143257
Submitted on 29 May 2019
HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.
A New Tool for the Quantification of MicroorganismInteraction Dynamics
Claire Albasi, Panagiotis Tataridis, Edgar Salgado Manjarrez, PatriciaTaillandier
To cite this version:Claire Albasi, Panagiotis Tataridis, Edgar Salgado Manjarrez, Patricia Taillandier. A New Tool for theQuantification of Microorganism Interaction Dynamics. Industrial and engineering chemistry research,American Chemical Society, 2001, 40 (23), pp.5222-5227. �10.1021/ie001060g�. �hal-02143257�
Open Archive Toulouse Archive Ouverte (OATAO) OATAO is an open access repository that collects the work of some Toulouse
researchers and makes it freely available over the web where possible.
This is an author's version published in: https://oatao.univ-toulouse.fr/22842
Official URL : https://doi.org/10.1021/ie001060g
To cite this version :
Any correspondence concerning this service should be sent to the repository administrator:
Albasi, Claire and Tataridis, Panagiotis and Salgado Manjarrez, Edgar and Taillandier, Patricia A New Tool for the Quantification of Microorganism Interaction Dynamics. (2001) Industrial & Engineering Chemistry Research, 40 (23). 5222-5227. ISSN 0888-5885
A New Tool for the Quantification of Microorganism InteractionDynamics
Claire Albasi,*,† Panagiotis Tataridis,† Edgar Salgado Manjarrez,‡ andPatricia Taillandier†
Laboratoire de Genie Chimique UMR, CNRS 5503, INP-ENSIACET, 18, chemin de la Loge,31078 Toulouse Cedex 4, France, and Departmento de Bioingenierıa, Unidad ProfesionalInterdisciplinaria de Biotecnologıa, IPN, Avenida Acueducto de Guadalupe S/N,Mexico D.F. 07340, Mexico
A two-reservoir, membrane bioreactor for carrying out studies of mixed population dynamics inbatch fermentation is presented. Mixing requirements and design aspects for the validity of theapproach are presented and discussed. Equations describing mixing times between the reservoirsare given and compared to the experimental results. The validity of the approach is demonstratedby the study of several types of situations in the bioreactor. The main validation consists of thecomparison between the results obtained in an actual mixed culture and the results obtainedby keeping the strains separated. Finally, this new tool is used to study the interaction kineticsbetween two oenological bacteria. These experiments in liquid media help to determine theseeding conditions for a chosen strain to overgrow another strain through a quantification ofthe interaction dynamics.
Introduction
In typical mixed culture studies, the routine proce-dures used for the tracking of individual strains aretime-consuming and costly. Very often, these proceduresinvolve the isolation of microorganisms and their de-velopment into colonies, which are subsequently identi-fied by their morphology or color, if possible. Frequently,the colonies are very similar, so a second procedure mustbe followed for their identification. In this case, abiochemical specificity, such as consumption of a givensubstrate, production of a given metabolite, or geneticdifferentiation, is pointed out.
Because the biomass concentration of each participat-ing microorganism in a mixed population is necessaryfor the quantitative study of the interactions, thesekinds of studies are limited by the availability andprecision of suitable differentiation methods.
This paper presents a membrane reactor specificallydesigned to carry out studies of mixed populationinteractions. It is based on the idea of keeping thestrains or species composing the mixed culture sepa-rated by way of a porous membrane, which allows thesubstrates and the metabolites to pass through and,therefore, the strains to interact.1 In this way, as thedifferent strains actually grow separated, they can besampled distinctly and there is no need for differentia-tion. Further, the individual strains can be tracked withthe methods used in pure culture studies.
The first part of the paper is devoted to a synthesisof the requirements for the validity of the proposal andthe successive steps of validation. The second part isan example of applications of the method.
Materials and Methods
As the experiments in the two parts of the paperinvolved yeast and then bacteria, this section will bedivided in the same way, after the bioreactor andmembrane are described.
Bioreactor. A scheme of the proposed reactor isshown in Figure 1. It is composed of two jars intercon-nected by a hollow fiber-membrane module, which isimmersed in the liquid of one of the jars. Medium flowand mixing are induced by alternately applying pressureto the headspace of each of the vessels. Compressed,filter-sterilized gas is used to apply pressure, and asystem of valves controls the admission and expulsionof the gas as a function of the liquid levels, which aremeasured with conductivity probes. A filtration swingis required to maintain the same mean liquid volumein both reservoirs. An extra benefit of this swing is thecontrol of fouling by backflushing. This device has beenpatented.2
Polymem (France) provided the hollow fiber modulefollowing a design specified by us. The fibers areU-shaped, and they are held together in their upperparts with an epoxy resin. The upper part of the bundleis contained in a stainless steel receptacle, and thefiltering part immerses freely in the medium. The fibersare made of polysulfone and present a nominal pore sizeof 0.1 µm; the internal and external diameters are 0.25mm and 0.43 mm, respectively. The water permeabilityof the fiberswas 3.5 × 10-9 m3 m-2 s-1 Pa-1, asestimated in the module. The total filtering surface was0.1 m2, as it will be explained further (see DesignAspects).
Validation: Yeast Fermentation. Salt Concentra-tion. Sodium chloride was used as the tracer in themixing studies. Conductivity measurements (Metro-Ohm, Switzerland) were used to assess the salt concen-tration.
Strains and Media. Saccharomyces cerevisiae K1 andSaccharomyces cerevisiae S6 (Lallemand, France) in-
* Author to whom correspondence should be addressed.E-mail: [email protected].
† INP-ENSIACET.‡ Unidad Profesional Interdisciplinaria de Biotecnologıa.
where c(t)i is the vector of concentrations of all thesolutes that can pass through the membrane and r(t)iis the vector of reaction rates for each solute. Thesubscripts A and B indicate the reservoirs.
The flow entering and leaving each reservoir recallsthe concept of dilution used in continuous flow systems;because of this similarity, we defined the ratio F/V asthe internal dilution rate, D.
Mixing without Reaction. In this case, rA ) rB ) 0.The solution of the previous equation at t ) 0, cA )
cA0, and cB ) cB0 is
From this equation, the relative degree of homogeneitycan be defined as
This equation can be used to define the mixing time,which is defined as the time required to achieve a givendegree of mixing. We took the value of 95% homogene-ity, i.e., a value of 0.95 for the right-hand expression ineq 3. The mixing time is thus defined by the equation
Therefore, the mixing time is proportional to the recip-rocal of the internal dilution rate.
Mixing with Reaction. When a reaction takes placein one or both of the reservoirs, no general solution toeq 1 exists because the expressions needed to calculatethe reaction rates can be quite complex. However, inthe case where the reaction rates are constant, asolution can be obtained as follows: The concentration
difference between the reservoirs is
The solution of eq 5 is
It can be noted that, for Dt f ∞
This means that, whenever a reaction takes place atdifferent rates in the two reservoirs, a concentrationdifference will be generated. Thus, a zero difference, thatis, complete mixing, cannot be achieved. This is trueindependently of the internal dilution rate and even ifreaction rates are not constant. Then, it can be shownthat, in this case, the mixing time is also given by eq 4.
Design Aspects
The theoretical studies on mixing rate led us to theconclusion that the approach would be valid only if avery high mixing rate was used. As explained previ-ously, we fixed, somewhat arbitrarily, a mixing time lessthan 10 min as our goal. From the equations presentedin the preceding section, it can be shown that thisrequires a dilution rate higher than 8 h-1 and given by
where VE is the exchanged volume in the pressurizationof reservoirs A and B and tA and tB are the timesrequired for the exchange from reservoir A to reservoirB and vice versa, respectively. VE is the volume of acomplete cycle, with the flow once in each direction.
Now, the parameters in eq 8 depend on the moduledesign. Several designs were tested, but it was con-cluded that the design best suited for our purposesconsisted of a bundle of hollow fibers immersed freelyin the liquid of one of the reservoirs (Figure 1). For sucha system, if VD represents the displaced volume actuallyseen as a variation in the reservoirs levels and VMrepresents the dead volume, then the exchange volumeVE and the filtered volume VF,i are given by
The expression for the filtered volume is necessary forcalculating the fouling of the membranes and, thus, thetimes tA and tB. To perform an estimation of thisquantity, it was assumed that, because of the very highsurface area, surface fouling was small and thus thecurvature effects on the cake arrangement could beneglected. Consequently, by using the cake model fordead-end filtration on flat membranes and the volumes
dc(t)A
dt) F
VA[c(t)B - c(t)A] + r(t)A
dc(t)B
dt) F
VB[c(t)A - c(t)B] + r(t)B
(1)
cA(t) )cA0 + cB0
2+
cA0 - cB0
2exp{-2Dt}
cB(t) )cA0 + cB0
2-
cA0 - cB0
2exp{-2Dt}
(2)
(cA0 - cB0) - (cA - cB)cA0 - cB0
) 1 - exp{-2Dt} (3)
θm ) -ln(0.05)
2D(4)
dcA
dt-
dcB
dt) 2D(cB - cA) + (rA - rB)
d∆cdt
) -2D∆c + ∆r(5)
∆c ) ∆r2D
+ (∆c0 - ∆r2D) exp{-2Dt} (6)
∆c f∆r2D
(7)
F )VE
tA + tB(8)
VE ) 2VD - VM
VF,A ) VD (9)
VF,B ) VD - VM
flow rate between the reservoirs, denoted A and B, and let V be the liquid volume of each reservoir. F is taken as the average flow through the membrane, calculated from the exchange volume and the duration of a filtration period. This concept is correct in the present case, because the exchange duration is small compared to the system dynamics, so that it can be written as a constant in the mass balance. Independently of how the medium transfer between the reservoirs is actually carried out, the mixing can be theoretically studied by using material balances. F, correlated with the mem-brane permeability and the applied pressure, is the membrane flow, continuously feeding each reservoir. Then, at any time t, the global material balances for the substances capable of crossing through the mem-brane are
defined by eqs 9, the following equation for D wasobtained
Equation 10 was used for the final sizing of the hollowfiber module. In Figure 2, the expected dilution rate ispresented as a function of the displaced volume andtotal filtering surface. The parameters used are reportedin Table 1; they are average values usually met in cellfiltration experiments. It can be seen that a dilution rategreater than about 8 h-1 can be obtained with a surfacearea greater than 0.05 m2. Thus, we fixed the finalsurface area at 0.1 m2.
Results
Mixing Quality. Mixing without Reaction. Experi-ments consisted of the mixing of a saline solutioncontained in one of the reservoirs with water containedin the other reservoir. A comparison of the experimentaland theoretical values showed good agreement. Its valuewas necessary to compute the value of the dilution rateD. It did not change significantly for displaced volumesbetween 100 and 200 mL. Therefore, the value for thisparameter was fixed at about 130 mL for the followingexperiments.
Mixing with Reaction. Mixing experiments with fer-mentation taking place in one of the reservoirs were thefirst performed (data not shown6). The observed evolu-tions of the consumption of the main substrate were
very similar in the two reservoirs. The greatest differ-ence observed in substrate concentration between thereservoirs was less than 1 g/L, coinciding with themaximum substrate consumption rate of about 6 g/(Lh). This result is higher than expected from eq 7, butthat equation is for constant reaction rates. As this wasnot the case, the observed difference was probably amanifestation of the cumulated effects of changes in thereaction rate. The suitability of the device was thussupported by these results.
Validation: Yeast Fermentation. As stated previ-ously, the proposed bioreactor can be considered ad-equate only if the behavior of a pair of strains is thesame when they are grown in the same reservoir (actualmixed culture) and in different reservoirs (simulatedmixed culture).
Two heterogeneous cultures exhibiting two kinds ofinteractions7 were used to test the validity of thebioreactor. The results obtained with the first of thesecultures, composed of Kluyveromyces marxianus ATCC28912 and a Saccharomyces cerevisiae strain competingfor the substrate, are reported elsewhere.6 Essentially,the expected similarity between the actual and simu-lated mixed cultures was found. Minor differences,probably due to small differences in the inoculationratio, were observed. As the interaction took place onlythrough small molecules, such as glucose and ethanol,we decided to test the validity with a protein-mediatedinteraction. Thus, the second mixed culture used forvalidation and reported here was the so-called “killerinteraction” between two strains of Saccharomycescerevisiae.
The killer phenomenon is established between a yeaststrain producing a killer protein and a yeast strainsensitive to this protein.8 Several killer proteins havebeen identified.9 The particular strain used in this workproduces a toxin of type K2, which has been identifiedas a glycoprotein of about 16 kDa.9 A period of time froma few minutes to several hours is observed between theadsorption of the protein and the loss of viability.
The killer protein activity is very sensitive to the pH.Therefore, we used buffered media during our studiesto keep the pH values between 4.5 and 4.8, near thepH of maximum activity reported by Ramon-Portugal4
for the strain used.Actual Mixed Culture. The results of the reference
fermentation, i.e., in the actual mixed culture, areshown in Figure 3. This fermentation was carried outin the proposed reactor with both strains growing in
Figure 2. Internal dilution rate, D, as a function of displacedvolume, VD, and filtration area.
Table 1. Parameters Used in the Estimation of D
parameter value
VD variableS variableVM variableV 2 LRm 2.7 × 1011 m-1
η 0.001 Pa s∆p 50 kPaRgi 3 × 1012 m kg-1
XA, XB 5 kg m-3
D ) 1V
×
2VD - VM
2Rmη∆p (VD - 0.5VM
S ) + η2∆p[RgAXA(VD
S )2
+ RgBXB(VD - VM
S )2](10)
Figure 3. Evolution of glucose and estimated evolution of S.cerevisiae K1 and S. cerevisiae S6 in an actual mixed culture.
only one of the reservoirs but with the second reservoirconnected and containing fermentation medium. Thiswas necessary to distribute the substrates and productsin the same way as they distribute when the strainsare grown separated and, thus, to avoid any possibledifferences in the behavior of the culture due to thesolute concentrations. The viable BDW of each strainwas calculated from the total cell number, the percent-age of S6 determined by the “killer test”, and thecorrelation between cell number and BDW.
Simulated Mixed Culture. The results obtained forthe strains grown separately in both reservoirs areshown in Figure 4. Viable BDW was calculated fromcolony counting of diluted aliquots of the mediumcontained in each of the jars. This was done to maintainsimilar analytical methods for the actual and simulatedmixed cultures. The percentage of S6 was thus calcu-lated from colony counts and not from killer tests.
Similar evolutions were observed for the two fermen-tations for the different parameters. The evolutions ofthe two strains were also very similar. Glucose was alsodepleted at roughly the same time, and the evolutionsof the total viable count and percent S6 were likewisesimilar. A discussion has been developed elsewhere;6given some details on the interpretation of these results,it was concluded that the apparatus is suitable for thestudy of an interaction dynamic.
Application: Bacterial Interactions. The secondpart of this paper is an example of the use of themembrane reactor described and validated above. Theindustrial background is wine making. Two fermentingsteps are required, occurring in nonsterile medium.10
To control the production rate and also the quality ofthe final product, a seeding with selected microorgan-
isms must be performed in order to ensure the implan-tation of the convenient strain.11 This study deals withlactic bacteria involved in the second fermentation ofwine making. It has been pointed out that the strainOenococcus oeni involved here is a producer of aninhibitory substance for other bacterial strains.12
Pure Culture. The first experiments consisted ofcultivation of the pure strains in order to obtain thekinetics under these conditions. The experimental re-sults are the biomass concentration and the consump-tion of the substrate versus time. The experiments werestopped when there was no more substrate in themedium (L-malic acid). The logistic equation13 givesanalytical expressions for the biomass (eq 11) and thegrowth rate (eq 12) as functions of time. By parametricidentification with the Excel solver, µmax and Xf wereobtained with good agreement for the pure cultures.
Mixed Cultures. For the mixed culture, to limit theeffect of competition for the substrate, the ratio betweenthe initial seedings of the two strains was calculatedby comparing the maximum values obtained for thebiomass growth rates from logistic eq 12. Then, themixed culture was performed in the membrane biore-actor, and the same experimental parameters werecollected, i.e., biomass and substrate concentration.These results are reported in Figure 5. The experimen-tal results, represented by the plain dots, are comparedto the growth of the pure culture under the sameconditions of initial seeding, given by the above-definedequation (eq 11) and represented by the dotted lines.
In Figure 5, the inhibition appears clearly, as thebiomass concentration of the sensitive strain increasesmore slowly and reaches a lower level in the mixedculture than in the pure culture. At the same time,Figure 5 allows us to verify that the dynamics of theinhibitory strain was not affected by its growth in themixed culture.
Figure 6 reports the specific growth rate of thesensitive strain in the pure and mixed cultures. Acomparison of the two curves shows that the inhibitory
Figure 4. Evolution of glucose and estimated evolution of S.cerevisiae K1 and S. cerevisiae S6 in a simulated mixed culture.
Figure 5. Comparison of the biomass concentrations of strainsin pure and mixed cultures 21A3-17A3.
Figure 6. Quantification of the inhibition by comparing growthrates in pure and mixed cultures.
X )X0 exp(µmaxt)
1 -Xo
Xf[1 - exp(µmaxt)]
(11)
µ )µmax(1 -
Xo
Xf)
1 -Xo
Xf[1 - exp(µmaxt)]
(12)
