On-Line Monitoring and Control of
`Methanol Concentration in Shake-Flask
`Cultures of Pichiapastoris
`
`M. M. Guarna,1 G. J. Lesnicki,1 B. M. Tam,2 J. Robinson,1,3
`C. Z. Radziminski,1,3 D. Hasenwinkle,1 A. Boraston,1,3 E. Jervis,1,4
`R. T. A. MacGillivray,2 R. F. B. Turner,1,5 D. G. Kilburn1,3
`
`1TheBiotechnologyLaboratory,TheUniversityofBritishColumbia,
`Vancouver,BCV6T1Z3,Canada;telephone:604-822-4182;fax:
`604-822-2114;e-mail:Kilburn@unixg.ubc.ca
`2DepartmentofBiochemistry,TheUniversityofBritishColumbia,
`Vancouver,BCV6T1Z3,Canada
`3DepartmentofMicrobiologyandImmunology,TheUniversityofBritish
`Columbia,Vancouver,BCV6T1Z3,Canada
`4DepartmentofChemicalEngineering,TheUniversityofBritishColumbia,
`Vancouver,BCV6T1Z3,Canada
`5DepartmentofElectricalEngineering,TheUniversityofBritishColumbia,
`Vancouver,BCV6T1Z3,Canada
`
`Received6September1996;accepted13March1997
`
`Abstract: The methylotrophic yeast Pichia pastoris can
`be used to express recombinant genes at high levels un-
`der the control of the methanol-inducible alcohol oxidase
`1 (AOX1) promoter. Accurate regulation of the methanol
`concentration in P. pastoris cultures is necessary to
`maintain induction, while preventing accumulation of
`methanol to cytotoxic levels. We developed an inexpen-
`sive methanol sensor that uses a gas-permeable silicone
`rubber tube immersed in the culture medium and an or-
`ganic solvent vapor detector. The sensor was used to
`monitor methanol concentration continuously through-
`out a fed-batch shake-flask culture of a P.pastorisclone
`producing the N-lobe of human transferrin. The sensor
`calibration was stable for the duration of the culture and
`the output signal accurately reflected the methanol con-
`centration determined off-line by HPLC. A closed-loop
`control system utilizing this sensor was developed and
`used to maintain a 0.3% (v/v) methanol concentration in
`the culture. Use of this system resulted in a fivefold in-
`crease in volumetric protein productivity over levels ob-
`tained using the conventional fed-batch protocol. © 1997
`John Wiley & Sons, Inc. BiotechnolBioeng 56: 279–286, 1997.
`Keywords: methanol sensor; methanol monitoring and
`control; methylotrophic yeast fermentation; Pichia pas-
`toris; transferrin; shake-flask cultures
`
`INTRODUCTION
`
`Pichia pastoris has proven to be an effective host for the
`production of recombinant proteins (reviewed by: Buckholz
`and Gleeson, 1991; Cregg et al., 1993; Faber et al., 1995;
`Romanos, 1995; Romanos et al., 1992). This methylotro-
`phic yeast combines several advantages of both prokaryotic
`
`Correspondence to: D. G. Kilburn
`Contract grant sponsor: The National Institutes of Health
`Contract grant number: RO1 DK 21739
`
`and animal cell expression systems. It is readily amenable to
`genetic manipulation, can be easily grown to high cell den-
`sities using minimal media, and is able to introduce some
`eukaryotic posttranslational modifications. P. pastoris is
`also capable of expressing heterologous genes to high levels
`under the control of the strong and tightly regulated alcohol
`oxidase 1 (AOX1) promoter (Tschopp et al., 1987). Metha-
`nol serves as the inducer and as the carbon source during the
`production phase of P. pastoris cultures. A challenge asso-
`ciated with methanol feeding is that the methanol concen-
`tration must be maintained within a relatively narrow range.
`We have observed that recombinant protein production de-
`creases when the methanol concentration exceeds ca. 1.0%
`(v/v); this concentration is only ca. 0.5% (v/v) higher than
`that suggested in the Pichia Expression Kit (Invitrogen, San
`Diego, CA) to initiate induction.
`While it is possible to monitor methanol concentration
`using HPLC or GC, these methods are expensive and not
`easily implemented on-line. Alternatively, methanol con-
`centration can be estimated indirectly by interrupting the
`feed and measuring the time required for methanol to be
`exhausted, as indicated by a sharp increase in dissolved
`oxygen (i.e., ‘‘DO spike’’). This approach, however, repeat-
`edly exposes the cells to potentially noninducing levels of
`methanol and it is restricted to clones having a wild-type
`phenotype for methanol utilization (Mut+). Clones with a
`disrupted AOX1 gene, exhibiting a slow methanol utiliza-
`tion phenotype (MutS), consume methanol at a rate that is
`too slow for this technique to be practical. Furthermore, in
`shake-flask cultures, DO monitoring is difficult to imple-
`ment using currently available probes.
`The protocol for shake-flask cultures provided with the
`Pichia Expression Kit recommends 0.5% (v/v) methanol to
`
`© 1997 John Wiley & Sons, Inc.
`
`CCC 0006-3592/97/030279-08
`
`Motif Exhibit 1032, Page 1 of 8
`
`Case No.: IPR2023-00321
`U.S. Patent No. 10,689,656
`
`

`

`Air In
`
`C
`
`60mm
`
`c 1n1e1 +
`
`Ple,aglass
`Housing
`
`! ...... .
`
`Figure 1. Methanol monitoring and control system for shake-flask cultivations of Pichia pastoris. (A) Schematic diagram of instrumentation and
`fluid/sample handling system components. (B) Photograph of silicone rubber tubing sensor probe showing stainless steel support fixture for shake-flask
`applications. (C) Drawing of in-house fabricated flow cell and detector housing.
`
`start induction and the additions of methanol to 0.5% (v/v)
`every 24 h. However, the methanol consumption rate is
`strongly dependent on culture variables such as cell density,
`and hence the actual methanol concentration in a shake-
`flask culture after methanol adaptation is normally not
`known with any precision. A simple, inexpensive, and re-
`liable on-line system for monitoring and controlling metha-
`nol concentration would be advantageous for optimizing
`and maintaining induction conditions in P. pastoris cultures
`both in shake-flask cultures and large-scale fermentations.
`We designed and characterized a practical closed-loop
`methanol control system based on a commercially available
`organic solvent vapor detector, an in situ silicone rubber
`tubing probe, and a computer-actuated methanol feeding
`system. In this report we describe the components and op-
`eration of this control system and present data demonstrat-
`ing its utility and performance in cell-free aqueous methanol
`solutions and in fed-batch shake-flask cultures of P. pastoris
`expressing recombinant human transferrin N-lobe (hTF/
`2N). Recombinant protein levels observed in the methanol-
`regulated cultures are compared to non-methanol-regulated
`cultures as a first step toward assessing the advantages of
`on-line control of methanol concentration in cell culture
`environments.
`
`MATERIALS AND METHODS
`
`Methanol Monitoring and Control System
`
`The methanol sensor and methanol feeding system with
`associated instrumentation developed for use in this work
`are shown schematically in Figure 1A. The sensor itself
`comprises two components: a probe with a segment of sili-
`cone rubber tubing that remains immersed in the culture
`medium and facilitates conversion of the solution-phase
`methanol concentration to a partial pressure of methanol
`vapor in a carrier stream of air perfusing the lumen of the
`tubing (Fig. 1B); and a solid-state methanol vapor detector
`connected in-line with the outlet of the methanol:air flow
`stream (Fig. 1C). The voltage output signal from the sensor
`is digitized and fed to a personal computer where it is com-
`
`pared to a manually entered set-point in order to generate an
`actuating signal to drive the methanol feed pump. The head-
`space of the shake flask is continuously flushed with sterile
`filtered air to maintain a constant gas-phase oxygen con-
`centration.
`The silicone rubber membrane probe was designed based
`on principles described earlier in work involving measure-
`ments of pCO2 (Kilburn, 1967), pO2 (Phillips and Johnson,
`1961), and ethanol (Yamada et al., 1987; Yamane et al.,
`1981) using silicone rubber and polytetrafluoroethylene tub-
`ing probes. Physically, the methanol probe consists of a
`20-cm length of silicone rubber tubing (Sani-Tech, Lafay-
`ette, NJ) with a lumen diameter of 0.31 mm and a wall
`thickness of 0.15 mm supported by a stainless steel wire
`mesh, as shown in the photograph of Figure 1B. The probe
`remains immersed in the liquid phase while the lumen of the
`probe is continuously perfused with filtered air. The flow
`rate is regulated by a needle valve and monitored using an
`Omega model FMA 1804 mass flow meter (Omega Engi-
`neering, Stamford, CT). Methanol dissolved in the culture
`medium diffuses across the gas-permeable silicone rubber
`‘‘membrane’’ of the probe. Vapor-phase methanol in the
`lumen of the probe tubing, at a partial pressure determined
`by the concentration in the bulk liquid phase, is carried by
`the flowing stream of air to the downstream methanol de-
`tector element. The methanol–air sample mixture passes
`through a 0.2-mm pore size hydrophobic filter (Gelman Sci-
`ences, Ann Arbor, MI) inserted upstream of the detector to
`remove aerosol moisture from the flow stream, which can
`lead to instability in the detector signal. The carrier stream
`can be switched to bypass the methanol probe using a manu-
`ally actuated three-way valve to provide a methanol-free
`sample to the detector at any time to check the baseline. For
`all media tested, the detector baseline voltage measured in
`methanol-free medium was within ±1% of the value mea-
`sured in the bypass air stream.
`The methanol vapor detector utilizes a Figaro model TGS
`822 SnO2 organic vapor sensor (Figaro Engineering, Wil-
`mette, IL), which is sensitive to a broad range of alcohols
`and organic solvent vapors. The operating characteristics
`and instrumentation required for the detector element are
`
`280
`
`BIOTECHNOLOGY AND BIOENGINEERING, VOL. 56, NO. 3, NOVEMBER 5, 1997
`
`Motif Exhibit 1032, Page 2 of 8
`
`Case No.: IPR2023-00321
`U.S. Patent No. 10,689,656
`
`

`

`described in the manufacturer’s products catalog (Figaro
`Engineering, Inc., 1990). The detector is mounted in a
`Plexiglas housing as shown in Figure 1C. The methanol–air
`sample mixture flows through an inlet port (1.7-mm diam-
`eter) on the side wall of the housing to an interior ‘‘head-
`space’’ (approximate volume 1.9 cm3) where it is exposed
`to the detector; the sample exhaust is vented through a port
`in the base of the detector element. The power supply for the
`detector, detector heater, and the conductance measuring
`circuit were designed and fabricated according to the manu-
`facturer’s specifications and application notes.
`The sensor (probe plus detector) was calibrated on-line
`by adding prescribed aliquots of 100% methanol to a known
`volume of distilled water or culture medium in an agitated
`thermostated shake flask. Calibration was verified by with-
`drawing samples and determining the methanol concentra-
`tion off-line by HPLC using a Shimadzu model SCL 6B
`auto-injector and LC 6A pump (Shimadzu Inc., Kyoto, Ja-
`pan) with an HPX-87H column (BioRad, Hercules, CA).
`In fed-batch operation, the methanol feed stock (100%
`CH3OH) was delivered using a Gilson Minipuls 3 peristaltic
`pump (Gilson, Villiers-Le-Bel, France) with an in-line
`check valve inserted between the pump and shake flask.
`Because the methanol consumption rate is slow compared to
`the slowest practical feeding rate, a simple on/off pulsed-
`feed control algorithm was used. When the methanol con-
`centration, as determined by the sensor output signal, fell
`below the set-point value, the controller actuated the feed
`pump for a prescribed (user-defined) time interval to deliver
`a precise dose of methanol. The controller then entered a
`wait state for 5–7 min (a time slightly longer than the char-
`acteristic response time of the probe) before reevaluating
`the concentration. These feed-pulse + delay cycles were
`repeated until the methanol concentration exceeded the set-
`point value.
`
`Pichiapastoris Cultures
`
`To evaluate the methanol control system in P. pastoris cul-
`tures, two parallel 2-L baffled shake flasks, each containing
`500 mL of BMGY medium (buffered minimal glycerol
`complex medium, Pichia Expression Kit) and 0.1 mL of
`antifoam 289 (Sigma Chemical Co., St. Louis, MO), were
`inoculated. The inoculum consisted of 20 mL of an over-
`night culture (6.2 optical density units at 600 nm, OD600) of
`a Mut+ P. pastoris GS115 clone producing hTF/2N (Mason
`et al., 1996). The cultures were grown in BMGY for 24 h to
`a cell density of 28 OD600. The cells were then separated by
`centrifugation and resuspended in 500 mL of basal salts
`medium (26.7 mL L−1 85% o-phosphoric acid, 0.82 g L−1
`calcium sulfate anhydrous, 18.2 g L−1 potassium sulfate,
`14.9 g L−1 magnesium sulfate-7H2O, and 4.13 g L−1 potas-
`sium hydroxide), containing 1 mL of PTM4 trace salts with
`biotin (2 g L−1 cupric sulfate-5H2O, 0.08 g L−1 sodium
`iodide, 3.0 g L−1 manganese sulfate-H2O, 0.2 g L−1 sodium
`molybdate-2H2O, 0.02 g L−1 boric acid, 0.5 g L−1 cobalt
`
`chloride, 7.0 g L−1 zinc chloride, 22.0 g L−1 ferrous sulfate-
`7H2O, 1 mL L−1 sulfuric acid, and 0.2 g L−1 D-biotin).
`Two different strategies were employed for feeding
`methanol to the cultures. For one culture, the standard (In-
`vitrogen) protocol was followed: 2.5 mL of pure methanol
`was added daily. In the parallel culture, methanol was main-
`tained at 0.3% (v/v) using the closed-loop control system
`based on the feedback signal provided by the methanol sen-
`sor. Culture samples were taken periodically to determine
`the OD600. Additional samples were collected and filtered
`through a 0.2-mm pore size filter; the filtrate was used for
`off-line analysis of methanol and hTF/2N concentrations.
`Culture supernatants were analyzed by sodium dodecyl sul-
`fate-polyacrylamide gel electrophoresis (SDS-PAGE), and
`protein bands were visualized with Coomassie brilliant blue
`stain and quantified by scanning densitometry using a Fast-
`Scan Computing Densitometer (Molecular Dynamics).
`
`Chemicals and Reagents
`
`All chemicals used in the characterization of the sensor and
`in the preparation of culture media were obtained from
`Fisher Scientific (Nepean, ON), BDH Chemicals (Toronto,
`ON), Anachem Science (Montre´al, PQ), and Sigma Chemi-
`cal Co. and were used as received. All solutions were pre-
`pared using distilled water.
`
`RESULTS AND DISCUSSION
`
`Methanol Probe Characterization
`
`SensitivitytoMethanol
`
`A typical strip-chart record of the time-dependent response
`to methanol additions and the resulting methanol calibration
`curve are shown in Figure 2A and 2B, respectively. The
`baseline sensor voltage (i.e., the reading in a methanol blank
`solution) under the conditions used throughout this work
`was 1.39 ± 0.01 V; the baseline voltage periodically mea-
`sured using the bypass stream of air remained within this
`range during all experiments. The sensor response time (to
`reach 90% of steady state) at 30°C in methanol–water so-
`lutions was ca. 4 min, using a carrier air-stream flow rate of
`5 mL min−1 with probe dimensions as indicated in the Ma-
`terials and Methods section. In terms of the calibration char-
`acteristic and response time, the behavior of the probe was
`virtually the same whether measuring in methanol–water
`solutions or culture media, with or without cells. In the
`present configuration, the methanol concentration detection
`limit (defined as the concentration corresponding to a signal
`3 standard deviations of the noise above the mean baseline
`response) at 30°C was approximately 1.2 × 10−4% (v/v).
`The nonlinearity evident in the sensor response function
`results from a combination of two effects: one is an artifact
`of the electronic measuring circuit that employs a simple
`voltage divider configuration to measure conductance
`
`GUARNA ET AL.: METHANOL IN PICHIAPASTORALIS CULTURES
`
`281
`
`Motif Exhibit 1032, Page 3 of 8
`
`Case No.: IPR2023-00321
`U.S. Patent No. 10,689,656
`
`

`

`5.0
`
`- 4.0
`>
`
`A
`
`0.1 %
`
`Cl)
`Cl
`~
`0
`>
`'5
`
`C. -::J
`0 ...
`
`0
`C/l
`C:
`Cl)
`(/)
`
`3.0
`
`2.0
`
`1 .0
`
`0.0
`
`0
`
`1 0
`
`20
`
`30
`
`40
`
`Time (minutes)
`
`-'5
`> 5 .0
`C. - 4 .0
`0 ...
`0
`rn
`C:
`Cl)
`U)
`"C
`
`3.0
`
`2.0
`
`:::I
`
`Cl) -u
`Q) ... ...
`
`0
`()
`
`1 .0
`
`Q)
`
`.5
`QI
`rn
`~ 0.0
`m
`
`0
`
`0.1
`
`0.2
`
`0 .3
`
`0.4
`
`0 .5
`
`Methanol Concentration (% v/v}
`
`Figure 2. Methanol sensor response characteristics measured at 30°C
`with a carrier air flow rate of 5 mL min−1. (A) Strip-chart record showing
`characteristic response time to 2.5 mL methanol injections (arrows); com-
`ponent T1 (0.71 ± 0.19 min) is the time for the sensor to initiate a response
`and component T2 (3.15 ± 0.63 min) is the additional time required to
`reach 90% of the steady-state signal. (B) Sensor response versus methanol
`concentration in methanol–water solution.
`
`changes in the detector element, yielding a hyperbolic rela-
`tionship between output voltage and conductance. The other
`is itself a composite effect that results from the fundamental
`physical chemistry and physical electronic principles of
`semiconducting metal-oxide gas sensors. These principles
`have been described in detail elsewhere (e.g., Heiland,
`1982; Jones et al., 1985; Morrison, 1982; Watson, 1984);
`however, a summary of some of the salient points may be
`helpful here. Briefly, the conductivity of sintered metal ox-
`ides is mediated primarily by electrons populating surface
`states that arise due to a deficit of lattice oxygen at the grain
`
`boundaries and lie energetically near the edge of the con-
`duction band. The conductivity is modulated by any process
`that alters the availability of these surface states. In the case
`of the methanol probe described here, chemisorption of oxy-
`gen in the absence of methanol depletes the conduction
`electrons available at the surface and lowers the macro-
`scopic conductivity. In the presence of methanol, which
`coadsorbs to the surface, the semiconductor catalyzes the
`reduction of surface-bound oxygen, forming the more
`weakly interacting formaldehyde that desorbs and frees sur-
`face conduction band electrons that, in turn, increases the
`observed conductivity. All of the processes involved are
`nonlinear over an appreciable range of methanol concentra-
`tions and hence so is the resulting response function, even
`though consideration of mass transfer limited by the diffu-
`sion of methanol across the silicone rubber membrane
`would lead to the expectation of a linear response.
`
`SensitivitytoOtherOrganicVapors
`
`The detector element employed in this work is not intrinsi-
`cally selective for methanol. Some extrinsic selectivity can
`be engineered in this family of sensors by the incorporation
`of appropriate dopant species in the sintered metal oxide
`pellet and by the choice of surface temperature (Gentry and
`Jones, 1986; Yamazoe et al., 1983). The particular model of
`detector element used here is specified for applications in-
`volving detection of a range of alcohols, as well as toluene,
`xylene, benzene, acetone, carbon monoxide, and several
`small hydrocarbons (Figaro Engineering, 1990). In the ap-
`plications considered here under aerobic conditions, none of
`these potential interfering species exist in the culture me-
`dium in significant concentrations. However, if mix feeding
`of glycerol and methanol is used or if the culture becomes
`oxygen limited, low levels of ethanol can be produced by P.
`pastoris. Although the sensor is slightly more sensitive to
`ethanol than to methanol, the maximum ethanol concentra-
`tion reported in mix-fed P. pastoris cultures was 0.02%
`(v/v) and only 0.003% in methanol-fed cultures (Brierley et
`al., 1990). Ethanol interference is thus of no practical con-
`cern when methanol is controlled at 0.3% (v/v), a level that
`seems adequate to maintain induction (see below).
`
`ProbeDimensionsandAirFlowRate
`
`The main physical parameters of the probe that dictate both
`the kinetics and the sensitivity of the response are the sili-
`cone rubber tubing length, tubing diameter and wall thick-
`ness, and the flow rate of the carrier stream of air. In the
`present work, the tubing diameter and wall thickness as
`specified in the Materials and Methods section were chosen
`to yield reasonable durability with a minimal compromise
`of sensitivity and response time. The tubing length and air
`flow rate were then optimized based on the performance
`needed for the intended application. The dependence of tub-
`ing length and air flow on the sensor response is shown in
`Figure 3A and 3B, respectively. Longer tubing lengths
`
`282
`
`BIOTECHNOLOGY AND BIOENGINEERING, VOL. 56, NO. 3, NOVEMBER 5, 1997
`
`Motif Exhibit 1032, Page 4 of 8
`
`Case No.: IPR2023-00321
`U.S. Patent No. 10,689,656
`
`

`

`7.0
`
`~ 6.0
`Q)
`0, 5.0
`m
`.....
`0
`>
`
`4.0
`
`'5 a. - 3.0
`
`:::J
`0
`....
`0
`U)
`C
`Q)
`C/J
`
`2.0
`
`1 ,0
`
`0.0
`
`7.0
`
`> 6.0
`---Q)
`0, 5.0
`~
`0 >
`
`4.0
`
`-:::J .s-:::,
`
`0
`
`L.
`0
`IJ'I
`C:
`Q)
`(/)
`
`3.0
`
`2.0
`
`1.0
`
`0.0
`
`A
`
`TemperatureandpHDependence
`
`The temperature and pH dependence of the sensor, mea-
`sured in a methanol–water solution, is shown in Figure 4A
`and 4B, respectively. Essentially all of the physical and
`chemical processes involved in the operation of the sensor
`are inherently temperature dependent. However, the detec-
`tor element is remotely located and thereby thermally iso-
`lated from the culture environment; the detector also incor-
`porates an integrated heating coil that maintains the surface
`temperature at ca. 400°C. No significant cooling of the de-
`tector is effected by the methanol–air gas flow through the
`
`0.3 %
`0.2 %
`
`0.1 %
`
`0.0 %.
`
`0
`
`50
`
`100
`
`150
`
`200
`
`Tubing Length (cm}
`
`8
`
`0.3 %
`0.2 %
`
`0.1 %
`
`------------➔ 0.0 %
`
`7.0
`
`~ 6.0
`0, 5.0
`
`Q)
`
`m -0
`a. - 3.0
`
`4.0
`
`>
`5
`
`:::J
`0
`
`L.
`0
`11'1
`C:
`Q)
`(/)
`
`2.0
`
`1 .0
`
`0.0
`
`7.0
`
`A
`
`- - - - - - - - - 0.0 %
`
`0
`
`20
`
`25
`
`30
`
`35
`
`40
`
`Temperature (°C)
`
`0
`
`5
`
`10
`
`1 5
`
`20
`
`25
`
`Carrier Air Flow (mUmin}
`
`Figure 3. Tubing length and carrier air flow-rate dependence determined
`at 30°C. (a) Sensor response versus silicone rubber sensor tubing length for
`a methanol blank solution and three nonzero methanol concentrations. (b)
`Sensor response versus carrier air flow rate for the same methanol con-
`centrations as in (a).
`
`clearly yield higher sensitivity and larger output signals,
`albeit with rapidly diminishing returns beyond ca. 50 cm;
`however, the response time beyond 40–50 cm increases
`(data not shown) to a point where the system becomes im-
`practical for use in P. pastoris fermentations. A length of 20
`cm was chosen as the optimal compromise between sensi-
`tivity and response time. Response time is also (inversely)
`dependent on the carrier air flow rate. However, as shown in
`Figure 3B, the sensitivity drops as a function of air flow
`above 5 mL min−1; hence, a rate of 5 mL min−1 was again
`chosen as the optimal compromise between sensitivity and
`response time.
`
`8
`
`-...
`
`...........
`
`~ 6.0 -
`
`Q)
`
`Cl 5 .0
`
`>
`
`4.0 •
`
`m -0
`-:::J a.
`:i
`0
`... 0
`f/J
`C
`Q) en
`
`3 .0
`
`2.0
`
`1.0
`
`0.0
`
`0
`
`'
`2
`
`4
`
`8
`
`1 0
`
`12
`
`6
`
`pH
`
`Figure 4. Effect of variations in temperature and pH determined at 30°C
`using a carrier air flow rate of 5 mL min−1 and a tubing length of 20 cm.
`(a) Sensor response versus temperature of the methanol–water analyte
`matrix. (b) Sensor response versus pH of the methanol–water analyte ma-
`trix with a methanol concentration of 0.1%.
`
`GUARNA ET AL.: METHANOL IN PICHIAPASTORALIS CULTURES
`
`283
`
`Motif Exhibit 1032, Page 5 of 8
`
`Case No.: IPR2023-00321
`U.S. Patent No. 10,689,656
`
`

`

`MW
`
`105
`70.8
`
`43.6
`
`28.3
`
`17.9
`
`2
`
`3
`
`4
`
`5
`
`6
`
`the methanol control system was used to maintain a metha-
`nol concentration of 0.3% (v/v) in the production phase. In
`contrast to the uncontrolled culture, recombinant protein
`accumulated continuously for 2 days, reaching a maximum
`concentration of 120 mg L−1.
`Under the experimental conditions used, the standard
`protocol of daily methanol additions was insufficient to
`maintain induction. The concentration of hTF/2N in the
`medium reached a maximum 10 h after the first methanol
`addition. Maintaining the culture for longer times with daily
`additions did not result in any further increase in volumetric
`protein yield. In contrast, control of methanol concentration
`
`Figure 5. SDS-PAGE analysis of hTF/2N produced by P. pastoris. Cul-
`ture medium after (lanes 1, 2) 10 h, (lanes 3, 4) 33 h, and (lanes 5, 6) 56
`h of production was sampled from (lanes 1, 3, 5) the flask receiving daily
`additions of methanol and (lanes 2, 4, 6) the flask with constant (ca. 0.3%)
`methanol concentration.
`
`housing as evidenced by the flat response as a function of
`flow rate for a methanol blank solution shown in Figure 3B.
`Furthermore, no temperature dependence is observed in the
`absence of methanol. The observed temperature dependence
`of the sensor is therefore dominated by effects relating to
`the mass transfer of methanol across the silicone rubber
`membrane of the probe. With regard to pH dependence, it
`was important to establish that the sensor response is inde-
`pendent of fluctuations in culture pH, as demonstrated by
`the data in Figure 4B, because pH is normally not controlled
`in shake-flask cultures.
`
`Increased Protein Productivity in P.
`pastorisCultures
`
`The results obtained using the two different methanol feed-
`ing protocols in parallel shake-flask cultures are shown in
`Figures 5, 6, and 7. SDS-PAGE showed that recombinant
`hTF/2N comprises the majority of the total protein in the
`medium under both conditions (Fig. 5). In the production
`phase of one 500-mL culture, the standard Invitrogen pro-
`tocol was followed with daily additions of 2.5 mL of metha-
`nol. After each dose, methanol was rapidly consumed such
`that its concentration was lower than 0.3% (v/v) for more
`than half of the production phase and the methanol was
`consistently exhausted by the culture before the next sched-
`uled feeding time was reached. In response to the first
`methanol dose, recombinant protein concentration in the
`culture medium increased to 32 mg L−1. However, as the
`run progressed, hTF/2N not only ceased to accumulate, but
`its concentration actually declined, presumably due to pro-
`tein degradation in the medium. After the third dose of
`methanol, however, recombinant protein accumulation re-
`sumed but only to 82% of the hTF/2N concentration ob-
`tained after the first methanol dose. In the parallel culture,
`
`A
`
`0
`
`1 0
`
`20 30 40
`
`50
`
`60
`
`70
`
`Time (hours)
`
`B
`
`0.7
`
`---~ 0.6
`
`>
`::.-!!
`~
`C:
`
`-~ ro ...
`c Q) u
`
`C:
`0
`(.)
`
`0
`C
`(1)
`.c
`
`iii ::
`
`0.5
`
`0.4
`
`0.3
`
`0 .2
`
`0. 1
`
`0.0
`
`0.7
`
`> > 0 .6
`
`0.5
`
`~
`~
`C:
`1ii .. 0.4
`.2
`c
`Q) u
`C:
`0
`(.)
`0
`C:
`IU
`.r::
`iii
`::
`
`0.3
`
`0. 2
`
`0.1
`
`0 .0
`
`0
`
`1 0
`
`20
`
`3 0
`
`40
`
`50
`
`6 0
`
`7 0
`
`Time (hours)
`
`Figure 6. Methanol concentration versus time in two parallel P. pastoris
`shake-flask cultures. (a) the 500-mL culture was fed daily with 2.5 mL
`methanol. (b) The closed-loop control system was set to maintain the
`methanol concentration at 0.3% (v/v), as determined by in situ calibration
`of the sensor.
`
`284
`
`BIOTECHNOLOGY AND BIOENGINEERING, VOL. 56, NO. 3, NOVEMBER 5, 1997
`
`Motif Exhibit 1032, Page 6 of 8
`
`Case No.: IPR2023-00321
`U.S. Patent No. 10,689,656
`
`

`

`140
`
`A
`
`~ 120
`'E
`:,
`0
`
`100
`
`0 "'
`C
`0
`~
`'<ii
`C:
`Q)
`0
`m
`-~
`ii
`0
`
`80
`
`60
`
`40
`
`20
`
`..n- -o
`
`-
`
`-
`
`- -0- -
`
`-0"" ,; ,
`
`140
`
`120
`
`100
`
`80
`
`60
`
`40
`
`20
`
`::::;
`oi
`.§.
`C:
`·;:
`'-
`Q)
`ui
`C:
`~
`I-
`
`0
`
`0
`
`0
`1 0 20 30 40 50 60 70
`
`Time (hours)
`
`140
`
`.. ~ 120
`C:
`:,
`
`~ 100
`C
`80
`2,
`~
`'iii
`C:
`Q)
`C
`
`60
`
`40
`
`20
`
`Ill
`
`-~ ii
`
`0
`
`B
`
`140
`
`120
`
`100
`
`BO
`
`60
`
`40
`
`2 0
`
`C) ........... --0•"'°
`
`-"'
`
`0
`
`0
`
`0
`1 0 20 30 40 50 60 70
`
`Time (hours)
`
`:::;-
`oi
`.§.
`·;: ...
`
`C:
`
`Q)
`
`-
`
`(/J
`C:
`Ill
`
`i=
`
`perature, pH) over their typical respective ranges. The probe
`is autoclavable, can be calibrated in situ, and is stable for
`several days under routine operating conditions for P. pas-
`toris fermentations.
`Shake-flask cultures in which the sensor was used to
`control the methanol concentration exhibited higher recom-
`binant protein productivity than simultaneously run cultures
`carried out using the conventional protocol. These experi-
`ments also provided the first quantitative data demonstrat-
`ing the drastic variability in the methanol concentration pro-
`file that can result from the conventional protocol. The
`higher productivity observed in the controlled cultures was
`attributed to elimination of this variability.
`On-line monitoring of substrate concentration is seldom,
`if ever, used in simple shake-flask cultures and we know of
`no similar example of controlled fed-batch operation in this
`type of system. These features have previously been limited
`to much more expensive and complex bioreactor systems
`not routinely available to most investigators. The ability to
`monitor and control methanol concentration in shake-flask
`cultures will expedite the early selection and characteriza-
`tion of recombinant clones of methanol utilizing organisms
`(e.g., P. pastoris, Hansenula polymorpha).
`Although beyond the scope of this work, the methanol
`sensor described here provides a valuable tool for further
`optimization of P. pastoris expression systems, particularly
`with respect to target methanol concentrations during the
`induction phase of the culture. This sensor system can easily
`be adapted for use in applications involving larger scale
`cultures in stirred-tank bioreactors of methylotrophic organ-
`isms or for monitoring production in biomass to ethanol
`processes. Continuing work in our laboratories includes ap-
`plication of this system to Mut+ and MutS mixed-feed fer-
`mentations of P. pastoris for enhanced productivity of re-
`combinant proteins.
`
`Figure 7. Effect of on-line control on protein productivity. (s- -s) Cell
`density and (j—j) recombinant protein production were analyzed in (a)
`a culture receiving daily doses of methanol and (b) in a controlled culture
`in which the methanol concentration was maintained at 0.3% (v/v).
`
`The authors wish to thank Mr. Randy Deane for assisting with
`the fabrication of electronic components used in this work. Fi-
`nancial support was provided by the Natural Sciences and En-
`gineering Research Council of Canada, The Medical Research
`Council of Canada, and Ciba–Geigy Corporation.
`
`at 0.3% (v/v) resulted in continuous culture growth and
`hTF/2N production, reaching levels fivefold higher than in
`the flask that received daily methanol feedings.
`
`CONCLUSIONS
`
`A practical and effective system for on-line monitoring and
`control of methanol concentrations in P. pastoris shake-
`flask cultures has been presented. The sensor element, based
`on a commercially available organic vapor detector and a
`custom designed gas-permeable methanol probe, has been
`optimized with respect to performance-determining physi-
`cal parameters (e.g., carrier stream flow rate, tubing length)
`and characterized as a function of the key fermentation en-
`vironmental variables (e.g., methanol concentration, tem-
`
`References
`
`Brierley, R. A., Bussineau, C., Kosson, R., Melton, A., Siegel, R. S. 1990.
`Fermentation development of recombinant Pichia pastoris expressing
`the heterologous gene: Bovine lysozyme. Ann. NY Acad. Sci. 589:
`350–362.
`Buckholz, R. G., Gleeson, M. A. G. 1991. Yeast systems for the commer-
`cial production of heterologous proteins. Bio/Technology 9:
`1067–1072.
`Cregg, J. M., Vadvick, T. S., Raschke, W. C. 1993. Recent advances in the
`expression of foreign genes in Pichia pastoris. Bio/Technology 11:
`905–910.
`Faber, K. N., Harder, W., Ab, G., Veenhuis, M. 1995. Review: Methylo-
`trophic yeasts as factories for the production of foreign proteins. Yeast
`11: 1331–1344.
`Figaro Engineering Inc. 1990. Figaro gas sensors. Figaro Products Catalog
`pp. 2–12.
`
`GUARNA ET AL.: METHANOL IN PICHIAPASTORALIS CULTURES
`
`285
`
`Motif Exhibit 1032, Page 7 of 8
`
`Case No.: IPR2023-00321
`U.S. Patent No. 10,689,656
`
`

`

`Gentry, S. J., Jones, T. A. 1986. The role of catalysis in solid-state gas
`sensors. Sensors Actuators 10: 141–163.
`Heiland, G. 1982. Homogeneous semiconducting gas sensors. Sensors Ac-
`tuators 2: 343–361.
`Jones, T. A., Firth, J. G., Mann, B. 1985. The effect of oxygen on the
`electrical conductivity of some metal oxides in inert and reducing
`atmospheres at high temperature. Sensors Actuators 8: 281–306.
`Kilburn, D. G. 1967. Some factors affecting growth of mammalian cells in
`suspension cultures. Ph.D. thesis. Department of Chemical Engineer-
`ing, University College, London, Chap. 2.
`Mason, A. B., Woodworth, R. C., Oliver, R. W. A., Green, B. N., Lin,
`L.-N., Brandts, J. F., Tam, B. M., Maxwell, A., MacGillivray, R. T. A.
`1996. Production and isolation of the recombinant N-lobe of human
`serum transferrin from the methylotrophic yeast Pichia pastoris. Pro-
`tein Expr. Purif., to appear.
`Morrison, S. R. 1982. Semiconductor gas sensors. Sensors Actuators 2:
`329–341.
`Phillips, D. H., Johnson, M. J. 1961. Measurement of dissolved oxygen in
`fermentations. J. Biochem. Microb. Technol. Eng. 111: 261–