Journal of
`Applied Bacteriology
`
`Edited by
`
`D.E. Stewart-Tull, G.I. Barrow
`
`and R.G. Board
`
`Volume 72, 1992
`
`UCB Biopharma SPRL (IPR2019-00400)
`Exhibit 2004
`Page 1
`
`UCB Biopharma SPRL (IPR2019-00400)
`Exhibit 2004
`Page 1
`
`

`

`Copyright © 1992 by The Society for Applied Bacteriology
`ALL RIGHTS RESERVED
`
`Nopart of this volume maybe reproduced in any form,by
`photostat, microfilm, or any other means, without written
`permission from the Society
`
`ISSN 0021-8847
`
`Published by
`Blackwell Scientific Publications Ltd
`OXFORD
`LONDON EDINBURGH BOSTON
`MELBOURNE
`PARIS
`BERLIN VIENNA
`
`Printed in Great Britain
`
`UCB Biopharma SPRL (IPR2019-00400)
`Exhibit 2004
`Page 2
`
`UCB Biopharma SPRL (IPR2019-00400)
`Exhibit 2004
`Page 2
`
`

`

`The bactericidal activity of a methyl and propyl parabens
`combination: isothermal and non-isothermal studies
`
`D. Gilliland, A. Li Wan Po and E. Scott
`The Drug Delivery Research Group, The School of Pharmacy, The Queen's University of Belfast, Northern Ireland
`
`3739/07/91: accepted 20 September 1991
`
`D. GILLILAND, A. L! WAN PO AND E. SCOTT. 1992. The effect of temperature on thekill rate of
`Escherichia coli by methyl and propyl parabens was studied. Thekill kinetics was first order.
`It was shownthat the Arrhenius equation provided a good model for describing the
`relationship between the first order rate constant and the temperature. The activation energy
`was found to be 274 kJ/mol for exponential phase cells and 168 kJ/mol for stationary phase
`cells. Exponential phase cells were much moresusceptible to the lethal effects of the parabens
`than werethe stationary phase cells. For example, at 34°C stationary phasecells, in
`chemically defined media, had a kill rate constant of 0-072/h while the corresponding value
`for exponential phase cells was 0-238/h. In water the rate of kill for exponential phasecells
`was even faster giving a rate constant of 5-25/h at 34°C. Non-isothermal kinetic testing was
`not found to be useful for modelling bacterial kill kinetics because we could not achieve the
`precision required in bacterial enumeration.
`
`INTRODUCTION
`
`In the presence of a bactericidal antimicrobial agent the
`rate of kill of microbial cells generally increases as the tem-
`perature increases (Lynn & Hugo 1983). The effect of tem-
`perature is often expressed in terms of a temperature
`coefficient (Pflug 1972), usually measured as the change in
`rate constant over a 10°C increase in temperature and
`referred to as the Q,,. Those compounds with high tem-
`perature coefficients exhibit greater increases in activity
`with increasing temperature. However,
`the temperature
`coefficient value tends to vary over the temperature range
`studied, with decreasing values at higher
`temperature
`ranges (Karabit ef a/. 1985). The Q. 19 values also vary from
`one organism to another for the same antimicrobial agent
`(Karabit et al. 1986).
`Both isothermal and non-isothermal methods were used
`to ascertain the effect of temperature on bacterialkill rates.
`
`THEORY
`
`Under standardized conditions bacterial death is often
`exponential as described by eqn.(1)
`
`N=WNoe *"
`
`(1)
`
`Correspondence to: Prof. A. Li Wan Po, The Drug Delivery Research
`Group, The School ofPharmacy, The Queen's University of Belfast, 97
`Lisburn Road, Belfast BT9, Northern Ireland.
`
`where N is the number of organisms surviving at time ¢,
`No is the initial number of organisms and k,
`is the rate
`constant describing thekill.
`One of the major factors affecting the kill rate of bacteria
`is the temperature. To describe the effect of temperature
`many authors have used the Arrhenius equation (eqn. (2))
`
`k = AeFe/(®T)
`
`(2)
`
`wherek is the rate constant, R is the gas constant, T is the
`temperature in degrees Kelvin, A is the pre-exponential
`constant and £,is the activation energy.
`Totest the validity of the Arrhenius equation, adherence
`of the kill curve to an appropriate rate equation is first
`established. Most commonly,
`the first order model (eqn.
`(1)) is appropriate. If validated, experimental verification of
`the Arrhenius equation then involves calculation of appro-
`priate rate constants (k) at a numberof different
`tem-
`peratures (7) and their statistical evaluation. Generally, the
`linear form of eqn. (2) is used and adherence to the model
`is shown bya linear relationship betweenlog, (k) and 1/7.
`Many authors have shown that when the range of tem-
`perature is wide, marked deviations from linearity are
`observed when log,(k) and 1/7 are plotted. Models put
`forward
`to
`represent
`the
`rate
`constant—temperature
`relationship in such cases have included the square root
`model (Ratkowsky et a/. 1983) and the Schoolfield model
`(Schoolfield et a/. 1981) represented by eqns. (3) and (4),
`respectively.
`(hE TtareTy
`
`UCB Biopharma SPRL (IPR2019-00400)
`Exhibit 2004
`Page 3
`
`UCB Biopharma SPRL (IPR2019-00400)
`Exhibit 2004
`Page 3
`
`

`

`BACTERICIDAL EFFECT OF PARABENS 253
`
`where T is the temperature in degrees Kelvin, T,,;, is the
`theoretical minimum temperature for growth, T.,,, is the
`theoretical maximum temperature for growth, b and c are
`constants and k is the rate constant.
`
`log. k = A + B/T — log, T
`(4)
`+ log, [1 + eft’ 4 efGtHITD)
`where A, B, D, F, G and H are constants, k is the rate
`constant and 7 is the temperature in degrees Kelvin.
`Data for testing temperature—kill rate relationships are
`collected as described above (eqns (2)-(4)) and each kill
`curve is
`recorded at a constant
`temperature. Such an
`experimental set-up is referred to as isothermal testing. An
`approach which has gained some measure of acceptance in
`the stability testing of pharmaceutical products is that of
`non-isothermal testing. In this case temperature is varied
`continuously during the experiment so that,
`theoretically,
`temperature effects on kill rates can be derived from one
`single experiment. The logic behind the method is as
`follows.
`Suppose that from an experiment, carried out at constant
`temperature, the kill curve adheres to eqn (1). A plot of
`log, (N) or log, (N/No) against time will be linear. If we
`now increase the temperature during the experimental run,
`the line will usually curve down as shown in Figs.
`la and
`lb. Starting the experiment below the optimum tem-
`perature for growth and increasing the temperature during
`the experiment will slow down the observed kill rate if the
`temperature effect on growth rate exceeds that on the
`preservative-induced kill rate. The two effects will counter-
`act each other.
`In that case the kill plot will show an
`upwardcurve.
`A curve such as that shown in Fig. 1b can usually be
`satisfactorily modelled by a low order polynomial equation
`which can be written as
`
`(5)
`loge(N) = f(t) = ao + ayt + age? + °°
`The derivative of this equation gives the rate constant k at
`the prevailing temperature 7 at time ¢
`
`d [log, (N)]/dt =k; =a, +2a,t+°::
`
`(6)
`
`Therefore, provided we have the temperature at time #,
`the corresponding rate constant can be calculated. The
`
`appropriate calculations can beeasily done using both stan-
`dard statistical computer packages and non-isothermal spe-
`cific programs (Li Wan Po et a/. 1983). More detailed
`descriptions of non-isothermalstability testing methodology
`are given elsewhere (Hempenstall et a/. 1983).
`
`MATERIALS AND METHODS
`
`Preparation of media
`
`(g/l):
`defined medium contained
`chemically
`The
`Na,HPO,, 11-45; KH,PO,, 1-4025; (NH,),SO,, 1-87;
`MgSO,,
`0-187; D-glucose,
`0-909; CaCl,
`(2H,0),
`1-245 x 107°; FeSO,(7H,O),
`5 x 1077. The pH was
`adjusted to 6:9 with dilute HCI. All chemicals were of ana-
`lytical reagent quality.
`
`Preparation of the inoculum of exponential phase
`cells
`
`Escherichia coli NCIB 8545 was maintained on Tryptone
`Soya Agar (Oxoid) slopes at 4°C. A loopful of the organism
`was added to 100 ml ofsterile media and grown overnight
`at 37°C in a shaking waterbath at 100 rev/min. Transfers of
`organism were madedaily for 2 d. On the third transfer the
`organisms were allowed to grow to an optical density
`reading of 0-1 at 540 nm (Corning colorimeter 254). This
`provided cells in the exponential phase of growth. The
`absorbance value of 0-1 at 540 nm was found to be approx-
`imately equal to 1 x 10° cfu/ml. An inoculum of 1 x 10°
`cfu/ml was prepared by filtering the culture (100 ml),
`under aseptic conditions, through a 0:45 ym membrane and
`washing with 100 ml of fresh, pre-warmed media. The
`organisms were then resuspended in 10 ml of media to give
`the final inoculum.
`
`Preparation of the inoculum of stationary phase cells
`
`The same procedure was carried out for the preparation of
`exponential phase cells except that instead of preparing a
`third transfer the overnight cells were used to prepare the
`
`Fig. 1 Theoretical first order kill curve for
`micro-organismsat (a) constant
`temperature and (b) when the temperature
`is increased continuously throughout the
`experiment
`
`(a)
`
`Lncfu/ml——
`
`(b)
`
`—_—r
`Lncfu/ml
`
`Time
`
`.
`
`Time ———
`
`UCB Biopharma SPRL (IPR2019-00400)
`Exhibit 2004
`Page 4
`—|
`
`UCB Biopharma SPRL (IPR2019-00400)
`Exhibit 2004
`Page 4
`
`

`

`254 D. GILLILAND ET AL.
`
`
`inoculum. The culture was filtered, under aseptic condi-
`tions, through a 0-45 zm membranefilter and washed with
`100 ml of pre-warmed media. The organisms were then
`resuspended in 10 ml of media and the density adjusted so
`that a 1: 10 dilution gave an absorbance reading, at 540
`nm, of approximately 0-1.
`
`Preparationof test solutions
`
`The appropriate weight of the methyl and propyl esters of
`p-hydroxybenzoic acid (Sigma) were added to 1
`1 of
`medium,or water, and placed in a sonic bath for up to 4h
`to aid solubilization. The solution was filter-sterilized and
`100 ml oftest solution dispensed into 250 mlflasks. Media
`without any parabens were employed as controls. Before
`inoculation,
`the test solutions were maintained at
`their
`respective test temperatures in a shaking waterbath for at
`least 18 h.
`
`Test procedure
`
`‘
`
`.
`
`.
`
`Lncfu/ml
`
`' ae
`
`16
`
`15
`
`Time (min)
`
`Fig. 2 Comparison ofthe rate of growth of exponential phase
`Escherichia colt cells in (], chemically defined media and @,
`water. (Error bars are s.£. for 3 days’ results.)
`
`In the experiments involving kill at various constant tem-
`peratures the test solutions were maintained at each tem-
`perature in separate shaking waterbaths (Grant SS40-D).
`In experiments involving a gradual increase in temperature
`over time the test solution was maintained in a Grant W14
`(1) and a positive rate constant was obtained (Fig. 2). As
`waterbath connected to a Grant temperature programmer
`expected, when the growth medium was substituted with
`PZ1. Since the Grant W14 waterbath is not a shaking
`water, no growth took place (Fig. 2). With 0:12% w/v
`waterbath, small sterile, magnetic teflon-coated fleas were
`methyl paraben and 0-012% w/v propyl paraben added to
`included in the test solutions and werestirred on an Inspin
`the growth medium bacterial kill was observed.
`2 (Baird and Tatlock). The temperature programmer was
`The kill curve could be satisfactorily modelled by first
`set to increase in temperature at the rate of 1°C/h, com-
`order kinetics (Fig. 3) as shown by the linear semi-
`mencing at 34°C andrising to 42°C. This was closely mon-
`logarithmic plot of the number of surviving organisms
`itored by a built-in thermometer and the inclusion of a
`against
`time. Figure 3 also shows that
`the rate ofkill
`thermometer in a control flask which was present in the
`increased as the temperature increased. Table1lists the kill
`waterbath.
`rate constants corresponding to the four different
`tem-
`peratures studied.
`
`Measurementof microbial numbers
`
`inoculation of the test solutions viable counts of
`After
`microbial numbers were madeat regular intervals. At each
`timeinterval a 1 ml sample was removed from thetest solu-
`tion and serial 10-fold dilutions were made in 0-1%
`peptone water. One ml volumes of the dilutions were
`plated, by the pour plate method, with Isosensitest agar
`(Oxoid). After incubation at 37°C for 20 h colonies were
`counted and the numberof cfu/ml evaluated.
`
`RESULTS AND DISCUSSION
`
`When exponential phase cells were placed in their growth
`medium the growth could be satisfactorily described by eqn
`
`Table 1 The effect of temperature on thekill rate constants for
`inocula prepared from exponential and stationary phase
`Escherichia coli cells in chemically defined media in the presence
`of 0:12% w/v methyl paraben and 0-012%w/v propyl paraben
`
`Rate constant/h
`(mean +5s.E.)
`
`Temperature(°C)
`
`Exponential
`phase cells
`
`Stationary
`phase cells
`
`—0-072 + 0-016
`—0-243 + 0-026
`34
`—0-141 + 0-027
`—0-510 + 0-008
`37
`—0:243 + 0-026
`— 1-546 + 0-173
`40
`
`42 — 3-442 + 0-097
`
`
`
`UCB Biopharma SPRL (IPR2019-00400)
`Exhibit 2004
`Page 5
`
`UCB Biopharma SPRL (IPR2019-00400)
`Exhibit 2004
`Page 5
`
`

`

`BACTERICIDAL EFFECT OF PARABENS 255
`
`
`18
`
`I7
`
`Lncfu/ml io
`
`0
`
`100
`
`200
`
`300
`
`Time (min)
`Fig. 3 Rate ofkill ofan inoculum ofexponential phase cells of
`Escherichia coli in chemically defined media in the presence of
`0-12% methyl +0-012% propyl parabens at 11, 34°C; @, 37°C;
`Wi,40°C and ©, 42°C.(Error barsare s.£. for 3 days’ results.)
`
`The plot of the logarithm ofthe first order rate constant
`against the reciprocal of the temperature was linear (Fig. 4)
`thus showing that the Arrhenius equation provided a good
`model
`for
`the temperature effect on the kill
`rate and
`obviated the need to use more highly parameterized equa-
`tions or equations which are not transformedinto a linear
`form. Simple linear regression provided an estimate for the
`activation energy of 274 kJ/mol.
`To investigate whether the activation energy was affected
`by the state in which the cells were in,
`the experiments
`were repeated using stationary phase cells instead of expo-
`nential phase cells. The data showed that the kill rate was
`first order (Fig. 5) and the Arrhenius plot was again linear
`(Fig. 6). The results were surprising in terms of how mark-
`edly different the kill rate constants for the stationary phase
`cells were from those relating to exponential phase cells
`(Table 1). The kill rate constants were in termsof hours for
`stationary phase cells and minutes for exponential phase
`cells. The activation energy (mean +5.£.) for the kill of
`stationary phase cells was low relative to exponential phase
`cells, 168 (34) kJ/mol vs 274 (17) kJ/mol. The practical
`implications of these observations are that stationary phase
`cells are much more resistant
`to the preservative com-
`bination at any given temperature than are exponential
`phase cells; a feature which must be borne in mind during
`pasteurization orsterilization processes.
`
`
`
`
`
`Lnrateconstant(/min) |
`
`b eae
`
`+
`
`Lncfu/ml
`
`?
`
`|ie
`
` SEeeee i
`
`
`0-0032
`
`0-0033
`
`00-0032
`
`0-0032
`
`0-0032
`Q-0032
`1/7 (K)
`Fig. 4 Arrheniusplot for the rate ofkill of an inoculum of
`exponential phase Escherichia coli cells in chemically defined
`media in the presence of 0-12% methyl +0-012%propyl
`parabens. (Error bars are S.E., 1 = 3)
`
`20
`
`25
`
`30
`
`QO
`
`5
`
`10
`
`15
`Time (h)
`Fig. 5 Rate ofkill of an inoculum ofstationary phase Escherichia
`coli cells in chemically defined media in the presence of 0-12%
`methyl +0-012% propyl parabens at 1], 34°C; @, 37°C and Mi,
`40°C. (Errorbars are s.£. for 3 days’ results.)
`UCB Biopharma SPRL(IPR2019-00400)
`Exhibit 2004
`Page 6
`lil
`
`UCB Biopharma SPRL (IPR2019-00400)
`Exhibit 2004
`Page 6
`
`

`

`(°C) 34
`
`—3:5piafsShed
`O
`100
`200
`300
`400
`500
`90-0032
`0:0032
`0-0032
`QO: 0032
`0:-0033
`Time (min)
`1/7 (K)}
`
`Fig. 6 Arrheniusplot for the rate of kill of an inoculum of
`stationary phase cells of Escherichia coli in chemically defined
`media in the presence of 0-12% methyl +0:012% propyl
`parabens. (Error barsare S.E., 7 = 3.)
`
`Fig. 7 Rate ofkill of an inoculum of exponential phase
`Escherichia coli cells (+) in chemically defined media in the
`presence of 0:12%methyl +0-012% propylparabens during a
`gradual temperature increase from 34 to 42°C (CO). (Error bars
`are S.E.,n = 3.)
`
`Comparison of our activation energies with those in the
`literature indicates that the high values observed,
`in our
`study with the parabens and E£.colt, are of the same order
`of magnitude as those reported for other preservatives such
`as phenol, benzalkonium chloride and benzyl alcohol with a
`range of organisms
`(Table 2). Karabit
`et al.
`(1989),
`however, reported much more rapid kill than was observed
`in our studies. This can be explained by the fact that these
`authors used different preservative agents and experiments
`were carried out in standard phosphate buffers which are
`
`Table 2 Reported activation energies (Ea) for the effect of a
`series of antimicrobial agents
`
`Ea (kJ/mol)
`
`b
`
`a
`Micro-organism
`65
`45
`Aspergillus mger
`150
`59
`Candida albicans
`78
`93
`Escherichia colt
`84
`72
`Pseudomonas aeruginosa
`81
`91
`Staphylococcus aureus
`a, 0:5% phenol at pH 6:1 in buffer (Karabit et al. 1985).
`b, 1% benzyl alcohol in pH 7:1 buffer (Karabit et a/. 1986).
`c, 0-001%benzalkonium chloride in pH 6:1 buffer except for Asp.
`niger for which 0:014% benzalkonium chloride was used (Karabit
`et al. 1988).
`
`c
`134
`77
`86
`76
`92
`
`not expected to be suitable for growth. To investigate this
`aspect further the kill rate of exponential phase E. colt was
`measured in sterile water where there wasstill no growth.
`Much morerapid kill was observed in this medium than in
`the growth medium. Indeed bacterial kill was so rapid that
`the temperature had to be lowered from the range 34-42°C
`to the range 26-5-34°C to obtain practically measurable kill
`rates, At
`these lower temperatures the rates were 1-08/h
`(265°C), 2:15/h (29°C), 3-86/h (31°C) and 5-25/h (34°C).
`The activation energy was 164 + 26 kJ/mol. Orth (1979)
`found, when calculating D-values for Staphylococcus aureus
`in a lotion, that the addition of brain heart infusion broth
`increased the D-value. It was suggested that
`the broth
`might have protected thecells by inactivating some of the
`preservative or that
`the broth supplied the bacteria with
`nutrients which allowed the organism to be less susceptible
`to the strain imposed by the preservative system.
`The results discussed so far show that, within the tem-
`perature range studied,
`the Arrhenius equation satisfact-
`orily modelled
`the
`kill
`rate
`constant-temperature
`relationship and interpolation within the range is justifiable.
`Under these conditions, non-isothermalkinetic modelling is
`often worthwhile. To investigate whether this approach,
`which has been used successfully to model chemical decom-
`position,
`is applicable to microbial kill kinetics we under-
`took a series of experiments in which the temperature of
`the preservative test medium was continuously altered.
`Figure 7 summarizes ourfindings.
`UCB Biopharma SPRL (IPR2019-00400)
`Exhibit 2004
`Page 7
`
`40
`
`38
`
`Temperature
`
`Hh
`
`Lncfu/ml w
`
`—|-5
`
`ce
`
`= -2:0
`
`3 2oo
`
`O 2O
`
`c
`af]
`
`BS
`
`-3:0
`
`UCB Biopharma SPRL (IPR2019-00400)
`Exhibit 2004
`Page 7
`
`

`

`BACTERICIDAL EFFECT OF PARABENS 257
`
`
`
`
`enaerTITTEEE=STIIIETEISENTIEEEEETTTREESERSIESSEESEERE
`
`The temperature increase was linear with time but, as
`can be seen,
`the viable count showed increasingly large
`variability. Indeed that variability was too high to enable
`adequately precise rate constants to becalculated.
`It would therefore appear
`that although the non-
`isothermal method looked promising,
`the difficulty intro-
`duced by wide biological variation and low precision in
`bacterial enumeration,
`relative to chemical analysis, has
`contributed to making this method of little value in this
`instance.
`
`REFERENCES
`
`HEMPENSTALL, J.M., IRwin, W.J., L1 Wan Po, A. &
`AnpREws, A.H.
`(1983) Non-isothermal kinetics using a
`microcomputer: a derivative approach to the prediction of the
`stability of penicillin formulations. Journal of Pharmaceutical
`Sciences 72, 668-673.
`KaraBit, M.S.,
`JUNESKANS, O.T. & LUNDGREN, P.
`(1985) Studies on the evaluation ofpreservative efficacy I. The
`determination of antimicrobial characteristics of phenol. Acta
`Pharmaceutica Suecica 22, 281-290.
`KaraBIT, M.S.,
`JUNESKANS, O.T. & LUNDGREN, P.
`(1986) Studies on the evaluation of preservative efficacy II. The
`determination of antimicrobial characteristics of benzylalcohol.
`Journal of Clinical and Hospital Pharmacy 11, 281-289.
`Karasit, M.S.,
`JUNESKANS, O.T. & LUNDGREN, P.
`(1988) Studies on the evaluation of preservative efficacy III.
`
`The determination of antimicrobial characteristics of benz-
`alkonium chloride. International Journal of Pharmacy 46, 141—
`147.
`JUNESKANS, O.T. & LUNDGREN, P.
`Karapit, M.S.,
`(1989) Studies on the evaluation of preservative efficacy IV.
`The determination of antimicrobial characteristics of some
`pharmaceutical compounds in aqueous solutions. International
`Journal ofPharmaceutics 54, 51-56.
`Li Wan Po, A., Evias, A.N. & IRWIN, W.J. (1983) Non-
`isothermal and non-isopH kinetics in formulation studies. Acta
`Pharmaceutica Suecica 20, 277-286.
`Lynn, B. & Huco, W.B. (1983) Chemical disinfectants, anti-
`septics and preservatives. In Pharmaceutical Microbiology, 3rd
`edn. Hugo, W.B. & Russell, A.D. Oxford: Blackwell Scientific
`Publications.
`OrtnH, D.S. (1979) Linear regression method for rapid determi-
`nation of cosmetic preservative efficacy. Journal of the Society
`of Cosmetic Chemists 30, 312-332.
`Prius, I.J. (1972) Heat sterilization. In Industrial Steriltsation :
`International Symposium, Amsterdam, 1972, Ch. 14. ed. Phillips,
`G.B. & Miller, W.S. North Carolina: Duke University Press.
`Ratkowsky, D.A., Lowry, R.K., McMEEKIN, T.A.,
`Stokes, A.N. & CHANDLER, R.E. (1983) Models for bac-
`terial culture growth rate throughout the entire biokinetic tem-
`perature range. Journal of Bacteriology 154, 1222-1226.
`SCHOOLFIELD, R.M., SHARPE, P.J.H. & MAGNUSON,
`C.E.
`(1981) Non-linear regression of biological temperature-
`dependent rate models based on absolute reaction-rate theory.
`Journal of Theoretical Biology 88, 719-731.
`
`UCB Biopharma SPRL (IPR2019-00400)
`Exhibit 2004
`Page 8
`a
`
`UCB Biopharma SPRL (IPR2019-00400)
`Exhibit 2004
`Page 8
`
`