Electronic Journal of Polish Agricultural Universities (EJPAU) founded by all Polish Agriculture Universities presents original papers and review articles relevant to all aspects of agricultural sciences. It is target for persons working both in science and industry,regulatory agencies or teaching in agricultural sector. Covered by IFIS Publishing (Food Science and Technology Abstracts), ELSEVIER Science - Food Science and Technology Program, CAS USA (Chemical Abstracts), CABI Publishing UK and ALPSP (Association of Learned and Professional Society Publisher - full membership). Presented in the Master List of Thomson ISI.
Volume 11
Issue 1
Sip A. , Dembczyński R. , Białas W. , Juzwa W. , Czaczyk K. , Grajek W. , Jankowski T. 2008. SCALE-UP OF DIVERCIN PRODUCTION BY CARNOBACTERIUM DIVERGENS AS7 IN STIRRED TANK REACTORS, EJPAU 11(1), #04.
Available Online: http://www.ejpau.media.pl/volume11/issue1/art-04.html


Anna Sip1, Radosław Dembczyński1, Wojciech Białas1, Wojciech Juzwa2, Katarzyna Czaczyk1, Włodzimierz Grajek1, Tomasz Jankowski1
1 Department of Biotechnology and Food Microbiology, Poznań University of Life Sciences, Poland
2 Department of Biotechnology and Food Microbiology, The August Cieszkowski Agricultural University of Poznań, Poland



Divercin is bacteriocin which is very effective against Listeria sp. In order to study the feasibility of commercialising the production of divercin by Carnobacterium divergens AS7 strain using a large scale bioreactor, a scale-up study from 5 l to 1500 l was carried out using constant agitation power per unit of liquid volume as the scale-up criterion. The maximum specific growth rate as well as the maximum viable cell concentration decreased as the bioreactor scale increased, while enhanced divercin release to the fermentation medium was achieved in 1500 l scale comparing to the reference 5 l bioreactor. This was attributed to nutrient depletion in the large scale bioreactor caused by degradation of the culture medium compounds during the heat sterilisation cycle.

Key words: scale-up, bacteriocin, divercin, fermentation, bioreactor.

List of symbols
DI – impeller diameter (m)
DT – bioreactor vessel diameter (m)
HL – liquid height (m)
HT – bioreactor total height (m)
N – agitation speed (s-1)
NP – power number (= Po ρ-1N-3DI-5)
Po – ungassed agitation power (W)
Re – Reynolds number (= ρND2µ-1)
VL – working volume of bioreactor (l)
µ – broth viscosity (kg m-1s-1)
ρ – broth density (kg m-3)


Recently, increased interest has been observed in using bacteriocins as natural antimicrobial agents and food preservatives. Bacteriocins are proteinaceus compounds produced by strains of certain bacteria and exhibit activity against specific pathogenic bacteria [4,20]. These antimicrobial agents added to foods alone or in combination with traditional and modern preservation techniques can eliminate foodborne pathogens and enhance the safety and shelf life of foods [1,13]. The ability to synthesise bacteriocins is a common feature of many lactic acid bacteria [18]. Among the bacteriocins produced by lactic acid bacteria, divercin synthesised by Carnobacterium divergens appears to be very promising as a food preservative. It is a nonlantibiotic, class IIa, heat-stable bacteriocin, very effective against Listeria monocytogenes, Listeria innocua and Clostridium tyrobutyricum [19]. Studies on divercin synthesis by C. divergens carried out in a small bioreactor scale (<5 l volume) were encouraging and indicated divercin activity in the growth medium as high as 107 AU ml-1 [22,23]. Moreover, it was shown that the biomass of C. divergens isolated from the culture and added to fish diet, had a beneficial health effect on common carp larvae, reducing their mortality and increasing the mean weight [25]. In another our work [data not published], the optimisation of media compositions and process conditions for divercin biosynthesis by C. divergens AS7 was presented in the 5 l scale. In this study, we aimed at scaling-up divercin production from the small, laboratory 5 l bioreactor to a 30 l pilot scale bioreactor, and finally to a 1500 l production-scale system.


For divercin production, C. divergens AS7, a lactic acid bacteria strain from the departmental collection, previously isolated from the salmon digestive tract has been used [23]. Indicator bacteria, Listeria innocua F (E.N.I.T.I.A.A. Nantes, France), used for the determination of divercin activity, has been cultured at 30°C in a medium containing 1% glucose, 1% sodium chloride and 0.5% yeast extract.

Culture medium and method
During bacteria propagation and fermentation stages of C. divergens AS7, a culture medium with the following composition was used: 10 g of glucose (POCH, Poland), 5 g of peptone, 5 g of yeast extract, and 1 g of nutrient broth (all BTL, Poland) per litre of deionised water. Trace salts used in a culture medium were: 2 g of K2HPO4, 2 g of (NH4)2C6H6O7, 5 g of CH3CO2Na×3H2O, 0.57 g of MgSO4×7H2O, and 0.14 g of MnSO4×5H2O (all POCH, Poland) per liter. In the 5 l bioreactor the culture medium was autoclaved at 121°C for 30 min. The media in the 30 l and 1500 l bioreactors were sterilized in-place via steam jacket according to the bioreactors’ manufacturer’s specifications at a medium holding temperature of 124°C and a holding time of 35 and 45 min respectively. Cultures were performed batch-wise in semi-anaerobic conditions. For the 5 l laboratory bioreactor and 30 l pilot plant bioreactor, a fresh medium was inoculated with 2% of an exponential culture grown at 30°C for 24 h in Erlenmeyer flasks. For the 1500 l bioreactor, the size of the inoculum was 2% of its working volume and was taken from 24 h culture prepared in the 30 l scale. The pH was automatically controlled at 6.5 by injection of 5 M sodium hydroxide solution and the temperature was kept constant at 30°C. Samples of fermentation broth were withdrawn in specified periods of time to perform biomass viability and concentration, antibacterial activity assays, and glucose and lactic acid concentrations.

Determination of biomass
Viable cell concentration in fermentation broth was determined in triplicate by plating an appropriate cell suspension onto a MRS agar medium and counting after incubation for 72 h at 30°C and expressed as colony forming units (CFU) per ml.

Determination of physical properties of fermentation broth
The viscosity of the fermentation broth was measured for the samples taken from the 5 l laboratory bioreactor at the beginning and at the end of fermentation using a Reostres 1 viscometer (Haake, Germany) at 30°C, applying coaxial cylinder geometry DG 41. The density of the fermentation broth was estimated by glass picnometry at 30°C using water as a reference.

Antibacterial activity assay The activity of divercin was determined by the method of critical dilutions, described by Pilet et al. [19] using L. innocua F as a target organism in triplicate. The antibacterial activity was expressed as Activity Units (AU ml-1) which denote the lowest concentration of a sample which does not exhibit the ability to inhibit growth of indicator bacteria. The activity of divercin was separately tested in the fermentation broth (assayed as total divercin) and in the supernatant of the fermentation broth after biomass separation (assayed as free divercin). Before the assays, the samples were heated at 80°C for 15 min to inactivate proteolytic enzymes.

Determination of glucose and lactic acid
Glucose and lactic acid in the fermentation broth were quantified with HPLC (Merck-Hitachi) using an Aminex HPX-87H column (Bio-Rad) at 30°C. The eluent, 0.005 M sulphuric acid, was used at a flow rate of 0.6 ml min-1. The samples were passed through a 0.22 µm Millipore filter and loaded onto the column at a volume of 50 µl.

The reference, laboratory scale fermentations were performed in a conventional 5 l stirred tank bioreactor Bioflo III (New Brunswick Sci. Edison, N.J., USA). The larger scale processes were performed in a 30 l M-1136 Micros System bioreactor and a 1500 l IF-1500 bioreactor (both New Brunswick Sci. Edison, N.J., USA). The specifications of the bioreactors are summarised in Table 1.

Table 1. Specifications of the bioreactors


Scale (nominal volume)

5 l

30 l

1500 l

Working volume
Total height
Impeller type
Impeller diameter
Number of impellers
Bottom geometry

Rushton turbine

Rushton turbine

Rushton turbine


Scale-up of bioreactors
The objective of the scale-up of the bioprocesses is to reproduce at a larger scale the behaviour of fermentation performed and optimised in a small-scale model. This can be achieved by using one or more scale-up methods and approaches proposed by various authors [6,9,11,21]. The exact strategy used for a scale-up largely depends on the process conditions, type of bioreactor used and whether preliminary data exist to show that the procedure chosen is applicable [12].

A pre-requisite for applying established scale-up procedures is the geometric similarity of bioreactors [9]. Geometric similarity is expressed by the equation:


It also assumes similar impeller geometry and the number of impellers in scaled-up bioreactors as well as several dimensionless parameters, such as the impeller diameter ratio (DI/DT), total height ratio (HT/DT), and the liquid height ratio (HL/DT). Geometrical comparisons of laboratory and pilot scale bioreactors used in this study are shown in Table 2. The impeller diameter ratio of the 5 l and 30 l scales differed from that of the 1500 l scale, while the total height ratios were comparable. Although the impeller diameter ratio equal to 0.33 is the most frequently used in the studies of kinematics similarity of stirred vessels [2], a range of 0.3-0.45 was proposed by Einsele [8] as an accepted equivalent.

Table 2. Geometric comparisons for laboratory and pilot scale bioreactors

(nominal volume)



geometric similarity at VL

5 l (basis)
30 l
1500 l



1.0 at 5 l
0.97 at 22 l
0.94 at 1200 l

As shown in Table 2 the geometric similarity parameter calculated for the actual working volume of a particular vessel decreased as the scale increased. In order to bring the geometric similarity parameters of the larger bioreactors to that of the 5 l vessel, the working volumes of 30 l and 1500 l scale were decreased to 20 l and 1000 l respectively (Table 3). The result was comparable values of liquid height ratios for all bioreactors used.

Table 3. Scale-up of bioreactors based on geometric similarity and constant power per unit volume

(nominal volume)

volume actual (l)

volume used (l)


Impeller speed

Reynolds number*
Re × 10-4

5 l (basis)
30 l
1500 l






*The Reynolds number was calculated for the mean broth density ρ = 1022 kg·m-3 and the mean broth viscosity µ = 1.05 × 10-3 Pa·s on the samples taken from bioreactor at the beginning and at the end of fermentation at the 5 l scale.

The next step in the scale-up procedure was to provide hydrodynamic similarity in the stirred bioreactors. In the present work we applied a frequently used scale-up rule in ungassed stirred vessels, assuming equal agitation power per unit of liquid volume [2]. Equal agitation power (Po/VL) for geometrically similar bioreactors is expressed as:


where c was found to be 0.37 in practice, based on a survey of industrial plants of various scales using a variety of processes [8]. In a turbulent flow (Re > 10,000), the ungassed power input is given by the equation:


Since the power number generally remains constant with scale-up for a similar impeller and vessel design [12], the equation (2) can be rearranged to:


The above equation was applied for the calculation of the impeller speed at the 30 l and 1500 l scale using the 5 l scale as a basis (Table 3). As expected, the consequence of matching constant power per volume in three scales was the increase of the Reynolds number in the 30 l and 1500 l bioreactors.

Results of fermentation scale-up
Divercin synthesis scale up was performed on the basis of the working conditions determined for the 5 l bioreactor (N = 1.67 s-1, ungassed), keeping a constant Po/VL scale-up criterion. After the scale-up procedure, new fermentations were made at greater scales to validate the growth results and divercin synthesis by the C. divergens AS7. According to the scale-up results the impeller rotation speed was set to 1.17 and 0.50 s-1 for the 30 and 1500 l scale respectively, while the nutrient broth composition, the process temperature and the pH were the same as at the 5 l scale.

The effects of the scale-up on the kinetics of the C. divergens AS7 growth, divercin activity and the concentrations of glucose and lactic acid are shown in Fig. 1. As can be seen, the initial concentration of free glucose in the culture media was lower than 10 g l-1, indicated in the media composition. Moreover, the level of fermentable glucose decreased noticeably as the bioreactor scale increased and its concentration in the 1500 l scale was nearly one third of that in the reference scale (2.45 vs. 7.22 g l-1). This effect can be related to the Maillard-type reaction and the breakdown of nutrient constituents, which is a common problem in autoclaving of complex culture media containing reducing sugars and amino acids. The adverse impact of medium heat stress observed in this work was scale dependent. Indeed, the culture medium in the 1500 l bioreactor was darker than that in the 5 and 30 l scale, and the specific flavour was apparent. The problem of the medium sterilisation effects during a scale-up of the fermentation process was discussed by Junker [12]. It was revealed that increasing the scale results in the elongation of the heat up and cool down times which become a larger proportion of the overall sterilisation time cycle. This undoubtedly happened during the sterilisation of the culture medium in the 1500 l scale bioreactor, where the heat up time from ambient to sterilisation temperature took about 1 hour and the cool down time was another 2 hours, even using chilled water. These led directly to a rise in the level of sterilisation overkill and medium heat stress respectively.

Fig. 1. Cell growth (open circle), glucose concentration (filled triangle), lactic acid production (open triangle), and divercin activity in supernatant (filled circles) for Carnobacterium divergens AS7 cultivation in batch cultures: a 5 l scale, b 30 l scale, c 1500 l scale

Different initial conditions of the media used for divercin biosynthesis, resulting from different sterilisation conditions in the three investigated processing scales significantly affected the cell growth in each scale, but to a different extent. While the kinetics of the cell growth was similar in the 5 l and 30 l scales and an exponential growth phase of about 15 h was followed by a decrease in the growth rate and a stationary phase after 20 h of fermentation, when the glucose level became growth limiting, the exponential growth phase in the 1500 l scale exceeded 20 h (Fig. 1). The maximum specific growth rate as well as the maximum viable cell concentration decreased as the bioreactor scale increased (Table 4). In the 5 l scale the maximum viable cell number reached 1.0×1010 CFU per ml, while in the largest scale amounted only to 3.5×108 CFU per ml.

Table 4. Fermentation parameters at 5, 30 and 1500 l scales

Fermentation parameter

Scale (l)




Maximum viable cell concentration (CFU ml-1)
Maximum specific growth rate (h-1)
Final lactic acid concentration (g l-1)
Maximum total divercin activity (AU ml-1)
Maximum free divercin activity (AU ml-1)
Maximum free divercin productivity rate (AU ml-1 h-1)




As shown in Fig. 1, divercin production was growth associated and reached a maximum at the beginning of the stationary growth phase when glucose was exhausted, irrespectively of the bioreactor scale. In other studies, too, bacteriocins were synthesized mostly in the active growth phase [3,5,18]. However, the maximum divercin titer in a supernatant was scale dependent and varied between 51,000, 204,000 and 820,000 AU ml-1 for the 5 l, 30l and 1500 l scales respectively. At the same time, the maximum divercin activity measured in the fermentation broth containing cells was similar in all the three scales and amounted to 820,000 AU ml-1 (Table 4). After reaching a maximum, free bacteriocin activity in the culture medium rapidly decreased. A similar decline in amylovorin activity, concomitant with the depletion of the energy source, glucose, was reported by De Vuyst at al. [5] during batch fermentation of Lactobacillus amylovorus and was explained by the adsorption of the bacteriocin molecule to the producer cells. These authors speculated that the high levels of bacteriocin in the cellular environment might inhibit further bacteriocin production because of a limited immunity of the producer cells to their own bacteriocin. Other authors ascribed the disappearance of bacteriocin activity to proteolytic inactivation [10, 18,] and the aggregation of bacteriocin molecules due to their high hydrophobicity [17].

A comparison of free divercin release to the medium in terms of kinetic parameters showed that the maximum specific productivity rate calculated for the time at which bacteriocin was at a maximum titer differed among the scales. The highest productivity rate was observed for the 1500 l scale while the lowest for the 5 l scale. Moreover, when the divercin productivity was calculated as a relative value, i.e. as the ratio of free bacteriocin activity to the total activity, it was evident that divercin release to the medium was strongly affected by the conditions of the culture in a particular process scale (Fig. 2). There may be several reasons for the observed differences in free divercin productivity. One possible reason was nutrient depletion caused by the breakdown of the nutrient constituents and the formation of the products of the Maillard reaction during sterilisation of the culture broth in the 30 l and 1500 l bioreactors. It has previously been observed that the cultivation of Lactococcus lactis subsp. lactis at a low nutrient concentration resulted in a higher relative specific nisin concentration [14]. Other authors showed a similar behaviour of Lactobacillus sakei strain producing sakacin K, where nutrient exhaustion resulted in limitation of cell growth but favoured bacteriocin production [16].

Fig. 2. Cell growth (open circle), glucose concentration (filled triangle), lactic acid production (open triangle), and divercin activity in supernatant (filled circles) for Carnobacterium divergens AS7 cultivation in batch cultures: a 5 l scale, b 30 l scale, c 1500 l scale

The formation of the products of the Maillard reaction in sterilised growth media should also be considered in relation to the observed differences in divercin production among the bioreactor scales. Yet, the obtained data do not make it possible to estimate the effect of the products of the Maillard reaction itself or other heat degraded compounds on divercin synthesis in the present work, as too many factors were involved. However, evidence of the formation of antibacterial and antioxidative compounds in the Maillard reaction has been reported [7, 15] which could cause environmental stresses in the cultures performed in larger scales, in addition to a lower level of fermentable sugar.

Another reason for the different amount of divercin released to the culture medium among the bioreactors was the different cell concentration in a particular scale. It has previously been observed that the cultivation of C. divergens AS7 strain at a low cell density stimulated higher divercin production, compared to that with high cell concentration [24]. Moreover, in a high cell density culture most of the divercin was adsorbed to the cells while in diluted cultures the divercin was present in a free form in the supernatant.


The present work demonstrated that a successful scale-up of divercin synthesis by C. divergens AS7 from 5 l to 1500 l was possible. Although it is not known precisely how the growth conditions interfered with bacteriocin production, nutrient depletion in the large scale bioreactor was the main possible reason for the increased release of divercin to the culture medium. Since bacteriocins are considered as offensive products of cells, adverse environmental conditions cause competition for nutrients to be more severe and results in enhanced bacteriocin release from the cells.


  1. Abee T., Krockel L., Hill C., 1995. Bacteriocins: modes of action and potentials in food preservation and control of food poisoning. Int. J. Food Microbiol. 28, 169-185.

  2. Brauer H., 1985. Stirred vessel reactors. In: Rehm H-J, Reed G. (eds), Brauer H (vol ed) Biotechnology. VCG Verlagsgesellshaft, Berlin, 395-444.

  3. Callewaert R., De Vuyst L., 2000, Bacteriocin production with Lactobacillus amylovorus DCE 471 is improved and stabilized by fed-batch fermentation. Appl. Environ. Microbiol. 66, 606-613.

  4. Cleveland J., Montvik T.J., Nes I.F., Chikindas M.L., 2001. Bacteriocins: safe, natural antimicrobials for food preservation. Int. J. Food Microbiol. 71, 1-20.

  5. De Vuyst L., Callewaert R., Crabbé K., 1996. Primary metabolite kinetics of bacteriocin biosynthesis by Lactobacillus amylovorus and evidence for stimulation of bacteriocin production under unfavourable growth conditions. Microbiology 142, 817-827.

  6. Diaz A., Acevedo F., 1999. Scale-up strategy for bioreactors with Newtonian and non-Newtonian broths. Bioprocess Eng. 21, 21-23.

  7. Einarsson H., 1987. The effect of time, temperature, pH and reactants on the formation of antibacterial compounds in the Maillard reaction. Lebensm. Wiss. u-Technol. 20, 51-55.

  8. Einsele A., 1976. Scaling of bioreactors, theory and reality. Paper 4.13, V International Fermentation Symposium, Berlin.

  9. Einsele A., 1978. Scaling-up bioreactors. Process Biochem. 7, 13-14.

  10. Joerger M.C., Klaenhammer T.R., 1986. Characterization and purification of helveticin J and evidence for a chromosomally determined bacteriocin produced by Lactobacillus helveticus 481. J. Bacteriol. 167, 439-446.

  11. Ju L-K., Chase G.G., 1992. Improved scale-up strategies of bioreactors. Bioprocess Eng. 8, 49-53.

  12. Junker B.H., 2004. Scale-up methods for Escherichia coli and yeast fermentation processes. J. Biosci. Bioeng. 97, 347-364.

  13. Kalchayanand N., Sikes T., Dunne C.P., Ray B., 1994. Hydrostatic pressure and electroporation have increased bactericidal efficiency in combination with bacteriocins. Appl. Environ. Microbiol. 60, 4174-4177.

  14. Kim W.S., Hall R.J., Dunn N.W., 1997. The effect of nisin concentration and nutrient depletion on nisin production of Lactococcus lactis. Appl. Microbiol. Biotechnol. 48, 449-453.

  15. Lanciotti R., Anese M., Sinigaglia M., Severini C., Massini R., 1999. Effects of heated glucose-fructose-glutamic acid solutions on the growth of Bacillus stearothermophilus. Lebensm. Wiss. u-Technol. 32, 223-230.

  16. Leroy F., De Vuyst L., 2001. Growth of the bacteriocin-producing Lactobacillus sakei strain CTC 494 in MRS broth is strongly reduced due to nutrient exhaustion: a nutrient depletion model for the growth of lactic acid bacteria. Appl. Environ. Microbiol. 67, 4407-4413.

  17. Muriana P.M., Klaenhammer T.R., 1991. Purification and partial characterization of lactacin F, a bacteriocin produced by Lactobacillus acidophilus 11088. Appl. Environ. Microbiol. 57, 114-121.

  18. Parente E., Ricciardi A., 1999. Production, recovery and purification of bacteriocins from lactic acid bacteria. Appl. Microbiol. Biotechnol. 52, 628-638.

  19. Pilet M.F., Dusset X., Barre R., Novel G., Desmazaeud M., Piard J.Ch., 1995. Evidence for two bacteriocins produced by Carnobacterium piscicola and Carnobacterium divergens isolated from fish and active against Listeria monocytogenes. J. Food Prot. 58, 256-262.

  20. Schillinger U., Geisen R., Holzapfel W.H., 1996. Potential of antagonistic microorganisms and bacteriocins for the biological preservation of foods. Trends Food Sci. Technol. 7, 158-164.

  21. Schmidt F.R., 2005. Optimization and scale up of industrial fermentation processes. Appl. Microbiol. Biotechnol. 68, 425-435.

  22. Sip A., Grajek W. Boyval P., 1999. Production of bacteriocin by Carnobacterium divergens in batch and continuous culture. Pol. J. Food Nutr. Sci. 8/49, 27-38.

  23. Sip A., Grajek W., 1997. Production of divercin by Carnobacterium divergens in the presence of divercin-sensitive bacteria Carnobacterium piscicola and Listeria innocua. Pol. J. Food Nutr. Sci. 6/47, 59-64.

  24. Sip A., Grajek W., 2003. Biosynthesis of divercin by Carnobacterium divergens AS7 in continuous high cell density cultures. EJPAU 6(1), #02, http://www.ejpau.media.pl/volume6/issue1/biotechnology/art-02.html.

  25. Sip A., Mazurkiewicz J., Grajek W., Przybył A., 2004. Use of Carnobacterium sp. as a probiotic for common carp, (Cyprinus carpio L.). Proc Int Probiotic Conf “New Perspectives of Probiotics”. 15-19 September, Kosice, Slovakia.


Accepted for print: 30.10.2007

Anna Sip
Department of Biotechnology and Food Microbiology,
Poznań University of Life Sciences, Poland
Wojska Polskiego 48, 60-627 Poznań, Poland
Phone: +48 061 846 60 04

Radosław Dembczyński
Department of Biotechnology and Food Microbiology,
Poznań University of Life Sciences, Poland
Wojska Polskiego 48, 60-627 Poznań, Poland
phone: +48 61 8466026
email: rdembcz@au.poznan.pl

Wojciech Białas
Department of Biotechnology and Food Microbiology,
Poznań University of Life Sciences, Poland
Wojska Polskiego 48, 60-627 Poznań, Poland
email: wbialas@au.poznan.pl

Wojciech Juzwa
Department of Biotechnology and Food Microbiology,
The August Cieszkowski Agricultural University of Poznań, Poland
Wojska Polskiego 48, 60-627 Poznań, Poland

Katarzyna Czaczyk
Department of Biotechnology and Food Microbiology,
Poznań University of Life Sciences, Poland
Wojska Polskiego 48, 60-627 Poznań, Poland
Phone: +48 061 846 60 04

Włodzimierz Grajek
Department of Biotechnology and Food Microbiology,
Poznań University of Life Sciences, Poland
Wojska Polskiego 48, 60-627 Poznań, Poland

Tomasz Jankowski
Department of Biotechnology and Food Microbiology,
Poznań University of Life Sciences, Poland
Wojska Polskiego 48, 60-627 Poznań, Poland
Phone: 061 846 60 04

Responses to this article, comments are invited and should be submitted within three months of the publication of the article. If accepted for publication, they will be published in the chapter headed 'Discussions' and hyperlinked to the article.