UF Skim Milk Cheeses

8/16/2019 UF Skim Milk Cheeses

1/9

International Dairy Journal 17 (2007) 674–682

Rheological properties and microstructure during rennet induced

coagulation of UF concentrated skim milk

A.O. Karlsson, R. Ipsen, Y. Ardo ¨

Dairy Technology, Department of Food Science, Centre for Advanced Food Studies, The Royal Veterinary and Agricultural University,

Rolighedsvej 305, 1958 Fredriksberg C, Denmark

Received 20 December 2005; accepted 10 August 2006

Abstract

Rennet induced coagulation of ultrafiltrated (UF) skim milk (19.8%, w/w casein) at pH 5.8 was studied and compared with

coagulation of unconcentrated skim milk of the same pH. At the same rennet concentration (0.010 International Milk Clotting Units

g1

), coagulation occurred at a slower rate in UF skim milk but started at a lower degree of k-casein hydrolysis compared with the

unconcentrated skim milk. Confocal laser scanning micrographs revealed that large aggregates developed in the unconcentrated skim

milk during renneting. Following extensive microsyneresis the protein strands were shorter and thinner in gels from UF skim milk.

Moreover, during storage up to 60 days (13 1C), the microstructure and the size of the protein strands of the UF gel changed only

slightly. Hoelter–Foltmann plots suggested that the coagulation rate was reduced in the UF skim milk due to a high zero shear viscosity

of the concentrate compared with the unconcentrated skim milk.

r 2006 Elsevier Ltd. All rights reserved.

Keywords: Casein gel; Coagulation; Microstructure; Rennet; Rheology; Ultrafiltration

1. Introduction

Membrane processes are used in dairy plants to

standardize or increase the protein content in milk and to

separate bacteria and spores from milk. Shortly after the

introduction of membrane processes for milk, the use of

concentrated ultrafiltrated (UF) milk for cheese production

was initiated. A liquid pre-cheese (LPC) concept resulted in

minimal whey drainage and therefore almost all whey

proteins could be incorporated into the cheese and

significantly increase the cheese yield (Maubois, Mocqout,

& Vassal, 1969). However, cheese made from UF milk andespecially cast cheese made using the LPC concept

generally have very different sensorial and functional

properties compared with cheese from traditional cheese

production (Mistry & Maubois, 1993). The incorporation

of whey proteins have been suggested to cause many of the

differences between cheeses from UF milk and unconcen-

trated milk (Bech, 1993). Whey proteins act as inert fillers

in the casein matrix and increase the water binding of

cheese and causes UF cheeses to be softer than traditional

cheeses. The presence of whey proteins has also

been suggested to reduce the enzymatic proteolysis of

caseins during ripening of UF cheeses. UF is hence

not used in the production of most cheese varieties.

However, some cast-type cheese varieties with a low pH,

i.e., Feta and Camembert types, have been accepted by

consumers.

The rennet induced coagulation of skim milk constitutes

three phases: enzymatic hydrolysis of k-casein, aggregation

of renneted casein micelles and gel development. In thelatter phase a three-dimensional protein network develops

and micro- and macrosyneresis, i.e., fusion of casein

micelles and whey separation, respectively, occurs (Walstra

& van Vliet, 1986). With increased casein concentration the

coagulation properties of milk change. The coagulation

time decreases (Sharma, Mittal, & Hill, 1994), the elasticity

of the gel increases (Culioli & Sherman, 1978), the level of

hydrolysed k-casein at the coagulation point is lower

(Sharma et al., 1994) and less water and whey proteins are

expelled from the gel.

ARTICLE IN PRESS

www.elsevier.com/locate/idairyj

0958-6946/$- see front matterr 2006 Elsevier Ltd. All rights reserved.

doi:10.1016/j.idairyj.2006.08.002

Corresponding author. Tel.: +45 3528 3253; fax: +45 3528 3190.

E-mail address: [email protected] (A.O. Karlsson).

http://www.elsevier.com/locate/idairyjhttp://localhost/var/www/apps/conversion/tmp/scratch_1/dx.doi.org/10.1016/j.idairyj.2006.08.002mailto:[email protected]:[email protected]://localhost/var/www/apps/conversion/tmp/scratch_1/dx.doi.org/10.1016/j.idairyj.2006.08.002http://www.elsevier.com/locate/idairyj


2/9

Results presented by Green, Marshall, and Glover

(1981) and Hyldig (1993) show that the protein network

coarsens when the casein concentration is increased

in casein gels and cheese. However, no physical explanation

has been given for this behaviour and the microstructure

of milk concentrates has not been studied during coagula-

tion. Since the microstructure of mature casein gels(after syneresis) has been shown to be influenced by the

aggregation kinetics (Green, 1990; Wium, Pedersen, &

Qvist, 2003) it is important to study the microstructure

during coagulation. Especially, coagulation of highly

concentrated milk with no whey separation is

interesting because this determines the microstructure of

cast cheese. This has generally received very little attention,

and the changes occurring in the microstructure

during longer storage, i.e., months, has not been empha-

sized.

In the present work we have investigated differences

between rennet coagulation of highly concentrated

(19.8%, w/w casein) and unconcentrated skim milk. The

rheological properties during aggregation and coagulation

were related to the enzymatic hydrolysis of k-casein and the

microstructure of developing aggregates in both unconcen-

trated and concentrated skim milk. Development of

microstructure in gels from unconcentrated skim milk

after extensive microsyneresis and gels from UF concen-

trate after storage at 13 1C for up to 60 days were studied

and compared.

2. Materials and methods

2.1. Preparation of UF concentrate

The UF concentrate was produced as described by

Karlsson, Ipsen, Schrader, and Ardo ¨ (2005). The UF

process was stopped when the UF concentrate had reached

a Brix value of 36.31, measured using a handheld

refractometer (Atago Co., Tokyo, Japan). After produc-

tion, the UF concentrate was poured into bottles (100 mL),

heat treated (62 1C for 30 min), quickly cooled to approxi-

mately 4 1C in an ice–water bath and stored in a freezer

(23 1C). The frozen UF concentrate was used within one

month. Compared with fresh UF concentrate, the coagula-

tion properties of frozen UF concentrate stored or two

months was shown not to be significantly different (results

not shown).

Prior to use in experiments, UF concentrate and

unconcentrated skim milk was thawed in a water bath

(30 1C) for 1 h. Thimerosal (0.02%, w/w; Merck, Darm-

stadt, Germany) was added as a preservative to all samples

to prevent microbial growth. The samples were equili-

brated for 24h at 30 1C before glucono-d-lactone (GDL;

Acros Organics, Geel, Belgium) was added to samples of

UF concentrate and unconcentrated skim milk in order to

obtain a final pH of 5.8 after 24 h of storage at 30 1C after

GDL addition.

2.2. Chemical analysis of unconcentrated skim milk and the

UF concentrate

The pH was measured directly using a Knick Portamess

(Knick Elektronische Messgera ¨ te, Berlin, Germany)

equipped with a Hamilton Tiptrode (Hamilton Instru-

ments, Bonaduz, Switzerland). Total solid contents weredetermined according to the International Dairy Federa-

tion (IDF) standard method (IDF, 1991). Nitrogen was

determined using a Kjeltec System 1026 Analyzer (Tecator,

Ho ¨ gana ¨ s, Sweden). Total nitrogen (TN), non-casein

nitrogen (NCN) and non-protein nitrogen (NPN) were

determined according to the IDF standard methods (IDF,

1993). The protein content was estimated by multiplying

the nitrogen content for casein by 6.36 and whey protein by

6.28 (van Boekel & Ribadeau-Dumas, 1987). Determina-

tion of lactose was carried out with a Lactose/ D-Galactose

Enzymatic BioAnalysis-kit (Scil Diagnostica, Martinsried,

Germany) according to the manual of the manufacturer.

All chemical analyses were performed at least in duplicate.

Casein has been reported to be totally demineralized

from colloidal calcium phosphate at pH 2.7 (Le Grae ¨t &

Gaucheron, 1999). Thus, by adjusting the pH to the

interval 2.2–3.0 using 1 M HCl, all the colloidal calcium was

dissolved in the serum phase of unconcentrated skim milk

and UF concentrate. At pH 5.8, adjusted by GDL, only a

part of the colloidal calcium is dissolved in the serum phase

of skim milk and UF concentrate. Rennet (CHY-MAX

Extra Liquid, Chr. Hansen A/S, Hørsholm, Denmark) was

diluted 10 times and 2.5 mL g1 sample was added to

samples 24 h after addition of HCl or GDL. The samples

were then stirred for 1.5min. The samples coagulated in24 h (30 1C) and the serum phase could be separated from

the gels by ultracentrifugation (100000g for 60 min at

30 1C) using a Beckman L8-70M Ultracentrifuge with a

SW28 rotor (both from Beckman Instruments Inc.,

Palo Alto, CA, USA). Following centrifugation,

the supernatant was carefully removed and transferred to

plastic tubes, which were placed in the freezer (23 1C)

for later determination of calcium. Prior to analysis

the supernatants were thawed in a water bath (20 1C).

Two supernatants of every sample were individually

prepared.

The calcium concentration was determined using a

Perkin Elmer Atomic Absorption Spectrometer (Perkin

Elmer, Boston, MA, USA). A calibration curve was

created by measuring standard solutions of CaCl2. Super-

natant solutions from ultracentrifugation were diluted with

double deionized H2O. Diluted supernatant solutions of

samples and standard solutions contained 0.02654 M LaCl3.

Dilution of every solution was performed at two indepen-

dent times to quantify the error of the sample preparation.

2.3. Degree of k-casein hydrolysis during renneting

Different amounts of rennet [0.010–0.003 International

Milk Clotting Units (IMCU) g1

, CHY-MAX Extra

ARTICLE IN PRESS

A.O. Karlsson et al. / International Dairy Journal 17 (2007) 674–682 675


3/9

Liquid] were added to the samples (75 g each) of

UF concentrate and unconcentrated skim milk (both

pH 5.8) 24h after addition of GDL. After rennet

addition the samples were gently stirred for 1.5 min

with a Stir-Pak laboratory mixer (Cole-Parmer Instru-

ments, Chicago, IL, USA) equipped with a paddle blade

stirrer.Degree of k-casein hydrolysis in the UF concentrate and

the unconcentrated skim milk was determined at appro-

priate times after rennet addition as described by van

Hooydonk and Olieman (1982). Samples of ungelled or

gelled milk were dissolved in trichloroacetic acid and the

amount of glycomacropeptide (GMP), i.e., one of the

products released following hydrolysis of k-casein was

quantified after separation on a RP-HPLC column

(Nucleosil C18, Macherey-Nagel GmbH, Du ¨ ren, Ger-

many). There was a higher concentration of casein in our

UF concentrate as compared with the samples of van

Hooydonk and Olieman (1982), and the amounts of added

chemicals and levels of dilution were consequently ad-

justed.

2.4. Rheological properties during renneting

The rheological properties during rennet coagulation

were monitored by small amplitude oscillatory shear

(SAOS) in a stress controlled rheometer (Bohlin C-VOR,

Malvern Instruments, Malvern, England) equipped with

a measuring system consisting of two coaxial cylinders

(14.0 and 15.4 mm in diameter, Malvern Instruments).

The instrument recorded the elastic modulus (G 0),

the viscous modulus (G 00

) and the phase angle (d) a t astrain (g) of 0.002 and a frequency ( f ) of 1 Hz. This strain

was found to be well within the visco-elastic region for

both the UF concentrate and the fully developed casein

gels. All curves of the three measured parameters versus

time were smoothed using the loess function in Mathcad

2001i Professional (MathSoft Engineering & Education,

Inc., 1986–2001, Cambridge, MA, USA) to compensate

for small fluctuations defined as noise. The loess function

generates a set of second order polynomials to fit the

data over time (MathSoft Engineering & Education, Inc.,

2001). An appropriate number of second order polyno-

mials were chosen in order not to over-fit the data.

The coagulation time was defined as the time when

G 0 exceeded G 00 (Curcio et al., 2001). An inflection point

in G 0 appeared before the coagulation point and the

time when this inflection point occurred was also deter-

mined. After the coagulation point a maximum value in

(dG 0/dt) was detected. At (dG 0/dt)max, d did not change

with time and the mean value in this region was taken

as d during gel firming. Means and standard deviations

of coagulation time, time for inflection in G 0 before

coagulation and d at gel firming were all calculated

from three measurements. Representative coagulation

curves of some selected samples are graphically shown in

this paper.

2.5. Microstructure by confocal laser scanning microscopy

(CLSM)

For samples investigated with CLSM, 0.005% (w/w)

Rhodamine B (Merck) was mixed with the milk sample

after pH adjustment with GDL but before addition of

rennet. After rennet addition and 1.5 min of mixing(equipment as described above), smaller proportions of

the samples were transferred to a microscope slide with a

polished cavity. The exposed sample surface was then

covered by a cover slip and a thin sticky plastic film (i.e., to

prevent evaporation) before the sample was left to

coagulate at 30 1C for 24 h and then further storage at

13 1C (up to 60 days).

The microstructures of samples were examined during

and after rennet coagulation by a Leica TCS 4D confocal

laser scanner with a Leitz DM RB/E* microscope (Leica

Microsystems, Heidelberg, Germany) equipped with a

water-immersion optics (63 times magnification, numerical

aperture 1.4). Two-dimensional images were recorded at a

distance below the cover slip where the protein micro-

structure did not appear to be affected by surface

phenomena due to the presence of the cover slip. A

568 nm laser was used to detect Rhodamine B stained

protein and emitted light of wavelength 580–623 nm was

recorded. Light intensity at recording was standardized by

using the Glow-over function in the software (Leica TCS

NT, version 1.6.587, Leica Microsystems).

3. Results

The chemical composition of skim milk and UFconcentrate confirmed that not only the caseins but also

the whey proteins were fully retained by the UF membrane

(Table 1). The UF permeate contained no proteins but a

low content of NPN (results not shown). The UF

concentrate had a slightly lower pH (6.52) than the

unconcentrated raw material (6.66), which is in accordance

with earlier observations (Walstra, Geurts, Noomen,

Jellema, & van Boekel, 1999).

3.1. Rennet coagulation and hydrolysis of k-casein

All coagulation experiments were made at pH of 5.8

because the unrenneted UF concentrate has a low viscosity

at this pH (Karlsson et al., 2005). In un-renneted UF

concentrate at pH 5.8 the measured G 00 was larger than G 0

(results not shown). In the interval pH 6.1–5.2, the

rheological properties of UF concentrate are very variable

and G 00 does not always exceed G 0 (Karlsson et al., 2005).

With increased rennet concentration the coagulation

time decreased and the maximum gel firming rate, i.e.,

(dG 0/dt)max, after coagulation increased (Table 2). The

maximum gel firming rate increased linearly (R240:99)

with the rennet concentration. Initially d changed during

renneting but it remained constant for all samples

after (dG 0/dt)max was reached and regardless of rennet

ARTICLE IN PRESS

A.O. Karlsson et al. / International Dairy Journal 17 (2007) 674–682676


4/9

concentration this constant value was not significantly

different between samples (Table 2). Similar data have been

found in previous studies on both unconcentrated and

concentrated skim milk (Culioli & Sherman, 1978; Zoon,

van Vliet, & Walstra, 1988). For comparison, 0.010 IM-

CU g1 of rennet was added to both UF concentrate and

unconcentrated skim milk (Fig. 1). The unconcentrated

skim milk had a shorter coagulation time and a lower gel

firming rate (Fig. 1) than the UF concentrate. In a

relatively short time (i.e., 50 min), G 0 and G 00 of the

unconcentrated skim milk reached plateau values.

For all samples of UF concentrate in Table 2, an

inflection point in both moduli could be detected prior to

the time where G 0 exceeded G 00. This is illustrated for a

sample of UF concentrate shown in Fig. 1. A similar

inflection point in G 0 could not be detected for unconcen-

trated skim milk (Fig. 1, insert).

It is expected for k-casein to be hydrolysed to varying

degrees in milks with different casein concentrations before

a gel develops (Curcio et al., 2001). By plotting G 0 and G 00

as function of k-casein hydrolysis for the same samples as

in Fig. 1, it was estimated that less than 20% of the

k-casein was hydrolysed by the rennet when the UF

concentrate reached the coagulation point (Fig. 2). For the

unconcentrated skim milk with the same concentration of

rennet approximately 50% of k-casein was hydrolysed

when G 0

4G 00

.For the samples of UF concentrate, as well as for those

of unconcentrated skim milk, with different concentrations

of added rennet, the time at inflection point in G 0 and

coagulation times were plotted as function of the inverse of

the rennet concentration (Fig. 3). Such plots are known as

Hoelter–Foltmann plots (Foltmann, 1959). Linear regres-

sions of data for time at inflection point in G 0 and

coagulation times for the UF concentrate resulted in

similar intercepts on the y-axis but in a higher slope for

the linear equation of the coagulation time data. The

coagulation times (i.e., when G 0 became larger than G 00) of

the unconcentrated skim milk resulted in a linear equation

with a lower slope and lower value of the intercept with the

y-axis.

3.2. Microstructure

When the microstructure of unconcentrated skim milk

and the UF concentrate was monitored during renneting

using CLSM, major differences could be observed (Fig. 4).

Larger aggregates of casein micelles were formed more

rapidly and within a shorter time interval in the unconcen-

trated skim milk. Also, due to the lower protein

concentration, the non-protein areas constituted a larger

fraction in the gels of the unconcentrated skim milk. Even

ARTICLE IN PRESS

Table 1

Composition of skim milk and UF concentrate

Skim milk UF concentrate

Total solids % (w/w) 9.1970.00a 30.8670.04a

pH (30 1C) 6.66b 6.52b

Casein (g 100 g1) 2.75b 19.81b

Whey protein (g 100g1) 0.54b 3.48b

Lactose (g 100g1) 5.1270.13c 4.2770.15c

Total Ca (mg kg1) 1195716d 68077125d

Ca in serum phase (pH 5.8) (mg kg1) 625725d 1423778d

aValues are means and standard deviations of two replicates.bValues are means of three replicates.cValues are means and standard deviations of three replicates.dValues are means and standard deviations of two replicates measured

two times.

Table 2

Time for inflection point in G 0 before coagulation point, coagulation time

(i.e., time when G 0 became larger than G 00), maximum gel firming rate(dG 0/dt)max after coagulation point and value of phase angle after

coagulation point (i.e., when phase angle did not change with time) at

renneting of UF concentrate.a See text for further details

Rennet

concentration

(IMCU g1)

Time at

inflection

point in G 0

(min)

Coagulation

time (min)

(dG 0/dt)max(Pamin1)

Phase angle

at gel firming

(1)

0.016 22.775.2 32.370.7 240.176.1 16.970.2

0.013 23.873.1 35.770.8 199.873.3 16.970.1

0.010 27.270.8 44.371.6 168.371.2 16.770.2

0.008 42.271.5 57.270.3 134.375.6 17.3770.4

0.005 45.971.8 79.475.1 95.2572.4 17.370.2

0.003 71.970.4 131.277.2 66.871.4 16.870.4

aValues are means and standard deviation of three replicates.

Parameters are defined in the text.

Fig. 1. Development of the elastic modulus (G 0, filled symbols) and

viscous modulus (G 00, open symbols) during rennet coagulation of skim

milk (circles) and UF concentrate (squares) at pH 5.8 after rennet addition

(0.010IMCU g1

). Arrow indicates the approximate position of theinflection point (i.e., 27min) in G 0 for UF concentrate before the

coagulation point (G 04G 00). The coagulation temperature was 30 1C.

The insert is a close-up showing data points ( G 0 and G 00) in the time region

of skim milk coagulation.



5/9

after 45 min, which is close to the measured coagulation

time of 44 min, no large aggregates were observed in the

UF concentrate (Fig. 4F). At longer times (i.e., 54 min;

Fig. 4G) the individual aggregates in the UF concentrate

still appeared smaller and less dense compared to the

aggregates found just after aggregation in the unconcen-

trated skim milk. However, even though only small

aggregates had formed in the UF concentrate a very firm

gel had developed (Fig. 4G).

After 24 h, extensive micro- and macrosyneresis resulted

in a coarse protein network in the gel made from

unconcentrated skim milk (Fig. 5A). The fusion of casein

micelles and casein strands resulted in a microstructure

where structures of casein, with a size of roughly 20–40 mm,

were linked together. The size of the casein structures after

24 h were approximately 4 times larger than the aggregates

of casein micelles at the coagulation point (Fig. 4C). Such

extensive changes in the microstructure were not apparent

during firming of the gel made from UF concentrate. Infact, no visible differences were discerned between the

microstructures of gels observed 215 min (Fig. 4H) and 24 h

(Fig. 5B) after rennet addition. The gel microstructure had

undergone microsyneresis but no macrosyneresis, i.e., whey

separation from the gel, in the gels from UF concentrates

after extended storage (i.e., 60 days; Fig. 5C). After 60 days

(13 1C) larger structures had formed in the gel from UF

concentrate but they were still much smaller than the

structures in the gel from the unconcentrated skim milk

24 h after rennet addition.

4. Discussion

4.1. Aggregation of casein micelles in UF concentrate

In UF concentrate the distance between unrenneted

casein micelles is smaller than the diameter of the micelles

and they interact strongly over short distances, i.e.,

approximately 7.9 nm at pH 5.8 (Karlsson et al., 2005).

The moduli G 0 and G 00 of the UF concentrate can hence be

expected to be highly sensitive to changes in the sterical and

electrostatic stability of the casein micelles caused by

hydrolysis of k-casein prior to coagulation (Karlsson et al.,

2005). This was confirmed when both G 0 and G 00 initially

decreased after rennet addition (Fig. 1). At approximately25 min G 0 then started to increase slightly (Fig. 1). The

slight increase in G 0 can probably be referred to an

aggregation and a creation of a loose network at a certain

degree of k-casein hydrolysis (Fig. 2). This resulted in the

characteristic inflection point in G 0, which could be referred

to an aggregation point although aggregation probably

started somewhat earlier. A similar trend could not be

detected in unconcentrated skim milk, in which the casein

micelles interact over much longer distances (Walstra &

Jenness, 1984).

4.2. Coagulation of UF concentrate and unconcentrated

skim milk

Much research has been performed on the coagulation of

different milk concentrates but only very limited work has

been done on concentrates with casein concentrations

approaching 20% (w/w), as in this study. Although

previous studies have reported elastic properties, i.e., G 0,

for rennet gels with a lower casein concentration, it has

been clear that increased casein concentration results in

larger G 0 (Hyldig, 1993). Thus, the higher casein concen-

tration made G 0 of the UF concentrate to rapidly exceeded

G 0 of the gel from unconcentrated skim milk once the

coagulation occurred (Fig. 1).

ARTICLE IN PRESS

Fig. 2. The elastic modulus (filled symbols) and viscous modulus (open

symbols) as function of the degree of k-casein hydrolysis during renneting

at 30 1C for unconcentrated skim milk (circles) and UF concentrate

(squares) of pH 5.8.

Fig. 3. Hoelter–Foltmann plots of the time of inflection point in G 0

(i.e., aggregation time, J) and coagulation time (K) for UF concentrate;

and coagulation time for unconcentrated skim milk (.) for different

additions of rennet. All samples of pH 5.8. Linear regressions of the data

showed in all cases R240:98. The time of the inflection point in G 0 and

coagulation times where extracted from coagulation curves (such as in Fig.

1). The coagulation time was when G 04G 00. Bars indicate standard

deviation of three replicates.



6/9

ARTICLE IN PRESS

Fig. 4. Confocal images of skim milk (A–D) and UF concentrate (E–H) during renneting. The rennet concentration was 0.010IMCU g1 and the

coagulation temperature 30 1C for both samples. Below each image, time after rennet addition and elastic modulus at that time is indicated. Proteins were

labelled with Rhodamine B and appear white on micrographs.

Fig. 5. Confocal images of rennet induced casein gels made from: (A) unconcentrated skim milk 24 h after rennet addition; (B) UF concentrate 24 h and

(C) 60 days after rennet addition. Gels were let to develop at 30 1C for the first 24 h. At storage longer than 24 h, the temperature was 13 1C. Rennet

concentration in all gels (A–C) was 0.010 IMCU g1. Proteins were labelled with Rhodamine B and appear white on micrographs.



7/9

Rennet coagulation of casein micelles in unconcentrated

skim milk is known to involve three reactions: enzymatic

hydrolysis of k-casein, aggregation of casein micelles and

creation of a gel network (Walstra and van Vliet, 1986).

The three reactions are not separated in time at cheese

making (Walstra and van Vliet, 1986) and this was seen for

the unconcentrated skim milk (pH 5.8) in Fig. 2, whereonly 50% of the k-casein had been enzymatically

hydrolysed when the coagulation occurred, i.e., G 04G 00.

In a UF concentrate the same three reactions, i.e.,

enzymatic reaction, aggregation and creation of a gel

network, will take place as a consequence of renneting.

Complete hydrolysis of the k-casein took more time due to

lower rennet–casein ratio in the UF concentrate than in the

unconcentrated skim milk. This explained why coagulation

is achieved later for the UF concentrate (Fig. 1). However,

it is evident that the coagulation point occurs at a lower

degree of k-casein hydrolysis in the UF concentrate

compared to the unconcentrated skim milk (Fig. 2). This

has been proposed by Sharma et al. (1994) to be caused by

a smaller mean free distance between micelles, which

increases the collision frequency, and higher concentrations

of ionic calcium, which reduces the electrostatic repulsions,

in UF concentrates.

Hoelter–Foltmann plots (Foltmann, 1959) were used to

estimate whether the coagulation was influenced by factors

other than the hydrolysis of k-casein. If the intercept in a

Hoelter–Foltmann plot (Fig. 3) is zero, this indicates that

coagulation occurs instantaneously at infinitely high rennet

concentrations, i.e., only the enzymatic hydrolysis of k-

casein is the rate limiting reaction (Foltmann, 1959). Large

values of the intercept indicate that other factors (e.g.,collision frequency and repulsion between micelles) are rate

limiting for coagulation. Hoelter–Foltmann plots of our

data indicated that coagulation was more limited by other

factors than k-casein hydrolysis in UF concentrate than in

unconcentrated skim milk (Fig. 3). Sharma et al. (1994)

suggested somewhat the opposite that factors like the small

mean free distance between renneted micelles (i.e., giving

high collision frequency) and the high concentration of

ionic calcium (i.e., screening of charges) in a concentrate

would result in decrease of coagulation rate. However, the

UF concentrate has a substantial zero shear viscosity

(Karlsson et al., 2005) and hence a limited movement of

micelles and a concomitant reduced rate of coagulation.

On the other hand, since coagulation took place at a lower

degree of k-casein hydrolysis in the UF concentrate

(Fig. 2), the proportion of collisions between micelles

resulting in aggregation during renneting can be assumed

to be higher in the UF concentrate than in the unconcen-

trated skim milk.

There was significant difference in the microstructural

appearance of the formed aggregates and protein network

(Fig. 4) at the coagulation point between unconcentrated

and concentrated skim milk. The microstructure of the

unconcentrated skim milk during coagulation was clearly a

result of diffusion of micelles and microsyneresis in the

protein network (Fig. 4A–D). In the UF concentrate, the

high zero shear viscosity and the early development of a

loose network (as indicated by the increasing G 0 prior to

actual coagulation) resulted in a restricted diffusion of

micelles and aggregates. The micrographs clearly support

that impaired diffusion hindered the growth of aggregates

and incorporation of material into the protein network.

4.3. Syneresis of gels from UF concentrate and

unconcentrated skim milk

Syneresis in casein gels comprises microsyneresis, i.e.,

rearrangements of the casein network on a microstructural

level, and macrosyneresis, i.e., separation of whey from the

gel on a macroscopic level (Mellema, 2000). In the

unconcentrated skim milk microsyneresis started just

minutes after coagulation (Fig. 4C and D) whereas the

results of the same process in the gel from UF concentrate

were first apparent later, i.e., some 10min after the

coagulation time (Fig. 4G). The reason for this observed

delay in microsyneresis could be the reduced rate of

hydrolysis of k-casein in the UF concentrate and the

occurrence of coagulation at a lower degree of k-casein

hydrolysis when the micelles were still partially stabilized

(Fig. 2). It took, in fact, approximately 9 h for all the k-

casein to be hydrolysed (results not shown). If the surface

of casein particles are only partially covered with k-casein

the reactivity between them, i.e., ability to bind to each

other, will be low (van Vliet, van Dijk, Zoon, & Walstra,

1991). This will in turn decrease the endogenous syneresis

pressure in the gel, which is the main factor determining the

rate of rearrangements in a casein gel (van Vliet et al.,1991). Rearrangements of the microstructure in a rennet

gel gradually strengthens the protein strands and causes an

increase in G 0 after coagulation (Zoon et al., 1988). A

strong correlation (R240:99) between the maximum gel

firming rate (dG 0/dt)max and the rennet concentration well

illustrated that in our experiments the hydrolysis of

k-casein did indeed strongly influence the rate of rearran-

gements (Table 2). This type of strong linear relationship

has not been shown for unconcentrated skim milk at native

pH where the structure of aggregates during the initial

aggregation and gel formation has been suggested to

influence the gel firming rate (Lomholt & Qvist, 1999).

After one day of storage when all k-casein was

hydrolysed, the rearrangements in the protein network

seemed to have ceased at a stage where the strands of

casein formed in the gel from UF concentrate appeared

smaller (Fig. 5B) than in the gel from unconcentrated skim

milk (Fig. 5A). Macrosyneresis was not detected in the gel

from UF concentrate as is was in the gel from unconcen-

trated skim milk.

The endogenous syneresis pressure in the gel can break

existing bonds within and between strands of proteins and

facilitate formation of new bonds. When the bonds within

the casein network are strong enough to resist the stress

from the endogenous syneresis pressure, the pressure in the

ARTICLE IN PRESS



8/9

gel will decrease and the rearrangements cease. The

development of an endogenous syneresis pressure is

dependent on the reactivity between casein particles,

bending stiffness of protein strands and the relaxation

times and number of bonds in the protein strands of a

casein gel (van Vliet et al., 1991). The temperature in our

experiments was always 301

C and pH 5.8 of gels from bothUF concentrate and unconcentrated skim milk, thus, the

reactivity between casein particles after all k-casein had

been hydrolysed was the same. Zoon et al. (1988) has

shown that d (and hence the relaxation time) does not

change due to increasing casein concentration and our

results also indicated no significant difference in d (at 1 Hz)

between gels from UF concentrate and unconcentrated

skim milk (Table 2). The higher density of casein, however,

can be assumed to have caused a very rapid formation of

many strong bonds between strands of casein with a high

bending stiffness in the gel from UF concentrate. The

resultant strong protein strands could then presumably

resist the endogenous syneresis pressure and hence the

rearrangements ceased earlier than in unconcentrated skim

milk.

The microstructure was also well preserved upon longer

storage of the gel. During 60 days of storage at 13 1C, some

rearrangements in the casein network occurred (Fig. 5C).

The very fine stranded protein network, i.e., large surface

to volume ratio, in the gel made from UF concentrate

could explain why cheese from concentrates can contain a

relatively large amount of water.

5. Conclusions

In skim milk UF concentrate (pH 5.8) with a casein

concentration of 19.8% (w/w), rennet coagulation occurred

at a lower degree of hydrolysis of k-casein than in

unconcentrated skim milk with the same amount of rennet

added and it took a longer time for the gel to develop

maximum firmness. In a concentrate, it takes a longer time

for all k-casein to hydrolyse, and this prolongs the stability

of the casein micelles and delays completion of the gel

ageing. The aggregation phase of the coagulation reaction

was extended in the UF concentrate, possibly due to the

higher zero shear viscosity retarding the diffusion rate and

collision frequency between casein micelles.

Compared to unconcentrated skim milk, the aggregates

formed in the UF concentrate were much smaller and

could not be detected with CLSM prior to coagulation

point. Only about 10 min after the coagulation time, larger

structures could be detected. Due to the higher volume

fraction of casein micelles in the UF concentrate than in

unconcentrated skim milk, the bonds between casein

strands will strengthen rapidly preventing further rearran-

gements in the gel microstructure. Thus, the casein network

was less coarse in gels made from UF concentrate and the

microstructure was preserved during storage at 13 1C for 60

days. This is proposed to be one of the reasons for the high

water content of cast cheese made using UF.

Acknowledgements

This study was initiated by Karsten Bruun Qvist (now

present at Danisco A/S, Copenhagen) and was financially

supported by the Danish Dairy Research Foundation and

the Danish Governmental Research Program (FØTEK 3).

The technical assistance from Anni Nielsen and VivianPedersen is gratefully acknowledged.

References

Bech, A-. M. (1993). Characterising ripening in UF-cheese. International

Dairy Journal , 3, 329–342.

Culioli, J., & Sherman, P. (1978). Rheological aspects of the renneting of

milk concentrated by ultrafiltration. Journal of Texture Studies, 9,

257–281.

Curcio, S., Gabriele, D., Giordano, V., Calabro ` , V., de Cindio, B., &

Iorio, G. (2001). A rheological approach to the study of concentrated

milk clotting. Rheologica Acta, 40, 154–161.

Foltmann, B. (1959). On the enzymatic and coagulation stages of therenneting process. Proceedings of the 15th International Dairy

Congress, 2, 655–661 London, UK.

Green, M. L. (1990). Cheddar cheese making from whole milk

concentrated by ultrafiltration and heated to 90 1C. Journal of Dairy

Research, 57 , 559–569.

Green, M. L., Marshall, R. J., & Glover, F. A. (1981). Influence of

homogenisation of concentrated milks on the structure and properties

of rennet gels. Journal of Dairy Research, 50, 341–348.

Hyldig, G. (1993). Rennet coagulation. Influence of technological para-

meters on the enzymatic reaction and gel formation in milk and UF

retentates. Ph.D. thesis, The Royal Veterinary and Agricultural

University, Copenhagen, Denmark.

IDF. (1991). Sweetened condensed milk: Determination of the total solids

content (reference method). IDF Standard 15B . Brussels, Belgium:

International Dairy Federation.IDF. (1993). Nitrogen content of milk and milk products. IDF Standard

20B . Brussels, Belgium: International Dairy Federation.

Karlsson, A. O., Ipsen, R., Schrader, K., & Ardo ¨ , Y. (2005).

Relations between physical properties of casein micelles and

rheology of skim milk concentrate. Journal of Dairy Science, 88,

3784–3797.

Le Grae ¨ t, Y., & Gaucheron, F. (1999). pH-induced solubilization of

minerals from casein micelles: influence of casein concentration and

ionic strength. Journal of Dairy Science, 66 , 215–224.

Lomholt, S. B., & Qvist, K. B. (1999). Gel firming rate of rennet curd as a

function of rennet concentration. International Dairy Journal , 9,

417–418.

MathSoft Engineering & Education, Inc. (2001). Mathcad 2001i User’s

guide with reference manual (p. 348). Cambridge, MA, USA: MathSoft

Engineering & Education, Inc.Maubois, J.-L., Mocqout, G., & Vassal, L. (1969). Procedure for treating

milk and milk products. French Patent no. 2052121.

Mellema, M. (2000). Scaling relations between structure and rheology of

ageing casein particle gels. Ph.D. thesis, Wageningen University,

Wageningen, The Netherlands.

Mistry, V. V., & Maubois, J.-L. (1993). Application of membrane

technology to cheese production. In P. F. Fox (Ed.), Cheese: chemistry,

physics and microbiology, Vol. 1: General aspects (2nd ed.,

pp. 493–522). London, UK: Chapman & Hall.

Sharma, S. K., Mittal, G. S., & Hill, A. R. (1994). Effect of milk

concentration, pH and temperature on k-casein hydrolysis at

aggregation, coagulation and curd cutting times of ultrafiltrated milk.

Milchwissenschaft, 49, 450–453.

van Boekel, M. A. J. S., & Ribadeau-Dumas, B. (1987). Addendum to the

evaluation of the Kjeldahl factor for the conversion of the nitrogen

ARTICLE IN PRESS



9/9

content of milk and milk products to protein content. Netherlands

Milk and Dairy Journal , 41, 281–284.

van Hooydonk, A. C. M., & Olieman, C. (1982). A rapid and sensitive high-

performance liquid chromatography method of following the action of

chymosin in milk. Netherlands Milk and Dairy Journal , 36 , 153–158.

van Vliet, T., van Dijk, H. J. M., Zoon, P., & Walstra, P. (1991). Relation

between syneresis and the rheological properties of particle gels.

Colloid and Polymer Science, 269, 620–627.Walstra, P., Geurts, T. J., Noomen, A., Jellema, A., & van Boekel, M. A.

J. S. (1999). Dairy technology—principles of milk, properties and

processes. New York, NY, USA: Marcel Dekker, Inc.

Walstra, P., & Jenness, P. (1984). Dairy chemistry and physics (p. 232).

New York, NY, USA: Wiley.

Walstra, P., & van Vliet, T. (1986). The physical chemistry of curd

making. Netherlands Milk and Dairy Journal , 40, 241–359.

Wium, H., Pedersen, P. S., & Qvist, K. B. (2003). Effect of coagulation

conditions on the microstructure and the large deformation properties

of fat-free Feta cheese made from ultrafiltrated milk. Food Hydro-

colloids, 17 , 287–296.Zoon, P., van Vliet, T., & Walstra, P. (1988). Rheological properties of

rennet-induced skim milk gels. 1. Introduction. Netherlands Milk and

Dairy Journal , 42, 249–269.

ARTICLE IN PRESS


Documents

UF Skim Milk Cheeses