Peptides Produced by Selected Lactose-Positive and Lactose-Negative Lactococci in a Model Cheese Ripening System1

Peptides Produced by Selected Lactose-Positive and Lactose-Negative Lactococci in a Model Cheese Ripening System ~ H. M. ABU-TARBOUSH, 2 R. T. MARSHALL, and H. HEYMANN Department of Food Science and NutriUon University of Missouri Columbia 65211

A~A~

Pastemized skim milk was placed in a dilution bottle with buffer (pH 5.1:1: .1) that had been immobilized in agar. Rennet and lactoeocci (109/ml) were added. Microbial growth was inhibited with penicillin, natamycin, and nalidixic acid. Following incubation at 25"C for 12 d, proteinase activities were quantified by the o-phthaldialdehyde assay and peptides produced were determined by reversedphase HPLC. Four lactose-negative mutants were among the 14 strains tested. Lactose-negative mutants were significantly less proteolytic than lactose-positive strains with one exception, but all strains were proteolytic. Three of the lactose-positive strains were two to five times more proteolytic than the other seven. Peptide profiles from reversed-phase HPLC were separated by cluster analysis into four groups with similarities greater than .64. The most proteolytic strains, which were lactose-positive, produced numerous hydrophitic peptides with chromatographic retention times of 5 to 25 rain, whereas the lactose-negative strains produced numerous hydrophobic peptides with retention times greater than 50 min.

sess surface-bound proteinase, lactose-negative (Lac-) mutants generally possess little or none (15). Numerous studies on proteinases and peptidases of lactococci have been reported (6, 7, 10, 11, 12, 18, 21, 22, 33, 34, 37), and numerous methods (2, 4, 13, 14, 17, 30) have been used to quantify proteolysis. Several methods exist for extraction of peptides from cheese or protein hydrolysates (19). Extraction with ethanol appears to be most satisfactory (20, 29). Furthermore, HPLC is now available to replace the less satisfactory methods of assaying for peptides and amino acids, e.g., chromatography by ion exchange or gel filtration methods or electrophoresis. In the present research, the resting cells of 14 strains of lactococci were incubated singly in skim milk that had been treated with the amount of rennet used in cheese making. This proteinaceous substrate was held in a pH range characteristic of Cheddar cheese. As proteolytic activity peaked, samples were analyzed for peptides. The objective was to provide peptide profiles for use in selection of lactococci suitable for use in experiments on accelerated cheese ripening. MATERIALS AND METHODS

Identities, lactose-fermenting ability, and sources of the 14 cultures studied are shown in Lactococci possess several proteinases and Table 1. Twelve were Lactococcus lactis ssp. peptidases that hydrolyze proteins to peptides cremoris and 2 were L. lactis ssp. lactis as and amino acids producing desirable cheese reported by sources. Other work in this laboraflavor (27, 28) and sometimes bitterness (35). tory suggests some of the swains called L. Whereas lactose-positive (Lac+) lactocoeci pos- cremoris are actually L. lactis. Lactose-positive strains were grown in M17 medium (36), whereas Lac- strains were grown in modified M17 medium in which glucose was Received December 19, 1988. substituted for lactose (MM17). Tubed media Accepted June 26, 1989. were sterilized at 1210C for 15 rain. Cultures IContribution from the Missouri Agricultural Expcximgat were transferred weekly during experiments Station. Journal Series Number 10718. and were propagated at 300C for 16 h. 2King Saud University, Riyadh, Saudi Arabia. INTRODUCTION

1989 J Dairy Sci 72:3143-3148

3143

3144

ABU-TARBOUSH ET AL.

TABLE 1. Protoalytic activities of lactococci against skim milk proteins alter 12 d of incubation in the model system

T m Monet Symm

The model system was made as follows: a 25-ml mixture o f B-glycerophosphate (19 g/L) and agar (20 g/L) was added to each dilution LacWcocci Source1 A3,10 bottle before sterilization (121"C for 15 rain). cremoris The buffer system was allowed to solidify (to NCDO924 BAL + 3.95" immobilize the buffer) on a 45* angle in the B1M UMC + 2.65c bottle to maximize the surface area for contact KH TK + 1.14d 145D CHL + 1.13d with 25 ml of skim milk, which was then lip TK + 1.09d ad,:led. 177 CHL + 1.05d Skim milk was prepared from fresh grade A 5P CHL + 1.01de raw milk from the University dairy herd. It L CHL + .93e 843 CHL + .72f consisted of a mixture of about 20% Guernsey KHA1 TK .(38 f g and 80% Holstein milk, The skim milk was I-IPA3 TK .41g pasteurized in the dilution bottles at 63"C for B1BC LLM .38g 30 min. After cooling to 25"C, 1% cell concen/acas trate was added to provide about 109 lactococci SLM1 LLM + 3.15b ml. To inhibit microbial growth, penicillin G, LMO231 WES .53g Delvocid (Natamycin), and nalidixic acid were a'b'¢'d'©'t'gMcans bearing the same superscript are not added at 1 IU, 3 mg, and 200 IU/ml, respectivesigniticandy different (P<.01). IBAL = B. A. Law, University of Reading, England; ly. Finally, 1:5000 single strength rennet (Chr. CHL = Chr. Hansens Laboratory, Inc., h/lilwm3k~¢,Wl; Hansen's Laboratory, Inc., Milwaukee, WI) LLM = L. L. McKay, University of Minnegxa, St. Paul; was added. The mixture was incubated at 25"C TK = T. Klaenlaammer, North Carolina State University, for 12 d in a shaking water bath. Although the Raleigh; UMC = University of Missouri-Columbia; and buffer system restricted reductions in pH, it was WES = W. E. Sandine, Oregon State University, Corvallis. necessary to adjust the pH periodically to above 2Measured by o-phthaldialdehyde assay. A340= Absor- 5.0 with sterile .1 N NaOH. bance at 340 m . SD = :1:.11; n = 2. Values are differences between treated and control samples. Proteinase Activity in the Model System Mean

Strain of

Ferment lactose

value 2

Stability of mutants was checked routinely by observing colonies produced by streaking and incubating lactose indicator agar (24). Purity o f cultttres was checked by streaking on blood agar and making Gram stains of broth cdtures. P m l m m t i o n o! Cell Concentrates

After three consecutive daily transfers, cells were subeultured with 1% inoculum in 250 ml of M17 or MM17 broth. Cells were centrifuged at 5000 x g for 30 rain and resuspended to a concentration of approximately 1010/ml in sterile reconstituted skim milk (10%). Numbers of cells were determined by nephelometry at 600 nm in a Beckman 25 DB spectrophotometer (Beckman Instruments Fullerton, CA) with reference to a standard curve prepared for each culture. Journal of Dairy Science Vol. 72, No. 12, 1989

The o-phthaldialdehyde (OPA) spectrophotometric assay (3) was used to monitor proteolysis in the model system. The O P A reagent was prepared essentially as described by Goodno et al. (14). The following reagents were combined and diluted to 100 ml with water: 50 ml of 100 mM sodium tetraborate, 5.0 ml of 20% (wt/vol) SDS, 80 mg of O P A dissolved in 2.0 ml of methanol, and 200 ~tl of B-mercaptoethanol. The OPA stock solution (40 mg OPA/ml methanol) was prepared every 2 wk and stored in an amber bottle. Working OPA reagent was prepared daily. To measure proteolysis, 5.0 ml of sample from the model system was taken every 3 d for 12 d. To this sample was added 1 ml of water and 10 ml .75 N T C A while stirring vigorously. After 10 min, the solution was filtered (Whatman #2 filter paper, Whatman, Clifton, N J), and 100 ~tl of the filtrate was added to 2.0 ml of O P A reagent in a quartz cuvette. The thoroughly mixed solution was incubated for 2 rain

PEPTIDF_~FROM LAC'IX3CXX2CI at room temperature, and the absorbance at 340 um was measured in a Beckman 25 DB spectrophotometer (Beckman Instruments, Fullerton, CA). A blank was prepared by adding 100 of .05 M phosphate buffer to 2.0 ml OPA reagent. Analyses for Peptides Peptides were extracted from samples using 70% ethanol according to the method of Reville and Fox (29). Extractions were performed on inoculated and uninoculated samples that had been incubated 0 and 12 d. Separation of the peptide extracts was carried out by reversed-phase HPLC (RP-HPLC) by the method of Champion and Stanley (1) except that acetonitrile (90%) was used instead of methanol. Water and acetonitrile were of HPLC grade. The solvents were 1-120, .1% trifluoracetic acid (Solvent A), and 90% acetonitrile in water plus .1% trifluoracetic acid (Solvent B). A Vydac RP-C18 column (3000A pores, 5-lain particle size, 250 x 4.6 mm) from The Separations Group (Hesperia, CA) was used for analyses. A Perkin-Elmer Series 4 Chromatograph (Perkin-Elmer Corp., Norwalk, CT), equipped with an LCO95 UV/vis detector and an LCi-100 integrator, was used for separation. For injection of samples, a model 7125 injector was used (Rheodyne Inc., Cotati, CA). Extracts were filtered through a .2-pan membrane filter (Millipore Corp., Bedford, MA). Ten Mieroliters of the peptide extract was injected into the column. A linear gradient from 2 to 35% solvent B at a flow rate of 1.0 ml/min was used to elute the peptides for 70 rain. Separation was conducted at room temperature. Eluted peptides were monitored at 220 run. The column was cleaned after each injection by the use of a linear gradient from 35 to 100% solvent B developed over 10 min. Before the next injection, the column was allowed to equilibrate with 2% solvent B for 25 to 30 rain. Statlmi~l Analyses Data from proteinase assays were subjected to analysis of variance (32). Differences among means were determined using Duncan's new multiple range test (8). Type I error was set at 1%. Data from peptide assays were subjected to cluster analysis [K nearest neighbor; (9, 31)].

3145

RESULTS AND DISCUSSION Of the 14 strains of lactococci studied, the following are frequently used to make Cheddar cheese: NCDO924, 145D, 843, 177, 5P and L. Strain liP is known to cause bitterness in cheese. Lactose-positive strains SLM1 and B1M were isolated from Lac- strains. Lactosenegative strains KHA1, HPA3, B1BC, and LMO231 were chosen for comparative purposes. Strain KH is the parental strain of KHA1. The model system was prepared using rennet-treated pasteurized high quality skim milk adjusted to pH 5.0 to 5.2. During incubation at 25"C, the resting cells of lactococci hydrolyzed milk proteins relatively rapidly because of the high temperature and moisture content. Proteolytic activities varied widely among the 14 strains (Table 1). Lactococcus cremoris strains NCDO924 and B1M plus L. lactis strain SLM1 were 2 to 10 times more active, as evidenced by the OPA reaction, than were the other strains. Five L. cremoris strains, KH, 145D, HP, 177, and 5P, were moderately and insignificantly different in their activities. Exterkate (12) found both an acid and a neutral proteinase in L. cremoris HP. All Lac- swains were weakly proteolytic and did not differ significantly among themselves in this regard. Activities of these Lac- mutants, compared with that of their Lac+ counterparts, were 60% (KHA1 vs. KH), 38% (HPA3 vs. HP), 17% (LMO231 vs. SLM1), and 14% (B1BC vs. B1M). Exterkate (11) observed that a slow variant of HP had lower proteolytic activity than its parent. Increases in free amino groups over time are shown for 8 of the isolates (Figure 1). Rates of hydrolysis appeared to decrease slowly with time, reflecting decreases in active enzyme from the inhibited cells. Following determinations of quantitative differences in proteolytic activities among Lac + and Lac- lactococci, assays were made of the amounts and types of peptides produced. Based on the latter data, cluster analysis (9) was used to classify the 14 strains of lactococci. The dendrogram in Figure 2 indicates that a neighboring pair of groups of strains linked with a solid line has a high correlation coefficient; e.g., r for HPA3 vs. B1BC is .99 and for SLM1 vs. liP is .92. Furthermore, r for swain KH with Jouraal of Dairy Science VoL 72, No. 12, 1989

3146

ABU-TARBOUSH El" AL.

3.0-

.250

SLM1 B1M

SLMI

2.5. E

.125

~2.0~ L5~ .000

I

.........I.........I.........I:........t.........,,I................I

~I.0"

HP KHAI

LM0231 HPA3 B1BC

.5'

3

i 9

6 DAYS

i I~

.250

.125

Figure 1. Protcinase activities (change in absorbanee at 340 am in o-phthaldialde.hyd¢ test) of lactococci (parents and mutants) against skim milk proteins in the model ~ .

.000

0

i0

20

30 TIME

CORRELATIONCOEFFICIENT .9 .8 .7 .6 I I I l

.5 I

40

50

60

70

(min)

Figure 3. Chromatogramsof hydrolysates of skim milk proteins generated by lXoteases of Lactococcus lactis SLMI (group A) and Lactococcus cremoris NCDO924 (Stoup B) during incubation in the model system at 25"C for 12 d. SLM1 produced numerous peptides retained 30 to 60 rain on the column; both SLMI and NCDO924 produced large amounts of hydruphilic pel~ides retained 5 to 30 rata).

KH

B1M

}

SLM1

,

924

{

------7.

177

5P

KHAI HPA3

~

M0231F

145D

I Ii

Figure 2. Dcnda'ogramof lactococci classified by analysis of protease and pcptidase activities against rcmacted skim milk proteins in the model system at pH 5.1 :l: .1. Jomnal of Dairy Science VoL 72, No. 12, 1989

strains B1M, SLM1, and HP is .77. Strains with mutual correlation coefficients higher than .64 were classified into single groups. Thus, four groups were established. Cluster analysis is a generic term for a set of techniques that produce classifications from initially unclassified data. Objects in a given cluster tend to be similar to each other in some sense, and objects in different clusters tend to be dissimilar. Cluster analysis can be used to summarize data rather than to find natural or real clusters (9). Chromatograms representative of each group illustrated differences in concentration, number, and retention time (RT) of peptides eluted from the RP-HPLC column. The chromatograms o f peptides generated from skim milk proteins by L. lactis SLM1 (group A) and L. cremoris NCIX)924 (group B) are shown in Figure 3, and those of L. cremoris B1BC (group C) and L. cremoris L (group D) are shown in Figure 4.

PEPTIDES FROM LhCTOCX~CI

3147

.250

RT) and fewer hydrophobic peptides (greater RT) than did group C, represented by swain B1BC (Figure 4). Seventy percent ethanol was used to extract .125 the peptides generated by the proteolytic systems of lactococci from skim milk proteins (mostly casein). Ethanol exwaction is suitable if o the extract is to be analyzed by HPLC because .000 ......... t ......... I ......... I ......... t ......... I ......... I ......... I ethanol can be readily removed by evaporation. Moreover, 70% ethanol more efficiently recovers low molecular weight peptides than other ~ . 250 precipitants (29). Separation of the peptides was conducted at room temperature. Cohen et al. (5) and Hermodson and Mahoney (16) stated that temperature seems to have a minor effect in separation of peptides by RP-HPLC. However, elevated temperature can increase the incidence of artifacts and multiple peaks from a single peptide .000 (24) and it can lower the resolution of the 0 10 20 30 40 50 60 70 method (16, 23). In RP-I-IPLC separation, small and hydrophilic peptides normally elute early from the Figure 4. C~omatograms of hydrolysatesof skim milk column. Large peptides elute later from the p¢oteins generated by proteascs of Lactococcus cremori~ column because they usually have more hydroBIBC (group C) and L. cremoris L (group D) during phobic amino acids than shorter ones. Howevincubation in the model system at 25"C for 12 d. Although both strains produced less hydropl~ilicpeptides than strains er, in some cases, larger hydrophilic peptides SLM1 and NCDO924 (Figure3), BIBC produr~dthe least may elute before shorter, more hydrophobic ones (16, 23). The sequence of amino acids in Amount of hydrophilic peptides. large peptides can be more important than peptide length (16). Also, larger peptides may retain some partly folded configurations that influence which residues are available for Extent of proteolysis, measured by the OPA interaction with the column (16, 23). Retention test, was generally associated with the cluster times for small linear peptides also may be into which the strains fell based on peptide affected by the sequence of amino acids (25). profiles. With a few exceptions, the rank from Results described herein demonstrated large greatest to least proteolysis was A to B to D to variations among the 14 strains of lactococci C. However, the most proteolytic strain, with the respect to the peptides generated by NCDO924, fell into group B along with the their proteases and peptidases. Lac- strain KHA1 that was about 20% as proPham and Nakai (26) used RP-HPLC to teolytic. The two other highly proteolytic analyze water extracts from 41 Cheddar cheeses strains fell into group A and the other Lac- of different ages (mild, medium, aged, and weakly proteolytic strains fell into group C. extra aged). Eight to 13 chromatographic peaks The NCDO924 appeared to have been sepa- with different concentrations correlated with rated from the other highly proteolytic strain, ages. Differences in cheese flavor correlated in SLM1, because of the lower number of pep- pan with the peptides formed in the cheese. tides retained 30 to 70 min on the column Concentrations of some of these peptides in(Figure 3). Both strains generated high amounts creased with age, but those of others decreased. of hydrophflic peptides with RT of 5 to 25 rain. Moreover. peak K was observed only in the SWains of groups C and D had numerous large extra aged cheeses. peaks with RT over 60 rain. Thus, many large We submit that information gained by exampeptides remained after 12 d of hydrolysis. ining the proteolytic activities and the peptide However, group D, represented by sla'ain L profiles of cheese cultures in our model system (Figure 4), produced more hydrophilic (lesser should be useful in predicting effects of the BIBC

L

• 125

......... i ......... L ......... l ......... i ......... J . . . . . . . . i .........

TIME

(min)

Journal of Dairy Science Vol. 72, No. 12, 1989

3148

ABU-TARBOUSH E'I" AL.

cultmes on cheese ripening. The next step is to relate activities in the model system to activities in ripening cheese. ACKNOWLEDGMENTS

The authors express appreciation to Steven Mitchell, Eugene Iannotti, and James Porter for technical assistance and to Tammy Holliday for help with the manuscript. This work was partially supported by King Saud University of Saudia Arabia, Dairy Research Foundation, and the Wisconsin Milk Marketing Board. REFERENCES 1 Champion, H. M., and D. W. Stanley. 1982. HPLC separation of bluer peptides from Cheddar cheese. Can. InsL Food Sci. Technol. J. 15:283. 2 Chert, R. F., C. Scott and E. Trepman. 1979. Fluoresc, nc¢ properties of o-phthaldialdehyds derivatives of amino acids. Biochim. Biophys. Acta 576:440. 3 Church, F. C., H. E. Swaisgood, D. H. Porter, and G. L. Cati¢,nani. 1983. Spectrophotometric assay using ophthaldialdehyde for determination of proteolysis in milk and isolated milk proteins. J. Dairy Sci. 66:1219. 4 Cliffe, A. l., and B. A. Law. 1982. A new method for the detection of microbial proteolytie enzymes in milk J. Dairy Res. 49:209. 5 Cohen, K. A., K. Sch©llenberg, K. Benedek, B. L. Karger, B. Grego, and M.T.W. Heam. 1984. Mobilephase and temperatta¢ effects in the reversed phase chromatographic separation of proteins. Anal. Biochem. 140:223. 6 Demm2t, aud, M. J., and J. C. Gripon. 1977. General mechanism of protein breekdown cheese ripening. Milchwissensehaft 32:731. 7 l~mmzeaod, M. J., and C. Zevaco. 1970. General p¢operties and substrate specificityof an intracellular neutral p r o m from Streptococcus diacetilactia. Ann. Biol. Anita. Biochem. Biophys. 16:851. 8 Duncan, D. B. 1955. Multiple range and multiple F tests. Biometrics 11:1. 9 Everitt, B. S. 1980. Cluster analysis. 2nd ed. Hcinemann Educational Books, Loadon, Engl. 10 Exterkate, F. A. 1975. An introductory study of the proteolytic system of Streptococcus cremoris strm HP. Neth. Milk Dairy J. 29:303. 11Faterkate, F. A. 1976. The proteolytic system of a slow lactic-acid-producing variant of Streptococcus cremoria liP. Neth. Milk Dairy J. 30:3. 12 Exterkaw, F. A. 1976. Comparison of su'ains of Streptococcus cremoris for proteolytic activities associated with the cell wall. Neth. Milk Dairy J. 30:95. 13 Folin, O., and V. Ciocalteau. 1927. On tryosine and tryptophan determinations in proteins. J. Biol. Chin. 73: 627. 14 Cmodno, C. C., H. E. Swaisgood, and G. L. Catignani. 1981. A fluodmettic assay for available lysine in proteins. Anal. Biochem. 115:203. 15 Grieve, P. A., B. A. Lockie and L R. Dulley. 1983. Use of Streptococcus/act/s lac- mutants for accelerating Cheddar cheese ripening. 1. Isolation, growth and properties of a C2 lac- variant. Aust. J. Dairy Technol. 38:10. 16 Hermodson, M., and W. C. Mahoney. 1983. Separation Journal of Dairy Science VoL 72, No. 12, 1989

of peptides by reverse~-plttsehigh-performance liquid chromatography. Page 352 in Methods in enzymology. Vol. 91. C.H.W. Hirs and S. N. Timasheff, cd. Academic Press, New York, NY. 17 Hull,M. E. 1947. Studieson milk proteins.II.Colorimctticdeterminationof the partialhydrolysisof the proteins in milk. J. Dairy Sci. 30:881. 18 Kolstad, J.,and B. A. Law. 1985. Comparative peptide specificity of cell wall, membrane and intracellular peptidases of group N streptococci.J. Appl. Bacte~iol. 58:449. 19 Kuchroo, C. N., and P. F. Fox. 1982. Soluble nitrogen in Cheddar cheese: comparison of extraction procedures. Milcbwissemchaft 37:331. 20 Kuchroo, C. N., and P. F. Fox. 1982. Fractionationof the water-soluble-nitrogen from Cheddar cheese: chemical methods. Milchwissenschaft 37:651. 21 Law, B. A., and L Kolstad. 1983. Proteolyticsystems in lacticacid bacteria.Anionic Leeuwenhuek. J. Microbiol. S¢~ol. 49:225. 22 Law. B. A., M. E. Sharpe, and B. Relier. 1974. The release of intracellular dipeptidase from starter streptococci during Cheddar cheese ripening. J. Dairy Res. 41: 137. 23 Mac|cod, P., and D. F. Gordon, Jr. 1961. Peptidase as som-ces of essential amino acids for lactic streptococci. J. Dairy Sci. 44:237. 24 McKay, L. L., K. A. Baldwin, and E. A. Zottola` 1972. Loss of lactose metabolism in lactic streptococci. Appl. Microbiol. 23:1090. 25 Meek, J. U 1980. Prediction of peptide retention times in high-pressure liquid chromatography on the basis of amino acid composition. Pro¢. Natl. Acad. Sci.77:1632. 26 Pham, A. M., and S. Nakai. 1984. Application of stepwise ~rimin~'t analysis to high-pressm'e ~iquid chromatography profiles of water extract for judging ripening of Cheddar cheese. I. Dairy Sci. 67:1390. 27 Reiter,B.,T. F. Fryer,M. E. Sbarpe, and R. C. Lawrence. 1966. Studies on cheese flavor. J. Appl. Bacteriol. 29: 231. 28 Reiter, B., Y. Sorokin, A. Picketing, and A. J. Hall. 1969. Hydrolysis of fat and protein in small cheese made under aseptic conditions. J. Dairy Res. 36:65. 29 Reville, W. J., and P. F. Fox. 1978. Soluble protein in Cheddar cheese: a comparison of analytical methods. Ir. J. Food Sci. Technol. 2:67. 30 Samples, D. R., R. L. Richter, and C. W. Dill. 1984. Measuring protcolysis in Cheddar,cheese slurries: comparison of Hull and trinitrobenzene sulfonic acid procedures. J. Dairy Sci. 67:60. 31 SAS Institute, Inc. 1985. Page 262 in SAS Users guide: statistics, version 5. SAS Inst., Inc., Cary, NC. 32 Snedccor, G. W., and W. G. Cochran. 1980. Statistical methods. 7th ed. The Iowa State Univ. Press, Ames. 33 Sorhang, T., and P. Solberg. 1972. Dipeptidase activities as some lactic acid bacteria. J. Dairy Sci. 55:675. 34 Sorhaug, T., and P. Solberg. 1973. Fractionation of dipq~tidase activities of Streptococcus lactia and dlpoptidase specificity of some lactic acid bacteria` Appl. Microbiol. 25:388. 35 Sullivan, J. J., and G. R. Jago. 1972. The structure of bitter poptides and their formation from casein. Aust. J. Dairy Tecimol. 27:98. 36 Terzaghi, B. E., and W. E. Sandine. 1975. Improved medium for lactic streptococci and their bacteriophages. Appi. Microbiol. 29,807. 37Thomas, T. D., and O. E. Mills. 1981. Proteolytic enzymes of stan~ bacteria. Neth. Milk Dairy J. 35:255.