| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
* Western Dairy Center and Department of Nutrition and Food Sciences, Utah State University, Logan, 84322-8700
Wisconsin Center for Dairy Research, University of Wisconsin, Madison, 53706
Department of Microbiology, Weber State University, Ogden, UT 84408-2506
Corresponding author: J. R. Broadbent; e-mail: broadbnt{at}cc.usu.edu.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Key Words: Lactobacillus • nonstarter lactic acid bacteria • cheese flavor
Abbreviation key: NSLAB = nonstarter lactic acid bacteria, RAPD = randomly amplified polymorphic DNA
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
While the types and numbers of starter (or adjunct) bacteria in cheese can be readily controlled, the complexity of NSLAB populations is still largely a matter of chance. The NSLAB biota of cheese is influenced by several factors including process and equipment sanitation in the manufacturing plant and milk heat treatment (Fox et al., 1998; Berthier et al., 2001; Somers et al., 2001). Several studies have demonstrated that substantial heterogeneity exists in the ratio of species and strains that dominate in a given cheese, even between products made in the same processing facility (Sherwood, 1939; Naylor and Sharpe, 1958; Jordan and Cogan, 1993; Williams and Banks, 1997; Fitzsimons et al., 1999 and 2001; Crow et al., 2001; Swearingen et al., 2001). Moreover, a succession of different Lactobacillus spp. and strains may dominate the NSLAB biota of Cheddar cheese at various stages of ripening (Crow et al., 2001; Fitzsimons et al., 2001).
Because the impact of NSLAB on cheese flavor is ultimately determined by the metabolic activities of bacteria that predominate during ripening, individual strains may, in theory, have a profound effect on cheese flavor and body characteristics (Williams et al., 2000). The unpredictable and dynamic nature of NSLAB communities is, therefore, believed to be an important source of cheese flavor defects and inconsistencies (Sherwood, 1939; Crow et al., 2001). If this hypothesis is correct, stringent control of NSLAB populations during ripening should facilitate industry efforts to produce more uniform, high-quality cheese.
One strategy to accomplish this goal is the use of Lactobacillus adjunct cultures that grow in ripening cheese (but do not affect the manufacturing process), possess desirable flavor-producing attributes (or at least have no negative effect on flavor), and suppress the emergence of "wild" NSLAB (Broome et al., 1990a; Crow et al., 2001). Although efforts to enhance flavor development with NSLAB adjuncts have produced mixed results (Peterson and Marshall, 1990), the overall efficacy of this approach is supported by academic studies that showed some adjuncts dominated wild NSLAB counterparts during ripening of Cheddar cheese (Broome et al., 1990b; Crow et al., 2001; Ryan et al., 2001; Swearingen et al., 2001). Furthermore, adjunct addition may help to control culture-related cheese quality defects (Layele et al., 1990; Frohlich-Wyder et al., 2002). Although adjuncts offer good promise as a tool to control cheese NSLAB and improve cheese quality, the dynamic environment of ripening cheese and the variability imparted by differences in cheese composition, added cultures or enzymes, and manufacturing or ripening regimens, suggests NSLAB control may be a difficult goal. Thus, industry efforts to utilize adjunct technology for NSLAB control would profit from a more fundamental understanding of the relationship between cheese environment, adjunct use, and NSLAB population dynamics.
Among the specific protocols used to make cheeses, curd washing or whey dilution steps offer the potential to produce dramatic differences between cheese varieties. Colby cheese is a washed, stirred-curd variety of Cheddar, and both cheeses are commonly manufactured with identical strains of Lc. lactis starter. In Colby manufacture, cold water is added to the curd after most of the whey has been drained. The wash treatment serves two purposes; it removes lactose and lactic acid from the curd, and cools the curd. Removal of lactic acid and lactose gives the finished cheese a lower acid content and a slightly higher pH. Cooling produces a slight increase in cheese moisture content and an initially firmer curd, which in turn produces a more open texture (Johnson, 2001). Whereas Cheddar can be aged for years, Colby is typically consumed within 3 mo since more aged Colby develops a pasty body and is often accompanied by undesirable flavor development.
Although differences in the flavor or flavor intensity of washed versus nonwashed (but otherwise identically produced) cheeses are commonly noted, the basis for this observation is unknown. One widely accepted theory is that the environment of washed curd cheese is not conducive to the development of particular flavor compounds, but many cheese technologists also believe that the changes in acid content and pH somehow alter cheese microbiology. In this study, we investigated NSLAB population dynamics in reduced-fat Cheand Colby cheese, and examined the effect of a Lb. casei adjunct on NSLAB populations.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Cheese Manufacture
Duplicate vats of 50% reduced-fat Cheddar and Colby cheese were manufactured at the University of Wisconsin-Madison from 250-kg lots of pasteurized milk (1.3% fat). The milk was inoculated with 1% Lc. lactis ssp. cremoris SCO213 starter grown at 30°C for 12 to 14 h in steamed (88°C for 45 min) skim milk. An additional set of duplicate vats for each cheese type was inoculated with starter plus 4 ml of Lb. casei LILA (108 cfu/ml).
Fifteen minutes after starter addition, 49 ml of calcium chloride (Rhodia, Inc., Madison, WI) and 19 ml of double-strength fermentation-produced chymosin (Chr. Hansen, Inc., Milwaukee, WI) were added. The coagulum was cut at a milk pH of 6.5 with 0.95-cm knives and allowed to heal for 5 min. After 10 min of gentle agitation, the temperature of the curd and whey slurry was raised from 32°C to 37.8°C over 25 min, then the whey was slowly drained. Curd for the manufacture of Cheddar was formed into slabs, cheddared, then milled when the curd reached pH 5.95. Fifteen minutes after milling, curd for Cheddar cheese manufacture was salted in three additions, 5 min apart, with 0.275% (wt/wt) flake salt (calculated from the original milk weight). For the manufacture of Colby, the curd was washed with 32°C water for 15 min before salt addition (pH 5.86). After salt addition, the curd was packed into 9-kg rectangular stainless steel hoops, pressed for 4 h at ambient temperature, vacuum-packaged, and stored at 7°C for ripening. Cheese pH, fat, moisture, salt, and lactic acid contents were determined as described previously (Weimer et al., 1997).
Microbiological Sampling
In an effort to identify possible sources of cheese NSLAB, microbiological surface samples were collected as described by Somers et al. (2001) immediately before cheese manufacture from inside and outside of cheese vats, floor drains between vats, cheese mill, lab benches, milk lines, and from the stirrer motor. Samples were also collected from pasteurized (uninoculated) cheese milk, then appropriate dilutions were plated on Elliker’s and Rogosa SL agars (Difco) and incubated for 2 d at 30°C.
Microbiological sampling of experimental cheese was performed at d 1, 2 wk, and once per month thereafter. Cheese samples (approx. 20 g) were collected, homogenized in sterile 2% sodium citrate (45°C) using a Seward (London, UK) model 400 Stomacher, then total bacterial counts were enumerated by the pour plate method with Elliker agar. Nonstarter and adjunct lactobacilli numbers were determined using Rogosa SL agar. Plates were incubated anaerobically for 2 d at 32°C (Elliker agar) or 37°C (Rogosa agar), and numbers of starter cfu were determined by subtracting the Lactobacillus ssp. count from the total bacterial count.
DNA Fingerprinting of Bacteria
Twenty isolates were selected at random from Rogosa and Elliker plates (10 per plate) collected from each cheese for microbiological counts at d 1, and at 2, 4, and 6 mo of ripening, and propagated overnight in their respective broth media. Frozen stock samples were prepared in quadruplicate by addition of 0.2-ml overnight culture to 1.8 ml of 9% reconstituted skim milk with 17% glycerol, then stored at -80°C in 2-ml cryotubes. Template DNA for PCR was isolated from 250 µl of the overnight culture as described previously (Broadbent et al., 1998) and stored at -80°C until needed.
Randomly amplified polymorphic DNA (RAPD) fingerprinting (Bassam et al., 1992) was used to investigate NSLAB population dynamics during ripening. The RAPD was performed in 25-µl reaction volumes that contained 2.5 mM Mg2+, 0.4 mM dNTP blend, 1 U Taq DNA polymerase (Invitrogen, Carlsbad, CA), 4 µM oligonucleotide primer (5'-TCAGCAGCCGCGG TAATTC-3'), and 4 µl of template DNA. Negative controls were prepared with sterile water as a substitute for template DNA. Samples were overlaid with 30 µl of PCR-grade sterile mineral oil (Perkin Elmer, Foster City, CA), then 40 PCR cycles were run in a Perkin-Elmer DNA Thermal Cycler model 480 set to the following parameters: denaturation at 94°C for 1 min, annealing for 1 min at 38°C, and extension at 72°C for 2 min. When the cycles were complete, the reaction was finished by incubation at 72°C for 5 min then stored overnight at 4°C. Individual band patterns were compared by visual examination after electrophoresis in 1.5% agarose gels, and representative strains (from each cheese at each sample time) that showed unique RAPD fingerprints were selected for further characterization. When necessary, additional RAPD experiments were performed to determine the similarity between isolates collected at various sample times or from different cheeses.
Identification of Bacteria
Cheese or environmental isolates that displayed a unique RAPD fingerprint were identified to the nearest known phylogenetic relative by partial nucleotide sequence analysis of the 16S rRNA gene (Baruzzi et al., 2000). One set of universal primers, UF1 (5'-AGAGTTTGATCCTGGCTCAG-3') and UR1 (5'-GCTGGCACG TAGTTAGCC-3'), were used to amplify 520 bp from the 5' end of the 16S rRNA gene by PCR. The reactions were performed in 50-µl volumes that contained 1.5 mM Mg2+, 0.4 mM dNTP blend, 1.75 U of Expand high fidelity DNA polymerase (Roche Diagnostics, Indianapolis, IN), 1 µM of each oligonucleotide primer, and 5 µl of template DNA. Negative controls were prepared with sterile water as a substitute for template DNA. Samples were overlaid with 70 µl of PCR-grade sterile mineral oil, then 40 PCR cycles were performed with the following parameters: 2-min soak at 94°C, then 30 cycles of 15 s at 94°C, 1 min at 55°C, and 1.5 min at 72°C. The reaction was finished by incubation at 72°C for 7 min then stored overnight at 4°C. The presence of amplicon was confirmed by electrophoresis in 1.5% agarose gels, then purified with a Bio-Rad Prep-a-Gene kit (Hercules, CA) and bidirectionally sequenced by fluorescent dideoxy chain termination on a Perkin Elmer Applied Biosystems automated DNA sequencer (model 373A). Nucleotide sequence similarity searches were performed using BLAST tools available through the National Institutes of Health Center for Biotechnology Information (www.ncbi.nlm.nih.gov) to determine the species from which each amplicon was most likely derived. In some cases, nucleotide sequence analysis of the 5' 16S rRNA gene fragment proved inconclusive, so a second set of universal primers (UF2 5'-GCACAAGCGGTGGAG-3'; and UR2 5'-TTGTCACCGGCAGTCT-3') were used to amplify a 235-bp distal region of the16S rRNA gene. The PCR, nucleotide sequence determination, and homology searches were performed as described above.
Sugar Fermentation Profiles of NSLAB
The ability of selected NSLAB isolates to produce acid from different carbohydrates was determined with API 50 CHL test kits (bioMérieux, Hazelwood, Mo). The API test strips were prepared as recommended by the kit supplier and scored after incubation for 24 h at 30°C. Test results were communicated to the bioMérieux technical service department, which used the phenotypic data to predict a species identity for each isolate.
Sensory Evaluation and Statistics
Trained sensory evaluation of cheese was performed at 3 and 6 mo by six to 10 experienced judges in a randomized blind design as described previously (Weimer et al., 1997). Cheeses were judged for qualities that included Cheddar flavor intensity (1 = none, 7 = aged), lipase/rancid flavor intensity (1 = none, 7 = pronounced), off-flavor intensity (1 = none, 7 = pronounced), and overall flavor preference (1 = dislike very much, 7 = like very much). Statistical evaluation of cheese composition and sensory data was performed by ANOVA using Minitab software version 9.1 (Minitab Inc., State College, PA), and differences were declared to be significant when P < 0.05.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
. As expected, the washed Colby cheese contained significantly (P < 0.05) less residual lactose at press, and had a significantly higher pH and a significantly lower total lactate content (and thus a significantly lower lactic acid to moisture ratio) throughout ripening versus the nonwashed Cheddar cheese. Experimental Colby also had significantly higher average moisture and salt contents than the Cheddar cheese, but mean percentages of salt-in-moisture or fat in each cheese type were not significantly different (Table 1).
|
, but cheeses made with the adjunct did have a significantly (P < 0.05) lower D-lactate content than control cheeses at 1, 2, and 4 mo of age. The percentage of D-lactate in the total lactate content of Cheddar and Colby cheese made with Lb. casei LILA at d 1, and 1, 2, 3 and 4 mo of age was 2.8, 2.9, 18.6, 23.6, and 28.8%, respectively. Levels for control cheeses at the same time points were 2.7, 7.8, 30.4, 21.4, and 33.3%, respectively.
Sensory Scores
The ANOVA of sensory scores collected at 90 and 180 d by experienced judges in a randomized blind design showed neither cheese type nor adjunct addition had a significant effect (P > 0.05) on Cheddar flavor intensity, lipase/rancid flavor intensity, off-flavor intensity, overall flavor preference, or body/texture preference (data not shown).
Microbiological Counts
No colonies were detected on Rogosa agar from microbiological samples collected from the processing plant immediately before cheese manufacture. Plate counts on Elliker agar were also negative for most areas, but 6.8 x 102 cfu were recovered from the floor drain between cheese vats, and <10 colonies were isolated from the lab bench and stirrer motor. Similar analysis of pasteurized milk samples yielded no colonies on Rogosa agar, while counts on Elliker agar detected 2.9 x 102 cfu/ml.
Enumeration of starter and nonstarter bacteria in experimental reduced-fat cheeses showed all cheeses contained approximately 2.0 x 109 starter cfu/g at pressing, and that the number of viable starter bacteria declined about one order of magnitude during the 6 mo ripening period (Table 2). No NSLAB were detected on Rogosa agar plates from d 1 Cheddar or Colby cheeses that did not contain Lb. casei LILA adjunct, but numbers of adventitious lactobacilli in these cheeses was greater than 104 cfu/g by 4 mo (Table 2). Lactobacillus spp. counts in cheeses that contained Lb. casei LILA adjunct were greater than 106 cfu/g at press and remained above 107 cfu/g throughout the ripening period.
|
and 2). Representative strains for each unique DNA fingerprint were then selected and identified to the nearest taxonomic relative by nucleotide sequence analysis of 16S rRNA gene fragments. As noted in the previous section, no colonies were detected on Rogosa agar from environmental samples or pasteurized cheese milk, and no facultatively heterofermentative lactobacilli were identified among the strains isolated from either source on Ellikers agar. Instead, representative strains of bacteria collected from environmental surface swabs prior to cheese making (Figure 1A) included Bacillus licheniformis, Enterococcus durans, Escherichia coli, and Klebsiella sp. Isolates obtained from pasteurized cheese milk before starter inoculation (Figure 1B) included one strain of Lc. lactis spp. lactis and several strains of Streptococcus spp. phylogenetically related to either S. salivarius or S. bovis.
|
|
and 4). By 4 mo, NSLAB isolates from control reduced-fat Cheddar cheese were all identified as Lb. casei, and this species was also the only one found at 6 mo (Table 3). In contrast, 4-mo-old control reduced-fat Colby contained a variety of NSLAB species including Lb. curvatus, Lb. casei, and Enterococcus spp., and species heterogeneity was also noted in 6-mo-old cheese (Table 4). Addition of Lb. casei LILA appeared to reduce NSLAB species and strain heterogeneity in ripening Colby and young Cheddar. As shown in Figure 2, the RAPD profile of the adjunct was the only pattern observed among NSLAB isolates from d 1 and 2-mo-old cheeses (Figure 2C). Although the adjunct was not detected in 4- or 6-mo-old reduced-fat Cheddar and Colby cheeses, both cheese types yielded fewer strains of wild Lb. casei than did their respective control cheeses (Tables 3 and 4).
|
|
and 4). Conversely, Lb. casei isolates 6.5.2 and 6.5.4 were recovered from 6-mo-old Cheddar and Colby cheeses made with Lb. casei LILA, but were not found in control cheeses.
Carbohydrate Utilization by NSLAB
Analysis of carbohydrate fermentation patterns by 39 dominant NSLAB isolates collected from 2- (8 strains), 4- (16 strains), or 6- (15 strains) mo-old, reduced-fat Cheddar or Colby cheese showed all isolates fermented galactose, glucose, fructose, mannose, and N-acetyl glucosamine. Lactose, esculine, and tagatose were also utilized by at least 75% of the NSLAB isolates collected for characterization from 2-, 4-, or 6-mo-old cheeses, while ribose was fermented by 63 to 75% of these strains, depending on the sampling time. Finally, the ability to ferment mannitol, turanose, cellobiose, maltose, trehalose, and melezitose was slightly more prevalent in dominant strains from 4- and 6-mo-old cheese (50 to 75% positive, depending on the sampling time) than 2-mo samples (38 to 50% positive).
Phenotypic identification of dominant NSLAB isolates through API 50 CHL carbohydrate fermentation profiles showed little consensus with 16S rDNA sequence data, with taxonomic agreement only noted for 12 of 39 strains (31%) analyzed by both methods. Most of the discrepancy between these methods was due to 19 isolates (49% of total) that had metabolic traits most akin to Lb. curvatus, but whose closest phylogenetic relative by 16S rDNA sequence analysis was Lb. casei.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
One of the most promising strategies to control NSLAB populations is to employ well-characterized Lactobacillus spp. adjunct cultures that suppress the emergence of wild NSLAB (Broome et al., 1990a; Crow et al., 2001). Cheese is a dynamic environment, however, and the effectiveness of adjunct technology may be limited by differences in cheese composition or manufacturing and ripening regimens. Colby and Cheddar cheeses are commonly manufactured with identical strains of Lc. lactis starter, but Colby cheese is a washed, stirred-curd variety of Cheddar, and this treatment produces significant differences in the cheese microenvironment (Table 1). Because the intrinsic properties of Colby are less restrictive to microbial growth than those of Cheddar, it was our hypothesis that the environment in the former cheese would accommodate a greater diversity of NSLAB species and strains than the latter. To investigate this hypothesis, we examined NSLAB population dynamics in reduced-fat Cheddar and Colby cheese, and studied the effect of Lb. casei LILA on NSLAB population dynamics in each cheese.
Results showed the NSLAB biota in both cheeses changed over time, but NSLAB populations in Colby cheese were more heterogeneous for a longer period of ripening than those of Cheddar made in the same processing facility (Tables 3 and 4). Addition of the Lb. casei adjunct to Cheddar or Colby cheese did not influence sensory qualities or important compositional attributes, but it did appear to limit the NSLAB species and strain heterogeneity in both cheeses. In addition, even though the adjunct was not detected in 4- or 6 mo-old reduced-fat Cheddar and Colby cheeses, the dominant strains of Lb. casei in each cheese appeared to be most affected by adjunct treatment and not cheese variety. The reasons for these observations are unknown, but they do support the hypothesis that adjunct cultures can help to control cheese NSLAB composition during ripening.
Data from this study also indicated that phenotypic identification of cheese NSLAB may be problematic, and similar observations have been noted by other researchers (Liu et al., 1988; Nigatu, 2000). The most common discrepancy between phenotypic and 16S rDNA sequence data in this study was due to isolates that had sugar fermentation profiles most like Lb. curvatus, but whose 16S rDNA sequence was most closely related to Lb. casei. Because Lb. curvatus and Lb. casei are mesophilic, facultatively heterofermentative species with relatively similar sugar fermentation profiles (Kandler and Weiss, 1986), it is not surprising that efforts to identify these species through phenotypic data may lead to confusion.
The substrates used by NSLAB for growth in cheese have been the subject of considerable research interest and conjecture (Fryer, 1969). Previous workers have shown NSLAB can derive energy from compounds such as lactic acid, citric acid, carbohydrate moieties from glycoproteins, fatty acids, glycerol, and amino acids (Fryer, 1969; Thomas, 1986; Williams et al., 2000). Evidence also suggests that starter autolysis may be an important source of ribose and other nutrients for NSLAB growth (Thomas, 1987; Rapposch et al., 1999). Nonetheless, the substrates required by NSLAB to attain final populations levels of 107 to 109 cfu/g in ripening cheese have not been conclusively identified. Examination of carbohydrate fermentation patterns for NSLAB strains isolated in this study also did not reveal any obvious differences in the sugar utilization profiles for strains isolated at different ripening periods. Although the ability to ferment some sugars was more prevalent in dominant strains from 4- and 6-mo-old cheese than 2-mo samples, the differences were not striking (50 to 75% vs. 38 to 50% positive) and could have simply been due to variation in the sample size.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
|
ACKNOWLEDGEMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
|
FOOTNOTES |
|---|
Received for publication October 18, 2002. Accepted for publication February 26, 2003.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  H. Cai, B. T. Rodriguez, W. Zhang, J. R. Broadbent, and J. L. Steele Genotypic and phenotypic characterization of Lactobacillus casei strains isolated from different ecological niches suggests frequent recombination and niche specificity Microbiology, August 1, 2007; 153(8): 2655 - 2665. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. Upreti, L. E. Metzger, and K. D. Hayes Influence of Calcium and Phosphorus, Lactose, and Salt-to-Moisture Ratio on Cheddar Cheese Quality: Proteolysis During Ripening J Dairy Sci, February 1, 2006; 89(2): 444 - 453. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
