Three serious problems have been experienced in the manufacture of low moisture cheese using ultrafiltration (UF)- high fat-loss, excessive moisture retention, and poor cheese texture. In this work the causes of these problems were identified, and means of overcoming them were developed. Coagulation and cheese-making experiments indicated that UF concentration of milk shifts the control of rennet coagulation toward the casein micelle collision rate and away from rennet activity, resulting in formation of a rough-textured curd structure that resists syneresis. Use of 4x whole milk retentate, instead of 5x, improved rennet curd structure, syneresis, and UF cheese texture without reducing protein retention in the cheese. Use of increased rennet and reduced set temperature (26°C) also improved curd structure, syneresis, and cheese texture. Washing of the rennet curd prepared from 4x milk retentate during cheese-making, instead of diafiltration of retentate, was found to improve cheese texture, and cheese moisture below 39% was achieved. UF retentate was inconsistent as a starter medium because it offered no protection against bacteriophage proliferation, and the growth of some strains of Lactococcus lactis was impaired in UF retentate. Commercial, internally-buffered pH-controlled starter media were more consistent than fermented retentate starter when used for making cheese from 4x retentate. Low-pressure homogenization of milk at a temperature between 37°C and 45°C increased fat recovery in UF cheese made from 4x ultrafiltration concentrated milk with minimal damage to cheese texture and syneresis. A procedure was developed for the manufacture of quality, high-yield, low-moisture cheese from 4 times ultrafiltration concentrated whole milk. Fat retention in the cheese was 95% and protein retention was 85%

Whole milks were adjusted to pH 5.8, 6.2, or 6. 7 with HCl and batch pasteurized at 63°C for 30 min. Each was concentrated 5:1 (40% total solids) through a single tube polysulfone membrane Abcor ultrafiltration unit. Lactose (L), casein hydrolysate (CH), and one of two brands of yeast extract (YE1, YE2) were added into cooled retentates at 0.1, 0.3, 0.5, 0. 7 or 0.9% and equilibrated overnight at 4°C. Five percent proteinase positive (Prt+) Streptococcus cremoris UC 310+ (v/w) milk based culture was added. Unfortified retentate was also inoculated with 0.1, 0.3, 0.5, 0. 7 or 0.9% starter and pH readings were taken on all samples for 24 h during incubation at 23°C. Similar substrates were inoculated with proteinase negative (Prt-) S. cremoris UC 310-. Lactose had no significant effect on acid production. Casein hydrolysate had a slight positive effect. Yeast extract had a significant effect at all preacidification levels and a significant difference was also noticed between the brands. Mean times required for the proteinase positive culture to reach pH 5.1 in 5x retentate from milk acidified to pH 5.8 were 24, 12, 10, 10, and 24 h for L, CH, YE1, YE2, and the control respectively. Proteinase negative variants of this strain had mean times of >24 h, 14 h, 11 h, 11 h, and >24 h respectively. These time differences were significantly different between Prt+ and Prt- variants. A minimum concentration of 0.2% yeast extract produced the most stimulation while greater quantities provided no additional benefit. Taste panelists were unable to detect yeast extract in retentates fermented by either culture variant.