Monitoring complex media fermentations with near-infrared spectroscopy: comparison of different variable selection methods

Near-infrared spectroscopy (NIRS) is known to be a suitable technique for rapid fermentation monitoring. Industrial fermentation media are complex, both chemically (ill-defined composition) and physically (multiphase sample matrix), which poses an additional challenge to the development of robust NIRS calibration models. We investigated the use of NIRS for at-line monitoring of the concentration of clavulanic acid during an industrial fermentation. An industrial strain of Streptomyces clavuligerus was cultivated at 200-L scale for the production of clavulanic acid. Partial least squares (PLS) regression was used to develop calibration models between spectral and analytical data. In this work, two different variable selection methods, genetic algorithms (GA) and PLS-bootstrap, were studied and compared with models built using all the spectral variables. Calibration models for clavulanic acid concentration performed well both on internal and external validation. The two variable selection methods improved the predictive ability of the models up to 20%, relative to the calibration model built using the whole spectra.     

At-line monitoring of key parameters of nisin fermentation by near infrared spectroscopy,chemometric modeling and model improvement

An analytical procedure has been developed for at-line (fast off-line) monitoring of 4 key parameters including nisin titer (NT), the concentration of reducing sugars, cell concentration and pH during a nisin fermentation process. This procedure is based on near infrared (NIR) spectroscopy and Partial Least Squares (PLS). Samples without any preprocessing were collected at intervals of 1 h during fifteen batch of fermentations. These fermentation processes were implemented in 3 different 5 l fermentors at various conditions. NIR spectra of the samples were collected in 10 min. And then, PLS was used for modeling the relationship between NIR spectra and the key parameters which were determined by reference methods. Monte Carlo Partial Least Squares (MCPLS) was applied to identify the outliers and select the most efficacious methods for preprocessing spectra, wavelengths and the suitable number of latent variables (n _LV). Then, the optimum models for determining NT, concentration of reducing sugars, cell concentration and pH were established. The correlation coefficients of calibration set (R _c) were 0.8255, 0.9000, 0.9883 and 0.9581, respectively. These results demonstrated that this method can be successfully applied to at-line monitor of NT, concentration of reducing sugars, cell concentration and pH during nisin fermentation processes.     

Transfer of Sucrose-Fermenting Ability and Nisin Production Phenotype among Lactic Streptococci

Transfer of sucrose fermentation ability, nisin production, and nisin resistance from Streptococcus lactis to S. lactis and Streptococcus lactis subsp. diacetylactis occurred between cells immobilized on nitrocellulose filters in the presence of DNase. Transconjugants were able to act as donors to transfer the Suc-Nis phenotype in subsequent mating. No changes in sensitivity to lytic phage c2 were noted in S. lactis transconjugants. However, temperature-independent restriction of lytic phage 18-16 was noted in transconjugants of S. lactis subsp. diacetylactis 18-16. Adsorption studies with phage-resistant transconjugants showed that resistance was not due to lack of adsorption by the lytic phage. Physical evidence for the presence of introduced plasmid DNA was not found in lysates of transconjugants.     

An Ecological Study of Lactococci Isolated from Raw Milk in the Camembert Cheese Registered Designation of Origin Area

The genetic diversity of lactococci isolated from raw milk in the Camembert cheese Registered Designation of Origin area was studied. Two seasonal samples (winter and summer) of raw milk were obtained from six farms in two areas (Bessin and Bocage Falaisien) of Normandy. All of the strains analyzed had a Lactococcus lactis subsp. lactis phenotype, whereas the randomly amplified polymorphic DNA (RAPD) technique genotypically identified the strains as members of L. lactis subsp. lactis or L. lactis subsp. cremoris. The genotypes were confirmed by performing standard PCR with primers corresponding to a region of the histidine biosynthesis operon. The geographic distribution of each subspecies of L. lactis was determined; 80% of the Bocage Falaisien strains were members of L. lactis subsp. lactis, and 30.5% of the Bessin strains were members of L. lactis subsp. lactis. A dendrogram was produced from a computer analysis of the RAPD profiles in order to evaluate the diversity of the lactococci below the subspecies level. The coefficient of similarity for 117 of the 139 strains identified as members of L. lactis subsp. cremoris was as high as 66%. The L. lactis subsp. lactis strains were more heterogeneous and formed 10 separate clusters (the level of similarity among the clusters was 18%). Reference strains of L. lactis subsp. lactis fell into 2 of these 10 clusters, demonstrating that lactococcal isolates are clearly different. As determined by the RAPD profiles, some L. lactis subsp. lactis strains were specific to the farms from which they originated and were recovered throughout the year (in both summer and winter). Therefore, the typicality of L. lactis subsp. lactis strains was linked to the farm of origin rather than the area. These findings emphasize the significance of designation of origin and the specificity of “Camembert de Normandie” cheese.     

A novel hybrid GA/RBFNN technique for protein sequences classification

A novel hybrid genetic algorithm (GA)/radial basis function neural network (RBFNN) technique, which selects features from the protein sequences and trains the RBF neural network simultaneously, is proposed in this paper. Experimental results show that the proposed hybrid GA/RBFNN system outperforms the BLAST and the HMMer.     

On-line process monitoring and chemometric modeling with 2D fluorescence spectra obtained in recombinant E. coli fermentations

2D spectrofluorometry produces a large volume of spectral data during fermentation processes with recombinant E. coli, which can be analyzed using chemometric methods such as principal component analysis (PCA), principal component regression (PCR) and partial least square regression (PLS). An analysis of the spectral data by PCA results in scores and loadings that are not only visualized in the score-loading plots but are also used to monitor the fermentation processes on-line. The score plots provided useful qualitative information on four fermentation processes for the production of extracellular 5-aminolevulinic acid (ALA). Two chemometric models (PCR and PLS) were used to examine the correlation between the 2D fluorescence spectra and a few parameters of the fermentation processes. The results showed that PLS had slightly better calibration and prediction performance than PCR.     

Fermentation process tracking through enhanced spectral calibration modeling

The FDA process analytical technology (PAT) initiative will materialize in a significant increase in the number of installations of spectroscopic instrumentation. However, to attain the greatest benefit from the data generated, there is a need for calibration procedures that extract the maximum information content. For example, in fermentation processes, the interpretation of the resulting spectra is challenging as a consequence of the large number of wavelengths recorded, the underlying correlation structure that is evident between the wavelengths and the impact of the measurement environment. Approaches to the development of calibration models have been based on the application of partial least squares (PLS) either to the full spectral signature or to a subset of wavelengths. This paper presents a new approach to calibration modeling that combines a wavelength selection procedure, spectral window selection (SWS), where windows of wavelengths are automatically selected which are subsequently used as the basis of the calibration model. However, due to the non-uniqueness of the windows selected when the algorithm is executed repeatedly, multiple models are constructed and these are then combined using stacking thereby increasing the robustness of the final calibration model. The methodology is applied to data generated during the monitoring of broth concentrations in an industrial fermentation process from on-line near-infrared (NIR) and mid-infrared (MIR) spectrometers. It is shown that the proposed calibration modeling procedure outperforms traditional calibration procedures, as well as enabling the identification of the critical regions of the spectra with regard to the fermentation process.     

应用近红外光谱法估测小麦叶片糖氮比

糖氮比能够反映作物碳氮代谢的协调程度,及时、准确地监测糖氮比对于作物氮素营养诊断和调控具有重要意义.本研究以不同年份、品种、施氮水平的小麦大田试验为基础,获取鲜叶和粉末状干叶近红外(NIR)光谱及糖氮比信息,分别运用偏最小二乘法(partial least squares, PLS)、BP神经网络(back propagation neural network, BPNN)和小波神经网络(wavelet neural network, WNN)3种方法建立了小麦叶片糖氮比预测模型,并利用随机选择的样品集对所建模型进行测试和检验.结果表明： 小麦鲜叶光谱模型预测性能不佳;而干叶片预测模型表现了较好的准确性,在1655~2378 nm谱区范围内基于3种方法构建的干叶粉末糖氮比估算模型,其预测均方根误差均低于0.3%,决定系数均高于0.9.比较而言,WNN法表现最佳.总体显示,近红外光谱法可以准确预测小麦叶片糖氮比状况,为科学诊断糖氮比提供了理论基础和技术途径.     

Improvement of Fermentation Conditions for the Production of X-Prolyl-Dipeptidyl Aminopeptidase from Lactococcus Lactis

The quantitative effects of some fermentation conditions on the production of the enzyme X-prolyl-dipeptidyl aminopeptidase (PepXP)(EC 3.4.14.5) of Lactococcus lactis subsp. lactis and cremoris were studied. The PepXP activity was found both in the membrane and in the cytoplasm, suggesting the presence of multiple molecular forms. Both microorganisms showed higher PepXP activities when glucose (5 g/l) was used as the carbon source and the yeast extract in the culture medium was increased to 3.5 g/l. In these conditions, 226 mU/ml of PepXP activity were obtained with L. lactis subsp. lactis and 235 mU/ml with the subsp. cremoris after 6 h. The best fermentation temperature was in the 30–32 °C range. The enzyme activity remained stable even during the stationary phase.     

Development of a low-cost medium for production of nisin from Lactococcus lactis subsp. lactis

The goal of this project was to develop a lower-cost medium for nisin production, so this bacteriocin could be used in a broader range of industrial fermentation processes. The objectives included: (1) evaluating methods for controlling the inhibitory effect of lactic acid produced during fermentation, and (2) comparing two inexpensive complex media for nisin production. Lactococcus lactis subsp. lactis was cultured in shake flasks on Laurel–Tryptose broth to evaluate a range of buffers for pH control. NaHCO₃ proved to be an effective buffer for increasing nisin production. Subsequent trials then evaluated condensed corn soluble (CCS, a fuel ethanol production byproduct) and cheese whey as inexpensive growth media. CCS was shown to be an efficient, low-cost medium for high nisin titers and yields. These modifications reduced the medium costs for nisin production from $600/kg nisin (based on Laurel–Tryptose broth medium) to $35–40/kg nisin for the corn solubles medium.     

Growth and Energy Generation by Lactococcus lactis subsp. lactis biovar diacetylactis during Citrate Metabolism

Growth of Lactococcus lactis subsp. lactis biovar diacetylactis was observed on media with citrate as the only energy source. At pH 5.6, steady state was achieved in a chemostat on a citrate-containing medium in the absence of a carbohydrate. Under these conditions, pyruvate, acetate, and some acetoin and butanediol were the main fermentation products. This indicated that energy was conserved in L. lactis subsp. lactis biovar diacetylactis during citrate metabolism and presumably during the conversion of citrate into pyruvate. The presumed energy-conserving step, decarboxylation of oxaloacetate, was studied in detail. Oxaloacetate decarboxylase was purified to homogeneity and characterized. The enzyme has a native molecular mass of approximately 300 kDa and consists of three subunits of 52, 34, and 12 kDa. The enzyme is apparently not sodium dependent and does not contain a biotin moiety, and it seems to be different from the energy-generating oxaloacetate decarboxylase from Klebsiella pneumoniae. Energy-depleted L. lactis subsp. lactis biovar diacetylactis cells generated a membrane potential and a pH gradient immediately upon addition of citrate, whereas ATP formation was slow and limited. In contrast, lactose energization resulted in rapid ATP formation and gradual generation of a proton motive force. These data were confirmed during studies on amino acid uptake. α-Aminoisobutyrate uptake was rapid but glutamate uptake was slow in citrate-energized cells, whereas lactose-energized cells showed the reverse tendency. These data suggest that, in L. lactis subsp. lactis bv. diacetylactis, a proton motive force could be generated during citrate metabolism as a result of electrogenic citrate uptake or citrate/product exchange together with proton consumption by the intracellular oxaloacetate decarboxylase.     

Enhanced production of lactic acid through the use of a novel aqueous two-phase system as an extractive fermentation system

 In order to enhance the productivity of lactic acid and reduce the end-product inhibition of fermentation, the partitioning and growth of four different strains of lactic acid bacteria in three different aqueous two-phase systems were studied. Polyethyleneglycol/ dextran, polyethyleneglycol/hydroxypropyl starch polymer (HPS), and a random copolymer of ethylene oxide and propylene oxide (EO-PO)/HPS were used as polymer systems. One strain each of Lactococcus lactis subsp. lactis and of Lactobacillus delbrueckii subsp. delbrueckii partitioned completely to the interface and bottom phase in two-phase systems with low polymer concentrations of EO-PO/HPS100 and EO-PO/ HPS200. The growth and production of lactic acid by two of three L. lactis strains in a two-phase system with 5.5% (w/w) EO-PO and 12.0% (w/w) HPS100 were reduced by less than 10% compared with a reference fermentation in a normal growth medium. The viability of L. lactis subsp. lactis ATCC 19435 was maintained for at least 50 h and with four top-phase replacements during extractive fermentation in the EO-PO/HPS100 system. Moreover, when cell density reached the stationary phase in the first extractive fermentation, the lactate production in this aqueous two-phase system was maintained. Received: 2 October 1995/Received revision: 16 January 1996/Accepted: 22 January 1996     

Components of fermentation medium regulate bacteriocin synthesis by the recombinant strain Lactococcus lactis subsp. lactis F-116

The regulation of the synthesis of bacteriocin produced by the recombinant strain Lactococcus lactis subsp. lactis F-116 has been studied. The synthesis is regulated by the components of the fermentation medium, the content of inorganic phosphate (KH₂PO₄), yeast autolysate (source of amine nitrogen), and changes in carbohydrates and amino acids. The strain was obtained by fusion of protoplasts derived from two related L. lactis subsp. lactis strains, both exhibiting a weak ability to synthesize the bacteriocin nisin. Decreasing the content of KH₂PO₄ from 2.0 to 1.0 or 0.5% caused bacteriocin production to go down from 4100 to 2800 or 1150 IU/ml, respectively; the base fermentation medium contained 1.0% glucose, 0.2% NaCl, 0.02% MgSO₄, and yeast autolysate (an amount corresponding to 35 mg % ammonium nitrogen). The substitution of sucrose for glucose (as the source of carbon) increased the antibiotic activity by 26%, and the addition of isoleucine, by 28.5%. Elevation of the concentration of yeast autolysate in the low-phosphate fermentation medium stimulated both the growth of the lactococci and the synthesis of bacteriocin. Introduction of 1% KH₂PO₄, yeast autolysate (an amount corresponding to 70 mg % ammonium nitrogen), 2.0% sucrose, and 0.1% isoleucine increased the bacteriocin-producing activity of the strain by 2.4 times.     

Effective conversion of industrial starch waste to L-Lactic acid by Lactococcus lactis in a dialysis sac bioreactor

We describe here a simple technological process based on the direct fermentation of potato starch waste (PSW), an inexpensive agro-processing industrial waste, by a potential probiotic strain, Lactococcus lactis subsp. lactis, for enhancing L-lactic acid production. To maximize bioconversion and increase cell stability, we designed and tested a novel dialysis sac-based bioreactor. Shake flask fermentation (SFF) and fed batch fermentation in the dialysis sac bioreactor were compared for L-lactic acid production efficiency. The results showed that the starch (20 g/L) in the PSW-containing medium was completely consumed within 24 h in the dialysis sac bioreactor, compared with 48 h in the SFF. The maximum lactic acid concentration (18.9 g/L) and lactic acid productivity (0.79 g/L·h) obtained was 1.2- and 2.4-fold higher in the bioreactor than by SFF, respectively. Simultaneous saccharification and fermentation was effected at pH 5.5 and 30 °C. L. lactis cells were viable for up to four cycles in the fed batch fermentation compared to only one cycle in the SFF.     

Genome-enabled prediction of genetic values using radial basis function neural networks

The availability of high density panels of molecular markers has prompted the adoption of genomic selection (GS) methods in animal and plant breeding. In GS, parametric, semi-parametric and non-parametric regressions models are used for predicting quantitative traits. This article shows how to use neural networks with radial basis functions (RBFs) for prediction with dense molecular markers. We illustrate the use of the linear Bayesian LASSO regression model and of two non-linear regression models, reproducing kernel Hilbert spaces (RKHS) regression and radial basis function neural networks (RBFNN) on simulated data and real maize lines genotyped with 55,000 markers and evaluated for several trait-environment combinations. The empirical results of this study indicated that the three models showed similar overall prediction accuracy, with a slight and consistent superiority of RKHS and RBFNN over the additive Bayesian LASSO model. Results from the simulated data indicate that RKHS and RBFNN models captured epistatic effects; however, adding non-signal (redundant) predictors (interaction between markers) can adversely affect the predictive accuracy of the non-linear regression models.     

Synthesis of γ-Aminobutyric Acid by Lactic Acid Bacteria Isolated from a Variety of Italian Cheeses

The concentrations of γ-aminobutyric acid (GABA) in 22 Italian cheese varieties that differ in several technological traits markedly varied from 0.26 to 391 mg kg⁻¹. Presumptive lactic acid bacteria were isolated from each cheese variety (total of 440 isolates) and screened for the capacity to synthesize GABA. Only 61 isolates showed this activity and were identified by partial sequencing of the 16S rRNA gene. Twelve species were found. Lactobacillus paracasei PF6, Lactobacillus delbrueckii subsp. bulgaricus PR1, Lactococcus lactis PU1, Lactobacillus plantarum C48, and Lactobacillus brevis PM17 were the best GABA-producing strains during fermentation of reconstituted skimmed milk. Except for L. plantarum C48, all these strains were isolated from cheeses with the highest concentrations of GABA. A core fragment of glutamate decarboxylase (GAD) DNA was isolated from L. paracasei PF6, L. delbrueckii subsp. bulgaricus PR1, L. lactis PU1, and L. plantarum C48 by using primers based on two highly conserved regions of GAD. A PCR product of ca. 540 bp was found for all the strains. The amino acid sequences deduced from nucleotide sequence analysis showed 98, 99, 90, and 85% identity to GadB of L. plantarum WCFS1 for L. paracasei PF6, L. delbrueckii subsp. bulgaricus PR1, L. lactis PU1, and L. plantarum C48, respectively. Except for L. lactis PU1, the three lactobacillus strains survived and synthesized GABA under simulated gastrointestinal conditions. The findings of this study provide a potential basis for exploiting selected cheese-related lactobacilli to develop health-promoting dairy products enriched in GABA.     

Cole–Cole, linear and multivariate modeling of capacitance data for on-line monitoring of biomass

This work evaluates three techniques of calibrating capacitance (dielectric) spectrometers used for on-line monitoring of biomass: modeling of cell properties using the theoretical Cole–Cole equation, linear regression of dual-frequency capacitance measurements on biomass concentration, and multivariate (PLS) modeling of scanning dielectric spectra. The performance and robustness of each technique is assessed during a sequence of validation batches in two experimental settings of differing signal noise. In more noisy conditions, the Cole–Cole model had significantly higher biomass concentration prediction errors than the linear and multivariate models. The PLS model was the most robust in handling signal noise. In less noisy conditions, the three models performed similarly. Estimates of the mean cell size were done additionally using the Cole–Cole and PLS models, the latter technique giving more satisfactory results.