The scale up of the novel, pharmaceutically important pneumocandin (B(0)), from the filamentous fungus Glarea lozoyensis was successfully completed from pilot scale (0.07, 0.8, and 19 m(3)) to production scale (57 m(3)). This was accomplished, despite dissimilar reactor geometry, employing a combination of scale-up criteria, process sensitivity studies, and regime analysis using characteristic time constants for both oxygen mass transfer and bulk mixing. Dissolved oxygen tension, separated from the influence of agitation by gas blending at the 0.07 m(3)-scale, had a marked influence on the concentrations of pneumocandin analogs with different levels of hydroxylation, and these concentrations were used as an indicator of bulk mixing upon scale up. The profound impact of dissolved oxygen tension (DOT) (low and high levels) on analog formation dictated the use of constant DOT, at 80% air saturation, as a scale-up criterion. As a result k(L)a, Oxygen uptake rate (OUR) and hence the OTR were held constant, which were effectively conserved across the scales, while the use of other criterion such as P(g)/V(L), or mixing time were less effective. Production scale (57 m(3)) mixing times were found to be faster than those at 19 m(3) due to a difference in liquid height/tank diameter ratio (H(L)/D(T)). Regime analysis at 19 and 57 m(3) for bulk mixing (t(c)) and oxygen transfer (1/k(L)a) showed that oxygen transfer was the rate-limiting step for this highly shear thinning fermentation, providing additional support for the choice of scale-up criterion.

The filamentous fungus Glarea lozoyensis produces a novel, pharmaceutically important pneumocandin (B(0)) that is used to synthesize a lipopeptide which demonstrates cidal activity against clinically relevant pathogens. A range of unwanted pneumocandin analogs are also produced by the organism. To maintain the unwanted impurities to acceptable levels upon scaleup, a good understanding of the impact of chemical and physical environment on the cell physiology is required, which benefits downstream processing. Pilot-scale studies were performed to determine the impact of dissolved oxygen, temperature, pH, and carbon dioxide on the process. Experiments included multiple fermenters (up to seven) at 0.07 and 0.8 m(3) scale using single source medium sterilization and inoculum. Gas blending was used to separate effects of dissolved oxygen from agitation. The process was significantly influenced by dissolved oxygen level. The critical dissolved oxygen tension (C(crit)) for growth was below 2% air saturation. The C(crit) for production of pneumocandin B(0) was 20% air saturation, with a significant reduction of the specific production rate below this value. In contrast, low dissolved oxygen levels produced a substantial increase of pneumocandins B(1), B(5), and E(0), while high dissolved oxygen levels produced a disproportionate increase of D(5). This sensivity to dissolved oxygen was independent of agitation within a power range of 2-15 kW/m(3). Broth viscosity was impacted below 10% dissolved oxygen, suggesting an effect on morphology. The process was shown to be sensitive to temperature but relatively insensitive to pH and carbon dioxide (in the exhaust gas) within the ranges studied. This scaledown analysis explained phenomena seen at pilot scale and helped define operating boundary conditions for successful scale up to 19 m(3).

Robust in situ biochemical monitoring is essential for the development of substrate feed control to optimize fermentation processes. The scale up of the fermentation for the fungus Glarea lozoyensis can bene®t from such technology to improve the yield of the pharmaceutically important pneumocandin of interest and control the levels of unwanted analogues. A new in situ probe, using a diamond attenuated total re¯ection element, was evaluated at pilot scale for the quantitative measurement of fermentation analytes using Fourier transform mid-IR spectrometry. The new technology was shown to be stable, unaffected by reactor operation conditions of agitation, air¯ow, and backpressure, but sensitive to temperature control. Both glucose and phosphate were simultaneously monitored during a seed fermentation at 280 L pilot scale using complex medium with detection to 0.1 g/L for both analytes. Fructose, glutamate, and proline were monitored at 75 L scale using production media with detection limits of 0.1, 0.5, and 0.5 g/L respectively. Partial least squares calibration/prediction models were created for analytes of interest using off-line reference measurements and speci®c spectral regions. Good ®ts were obtained between off-line measurements and those predicted by in situ mid-IR. Standard errors of prediction (SEP) for glucose (range 18±0.1 g/L) and phosphate (range 11±7.5 g/L) were 0.16 and 1.8 g/L respectively with mean percentage errors (MPEs) around 2.5%. SEP values for the production process: fructose (range 20±0.1 g/L), glutamate (8±0.5 g/L), and proline (12±0.5 g/L) were 0.44, 0.6, and 0.5 g/L respectively with MPEs of 2.2, 5.3, and 10.1%. The technology effectively demonstrates quantitative multicomponent analysis of fermentation processes using in situ monitoring.