Quality and reproducibility
Quality aspects / reproducibility
For many screening projects, the amounts of biomass and cell products generated in Erlenmeyer shake flasks far exceed the amounts required for subsequent analysis or bioassay. Researchers in many areas have - of course - realised this for quite some time, but were hesitant to switch to microplates because of the worry for quality loss. This concern was justified by the following problems:
- Wells were not adequately closed so that evaporation rates were too high and cross-contamination could occur
- Well-to-well variability: wells in the outside rows received more oxygen and lost more water
- Oxygen transfer rates were often more than 10 fold lower than in shake flasks because of too low shaking frequencies and amplitudes (high shaking frequencies would have caused cross-contamination).
In our system, all of these problems have been adequately addressed; the wells of microtiter plates covered with our sandwich covers are now a fully mature alternative to Erlenmeyer flasks:
- All wells are hermetically closed by our sandwich covers; evaporation is reduced 20-fold (1-2% per well per day at 30 ºC) while still a sufficient degree of exchange of headspace air with the surroundings is maintained.
- At a working volume of 0.5 ml culture (in a well from a 96-square deepwell plate), the oxygen transfer rate at 300 rpm and a shaking amplitude of 50 mm is 38 mmol O2 l-1 h-1 , which is at the high end of that obtainable in Erlenmeyer flasks. Also for the other types of microtiter plates, suitable incubation conditions resulting in OTRs of 30-50 mmol O2 l-1 h-1 are achievable (for more details see hydrodynamics, oxygen transfer rates).
- All wells behave as individual - exactly identical - mini-reactors; identical hydrodynamics are secured by i) the well defined shape of the wells, ii) the exactly identical mixing force (the centrifugal force), and iii) the individual connections of the wells with the outer air.
Sources of statistical error
With identical hydrodynamics inside the wells, equal headspace exchange rates, and so equal evaporation rates (as secured by our sandwich covers and clamp systems), statistical deviations between independent duplicates come from other sources, most notably:
- Differences in the growth curves, due to biological variation, and different starting amounts of cells (inoculant): use of pre-cultures and “synchronization” of the cultures reduce this source of error.
- Pipetting errors (especially relevant with culture volumes below 0.5 ml): use of high-quality multipipettes and/or pipetting stations-robots, and optimization of pipetting protocols - or even special pipetting courses for technicians - may minimize this source of errors
- Analytical variation (preferably use analytical methods with a high reproducibility, such as HPLC-UV, or infrared-based methods)
Standard deviations obtainable for various types of microplates
On the basis of our own experience combined with feedback from users, the best obtainable standard deviations between independent duplicates for the different microplates are as follows*:
- Square 24-deepwell plates (2.5 ml cultures): 2 % SD (the easiest format to get good reproducibilities after a relatively short optimization project)
- Round 24- well plates (0.75 ml cultures): 2-4 % SD
- Square 96- deepwell and 96-half-deepwell plates (0.3- 0.75 ml cultures): 2-8 % SD (pipetting errors start to play a sigificant role: also inoculations procedures can be critical, especially for strains that do not grow fully homogeneous)
- Round 96-wells plates (0.15 ml cultures): 5-15 % SD (relatively large, especially due to pipetting errors): only applicable for screenings where the best mutants or constructs show titers of the desired molecule that are at least 20 % higher than the starting strain
* note: For parameters that show more biological varaiation than average, SD’s may be higher. Also, such low SD’s may be only be obtainable after optimization projects that can take several weeks