Robotics

ttere are four important aspects in the robotics of protein nanocrystallization:

1. Motionduringthedispensing

2. Motion involved in filling/refilling and cleaning

3. Dispensing control

4. Motion involved in inspection of plates tte last aspect, inspection of plates, is not discussed at length as most of the critical steps are found in the first three aspects. Several commercial systems storing, retrieving and moving the plates with reduced vibration, accurate alignment and good temperature control are available (e.g., the Bruker Nonius Crystal Farm).

Robotics required for dispensing should be fast in order to reduce evaporation and increase throughput. Massive parallel-dispensing systems, such as 1,536-nozzle inkjet systems, have almost eliminated the speed problem as all compounds can be dispensed in one step (Cast and Fiehn 2003). Cocktails can be freshly prepared in microliter to milliliter quantities using existing technology and can be "uploaded" to a "multinozzle" pipette and multiple 1,536 arrays can be processed with a very high throughput, tte overall speed is determined by the dispensing speed of the protein - the most precious component. As the amount of protein is often limited, the amount of protein needed for dispensing should be minimal. To reduce dead-volume losses, the optimal dispensing strategy is to dispense the protein sequentially from a highly reliable single nozzle system, which will then take most of the processing time, ttis time can only be reduced by accurate in-flight dispensing. For example, to optimize the protein stability, a chilled protein holder has been developed. One microliter can be dispensed to all wells of a 96-well plate in approximately 15 s (Cartesian synQUAD technology). Most pipetting stations use standard size microtitre plates and the spacing and sizes of future-generation plates can easily be extrapolated, tte plate height has not been considered too much in the early design considerations. When the standard footprint size is kept constant we will likely be faced with very flat plates that will easily bend, and as the assay volumes continue to decrease the plates will inevitable become lower, tte manipulation of plates is at present mainly mechanical, but advances in magnetic materials could lead to plates that are picked up by electromagnets covered by a small elastic buffer to dampen the vibrations upon contact. Another approach could be the use of vacuum tweezers if the arrays become really small.

As screening operations tend to use lower volumes and thus higher-density plates, not only reproducible reagent dispensing but also rapid and highly accurate positioning are required, tte accuracyneeded scales with the radius of the droplets e.g., with y1/3 and there are many robotic systems that achieve the desired resolution, sometimes at amazing speeds (and accelerations). In comparing different high-throughput concepts, measuring the average timing of individual dispensing steps is insufficient. For example, although most inkjet-type dispensing is very fast, this speed is only realized for multiple dispensing of a small number of components (four to eight colors). If washing is required between each transfer, the throughput is substantially reduced. Often, intermediate washing steps or the change of tips is rate-limiting for the overall speed of the dispensing process. For example in the Cartesian dispensing system the dispensing head is composed of either 4, 8, or 16 independent channels that each can be used to dispense a different reagent with a different volume and therefore there is no need to wash the channels during the dispensing process if the number of components is not too high.

tte precise positioning of the droplets in the nanowells as well as the reproducibility of the system can be checked using fluorescence. If the liquid to be dispensed is labeled with fluorophores, the trajectory, final location and the dispensed volume can be verified. Using the intensity of the fluorescence signal, we can calibrate the amount ofliquid that has been deposited in each well.

Moving the dispensing head over the well, stopping, dispensing the drop into the well and then moving to the next well in high-density plates is the usual practice in syringe-solenoid technology. In-flight dispensing, which increases throughput, as no stopping and starting of the dispensing head relative to the plate is required, is more problematic. Here, the actuation frequency and the table speed must be adjusted to deliver the correct number of drops to each well. Nevertheless, dispensing, a high-density plate using multiple nozzles can be completed in tens of seconds, ttis is more than sufficient to keep up with the protein purification steps. Data showing linearity, accuracy and dynamic range for syringe-solenoid dispensing technology are given in Fig. 1.7; the insert shows the linearity for low volumes. Note that the linearity is good both in the submicroliter and in the microliter range. Accuracy tests for nanodispensing have been described by Rose (1999) and Walter et al. (2003). With increasing viscosity of the liquids, the reproducibility of dispensing decreases. In the Microdrop system this has been remedied by equipping the dispensing head with a nozzle heater controlled by a built-in temperature sensor that limits temperature variations to less than 1°C. tte nozzle heater has a dual function: in the range 20-100 mPa s it keeps the temperature constant to avoid viscosity changes by variations of ambient temperature,; above 100 mPa s it is necessary to reduce the viscosity below 100 mPa s by increasing the liquid temperature (see http://www.microdrop. de). Protein solutions cannot be heated, but since these rarely have a viscosity in this high range, this is not a limitation.

0 2000 4000 6000 8000 10,000 12,000

Predicted volume (nl)

Fig. 1.7. Dispensing linearity. The linearity, accuracy and dynamic range of a syringe-valve system (PixSys 3200, Cartesian Technologies). Note the two different ranges that can be used in this system which allows the dispensing of both the protein component and the reservoir solution forvapor-diffusion crystallization experiments. (From Rose 1999)

0 2000 4000 6000 8000 10,000 12,000

Predicted volume (nl)

Fig. 1.7. Dispensing linearity. The linearity, accuracy and dynamic range of a syringe-valve system (PixSys 3200, Cartesian Technologies). Note the two different ranges that can be used in this system which allows the dispensing of both the protein component and the reservoir solution forvapor-diffusion crystallization experiments. (From Rose 1999)

Was this article helpful?

0 0

Post a comment