Elena M.
Lucchetta
,
Matthew S.
Munson
and
Rustem F.
Ismagilov
*
5735 S. Ellis, University of Chicago, Department of Chemistry, Chicago, IL USA. E-mail: r-ismagilov@uchicago.edu; Fax: 773 702-0805; Tel: 773 -5816
First published on 12th January 2006
This paper characterizes a microfluidic platform that differentially controls the temperature of each half of a living Drosophila melanogaster fruitfly embryo in space and time (E. M. Lucchetta, J. H. Lee, L. A. Fu, N. H. Patel and R. F. Ismagilov, Nature, 2005, 434, 1134–1138). This platform relies on laminar flow of two streams of liquid with different temperature, and on rapid prototyping in polydimethylsiloxane (PDMS). Here, we characterized fluid flow and heat transport in this platform both experimentally and by numerical simulation, and estimated the temperature distribution around and within the embryo by numerical simulation, to identify the conditions for creating a sharper temperature difference (temperature step) over the embryo. Embryos were removed from the device and immunostained histochemically for detection of Paired protein. Biochemical processes are sensitive to small differences in environmental temperature. The microfluidic platform characterized here could prove useful in understanding dynamics of biochemical networks as they respond to changes in temperature.
Using a simple ‘Y’ junction device, the temperature of different regions of a live Drosophila embryo can be controlled in space and in time by flowing two laminar streams of different temperature over the embryo.1 Previously, we have shown that the ‘Y’ junction device can be used to understand the dynamics of the biochemical network in a live Drosophila embryo. Here, we characterize the temperature distribution around the embryo in the device, and characterize the effect of positioning of the embryo in the channel on fluid flow and the temperature distribution around the embryo. Using numerical simulations, we predict the temperature distribution within the embryo. We also confirm previous results that the embryo feels the effect of the temperature, previously observed by staining for the protein Even-skipped. Here, we use histochemical immunostaining11 to detect the protein Paired after the embryo is removed from the channel and fixed to preserve internal structure.11 Paired is a pair-rule protein expressed after Even-skipped in a similar striped pattern along the antero-posterior axis of the embryo. We show that Paired stripes, like Even-skipped stripes, are formed out of order (stripes form in the warm half of the embryo first) but in the correct position.
![]() | ||
Fig. 1 Fabrication and assembly of the microfluidic device. (a) Multi-level polydimethylsiloxane (PDMS) molds were fabricated from multi-level machine-milled brass masters. The top channel was 850 µm × 500 µm (width × height) and the bottom channel was 850 µm × 250 µm (width × height). (b) The embryo was placed on double-sided tape over the bottom channel. Alignment posts of one PDMS mold were snapped into alignment holes of the second PDMS mold. (c) After assembly, one warm laminar stream and one cool laminar stream were flowed around the embryo, maintaining each half of the embryo at a different temperature. |
To create a temperature step across the main channel, each stream was maintained at a controlled temperature using EchoTherm™ digital electronic chilling/heating plates (model IC22, Torry Pines Scientific, Inc.). The fluid was flowed through Intramedic™ Clay Adams™ Brand polyethylene tubing (Becton Dickinson and Company) with an inner diameter of 1.14 mm and an outer diameter of 1.57 mm. The polyethylene inlet tubing was passed through an 80 cm long piece of copper tubing, coiled over the heating/chilling plates and insulated. To minimize heat loss, the length of the tubing connecting the chilling/heating plate and the device was kept to a minimum (typically ∼2 cm). Temperature of the fluid entering the main channel was calibrated at each inlet of the PDMS channel using a Digi-Sense® Thermistor Thermometer (Cole-Parmer Instrument Company). Temperature equilibration through the copper tubing was efficient, and heat loss through the connecting segment was minimized, as evidenced by the similarities of the set temperature of the plate and measured temperature inside the channels. When the set temperature of the plate was 28 °C, the temperature at inlet was 27 °C, and when the set temperature of the plate was 18 °C, the temperature at the inlet was 20 °C. Room temperature was ∼23 °C.
The temperature difference between the two streams of laminar flow was visualized using a suspension of thermochromic liquid crystals (Image Therm Engineering). The response time of the liquid crystals is ∼10 ms—at a flow velocity of 5 cm s−1, each point along the channel is represenative of 500 µm upstream of that point.
The temperature of the two different streams of flow could be rapidly switched on the monitor of the digital electronic chilling/heating plates during experiments in which the temperature of the two laminar streams was reversed. The response time after digitally changing temperatures was ∼5 minutes for the temperature at the inlets to go from 20 °C to 27 °C or from 27 °C to 20 °C. Room temperature was ∼23 °C.
![]() | ||
Fig. 2 Geometry of the channel and finite element mesh used in numerical simulations. All segments of the channel are 850 µm wide in the y-direction and 750 µm deep in the z-direction. The channel outline and the position of the double-sided tape and embryo are shown in the inset. |
To simulate fluid flow over the embyro, the incompressible steady-state Navier–Stokes equations were used [eqn (1)]
![]() | (1) |
To simulate heat transport, the convection-conduction equation was used [eqn (2)]
k∇2T + ρCpv∇T = 0 | (2) |
The channel walls were assumed to be thermal insulators. The temperatures at the inlets were fixed at 300 K and 293 K. The temperature at the surface of the embryo and surface of the tape was assumed to be continuous with the temperature of the fluid. We assumed that the heat flux at the channel outlet was dominated by the convective flux. Simulation convergence was determined using successive mesh refinement. Heat capacities [J kg−1 K−1] of 4180, 130, and 4180 were used for buffer, tape, and embryo, respectively. A thermal conductivity [W m−1 K−1] of 0.6, 0.2, and 0.6 was used for buffer, tape, and embryo, respectively. Densities [kg m−3] of 1000, 1160, and 1000 were used for buffer, tape, and embryo, respectively. Finally, viscosities [kg m−1 s−1] of 0.001 and 0.001 were used for buffer and embryo, respectively.
All models presented here assumed the system was operating in a steady state. This assumption was based on the relatively rapid time scale for heat transfer across the embryo in comparison to the time scale for heat transfer into the PDMS. To verify rapid heat transfer through the embryo, the thermal diffusivity through the eggshell surrounding the embryo and through the cytoplasm was calculated and compared. The thermal diffusivity, DT [m2s−1] was calculated using DT = k/ρCp, where k is thermal conductivity [W m−1K−1], ρ is density [kg m−3], and Cp is specific heat [J kg−1K−1]. The time scale t [s] of heat transport through the eggshell and cytoplasm was approximated by t = x2/2DT, where x [m] is the distance over which heat transport is taking place. The eggshell, which is composed of waxy layers, was approximated by paraffin wax, for which k = 2.5 × 10−3 kJ m−1 s−1 K−1, ρ = 930 kg m−3, and Cp = 2.1 kJ kg−1 K−1, giving DT = 1.3 × 105 µm2 s−1 , t = 0.9 × 10−5 s. The cytoplasm of the embryo was approximated by water, for which k = 6.0 × 10−3 kJ m−1 s−1 K−1, ρ = 1000 kg m−3, and Cp = 4.2 kJ kg−1 K−1, giving DT = 1.4 × 105 µm2 s−1, t = 0.9 s.
Embryos removed from the microfluidic device after being exposed to the temperature step, were hand dechorionated on double-stick tape, fixed in 3% formaldehyde in PEM buffer (0.1 M 1,4-piperazine diethane sulfonic acid (PIPES), 2.0 mM [ethylenebis(oxyethylenenitrilo)]tetraacetic acid (EGTA), 1.0 mM MgSO4, adjusted to pH 6.95) (Fisher Scientific, Fairlawn, New Jersey) and histochemically immunostained according to standard procedures.11 Protein expression was detected through oxidation of diaminobenzidine (DAB) by conversion of H2O2 by horseradish peroxidase. A mouse monoclonal antibody (DP311) that recognizes several Pax3/7 gene products including Paired and goat anti-mouse HRP IgG (H + L) secondary antibody were used to detect Paired protein.14
![]() | ||
Fig. 3 Experimental observation of flow and temperature profile around a live embryo in a microfluidic device. (a) Flow profile observed with a suspension of FluoSpheres. No eddies were observed around the embryo. (b) Temperature profile observed with a suspension of thermochromic liquid crystals. A tight thermal boundary was observed between the two laminar streams of flow around the embryo, confirming the presence of a temperature step (T-step) across the main channel. |
A step-like temperature profile was observed in the main channel by flowing a suspension of thermochromic liquid crystals. A tight thermal boundary was experimentally observed at the interface of the two laminar streams of flow (Fig. 3b). The temperature profile across the main channel and the embryo is therefore referred to as a temperature step (T-step).
![]() | ||
Fig. 4 Velocity and temperature profiles determined by numerical simulation. (a) Stream lines color-coded with fluid temperature mapped the paths of the fluid as it flows around the embryo. Eddies would appear as closed loops, and were not observed. (b) Velocity field at the vertical mid-plane of the channel. The fluid was observed to accelerate as it flowed around the embryo. (c) Temperature profile at the vertical mid-plane of the channel. Conduction through the embryo broadened the transition between the high and low temperatures. (d) Temperature distribution within the embryo. The temperature profile was taken at a cross-section through the middle of the embryo. |
The numerical simulations also considered heat conduction within the embryo. The steady-state temperature distribution for the cross section along the antero-posterior axis of the embryo is shown in Fig. 4d. The temperature profile within the embryo is also step-like.
![]() | ||
Fig. 5 Effect of embryo position on temperature within the embryo. (a) The layout of the computational domain is shown from above for the normal embryo position. (b) The temperature profiles along the hemisphere of the embryo are plotted as a function of embryo length. (c)–(h) The temperature on the fluid/embryo interface is shown projected onto the x–y plane for each of the simulated embryo positions. (i) Flow velocity profile around an embryo shifted 36 µm in the y direction. |
Normal position of the embryo centered across the main channel with the length of the embryo perpendicular to flow resulted in a step-like temperature profile (Fig. 5b, c). The effect of rotating the embryo 3° away from the perpendicular was negligible, shifting the temperature profile by ∼2% embryo length (Fig. 5b, d). Shifting the embryo to the edge of the tape upstream, towards the converging point of the two laminar streams, did not shift the temperature profile along the embryo length, but slightly sharpened the thermal boundary (Fig. 5b, e), presumably due to the shorter contact time between the two streams upon reaching the embryo. The effect of embryo offset on the temperature distribution within the embryo was more pronounced. The smallest offset detectible by visualization, 20 µm, caused the temperature profile to be shifted ∼8% embryo length (Fig. 5b, f), while the maximum shift of 36 µm caused the temperature profile to be shifted ∼16% embryo length (Fig. 5b, g). This shifting of the boundary between the two different temperature laminar streams can be explained by a change in the flow velocity between the tip of the embryo and the channel wall. The regions between the tips of the embryo and the walls are constricted. Shifting the embryo opens one constriction while closing the other, leading to the corresponding increase and decrease of the flow velocity, and the shift of the boundary between the two streams. Therefore, small changes in distance have a significant impact on flow rate.
Under normal conditions, Paired is expressed in a complex spatial and temporal pattern of seven stripes along the antero-posterior axis of the embryo.17,18 Normally, Paired stripe 1 is expressed first with Paired stripe 2 forming slightly afterwards. Paired stripe 7 is then formed, followed by Paired stripes 3-6.17,18 During early gastrulation, the 7 broader regions of Paired stripes form a 14 striped pattern with gooseberry, a segment-polarity gene.
Development of embryos exposed to a temperature step of 20 °C/27 °C was affected by the two different temperatures. Paired stripes formed out of order, with stripes forming in the warm half of the embryo first (Fig. 6a, c). This result was particularly striking in embryos in which the anterior half was warm and the posterior half was cool (Fig. 6c) in that stripes 3–5 formed before stripes 6–7 were expressed. This is highly unusual, since stripe 7 always forms before stripes 3–5 under normal conditions. While stripes were expressed in the wrong order (Fig. 6a, c), the later 14 striped pattern of Paired and Gooseberry expression was in the correct position (Fig. 6b, d). The developmental network that controls later pair-rule and segment-polarity gene expression remained robust even under the unnatural condition of the T-step.
![]() | ||
Fig. 6 Paired protein expression in embryos exposed to a T-step of 20 °C/27 °C. (a)–(b) Embryos exposed to a T-step with the anterior half at 20 °C and the posterior half at 27 °C. (c)–(d) Embryos exposed to a T-step with the anterior half at 27 °C and the posterior half at 20 °C. Paired stripes resolved in the warm half of the embryo first, shown by white arrows in a and c, but the later 14 striped pattern of Paired and Gooseberry expression was in the correct position. Anterior is to the left, posterior is to the right in each image. Fluid flow was from the top. |
This journal is © The Royal Society of Chemistry 2006 |