Trushal Vijaykumar
Chokshi
a,
Adela
Ben-Yakar
b and
Nikos
Chronis
cd
aDepartment of Electrical Engineering and Computer Science, University of Michigan, Ann Arbor, MI 48109, USA
bDepartment of Mechanical Engineering, The University of Texas at Austin, Austin, Texas 78712, USA
cDepartment of Mechanical Engineering, University of Michigan, Ann Arbor, MI 48109, USA. E-mail: chronis@umich.edu
dDepartment of Biomedical Engineering, University of Michigan, Ann Arbor, MI 48109, USA
First published on 21st October 2008
We present two microfluidic approaches for immobilizing the roundworm C. elegans on-chip. The first approach creates a CO2 micro-environment while the second one utilizes a deformable PDMS membrane to mechanically restrict the worm's movement. An on-chip ‘behavior’ module was used to characterize the effect of these methods on the worm's locomotion pattern. Our results indicate that both methods are appropriate for the short-term (minutes) worm immobilization. The CO2 method offers the additional advantages of long-term immobilization (1–2 hours) and reduced photobleaching, if fluorescent imaging during immobilization is required. We envision the use of these methods in a wide variety of biological studies in C. elegans, including cell developmental and neuronal regeneration studies.
The unique ability of microfluidics to handle small-size biological objects, motivated scientists to design a variety of microfluidic devices for the precise manipulation of C. elegans. Worm microtraps have been successfully used for correlating interneuronal activity with locomotion patterns as well as for imaging olfactory responses.16,17 Micro-arrays of fixed-size clamps for immobilizing large populations of individual worms have also been developed.18 A two-step single worm immobilizing approach utilizing suction posts19 and its improved version integrating a deformable membrane for stable worm immobilization has also been proposed.20 In all above studies, the worms were immobilized for short period of time (seconds) and the effect of immobilization on the behavior of the worm—as a direct indication of the worm's physiological condition—was not tested.
In this work, we developed a microfluidic device for immobilizing single worms and characterizing on-chip the effect of immobilization to the worm's behavior (figure 1). We explored two approaches for immobilizing worms on-chip: (i) the first approach creates a CO2 micro-environment to cease the worm's movement and it proved to be efficient for the long term worm immobilization (1–2 hours) (ii) the second approach utilizes a deformable membrane to mechanically restrict the worm15,20 and it is appropriate for immobilizing the worm for shorter periods of time (minutes).
![]() | ||
Fig. 1 (A) The microfluidic device consists of the behavior (pictures I, II and III) and immobilization modules (pictures IV and V). The saw shape channel (III) is used to facilitate the revitalization of the worm and the on-chip quantification of the worm's locomotion pattern. PDMS pillars (II) do not allow the worm to enter the position channel. When high pressure (25 psi) is applied to the immobilization channel the worm is compressed on the microfluidic sidewalls (V). Scale bar, 1 mm (left picture). Scale bars are 300 µm, 500 µm, 10 µm, 100 µm, 300 µm for pictures I–IV respectively. (B) Immobilizing the worm by passing a CO2 stream or by pressurizing the immobilization channel (control layer). |
Both techniques can be used to immobilize worms of different age groups (L4's to adults). The proposed techniques are easy to implement and allow worm recovery within a few seconds after immobilization. Moreover, the device architecture of the two techniques allow to optically access the worm through a glass coverslip and thus these techniques are compatible with high resolution optical microscopy.
The CO2 micro-environment is created by passing pure CO2 through the control layer and diffuses through the PDMS membrane into the flow layer. The high permeability of PDMS to nonpolar gases22 results in the fast replacement of air with CO2. The PDMS membrane is 30 µm thick and shows minimum deflection at moderate pressures (∼69 kPa (10 psi)) exerted by CO2.
The compressive immobilization approach utilized the deflection of a thin PDMS membrane to restrict the worm's movement, when high pressure (∼172 kPa (25 psi)) is applied to the control channel. At such high pressures, the thin membrane collapses onto the worm squeezing it into the side of the microfluidic channel (see ESI movie 1†). In order to achieve large membrane deflections, a 20:1 PDMS mixing ratio is used for fabricating the flow layer. This reduces the PDMS elastic modulus by a factor of two23 when compared to the 10:1 ratio used for the flow layer. The width of the microchannel in the immobilization module is made wider (110 µm) than the width of the saw-shape microchannel (90 µm) to allow large membrane deflections.
Worms are loaded into the main flow channel of the chip and manipulated by activating the integrated microfluidic valves (valves 1, 2 and 3 in figure 1) via the control channel. A separate channel (the ‘position’ channel) perpendicular to the main loading channel was used to position the worm inside the behavior and immobilization modules. In order to prevent the worm from entering the position channel, PDMS pillars (figure 1A (II)) are fabricated at its intersection with the loading channel.
![]() | ||
Fig. 2 Sequence of events (five steps) for the on-chip characterization of the worm locomotion activity: 1) the worm is loaded inside the flow channel (valve 2 is closed), 2) the worm is positioned inside the behavior module by controlling the flow in the position channel (valves 1 and 3 are closed), 3) the worm is pushed into the immobilization module (valve 3 is open). The worm is immobilized by pressurizing the immobilization channel in the control layer with air (25 psi) or by applying CO2 (10 psi), 4) the worm is released and sent back into the behavior module (valves 1 and 3 are closed), 5) the worm is forced out of the chip by applying positive pressure into the position channel (valve 3 is open). It should be mentioned that all valves are partially closed when activated, allowing flow through them. That ‘leaky’ operation is typical for valves of rectangular cross-section.20 |
We tracked the displacement over time of a small region (10 µm × 10 µm) of the worm body (figure 3) before, during and after immobilization. We avoided the use of fluorescent markers (e.g. GFP) for tracking the worm body motion to eliminate radiation damage that could cause locomotion abnormalities. Worms were active before immobilization and maintained completely immobile at the presence of CO2 or when pressure was applied (immobilization periods up to 1 hour were performed). During the immobilization period, abrupt displacement peaks of 1.6 µm were observed. We attributed these peaks to movement artifacts generated by the tracking software. None of the methods affected the worm locomotion pattern when short immobilization periods (∼1 min) were considered. The CO2 method proved to be superior for long immobilization periods (>30 min) as the worm showed similar locomotion pattern before and after immobilization.
![]() | ||
Fig. 3 Characterization of the worm's locomotion pattern after subjecting it to CO2 and compressive immobilization for a period of 1 min and 30 min. Graphs indicate the vertical displacement of a 10 µm × 10 µm rectangular area (shown in the upper left figure) in the mid-portion of the worm's body before (I), during (II) and after (III) immobilization. |
A second set of experiments was performed to quantify the off-chip worm behavior after immobilization. Worms were flushed out of the chip to a food-free agar plate after immobilizing them for different time intervals and their average speed was measured (figure 4). As no food was present during immobilization, worms were in their dispersal state while on the agar plate. The dispersal state is initiated after prolonged starvation (>5–10 min off food) and it is identified by the significant decrease in the number of short and long reversal as well as omega turns.24 The corresponding control experiment was performed by loading worms into the chip and flushing them out on an agar plate without immobilizing them.
![]() | ||
Fig. 4 Post-immobilization worm locomotion speed on a food-free agar. The horizontal axis represents the duration of the immobilization step. Worms that were subjected to compressive immobilization for 1 hour did not show any locomotion activity (zero speed). Worms before immobilization (control experiment) had an average speed of 95 µm/s (data not shown). Errors bars represent SEM from 8 worms. |
We observed that the average speed of the worms on the plate decreased as the immobilization period increased. Immobilization periods of 1 hour resulted zero average speed (immobile worms) and speed reduction of ∼70% for the compressive and CO2 immobilization methods respectively. It should be noted that neither methods resulted in the death of worms for the tested conditions. Worms appeared to slowly recover after left on the agar plate for a few hours. Furthermore, we investigated the effect of longer CO2 immobilization periods (up to 4 hours) to the worm's survival. Worms immobilized up to 2.5 hours recovered completely as they were observed to behave normally on the bacterial lawn. Worms immobilized for more than 3 hours did not show any recovery.
We also observed that worms were completely immobilized within 1–2 minutes upon CO2 application. Such a delay (1–2 min) was attributed to the time that is needed for CO2 to diffuse into the microfluidic chamber and partially remove the air content from the worm's body. To validate that hypothesis, we measured the concentration of CO2 and O2 in the immobilization module. The concentration of CO2 was measured using a pH sensitive dye (thymol blue) while the concentration of O2 was measured using an oxygen sensitive, fluorescent dye (tris(4,7-diphenyl-1,10-phenanthrolin) RU dichloride complex). Within 1 min of CO2 application, the CO2 and O2 concentrations reached an equilibrium level of ∼76% and 6–7% respectively. It should be emphasized that the immobilized channel (in the control layer) is closed at its end, not allowing a fresh CO2 stream to continuously circulate through the chip. We believe that such a design of the immobilization channel (closed at its end) along with the leakage through the fluidic interconnections of the setup and the open architecture of the chip (the chip is exposed to air) are the major sources for not achieving an oxygen-free environment.
Several studies have reported the use of CO2 as an anesthetic for invertebrates with an almost immediate recovery upon its removal,25 but the molecular mechanism underlying its anesthetic action is unclear. As CO2 is known to reduce the pH of an aqueous environment, it has been hypothesized that CO2 creates an intracellular acidic environment resulting in the deformation of transmembrane proteins and modification of the cell membrane permeability. However, a recent study in Drosophila melanogaster does not support that argument.26 The presence of CO2 was shown to inhibit the synaptic transmission at the neuromuscular junctions (NMJs) by reducing their sensitivity to glutamate receptors. Other studies hypothesized that CO2 has an effect on acetylcholine and quantify the effect on dopamine and octopamine neurotransmitters.27,28 As these molecular mechanisms have been conserved in C. elegans, we anticipate that the CO2 anesthesia directly affects the NMJ transmission properties.
At the sensory level, CO2 was recently implicated in affecting neuronal functionality in C. elegans through several regulatory molecules.29,30 A CO2 avoidance behavior in well-fed nematodes was found to be mediated by cGMP signaling through the DAF-11 receptor and the TAX-2/TAX-4 cGMP-gated channel. On the contrary, a reduced CO2 avoidance behavior in starved worms was found to be regulated by insulin and TGFβ pathways.
To identify whether the low-oxygen environment (6–7% O2) or the presence of CO2 resulted in the immobilization of the worms, we replace CO2 with pure N2. The presence of N2 considerably slowed down the worm's movement but it did not immobilize them completely. This is consistent with the fact that worms are still mobile at low-oxygen concentration environments.31 We thus believe that the presence of CO2 and the low-oxygen environment act synergistically in altering the functionality of the worm's neuromuscular system.
Furthermore, the absence of oxygen is known to reduce photobleaching of fluorescent markers32 and therefore the CO2 immobilization method can be appropriate for long-term fluorescence imaging experiments such as time-lapse imaging of cell development. Photobleaching due to the removal of O2 was quantified by fluorescent imaging of GPF-expressing motorneurons (figure 5). Worms were immobilized using the two proposed methods and the fluorescent intensity of the neuronal cell body was continuously recorded for 4 min. At the low-oxygen environment, a 20% reduction in the fluorescent intensity was observed, significantly smaller than the 55% reduction observed during compressive immobilization.
![]() | ||
Fig. 5 Photobleaching curves of GFP-expressing neurons during CO2 and compressive immobilization. Plot shows the percentage change in the cell body fluorescence intensity with respect to the intensity measured at time t = 0. Error bars represent SEM from 8 worms. The dashed white rectangle in the picture denotes the position of one of the VA motorneurons where fluorescence was measured. |
In order to show the applicability of CO2 immobilization method for long-term fluorescence imaging, we visualized the movement of bivalent chromosomes in the gonadal cells undergoing meiotic prophase. The bivalent chromosomes undergo restructuring during the late meiotic prophase33 and the ability to visualize the chromosomal movements during this cell cycle may facilitate the understanding of the chromosomal restructuring process. The chromosomal movements were observed in nematodes expressing GFP in the H2B histone proteins. The gonadal cells undergoing meiotic prophase were imaged continuously for a period of 10 minutes (see supplementary movie 2, ESI†). We tracked the movement of the bivalent chromosome (figure 6A) and measured its displacement relative to that of the cell over time (figure 6B). It should be noted that there was no significant photobleaching observed, despite the continuous 10-min illumination.
![]() | ||
Fig. 6 (A) Fluorescent images of gonadal cells undergoing meiotic prophase. Pictures I–IV represent the position of a bivalent chromosome at different time instants, (B) the graph indicates the total displacement of a single bivalent chromosome from its initial position (at t = 0 min) relative to the displacement of the cell. Images were obtained with a 100x oil immersion objective. Scale bar is 2 µm in all images. |
![]() | ||
Fig. 7 Images of worms of different ages (and thus of different sizes) before and during compressive immobilization. Scale bar is 500 µm. |
We anticipate that the proposed immobilization approaches can have an impact in a variety of applications, including neuron ablation for studying axon regeneration and neural circuits and time lapse imaging for studying cell development. Both methods can be further applied to accurately identify mutant phenotypes in genetic screening experiments10 or quantify gene expression over time. The CO2 method offers the advantage of long-term fluorescent imaging due to the creation of low-oxygen environment that reduces fluorescent photobleaching. Furthermore, the proposed methods can be automated to immobilize large number of worms in a serial manner,34 thus making them attractive tools for various high-throughput screening applications.
To obtain the calibration curve between the CO2 concentration and the absorption intensity of the dye (ESI figure S2†), we applied known CO2 concentrations (0.035% (air), 5%, 20% and 100% CO2) through a microfludic device that was hermetically sealed. A CO2-impermeable tubing was used to connect the CO2 tank supply to the microfludic device. The calibration data were curve-fitted using a logarithmic relationship.
Footnote |
† Electronic supplementary information (ESI) available: Supplementary figures, movies and table. See DOI: 10.1039/b807345g |
This journal is © The Royal Society of Chemistry 2009 |