SEARCH

SEARCH BY CITATION

Keywords:

  • anesthetic;
  • C. elegans;
  • dendritic transport;
  • Drosophila larvae;
  • intraflagellar transport;
  • L1 early larvae;
  • microfluidic device;
  • mitochondrial transport;
  • pre-synaptic vesicle transport;
  • Q neuroblast division;
  • synaptic growth

Abstract

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References
  8. Supporting Information

Microfluidic devices have been developed for imaging behavior and various cellular processes in Caenorhabditis elegans, but not subcellular processes requiring high spatial resolution. In neurons, essential processes such as axonal, dendritic, intraflagellar and other long-distance transport can be studied by acquiring fast time-lapse images of green fluorescent protein (GFP)-tagged moving cargo. We have achieved two important goals in such in vivo studies namely, imaging several transport processes in unanesthetized intact animals and imaging very early developmental stages. We describe a microfluidic device for immobilizing C. elegans and Drosophila larvae that allows imaging without anesthetics or dissection. We observed that for certain neuronal cargoes in C. elegans, anesthetics have significant and sometimes unexpected effects on the flux. Further, imaging the transport of certain cargo in early developmental stages was possible only in the microfluidic device. Using our device we observed an increase in anterograde synaptic vesicle transport during development corresponding with synaptic growth. We also imaged Q neuroblast divisions and mitochondrial transport during early developmental stages of C. elegans and Drosophila, respectively. Our simple microfluidic device offers a useful means to image high-resolution subcellular processes in C. elegans and Drosophila and can be readily adapted to other transparent or translucent organisms.

Neurons are highly differentiated cells with long axonal processes that extend from the cell body and end at synapses, structures that are essential for synaptic transmission. Most protein synthesis takes place in the cell body, making axonal transport a vital highway for synaptic development and function. Several cargoes such as synaptic vesicles and mitochondria are transported to synapses by microtubule-dependent molecular motors (1–4). Transport of such cargoes is necessary for synaptic development and function. Transport of cargoes also occurs in dendrites and cilia of neurons and is important for function. An important tool to study the dynamics of neuronal transport is high-resolution time-lapse imaging of fluorescently labelled organelles such as vesicles and mitochondria in culture (5–8), in dissected neurons (9–12) and in vivo(13–18). Axonal transport occurs in the time frame of seconds requiring acquisition rates of 3–8 Hz. There are several cellular processes such as neuronal whole-cell calcium responses to stimulus (19) and endocytic processes (20) that occur at similar time scales. Acquiring appropriate in vivo data for any such biological process is at once a challenge and an opportunity. A model organism easily accessible for in vivo cell biological studies is the nematode Caenorhabditis elegans. Owing to its transparency, short life cycle, simple neuronal circuit and well-characterized genetics, this is a useful model organism for studying many cellular processes including axonal transport in an intact animal.

Using C. elegans as a model, we wished to develop a simple microfluidic device to acquire in vivo data of fast cell biological events across all developmental stages. This requires immobilization of the animals at any developmental stage. Current techniques use glue or anesthesia. Both methods have their advantages and limitations: (i) Cyanoacrylate glue can completely immobilize the worm but is toxic and laborious as each animal needs to be individually glued and it is difficult to implement for very young, small animals (21,22). (ii) Anesthetics are convenient and can be used to assay multiple organisms in parallel experiments but anecdotal evidence suggests that they have effects on subcellular processes. Commonly used anesthetics include sodium azide (23–25), levamisole (14–16,26), tricaine (27,28), 1-phenoxy-2-propanol (1P2P) (29) and muscimol (30,31). Many of these anesthetics act on neuronal receptors and consequently they can influence synaptic transmission (32) and possibly other processes in neurons. We found that these anesthetics have significant effects on flux of pre-synaptic vesicles, the parameter that usually correlates best with observed steady-state cargo distributions. These effects are strongest during the early stages of development.

One means to bypass the problems of anesthetics or glue is the use of microfluidic devices (33). Miniaturized microfluidic devices have been used with intact C. elegans to study many neuronal processes like olfactory behavior (19), neuronal axotomy (34–37), synaptogenesis (38) and high-resolution imaging of green fluorescent protein (GFP) expressed in cells and neurons (24,39–41). However, none of the above studies imaged subcellular processes requiring accurate spatial resolution, which depends upon effective animal immobilization. Immobilization has been accomplished in microfluidic devices using a variety of techniques such as suction channels (35,39), tapered devices (19,38,42), thin membrane layers (24,36), a membrane layer with additional cooling (39), temperature-sensitive gel-based immobilization (43) and application of anesthetic gases (40). Hitherto such devices have been mostly used to image relatively large and stationary fluorescent objects over short time scales, but not to image fast events that require highly accurate spatial resolution.

Here, we illustrate the utility of microfluidic technology to collect biologically accurate high-resolution data for various cellular and subcellular processes occurring at different time scales. We show this by using very simple and easily accessible devices to image axonal (44,45), intraflagellar (14,15,46) and dendritic transport (17) and Q neuroblast migration (47,48) in C. elegans.

We have developed a modified membrane-microfluidic device to acquire subcellular millisecond time scale data by imaging GFP-tagged proteins known to be present on various cargoes. We use the same device with larger dimensions to immobilize and image axonal transport in intact young Drosophila larvae. Moreover, in C. elegans, axonal transport data acquired in this microfluidic device appear to be spared from anesthetic-induced changes in transport. Our simple design can be used for C. elegans at all developmental stages with small modifications. Further, we are able to dynamically image for the first time pre-synaptic vesicle transport during early larval development, hitherto not possible using anesthetic preparations. Using our device, we determined that transport of pre-synaptic vesicles increases as synaptic growth takes place.

The principles behind our device can be applied to study other transparent or translucent model organisms. Our device can be readily adapted to such animals simply by changing device dimensions.

Results

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References
  8. Supporting Information

A microfluidic device can be used effectively for subcellular imaging of organelle transport

Transport of pre-synaptic vesicles marked with GFP::RAB-3 was imaged in the posterior lateral mechanoreceptor/touch neurons, henceforth referred to as touch neurons (45,49,50). Each lateral touch neuron can be imaged in entirety but we selected the posterior touch neuron for this study. All the images were oriented such that the cell body of the touch neuron is on the right, with anterograde transport to synapses being directed to the left (Figure 1A,H).

image

Figure 1. In vivo imaging of GFP::RAB-3-marked vesicle transport in 1-day adult C. elegans mechanoreceptor neurons in a microfluidic device. A) GFP::RAB-3 movement was imaged in the posterior lateral touch neuron (PLM) out of the six touch neurons [PLM, anterior lateral touch neuron (ALM), posterior ventral touch neuron (PVM), anterior ventral touch neuron (AVM)]. B and C) Schematic of PDMS membrane device used for C. elegans immobilization. A deflected PDMS membrane using 14 psi nitrogen gas in the control channel in the PDMS2 layer was used for immobilizing C. elegans present in the flow channel in the PDMS1 layer. D) Shows life cycle of C. elegans with the associated body length and body diameter in parenthesis. E) Bright field image of a 1-day adult worm immobilized in a PDMS device. F) Montage of five successive frames acquired at 5 fps with the frame numbers mentioned in each image. Anterograde (solid ‘down’ arrowhead) and retrograde (solid ‘up’ arrowhead) moving vesicles can be clearly seen in the neuronal process (cell body on the right side). G) The image stacks (250 frames are shown) are analyzed using kymograph plugins of ImageJ to visualize particle displacement over time. H) Schematic representation of a neuron with a 120 µm shaded region near the cell body with anterograde and retrograde movements indicated. I) Representative contours for anterograde (solid ‘down’ arrowhead), retrograde (solid ‘up’ arrowhead) and stationary (solid ‘left’ arrowhead) particle tracks are plotted. The shaded box shows the 20 µm window that is used to calculate vesicle flux. J) Kymograph of an L1 worm imaged in a microfluidic device and anesthetized using 0.3 mm levamisole (lev), respectively. Scale bar is 200 µm (E), 10 µm (G and J) and 5 µm (F).

Download figure to PowerPoint

To develop anesthetic-free imaging, we adapted a recently published device by modifying the device size to accommodate larval stages (24,36). Although the fabrication of our device involves bilayer polydimethylsiloxane (PDMS) stacks, it does not require sensitive alignment nor clean room facilities. Our device is simple to use as, unlike pre-existing devices, the worms are fed manually into the microfluidic chip without complex flow valves. Further, we use a water column to apply pressure on the animal instead of using just gas. This allows us to use the same device for all developmental stages, prevents gas bubbles from interfering with data collection and even allows use of devices with small fabrication errors. The operation of the device is shown schematically in Figure 1B,C and a low magnification image of an immobilized worm is shown in Figure 1E. After imaging GFP::RAB-3 transport for 70 seconds in our microfluidic device, worm locomotion was found to be slightly uncoordinated for the first 10 min after release from immobilization compared to wild-type animals that are not immobilized (Movie S1).

Time-lapse movies of GFP::RAB-3-marked vesicles were converted into kymographs (Figure 1G) from which velocity, flux, pause time and pause frequency were calculated. Fluorescent puncta show three kinds of particles: stationary and those moving anterogradely or retrogradely (Figure 1F,I).

Imaging adult animals or animals at late larval stages (L4) is relatively easy using either anesthetics or microfluidic devices. However, it is challenging to image transport during early developmental stages such as L1 (Figure 1D), during which several synapses, notably those of the touch neurons, are formed (51). During this stage, synapse remodeling and synaptic growth take place in other C. elegans neurons (52,53). Thus, it would be valuable if one could robustly image subcellular events such as synaptic vesicle transport at early larval stages. Even under greatly reduced concentrations of anesthetics such that only 50% of the animals were partially anesthetized, we found that no transport of GFP::RAB-3-marked vesicles was observed in L1 animals (Figure 1J) and very little vesicle movement was observed in L2 animals (data not shown). In comparison, we found robust flux of vesicles in touch neurons of an L1 stage animal when immobilized in our microfluidic device (Figure 1J). Thus, our device offers for the first time a means to image pre-synaptic vesicle transport during early development.

Imaging other subcellular and cellular events in C. elegans and Drosophila

Microfluidic devices can be utilized for imaging transport of other neuronal cargo or other cell biological events (33,36). To expand the utility of our microfluidic devices, we imaged intraflagellar transport (IFT), dendritic transport and the migration of neuroblasts in C. elegans. We also imaged mitochondrial transport in intact first instar Drosophila larvae to extend the use of these devices beyond the C. elegans model.

To mark IFT cargo we used OSM-6::GFP expressed in amphid channel cilia (46,54) and to mark dendritic cargo we used ODR-10::GFP, an odorant receptor expressed in two chemosensory neurons (17). We captured time-lapse movies and observed robust anterograde movements of OSM-6::GFP with very few retrograde movements in both the middle and distal segments of the cilia (Figure 2A,B and Movie S2) (46). We observed many anterograde and retrograde events using the dendritic cargo marker ODR-10::GFP (Figure 2F–H and Movie S3). In young L1 larvae, ODR-10::GFP showed very little transport in levamisole and slightly greater transport in microfluidic devices (data not shown). On the other hand, IFT visualized in early larval L1 stages using OSM-6::GFP showed identical transport in levamisole and in microfluidic devices (data not shown).

image

Figure 2. Microfluidic devices can be used to image intraflagellar and dendritic transport in C. elegans and axonal transport in Drosophila larvae. Intraflagellar and dendritic transport of OSM-6::GFP and ODR-10::GFP transgenic C. elegans were imaged using L3 animals. A) Fluorescent micrograph with a corresponding schematic showing OSM-6::GFP anterograde movements in the distal and middle segments of amphid channel cilia. B) Kymographs of OSM-6::GFP for 350 frames of distal (D) and middle (M) segments. C) Bright field image of a first instar Drosophila larvae immobilized under a PDMS membrane. The box in the figure defines the region where movies were acquired. D) Kymograph of mitochondria marked with GFP from time-lapse imaging of a segmental nerve. E) Montage of eight successive frames acquired at 2 fps with the frame numbers mentioned on each image showing anterograde (solid ‘down’ arrowhead) and retrograde (solid ‘up’ arrowhead) moving mitochondria. F) Schematic of the AWB neuron showing the cell body, axon, dendrite and distal cilia. Anterograde and retrograde ODR-10::GFP movements shown in a montage (G) and a kymograph (H). I and J) Time-lapse images of Q neuroblast and its progeny undergoing cell division and migration in early L1 animals. QL divides into two daughter cells QL.a and QL.p and migrates ∼4.5 µm away from each other in 40 min after initial image acquisition of QL (at L1 age: ∼3 h after hatching shown as 0 min) (I). First QL.a and then QL.p divides into respective daughter cells, both divisions within 85 min after initial image acquisition of QL.a and QL.p (at L1 age: ∼4.5 h after hatching shown as 0 min) (J). Individual cells are marked in every panel. L, left; R, right; A, anterior and P, posterior. Scale bar is 5 µm (A, D, E, G, H, I and J), 2.5 µm (B) and 200 µm (C).

Download figure to PowerPoint

We also wished to determine if our device could be used for tracking cellular events that occur in slower time scales. As an example of such phenomena, we chose to image the migration of the Q neuroblast, which undergoes both cell division and migration during early L1 larval development (Figure 2I,J; Movies S4 and S5) (47,48). Sulston and Horvitz report that the QL neuroblast divides into the QL.a and QL.p cells 4 h after the L1 hatches from the embryo (47). We observe that the QL neuroblast divides into its daughter cells close to 4 h after hatching in both 0.3 mm levamisole (n = 4,data not shown) and the microfluidic device (n = 10, Figure 2I and Movie S4). The QL.a migrates away QL.p before it turns around and moves across QL.p (47). The rates of initial movement of QL.a away from QL.p in the microfluidic device and in 0.3 mm levamisole were not significantly different and were measured to be 5.0 ± 0.5 µm/h and 4.6 ± 0.6 µm/h, respectively (n = 6 and 4, respectively, p = 0.7). These numbers are likely to be more accurate compared to ∼2.5 µm/h migration rates estimated by Ou and Vale (48). These authors estimated the rate of migration for the first division using published sketches made from camera lucida drawings in Sulston and Horvitz (47). We are also able to observe the movement of QL.a toward QL.p as they cross (Movie S5) and subsequent division of QL.a and then of QL.p. Sulston and Horvitz have reported that the division of QL.a and QL.p into their respective daughter cells QL.aa + QL.ap and QL.pa + QL.pp takes place within 6 h of hatching (47). We too observed that all four daughter cells are born by 6 h after hatching (n = 4, Movie S6). We observed that the division of QL.a into QL.aa and QL.ap occurs approximately an hour after the first image of QL.a and QL.p is acquired after crossing (n = 4, Figure 2J). The QL.p divides into QL.pa and QL.pp 25–30 min after the birth of the QL.a daughters (n = 4, Figure 2J). Ou and Vale report the corresponding times for the above divisions as between 39 and 111 min after the first image acquisition of QL.a + QL.p and between 33 and 45 min after the birth of QL.a daughters (48). Thus, both immobilization techniques gave similar division times and migration rates and are compatible with the early findings of Sulston and Horvitz, which were obtained without anesthetics (47).

Further, we observe that all L1 animals stayed viable and healthy after imaging using the microfluidic device. By contrast, nearly 35–40% of the animals imaged using anesthetic died during the imaging session. These observations suggest that microfluidic devices may be able to improve the efficiency of data collection during early development.

We also extended the same device principles to first instar Drosophila larvae by altering only the device dimensions. We chose first instar larvae as they are more challenging to dissect for acquiring dynamic imaging data (10–12). We immobilized young first instar larvae in the microfluidic device and imaged mitochondrial transport in sensory neurons (Figure 2C). Mitochondria were visualized using a matrix-targeted GFP (12) expressed in cholinergic sensory neurons (55) of live animals (Figure 2D,E and Movie S7). Larvae released after imaging returned to normal crawling and locomotion within 10 min. The instantaneous velocities in anterograde and retrograde directions were between 0.2 and 1 µm/second and in the range of prior published observations in Drosophila segmental nerves (12).

In summary, our PDMS device can be used for imaging a variety of subcellular and cellular events in vivo in both C. elegans and Drosophila.

Comparison of transport characteristics obtained using microfluidic immobilization with those obtained using anesthetics

As stage 4 larvae or 1-day adult of C. elegans has been commonly used in studying behavior, neuronal signaling and cargo transport, we standardized the immobilization and imaging of 1-day adult worms. To make effective comparisons between anesthetized and immobilized worms (Figure 3A), it was important to ensure that we were capturing nearly all transport events. Thus, we imaged cargo transport of vesicles marked with GFP::RAB-3 at four different acquisition rates of 3, 5, 7 and 9 frames/second (fps) (Figure S1A–D). Increasing the acquisition rate increased the average velocity of the moving particles with the greatest increase occurring between movies acquired at 3 fps and movies acquired at 5 fps (Figure S1A). Although it was possible to acquire movies at higher frame rates and observe a few moving vesicles, it was difficult to analyze the data as the brightness and contrast of the kymograph line depends on the brightness of the particle (Figure S1E–H). Using the fastest acquisitions, we were unable to analyze faint moving vesicles that probably had fewer GFP::RAB-3 molecules on their surface. Thus, we standardized our acquisition rates for all comparisons at 5 fps.

image

Figure 3. Comparison of transport imaging in PDMS membrane device with anesthetic preparations. A) Kymographs of GFP::RAB-3-marked vesicles from a 1-day adult worm when immobilized in a microfluidic device or using 10 mm levamisole. B) Average anterograde and retrograde flux of L4 animals (n = 4) when repeatedly immobilized four times within an hour in the device. Significance was calculated in comparison with the first immobilization. C and E) Comparison of anterograde velocity and flux in worms immobilized using the microfluidic device, with four anesthetics [10 mm levamisole (L), 10 mm tricaine (T), 0.5% 1P2P and 6 mm muscimol] and both anesthetic-cum-microfluidic device. D and F) Velocity and flux of retrogradely moving vesicles under the same experimental conditions. G) Velocity and flux of anterogradely moving OSM-6::GFP L3 animals (n = 5) in the middle and distal segments of amphid channel cilia immobilized in the microfluidic device. H) Velocity and flux of anterograde and retrograde ODR-10::GFP movements (n > 6) immobilized in the microfluidic device. Scale bar is 10 µm (A). Data are mean +/− SEM. Statistical significance was tested for every assay with respect to the device values. p-value < 0.05 (*) and p-value < 0.005 (**).

Download figure to PowerPoint

Multiple anesthetics have been utilized to immobilize worms (25–27,29,31). Most are thought to act on post-synaptic targets (26,32,56,57) and hence would not necessarily be expected to have direct effects on transport in the pre-synaptic neurons. We compared transport parameters of GFP::RAB-3-marked vesicles in 1-day adult animals immobilized using our device versus those anesthetized using 10 mm levamisole, 10 mm tricaine, 0.50% 1P2P or 6 mm muscimol. We chose these concentrations as several published studies have reported the use of these anesthetics at these concentrations. The anterograde velocity showed a very small but statistically significant increase in all anesthetics compared to worms immobilized in the microfluidic device (Figure 3C). Anterograde flux showed larger changes and complex patterns with both increases and decreases in different anesthetics compared with animals immobilized in our microfluidic device (Figure 3E). Retrogradely moving vesicles showed increased velocities in most anesthetics, except in tricaine, when compared with animals immobilized in our microfluidic device (Figure 3D). Retrograde flux in animals immobilized in the microfluidic device differed from the flux in anesthetized animals only in tricaine and muscimol (Figure 3F). These data showed that the application of anesthetics and the use of the microfluidic-based immobilization gave comparable results for the speed of GFP::RAB-3-marked vesicles, but showed greater differences in the number of vesicles moving in the axon.

We also imaged the transport of IFT cargo and dendritic cargo in the microfluidic device using an acquisition rate of 5 fps. The velocity and flux values obtained from anesthetic and microfluidic immobilization were similar (Figures 3G,H and 4E–H). We found that our IFT velocity values (0.70 ± 0.01 µm/second in the middle segments and 1.28 ± 0.03 µm/second in the distal segments) were similar to those reported in the literature (0.68 ± 0.10 µm/second and 1.27 ± 0.19 µm/second in the middle and distal segments, respectively) (14). Anterograde and retrograde velocities of ODR-10::GFP measured in the microfluidic device were 1.46 ± 0.04 µm/second and 0.85 ± 0.02 µm/second, respectively (Figure 3H). These are compatible with the 0.3–2.1 µm/second range of values reported in Dwyer et al. (17).

image

Figure 4. Effect of different anesthetic concentrations on transport properties. Four anesthetics, levamisole, tricaine, 1P2P and muscimol, were tested at three different concentrations on transport parameters. A and B) Average anterograde and retrograde velocities of synaptic vesicles (GFP::RAB-3) transport for all four anesthetics, acquired at 5 fps (n > 5). The flux of particles for anterograde and retrograde moving particles are shown in (C and D). E and F) Average anterograde velocity and flux of IFT particles (OSM-6::GFP) vesicles in the middle and distal segments of amphid channel cilia are compared with different concentrations of levamisole (3, 7 and 10 mm). G and H) Average velocity and flux of odorant receptor carrying vesicles (ODR-10::GFP) in the dendrite of AWB neuron is compared for different concentrations of levamisole (3, 7 and 10 mm). Data represented as mean +/− SEM and are significantly different compared to the lowest concentration of anesthetic and are indicated by *(p-value < 0.05) and **(p-value < 0.005).

Download figure to PowerPoint

In summary, we find that microfluidic devices can be used to acquire time-lapse data of many neuronally transported cargoes with good spatial and temporal resolution. Further, imaging in microfluidic devices does not appear to detrimentally alter transport characteristics.

Microfluidic-based immobilization has minimal effects on in vivo transport

Microfluidic traps use pressure-based immobilization, which could arguably have strong effects on transport. We wished to test if this is the case, as it would limit the utility of our approach. To image transport, the organism has to be immobilized and as we used the lowest pressure possible to immobilize worms, any further reduction in pressure was not possible to acquire high-quality spatial data. To test the effect of pressure-based immobilization, we compared the transport in preanesthetized worms with and without the use of microfluidic traps on GFP::RAB-3-marked vesicles.

We immobilized animals that were pre-anesthetized with levamisole (most commonly used anesthetic) and tricaine in our microfluidic device. We observed no significant differences in values for anterograde velocity or flux and retrograde velocity or flux in these animals as compared to animals that were only anesthetized by the respective drug (Figures 3C–F and S2A,B,E,F). This showed that the application of pressure did not cause any changes in transport parameters in preanesthetized animals and may also not cause strong effects on endogenous vesicle transport.

Further, we are able to image GFP::RAB-3-marked vesicles in the same animal with repeated immobilization every 15 min for nearly 45 min with little effect on cargo transport (Figure 3B). On the contrary, anesthetics show strong effects over the same time scales, especially on anterograde flux, which shows an intermediate increase and then a large reduction when compared to the start of the imaging session, see below (Figure S2H). Such effects persist even in the lowest concentration of anesthetic namely 3 mm levamisole (Table 1). Moreover, imaging in the microfluidic device can be performed immediately after applying pressure, unlike the time lag required after anesthetic application. Taken together, imaging in our PDMS device appears advantageous to study neuronal transport as it permits acquisition of subcellular data and allows repeated imaging while retaining robust transport properties.

Table 1.  Time for immobilization of 1-day adult C. elegans and GFP::RAB-3 flux after immobilization
AnestheticConcentrationTime for immobilization (min)Imaging time until there is ∼30% reduction in flux compared to start of imaging session
Levamisole10 mm1030 min
 6 mm1550 min
 3 mm2590 min
Tricaine10 mm820 min
 8 mm1530 min
 6 mm2030 min
1P2P1.00%1010 min
 0.50%1515 min
 0.25%2520 min
Muscimol10 mm1545 min
 8 mm2590 min
 6 mm30120 min
PDMS device14 psi020 min continuous pressure

Some anesthetics have unexpected effects on neuronal transport in vivo

We and others had observed the cessation of pre-synaptic vesicle transport under certain anesthetic concentrations. Thus, we decided to systematically analyze the effects of anesthetics to determine if microfluidic devices offer a more useful means to image vesicle transport and possibly other subcellular phenomena. Toward this goal we varied the anesthetic concentration from the lowest possible to get sufficient worm immobilization to more commonly published concentrations and analyzed GFP::RAB-3-marked vesicle movies for transport parameters in 1-day adult animals.

The two major parameters we analyzed were velocity and flux. Of the anesthetics studied, levamisole and muscimol are potent agonists, respectively, of nicotinic acetylcholine receptors and GABA receptors present on the muscle at post-synaptic specializations (26,32,56,57). Tricaine affects body wall muscles of the nematode (27), although the exact molecular target is unknown. Owing to the reported sites of action, we expected no direct effects on vesicle transport in the pre-synaptic neuron. This was expected to hold true especially in touch receptor neurons, which do not directly synapse onto muscles (58). On analysis of cargo transport in different concentrations of anesthetics, we see mostly small, but highly reproducible and statistically significant effects on velocity (Figure 4A,B). Other parameters of transport such as pause frequency and pause time (Figure S2I–L) also show complex changes, which while statistically significant may not be biologically significant. Nonetheless, it is clear that upon the use of anesthetics, numbers of stationary particles appear to increase (Figure S2A–D,G). Anesthetic concentration had the strongest effects on the flux of GFP::RAB-3-marked vesicles moving in the neuronal process. Notable reductions in flux were observed in tricaine (Figure 4C,D). Unexpectedly, upon increasing 1P2P concentrations, there were increases in anterograde and retrograde flux. However, the most striking result was obtained with 10 mm levamisole, a common anesthetic concentration used in the literature. At this concentration, we observe that both anterograde and retrograde flux reduce almost by half compared to data acquired using animals anesthetized with 3 mm levamisole. We further observed that it was possible to begin imaging at 10 mm levamisole only after 10 min of exposure to anesthetic, since prior to this many animals are highly motile. The flux of vesicle transport in the animals changed with time, first increasing after 20 min of exposure to anesthetic and then dropping again after 40 min of exposure to anesthetic (Figure S2H). Thus, there appear to be significant differences in the flux of GFP::RAB-3-marked vesicles dependent on time of exposure to the anesthetic.

We also tested the effect of a recently reported gaseous anesthetic CO2 on GFP::RAB-3-marked vesicle transport in vivo(40). We saw that use of CO2 as the gas used to provide pressure in the device also significantly reduces anterograde flux (but not retrograde flux) with modest effects on retrograde velocity compared to animals immobilized using N2 gas (Figure S1I–O).

To determine if the effects of anesthetics are specific to GFP::RAB-3-marked vesicles or if anesthetics have effects on other types of transport as well, we measured velocity and flux of IFT cargo OSM-6::GFP in the middle and distal ciliary segments and dendritic cargo ODR-10::GFP at different anesthetic concentrations of levamisole. Velocities of OSM-6::GFP in the middle segment and of ODR-10::GFP show a small but statistically significant and reproducible reduction (∼8–15%) at higher concentrations of levamisole (Figure 4E,G). However, the flux of ODR-10::GFP, while showing a small reduction in higher concentrations of levamisole, is not as strongly affected as the flux of GFP::RAB-3-marked vesicles or OSM-6::GFP flux in the middle segment (∼40% reduction in both) under the same conditions (Figure 4F,H). By contrast, the anterograde flux measured in our microfludic device of both OSM-6::GFP in the middle ciliary segment and ODR-10::GFP are significantly higher that those measured in 10 mm levamisole (p < 0.05). The velocity and flux of OSM-6::GFP in the distal ciliary segment are unaffected by levamisole concentration (Figure 4E,F). Continued exposure to levamisole does alter ODR-10::GFP anterograde flux with time, showing an intermediate increase at 20 min after exposure followed by a small decrease from the elevated level at 40 min (Figure S2H). Taken together, these data suggest that levamisole has little effect on movements of IFT cargo in the distal segment and greater effects on ODR-10::GFP transport and OSM-6::GFP transport in the middle segment.

These data suggest that care needs to be exercised in acquiring and analyzing data from anesthetic preparations for axonally transported cargo. Similar caution may be appropriate for some other cargoes and some subcellular events. These data support the use of PDMS-based devices as a good alternative immobilization method for imaging certain subcellular events.

The number of transported synaptic vesicles increases with synaptic growth

Synaptic growth occurs in all organisms during development (59). We applied our microfluidic-based imaging protocols to study pre-synaptic vesicle transport during the process of synaptic growth. The main neuronal processes of the posterior and anterior touch neurons of C. elegans grow from nearly 100 µm in young (L1) animals to 400 µm in 1-day adults (60). The posterior touch neuron synapses form in the early L1 (51) and the vesicle cluster grows with developmental age. To study transport during the process of synapse growth required imaging early developmental stages. Imaging transport of GFP::RAB-3 vesicles was performed on synchronized populations of different larval stages using the microfluidic device.

The anterograde flux of GFP::RAB-3-marked vesicles approximately doubles from L1 to adult, while retrograde flux remained relatively unchanged throughout development (Figure 5C,D). The average anterograde velocity showed a small but statistically significant increase during development, while the average retrograde velocity of GFP::RAB-3-marked vesicles increased notably in L4 and adult animals (Figure 5A,B). The doubling of net flux in the anterograde direction correlates with the sevenfold increase in the area covered by GFP::RAB-3 in the synaptic regions (Figure 5E,F). The most parsimonious conclusion is that increased vesicle transport events contribute to synaptic growth or maturation.

image

Figure 5. Imaging GFP::RAB-3 transport in different stages of C. elegans in microfluidic devices. Vesicle transport was imaged in L1, L2, L3, L4 and 1-day adult (1D adult) animals in PDMS membrane devices. A and C) Compares average velocity and flux in the anterograde direction for all stages. Panels (B and D) show the average velocity and flux for the retrogradely moving vesicles in all developmental stages. E) Arrow indicates the GFP::RAB-3 vesicle cluster at synapses at different developmental stages. The size of the vesicle cluster was quantified and plotted as synapse area. The net anterograde flux was calculated as the difference between anterograde and retrograde flux (F, n > 6). Synapse sizes at all stages were significantly different (p-value < 0.005) compared to L1. Scale bar is 10 µm (E). Mean +/− SEM. All comparisons are made with L1 values and denoted by p-value < 0.005 (**).

Download figure to PowerPoint

These data show the utility of microfluidic-based devices to carry out early developmental and cell biological studies.

Discussion

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References
  8. Supporting Information

We have developed a microfluidic device that can be used for subcellular imaging in the millisecond to hour time scales. Using axonal, intraflagellar and dendritic transport as examples, we show both the utility and in some cases advantages of using a microfluidic device as an anesthetic-free preparation to acquire such data. Further, we are able to obtain biologically useful information. For instance, we observed changes in transport parameters that correlate with synaptic growth. Using a larger device utilizing the same immobilization principle, we are able to image axonal transport in young intact Drosophila larvae. The simplicity of our design makes it readily adaptable to other transparent or translucent model organisms and easily usable by laboratories that do not routinely work with PDMS microfluidic technology.

Microfluidic devices offer an alternate means to image cellular and subcellular events

Most of the commonly available anesthetics at different concentrations show strong effects on transport of GFP::RAB-3-marked vesicles in wild-type C. elegans, especially on flux of cargo, e.g. increasing concentration of the anesthetic 1P2P increases the flux (Figure 4C,D). Further, using a single anesthetic we observe changes in flux dependent on the length of exposure to the anesthetic (Figure S2H). Such effects may be mitigated upon using anesthetic combinations (44), nonetheless they too may need to be tested to determine if they have any effects on transport parameters. The strong reduction in flux observed in anesthetics is especially detrimental to collecting data from mutants (e.g. in molecular motors) that have greatly reduced cargo movement (45). Microfluidic devices may be particularly useful in collecting data in these genetic backgrounds.

Other cell biological processes appear to be affected to varying extents by the use of anesthetics. IFT in middle cilia and dendritic transport appear to be less sensitive to anesthetic concentrations and IFT in distal cilia seems to be unaffected (Figure 4E,F). In such cases, appropriately calibrated anesthetic immobilization continues to be a useful method to acquire data. Recently, anesthetics have been observed to delay axonal regeneration in C. elegans compared to regeneration time scales observed in microfluidic devices (36). Further, our own observations with Q neuroblast migrations show that anesthetic application results in significant lethality in very young animals during imaging, which is absent when microfluidic devices are used. These considerations support the use of microfluidic devices as a good option in imaging certain cellular and subcellular events in vivo.

Application of pressure in the microfluidic device can conceivably have detrimental effects on biological processes. But there are only subtle changes in axonal transport parameters in anesthetized animals that are imaged with pressure in the device (Figure 3C–F). Thus, we infer that the effect of the minimal applied pressure may not be as severe as that of anesthetics.

As we apply mechanical pressure and image transport in touch neurons, a major concern is that mechanical stimulation of this neuron may cause changes in transport parameters. However, it has been shown that upon application of pressure on the touch neuron, the mechanosensory/touch response occurs only in the first 0.5 seconds (61). We therefore extrapolate that the pressure applied by our device will probably stimulate the touch receptor neuron for only the first 0.5 seconds during which time imaging of transport is not initiated.

We would like to point out that the microfluidic device is compatible with acquisition of data at very high frame rates. Moreover, it allows repeated immobilization over the time scales up to 1 h without significant changes in transport parameters (Figure 3B). Both these advantages will allow long-term imaging of developmental and cell biological events such as growth cone extension or mitochondrial dynamics. Repeated short time scale experiments can also be performed, e.g. allowing an investigator to test the role of drugs (62,63) and to study mutants in various subcellular processes.

Microfluidic devices can be used to collect developmental data across all stages

Our attempts at imaging vesicle transport during early development using anesthetics proved impossible as at concentrations of anesthetics for which the worm was relatively immobile, no transport of GFP::RAB-3-marked vesicles was visible (Figure 1J). The effect of anesthetics on GFP::RAB-3-marked vesicles in part led us to develop our simple microfluidic device to study the contribution of vesicle transport to synaptic growth. We see that as development occurs, anterograde (but not retrograde) flux of GFP::RAB-3-marked vesicles increases (Figure 5C,D,F). During the development of touch neurons, microtubule tracks increase in number and length (60). This change is likely to have significant effects on the motion of individual cargo molecules. Indeed, we observe that both pause frequency and pause time, measures of uninterrupted cargo motion, are consistent with the increase in microtubule numbers and lengths as the animal develops (Figure S3).

The use of microfluidic devices for study of synaptic remodeling has been previously shown in later developmental stages in the time scale of hours (38). In our work, we have collected faster time scale data and showed for the first time the use of such devices in collecting data during very early developmental stages. Our device can also be used to study other subcellular and cellular events in neurons as well as other cells/tissues in C. elegans. As a further demonstration of the utility of our microfluidic device, we have imaged axonal transport of mitochondria in the model organism Drosophila and our method can be easily extended to study other subcellular events in this model organism. Moreover, the current design with mere change in dimensions can be easily adapted to other transparent/translucent model organisms such as zebrafish larvae.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References
  8. Supporting Information

C. elegans strains

We used C. elegans transgenic jsIs821 expressing GFP::RAB-3 in synaptic vesicles of the touch neurons for synaptic vesicle transport experiments (45,49,50). Other strains used in this study for IFT transport, dendritic transport and Q neuroblast migration were mnIs17 (OSM-6::GFP) (14) and kyIs156 (ODR-10::GFP) (17) and ayIs4 (soluble GFP) (64), respectively, and were all obtained from the CGC. Animals were grown and maintained as described previously (65). Synchronized 1-day adult animals were used for imaging by picking them as larval L4 animals. To carry out developmental imaging, animals of the appropriate age were picked from synchronized cultures, imaged and then discarded. Individual animals were not followed from L1 to 1-day adult. Animals were imaged only once unless stated otherwise in results.

Drosophila stocks

Drosophila stocks were raised on standard corn meal medium at 25°C. Mitochondrially targeted GFPs were selectively expressed in ‘cholinergic sensory neurons' using a UAS-GFP targeted to mitochondria stock (12) and a cha19bGal4 driver stock (55). The UAS-mitoGFP cha19bGal4 embryos were collected and maintained on 1% sucrose agar at 25°C for ∼30 h, and mitochondrial transport was imaged in the ‘segmental nerve axons' of the first instar larvae within 6–9 h after hatching.

PDMS microfluidic immobilization devices

We used double-layer PDMS fabrication technique for microfluidic devices for anesthetic-free transport imaging (66). In brief, we fabricated two different moulds for flow and control channels using photolithography of SU8. PDMS mixture (10:1) was either poured on control channel master (layer 2) or spin coated on flow channel master (layer 1) at 1500 rotation per min (rpm) for 30 seconds and baked at 80°C. Two layers were plasma bonded without any precise alignment arrangement to a glass coverslip to form the device (24,33,36). The devices for Drosophila larvae were prepared with the following modifications. The flow and control channels were 5 mm wide and ∼120 µm high. The membrane layer was spun at 500 rpm on the flow pattern to obtain thicker membranes.

Membrane deflection

Individual worms were pushed into the microchannel where they were pressurized with a column of water against the control channel using 14 psi compressed nitrogen gas. On releasing the pressure, the membrane relaxed back to uncompressed state and this can be repeated many times. Individual first instar Drosophila larvae were pushed into larger devices using 1× PBS and allowed to crawl into the flow chamber. The larvae were immobilized using ∼7 psi compressed gas applied through a column of water.

Anesthetic solutions

Anesthetic solutions were prepared in M9 buffer of levamisole (Sigma), tricaine (Sigma), 1P2P (Lancaster synthesis) and muscimol (Sigma). Worms were anesthetized on a thin agar pad using ∼10 µL of the appropriate concentration of anesthetic for high-resolution imaging.

Time-lapse imaging and analysis of cargo transport

In vivo fluorescence imaging experiments were performed using an inverted microscope (Olympus iX81) equipped with central spinning disc unit (CSU, Yokogawa) and charge-coupled device (CCD)-based camera (iXon, Andor). All imaging was performed using a 100× objective [oil immersion and numerical aperture (NA) 1.4] in the first 120 µm of the posterior touch cell neuronal process near the cell body. We used the NIH ImageJ kymograph plugins (http://www.rsbweb.nih.gov/ij/ and http://www.embl-heidelberg.de/eamnet/html/body_kymograph.html) for analysis. A moving particle was displaced by at least 3 pixels in successive time frames, while a stationary particle was immobile for more than three consecutive frames. The flux of particles was calculated as number moving in either direction within a specific region of the process in all 350 frames. The number of pauses and amount of time spent in each pause were represented by pause frequency and pause time. In general, we interpret movement characteristics with lower pause time and pause frequency as more processive transport. Velocity is a measure of how fast the cargo moves and reflects the characteristics of all the motors present on the cargo. Flux is a reflection of the number of actively moving cargo particles, which in turn is likely to be the property of the number of all motors engaged in transporting the imaged cargo. For each data point, we analyzed at least 300 ODR-10::GFP particles obtained by imaging seven animals, 300 OSM-6::GFP particles obtained by imaging five to seven animals and 500 GFP::RAB-3 particles obtained by imaging four to eight animals. Mitochondria velocities were calculated by using the manual tracking plugin http://rsbweb.nih.gov/ij/plugins/track/track.html in NIH ImageJ. We analyzed the motion of 12 mitochondria in movies made from seven animals.

Q neuroblast imaging

Development of Q neuroblast was imaged capturing images using a spinning disc confocal microscope every 5–20 min (as necessary) with a 400 millisecond exposure using a 100× 1.4 NA oil objective. The laser power was 26 mW at source, 15 mW at objective and 20–50% of the maximum power is used while imaging. L1 animals were anesthetized using 0.3 mm levamisole on a thin agar pad after ∼3 h of hatching. Immobilized L1s of same stage were imaged in microfluidic devices filled with M9 buffer as described earlier. The images were acquired manually to adjust the cell position at the center of the imaging field. The fluorescence images were adjusted for background intensity to make Q-neuroblast cells alone visible using ImageJ (NIH). The distance between the two fluorescently marked cell centers was used to calculate the relative migration distance. Relative cell migration velocity between QL.a and QL.p was calculated from six movies. The distance between the cell centers was calculated when the cells had not yet divided (0 µm) and when the cells had completed migration but the QL.a had not begun turning around to go cross QL.p.

Statistical analysis

Data were analyzed from at least five independent movies acquired from different animals unless stated otherwise. We used the two-tailed t-test with unequal variance to check the significance of the differences in the mean values among two different sets of experiments and presented as single and double asterisk marks (* and **) for p-value less than 0.05 and 0.005, respectively.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References
  8. Supporting Information

S. P. K. and S. M. designed the experiments and wrote the paper. S. M. started work on the device with K. R. V. V. provided infrastructure to design and fabricate devices. S. A. carried out most imaging experiments and analysis in C. elegans. We thank Tarjani Agrawal for maintaining a Drosophila cage. We thank Dr. Sreekanth Chalasani and Dr. Cori Bargmann for their helpful discussions and suggestions on technical aspects of the work. We thank Dr. Krishanu Ray for providing us necessary Drosophila stocks and CGC for C. elegans strains. We thank Dr. Krishna and CIFF at NCBS for use of the spinning disc confocal microscope supported by the Department of Science and Technology—Centre for Nanotechnology (No. SR/55/NM-36-2005). S. M. and S. A. thank Bikash Choudhary for helpful discussions. The work was supported by a DBT postdoctoral fellowship (S. M.), DST Fast Track scheme (S. M.) and DST grant to (S. P. K.). S. A. was supported by a DST grant to S. P. K.

References

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References
  8. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References
  8. Supporting Information

Supporting Information

Additional Supporting Information may be found in the online version of this article:

Figure S1: Effect of frame rate and two gaseous anesthetics on transport characteristics of GFP::RAB-3-marked vesicles. Average values of velocity (A), flux (B), pause frequency (C) and pause time (D) are plotted. Different frame rates for 3, 5, 7 and 9 fps were represented using different bar patterns for both anterograde and retrograde movements. E–H) Show representative kymographs acquired at 3, 5, 7 and 9 fps. The figure also shows the effect of compressed nitrogen and carbon dioxide gas when used in a PDMS device for immobilization. Average velocity (I), flux (J), pause frequency (K) and pause time (L) are plotted for anterograde and retrograde movements. Carbon dioxide gas showed a strong effect on transport. 14 psi CO2 application shows greater variation in transport compared to nitrogen immobilization. Kymographs show both enhanced (N) and reduced (O) transport compared to more uniform flux observed using nitrogen gas (M). Data represented as mean +&sol;− SEM. Statistical significances are denoted by asterisk for p-value < 0.05 (&ast;) and p-value < 0.005 (&ast;&ast;).

Figure S2: Kymographs and pause parameters of different immobilization preparations. All images were acquired at 5 fps for 1-day adult and kymographs are plotted with the cell body on the right. Animals were immobilized using 10 mM levamisole (A), 10 mM tricaine (B), 0.5% 1P2P (C) and 6 mM muscimol (D), combination of 10 mM levamisole in a microfluidic device (E), combination of 10 mM tricaine in a microfluidic device (F) and PDMS membrane immobilization alone (G). The white triangle indicates stalled particles in animals immobilized with 10 mM tricaine (B) and PDMS membrane immobilization (G). H) Flux of anterograde and retrograde GFP::RAB-3-marked vesicles and ODR-10::GFP in anesthetized C. elegans acquired at different time-points after immobilization using 10 mM levamisole. Pause frequency and pause time of anterogradely (I and K) and retrogradely (J and L) moving particles immobilized using different anesthetics. Data represented as mean +&sol;− SEM. Statistical significances are denoted by asterisk for p-value < 0.05 (&ast;) and p-value < 0.005(&ast;&ast;).

Figure S3: Pause parameters improve with development of C. elegans. Average anterograde and retrograde pause frequency (A and B) and pause time (C and D) decrease through development. Data represented as mean +&sol;− SEM. Statistical significances are denoted by asterisk for p-value < 0.05 (&ast;) and p-value < 0.005(&ast;&ast;).

Movie S1: Trapping, imaging of GFP::RAB-3-marked vesicles and release of a 1-day adult in a PDMS membrane-immobilization device. Movie acquired at 5 fps and played back at 15 fps. The movies are oriented such that the anterior of the animal is to the left.

Movie S2: Time-lapse imaging of an L3 animal expressing OSM-6::GFP in a PDMS membrane-immobilization device. Movie acquired at 5 fps and played back at 15 fps. The movies are oriented such that the anterior of the animal is to the left.

Movie S3: Time-lapse imaging of an L3 animal expressing ODR-10::GFP in a PDMS membrane-immobilization device. Movie acquired at 5 fps and played back at 15 fps. The movies are oriented such that the anterior of the animal is to the left.

Movie S4: Time-lapse imaging of an L1 animal in a microfluidic device showing division of the QL expressing soluble GFP into QL.a and QL.p. This occurs approximately 4 h after hatching. The maximum distance between the cell centers was 4.5 µm in frame 5 of the movie. After this frame, the QL.a begins to migrate toward QL.p. Images were acquired every 10 min and played back at 2 fps. Cells are marked with an &ast;. The movies are oriented such that the anterior of the animal is to the left and the right side of the animal is at the bottom.

Movie S5: Time-lapse imaging of an L1 animal in a microfluidic device showing migration of QL.a across QL.p. Images were acquired every 5–15 min and played backed at 2 fps. Cells are marked with an &ast;. The movies are oriented such that the anterior of the animal is to the left and the right side of the animal is at the bottom.

Movie S6: Time-lapse imaging of an L1 animal in a microfluidic device showing the division of QL.a into QL.aa and QL.ap (first division in the movie) and QL.p into QL.pa and QL.pp (second division in the movie). Both divisions are completed by 2 h after the birth of QL.a and QL.p and are completed by 6 h after hatching. Images were acquired every 4–5 min and played back at 2 fps. Cells are marked with an &ast;. The movies are oriented such that the anterior of the animal is to the left and the right side of the animal is at the bottom.

Movie S7: Time-lapse imaging of first instar Drosophila larvae expressing GFP targeted to mitochondria under a PDMS membrane-immobilization device. Movie acquired at 2 fps and played back at 15 fps. The movies are oriented such that the anterior of the animal is to the left.

FilenameFormatSizeDescription
TRA_1157_sm_FigureS1.tif9165KSupporting info item
TRA_1157_sm_FigureS2.jpg582KSupporting info item
TRA_1157_sm_FigureS3.jpg303KSupporting info item
TRA_1157_sm_MovieS1.wmv3436KSupporting info item
TRA_1157_sm_MovieS2.wmv223KSupporting info item
TRA_1157_sm_MovieS3.wmv351KSupporting info item
TRA_1157_sm_MovieS4.wmv94KSupporting info item
TRA_1157_sm_MovieS5.wmv102KSupporting info item
TRA_1157_sm_MovieS6.wmv143KSupporting info item
TRA_1157_sm_MovieS7.wmv126KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.