Circular compartmentalized microfluidic platform: Study of axon–glia interactions

Abstract

We describe a compartmentalized circular microfluidic platform that enables directed cell placement within defined microenvironments for the study of axon–glia interactions. The multi-compartment platform consists of independent units of radial microchannel arrays that fluidically isolate somal from axonal compartments. Fluidic access ports punched near the microchannels allow for direct pipetting of cells into the device. Adjacent somal or axonal compartments can be readily merged so that independent groups of neurons or axons can be maintained in either separate or uniform microenvironments. We demonstrate three distinct modes of directed cell placement in this device, to suit varying experimental needs for the study of axon–glia interactions: (1) centrifugation of the circular platform can result in a two-fold increase in axonal throughput in microchannels and provides a new technique to establish axon–glia interactions; (2) microstencils can be utilized to directly place glial cells within areas of interest; and (3) intimate axon–glia co-culture can be attained via standard pipetting techniques. We take advantage of this microfluidic platform to demonstrate a two-fold preferential accumulation of microglia specifically near injured CNS axons, an event implicated in the maintenance and progression of a number of chronic neuroinflammatory and neurodegenerative diseases.

---

1. Introduction

The mammalian central nervous system (CNS) is composed of several highly specialized cell types, including neurons and glial cells. Neurons communicate with each other via dendrites and axons, which respectively are responsible for integrating synaptic input and providing output. CNS axons, and to a lesser extent dendrites, often span substantial distances and are situated in compositionally distinct microenvironments as compared to their cell bodies. Axonal degeneration and axon–glia interactions have been strongly implicated in the pathogenesis of a growing number of neuroinflammatory and neurodegenerative diseases, but the underlying cellular and molecular mechanisms have yet to be identified.^1,2

Elucidation of mechanisms involved in axonal injury and axon–glia interactions can be difficult in the complex in vivo setting, and thus many investigators have turned to in vitro approaches. However, standard cell culture approaches do not compartmentalize axons from neuronal cell bodies, thus making it difficult to delineate axon-specific mechanisms. To address this problem, the Campenot chamber,³ which enables manipulation of axons independently from neuronal cell bodies, was developed. Studies using the Campenot chamber have led to important discoveries in peripheral nervous system (PNS) axonal development,⁴ degeneration,^3,5 and regeneration.⁴ However, the far smaller sizes of CNS neurons and axons have precluded the reproducible study of compartmentalized CNS neuronal cultures within Campenot chambers.

Recently, it has been recognized that lab-on-a-chip devices may provide unique solutions to study cellular and molecular aspects of neuronal and axonal function, with precise control over the cellular microenvironment.^3,6–13 In particular, microchannels have been shown to be powerful micro-features capable of passively guiding axons of both CNS and PNS neurons. Several novel microfluidic devices have utilized these features to fluidically isolate axons from their respective cell bodies.^{11,12,14–16} In these devices, cells are introduced by way of a loading inlet and the cells are randomly dispersed throughout a cell reservoir. Neuronal cells that are in close proximity to the microchannels extend axons through the channels and into adjacent compartments. Application of hydrostatic pressures between microchannel-connected compartments can induce fluidic isolation between axon and cell body.

These microfluidic approaches are, however, limited by the inability to precisely place cells within microscopic areas of interest and to control neuronal and axonal microenvironments. Here, we present a novel platform that advances existing device technologies to enable directed placement of neurons and glial cells within defined microenvironments in a compartmentalized microfluidic platform. Our circular device consists of multiple independent units arranged in a closed circular pattern, with each unit consisting of two compartments (somal and axonal) connected by an array of microchannels. Fluidic access ports are punched on both somal and axonal sides near the microchannels, allowing for direct pipetting of cells close to the microchannel interface. During the creation of fluidic access ports, adjacent somal or axonal compartments can be readily merged so that independent groups of neurons or axons can be maintained in either distinct or uniform microenvironments. We describe three different modes of directed cell placement in this device, to suit varying experimental needs for the study of axon–glia interactions. First, the circular geometry of the platform lends itself to centrifugation, or spinning, to optimize the positioning of both neurons and glial cells, thereby enhancing axonal throughput in microchannels and providing a new technique to establish axon–glia interactions. Second, patterned microstenciling can be utilized to directly place glial cells within areas of interest in the axonal compartment. Thirdly, intimate axon–glia co-culture can be attained via standard pipetting techniques as a result of the close proximity of large (≥3 mm) access ports to the microchannel interface.

To demonstrate the enabling power of the platform, we investigated whether microglia specifically respond to injured axons in the setting of CNS inflammatory disease. Microglial accumulation to sites of neurodegeneration plays a major role in maintenance and progression of a number of chronic neuroinflammatory and neurodegenerative diseases,^17,18 but it is not known whether microglia are specifically recruited toward degenerating CNS axons in the absence of signals from other neural cells. To address this question, we configured our device to have multiple somal compartments whose microchannels connect to a unified axonal compartment, thereby creating independent populations of neurons whose axons reside in a common microenvironment. By placing microglia in spatially defined areas between groups of axons, we were able to monitor microglial responses to populations of axons that were either healthy or degenerating. To our knowledge, this is the first demonstration of a differential microglial response to healthy versus injured CNS axons in a microfluidic platform.

2. Materials and methods

2.1 Fabrication of co-culture platform

The co-culture platform was fabricated in poly(dimethylsiloxane) (PDMS), a biocompatible silicone rubber, following well established replica molding procedures. Briefly, a master template was created using a two-layer microfabrication process. Silicon wafers (Fig. 1A; University Wafer, MA) were coated with a thin photoresist layer (SU-8 2002, Microchem, MA), soft baked, exposed with a high resolution DPI transparency mask (Fig. 1B; Cad/Art, OR), and developed to define a circular array of 1500 microchannels (Fig. 1C; dimensions: height = 2–4 μm, width = 8–10 μm, length = 500 μm, and minimum spacing = 15–25 μm). The process was immediately repeated with a thicker photoresist (Fig. 1D and 1E; SU-8 3050, Microchem, MA) to define multiple axonal (outer port; diameter = 4 mm) and somal (inner port; diameter = 3 mm) compartments (height = 150–200 μm). Details pertaining to photoresist (soft/hard/post exposure) bake, exposure, and development times can be found in the manufacturer's technical sheet (Microchem, MA).

Fig. 1 The device mold was constructed using standard SU-8 photolithography. (A) Beginning with a bare silicon wafer, (B) an initial thin-film resist layer (SU-8 2002; height = 2.5 μm) was spun, soft baked, and optically exposed. (C) Subsequently, the substrate was post exposure baked and immersed in developer to define the circular array of microchannels. (D & E) The thick-film resist (SU-8 3050; height = 150 μm) was processed similarly to define larger fluidic access ports. After (F) PDMS replication, (G) devices were customized through the use of commercially available dermal biopsy punch tools.

Reproducible replication of the device was done by soft polymer casting using Sylgard 184 PDMS (Fig. 1F; Dow Corning, MI) as described previously.^19–22 Once replicated, access ports were formed using commercially available punch tools (Fig. 1G; Huot Instruments, WI). Prior to cell loading, replicated devices were sonicated (Branson Ultrasonics, CT) in 70% ethanol for 5 min and dried with compressed air to remove PDMS debris and other surface contaminants. Cleaned devices were placed feature-side down onto 40 mm glass bottom petri dishes (Willco Wells, Netherlands) and sealed upon contact. If a tighter seal was required, both the device and glass substrate were exposed to a low-power (25 W) oxygen plasma treatment for 1 min (Harrick Plasma, NY) prior to contact. Devices were sterilized with 100% ethanol, washed three times with doubly deionized water (ddH₂0; Millipore, MA) to remove residual ethanol, and coated with poly-D-lysine hydrobromide (100 μg ml⁻¹, Sigma, MO). Devices were stored at 37 °C until needed for experimentation.

2.2 Cell preparation

Primary hippocampal neurons were derived from embryonic day 17 (E17) pups as previously described²³ and resuspended at a final density of 1 × 10⁷ mL⁻¹, unless otherwise stated. Primary microglial cultures were derived from postnatal rat pups as previously described²⁴ with minor modifications. Briefly, dissociated hippocampal cells were plated in DMEM/F12/10% fetal bovine serum (FBS), and media was changed every 3 days. Between days 10 and 14, cells were shaken at 225 rpm for 90 min, and non-adherent cells were collected and plated. After 30 min, the microglia cells had attached to the plate, and non-adherent contaminating cells were removed. This resulted in a population of microglia of >95% purity (data not shown). The BV2 microglial cell line, used in several co-culturing experiments, was maintained in DMEM/10% FBS.

Devices were washed 3× with ddH₂0, filled with serum-containing media, and placed in a standard humidified cell culture incubator set to 37 °C and 5% CO₂ (Thermo Scientific, MA) for 15–30 min. Primary neurons were loaded into the somal compartment of the device in increments of 5 μL. If spinning was required, devices were moved to a spinner (Laurell Technologies, MA) immediately after cell loading, centered, and spun for 1–2 min at 400–1200 rpm. For experiments in which cells were labeled with a fluorescent protein, dissociated neurons were nucleofected (Amaxa, MD) with a plasmid encoding the TdTomato gene as per the manufacturer's instructions. Efficiency of labeling was greater than 70%.

2.3 PDMS stencil patterning

PDMS stencils were prepared from unpatterned 3 mm PDMS slabs that were cut to yield thin 500–1000 μm membranes. Circular feature sizes (diameters) ranging from 500 μm to 1000 μm were then created by piercing the membrane with surgical needles (Fisher Scientific, PA). The stencils were sonicated with ethanol, dried, and placed within the device prior to cell seeding. During cell placement, glial cells were first concentrated and then slowly pipetted into the cavity of the PDMS stencil. After 1 h, the stencil was carefully removed and media was filled into the compartment without noticeable cell dispersion.

2.4 Restriction of diffusing analytes

To delineate the effects of inflammatory mediator treatment on either the cell body or axon, fluidic isolation was established to prevent the diffusion of analytes from one compartmentalized region of the neuron to the other. Fluid volumes were modified such that the somal compartment was of lower fluidic height than the axonal compartment. By doing so, a hydrostatic pressure was established between these microchannel-connected compartments, thereby creating a small flow to counteract diffusive forces. A 1 μL bolus of fluorescein isothiocyanate (FITC; Sigma, MO) dye was introduced to the somal compartment and was imaged over the course of 24 h.

Empirically, a fluid height difference ≥2 mm was sufficient to prevent the diffusion of a low-molecular weight (MW 700 Da) analyte to the axonal compartment for at least 24 h (ESI† Fig. 1). In principle, the ideal pressure required to maintain fluidic isolation could be determined analytically, however, empirical testing provides proof of principle that fluid isolation could be achievable for the desired pressures, times and proposed analytes. Continued maintenance of the height differential was achieved by adding 5 μL of media to the higher volume compartment daily. This allowed for restriction of small analytes (e.g. nitric oxide, MW 30 Da) over several days as verified by commercially available kits (Griess nitric oxide detection kit; Promega, Madison, WI).

2.5 Imaging and analysis

Images were obtained in a live-cell Zeiss inverted microscope (Axio Observer) using Zeiss Axiovision software. After data collection, images were exported to either Metamorph (MDS Analytical Technologies, Toronto, CA) or NIH ImageJ (Bethesda, MD) for further analysis. Statistical analysis was done in Microsoft Excel.

2.5.1 Surface density. The number of cells within a given surface area was measured under 10× magnification utilizing the grid function of Metamorph which overlays a two-dimensional mesh of user-defined size on top of the image. Bins of defined widths (10 or 50 μm) and varying lengths were used to calculate cell distribution and density.

2.5.2 Microglia chemotaxis and accumulation. Microglia were patterned within a circular microstencil (diameter = 500 μm) in an area between arrays of microchannels. After microglia attached (approximately 1 h later), the microstencil was carefully removed to minimize cell dispersion. Microglial movement was quantified parallel to the microchannel interface (arbitrarily defined as the y-axis) and information pertaining to cellular position in the orthogonal direction (x-axis) was not recorded. The furthest microglia (n = 6) along the y-axis but within 500 μm of the x-axis were quantified in terms of their average y-displacement from the start of the microchannels (y = 0). For each time point and experimental condition, the same procedure was repeated to determine the average position of the furthest group of microglia under those circumstances. Microglia accumulation was defined as the net displacement along the y-axis between 0 and 72 h of the same experimental condition (no axons, healthy axons, or degenerating axons; n = 4 independent experiments for each situation).

3. Results and discussion

3.1 Microfluidic device design

The default configuration of the circular platform consists of sixteen independent units (ESI† Fig. 2A). Each unit is defined by an axonal (outer) and somal (inner) compartment that are physically interconnected by an array of 90 microchannels. These independent units are arranged radially in a closed circular pattern and are separated from each other by a minimum spacing of 250 μm. Neurons are seeded into the somal compartment, and as axons extend they are passively guided through the microchannels and into the axonal compartment. Using this default configuration, multiple unique experiments can be performed within the same device. As proof of principle, we labeled consecutive axonal compartments with differing dyes, and demonstrate that retrograde labeling of populations of neurons is confined to each distinct somal compartment (ESI† Fig. 3).

Additionally, adjacent somal or axonal compartments can be readily merged, resulting in four major device configurations that differ in size and connectivity of the neuronal and axonal microenvironments (ESI† Fig. 2). These configurations consist of: (1) fully segmented somal and axonal compartments; (2) fully merged somal and axonal compartments; (3) fully merged somal and fully segmented axonal compartments; and (4) fully segmented somal and fully merged axonal compartments. Depending upon experimental needs, minor permutations to these four major configurations can yield up to 256 (16 × 16) device possibilities by simply merging neighboring compartments with commercially available punch tools during the creation of fluidic access ports.

3.2 Neuronal and glial cell positioning

Cell positioning can be modulated by several distinct methods in our platform, allowing for versatility when developing novel axon–glia co-culture studies. While several compartmentalized microfluidic platforms have utilized microchannels to guide neurites and establish co-culture conditions, no specific device has (a) integrated large access ports within close proximity to microchannel features to provide straightforward loading of cells near guidance features, (b) allowed incorporation of micropatterning techniques (e.g., microcontact printing, or microstencils) to manipulate co-culture conditions, and (c) optimized platform geometry so that centrifugal forces can be quickly applied to provide a robust mechanism to manipulate neuronal and glial cell distribution.

3.2.1 Neuronal cell positioning. In our culture platform, we have incorporated large fluidic ports placed within close (<500 μm) proximity to the microchannels. This enables the ability to pipette cells, analytes, and other chemical reagents directly, without requiring traditionally lengthy microfluidic input and output channels. When introduced into the device with a standard pipette (Fig. 2A and 2B), cells can consistently be placed within 20–30 μm of the microchannel features (Fig. 2C). As a result, the numbers of neurons extending axons through the microchannel features is maximized, with a very high percentage (nearly 100%) of channels filled with numerous axons after 7 days (Fig. 2E).

Fig. 2 (A) A false-colored 3D schematic of the PDMS device is shown to aide in visualization of the microchannels (red) and somal (blue)/axonal (green) compartments (not to scale). (B) Large (>3 mm) compartments enabled pipetting close (<500 μm) to the microchannels (drawing to scale). (C) Cell distribution was quantified for the first 100 μm on the somal side of the microchannels after seeding neurons (5 μL, 1500 cells mm⁻²). Cells did not pass through the microchannels (height = 2.5 μm, width = 10 μm) as their cross sectional area restricted cell migration under normal culture conditions. (D) The devices were spun at 600 rpm (4.5g) for 1.5 min to further enhance cell proximity to the microchannel–somal compartment interface. Red dashed lines correspond to microchannel interface. (E) Cell placement within the first 20 μm of the microchannel interface was increased 13-fold in spun devices as compared to control (n = 10 replicates). (F) In a separate experiment, devices containing low (120 cells mm⁻²), medium (300 cells mm⁻²), and high (1500 cells mm⁻²) density neurons were either left stationary or immediately spun to enhance cell position near the microchannel interface. After 7 days, neurite outgrowth was quantified for all conditions. Spinning resulted in up to 2-fold increases in axon throughput, or number of channels containing axon processes, in low (n = 5) and medium density cultures (n = 5). Under high seeding conditions (n = 5), no enhancement could be observed, but nearly 100% of the channels contained neurites as shown by 5 μM CalceinAM staining (G). Alignment was also achieved in the axonal compartment as shown by (H) the utilization of a PDMS wedge near the microchannel inteface. Microglia were distributed (1 μL, 3000 cells mm⁻²) along the axis of the wedge and the device was immediately spun (v = 1000 rpm, t = 30 s) to align the microglia along the edge of the wedge. (I) After removal of the PDMS piece, a band of microglia cells remained. Error bars on graphs correspond to standard errors. * = p-value < 0.05, ** = p-value < 0.01, unpaired Welch's t-test (2way for 2E, 1way for 2F).

3.2.2 Centrifugation. The circular geometry of the platform lends itself to centrifugation, or spinning, to enhance cell position with respect to physical features of the device. Although spinning has been previously shown to modulate cell placement,²⁵ our circular device enabled rapid centering and centrifugation of the device within 2 min of seeding cells. Thus, even highly adherent cells such as neurons can be effectively and uniformly centrifuged. Compared to non-spun devices (Fig. 2C), centrifuged devices (v = 600 rpm for t = 90 s; r = distance from axis of rotation = 11 mm; Fig. 2D) showed a 13-fold enhancement in neuronal cell placement within 20 μm of the microchannel interface (Fig. 2G) with only 4.5g of force applied, the lowest among reported devices. Furthermore, with the exception of high seeding densities, centrifugation resulted in up to a two-fold increase in the number of channels containing neurite processes (Fig. 2F).

Additionally, a novel approach using centrifugal forces (v = 1000 rpm, t = 60 s) can be exploited to align glial cells adjacent to physically imposed structures, such as PDMS wedges (Fig. 2H and 2I). The advantage of this approach is that glia can be compacted into a very thin (<350 μm) band near axonal processes and blocked from traversing into unwanted regions of the compartment. Cell viability, as assessed by calcein AM uptake and by trypan blue staining, was not adversely affected by centrifugation over the range tested (0–1200 rpm), as compared to non-spun cells (data not shown), and cells continued to remain healthy in the device for over 3 weeks under standard media conditions.

3.2.3 Co-culture. Our platform also allows directed co-culture of glial cells within microdomains of the axonal compartment. Since the device employs large access ports, PDMS microstencils²⁶ can be used to directly place glial cells within a defined microenvironment in the multi-compartment platform (Fig. 3A and 3B). As a result, populations of glial cells can be placed in defined regions either adjacent to or in between areas of axons.

Fig. 3 Co-culture conditions can be modulated by distinct methods in the multi-compartment platform. (A) PDMS microstencils were used to pattern microglia (1 μL, 3000 cells mm⁻²) in a spatially defined region within the axonal (outer) compartment. (B) After removal of the stencil and addition of media, no significant cell dispersion was seen under differential interface contrast (DIC) microscopy. (C) Neurons (60 μL, 1500 cells mm⁻²) were cultured for 11 days to allow dense axonal outgrowth (red) in the axonal compartments. (D) Primary rat microglia (CalceinAM labeled, green) were slowly introduced (1 μL, 3000 cells mm⁻²) to the axonal compartment to minimize cell dispersion. Intimate axon–glia interactions were observed when glial cells were added close to the microchannel interface, particularly within 250 μm of the microchannel–axonal compartment interface.

If intimate axon–glia co-culture is required, glial cells can be reproducibly pipetted along the axonal–microchannel interface, in large part due to directed flow and capillary action. This results in a large percentage of microglia in intimate contact with axons, as shown in Fig. 3C and 3D where over 93% of microglia were in intimate contact with axonal processes. The co-culturing procedure did not appear to adversely affect axonal viability, as no observable markers of axonal degeneration (e.g. increase in number of blebs or changes in axon morphology) were seen. Additionally, the cellular distribution of microglia in the axonal compartment was unaffected by the presence of axons, as the microglia distribution was similar in the absence of axons (data not shown).

3.3 Microglial response to axonal degeneration

We next took advantage of our multicompartment platform to determine whether microglia migrate specifically toward injured CNS axons. Accumulating evidence suggests that recruitment of microglia during CNS inflammatory disease can contribute to neurodegeneration,^27–29 although it is unknown whether such recruitment occurs via direct axon–microglial interactions. To answer this question, we configured the device such that four consecutive axonal compartments were merged to allow sites for microglial stencil patterning near axons (Fig. 4A). Neurons were plated in the somal compartments, and centrifuged (v = 600 rpm, t = 90 s) to enhance axonal throughput. After dense bundles of axons arising from distinct somal compartments had extended into the unified axonal compartment, primary rat microglia were directly placed in between areas of axonal outgrowth (Fig. 4B). By placing microglia in defined regions within the device, a common starting point for cellular migration could be assessed. To induce oxidative stress, which has been implicated in the pathogenesis of many neuroinflammatory diseases including multiple sclerosis and human immunodeficiency virus (HIV) dementia, we added H₂0₂ selectively to the somal compartment of only one group of neuronal cell bodies. This resulted in induction of neuronal cell death and accompanying axonal degeneration over the course of 24–72 h. Quantification of microglial chemotaxis and accumulation demonstrated that, over the course of 72 h, microglia migrated an average of 70 ± 21 μm (mean ± standard error; n = 4) when placed in areas devoid of axons, reflecting basal rates of migration (Fig. 4C). Migration distance was increased when microglia were placed adjacent to healthy axons 179 ± 58 μm (Fig. 4D), demonstrating endogenous levels of migration under standard cell culture conditions. Strikingly, however, microglia migration toward degenerating axons (388 ± 81 μm) was further enhanced two-fold on average as compared to healthy axons, demonstrating preferential accumulation near degenerating, as compared to healthy, axons (Fig. 4E).

Fig. 4 Microglial response to degenerating axons. (A) The device was configured so that it consisted of 4 somal compartments connected to a unified axonal compartment. Neurons were added to 3 of the 4 somal sectors (one intentionally left empty as a negative control; 5 μL per sector; 1500 cells mm⁻²). (B) On day 14, primary microglial cells (false-colored green, 2 μL, 1500 cells mm⁻²) were selectively placed between bundles of axon outgrowth in the unified axonal compartment through the use of PDMS microstencils. Neurons were treated with control or peroxide containing media (2 mM) and volumes were adjusted to prevent diffusion into the axonal compartment. Chemotaxis and accumulation of microglia in response to signals from (C) no axons, (D) healthy axons, or (E) degenerating axons was quantified by measuring the total displacement (net upward or downward movement) of microglia along the microchannel interface in the axonal compartment prior to, 48 h after, and 72 h after insult with respect to their initial seeding position. Axon morphology is shown for 72 h time point. *p-value < 0.05 compared to healthy axons and <0.01 compared to no axons at 72 h, unpaired 1way Welch's t-test. Error bars on graphs correspond to standard errors (n = 4 replicates).

4. Discussion

The precise positioning of cells between independent groups of CNS axons in our multi-compartment platform allowed us to investigate microglial responses to degenerating axons. While standard culture platforms and commercially available kits can quantify cellular chemotaxis, they cannot address whether microglial accumulation occurs specifically in response to CNS axonal injury. We found that microglia preferentially accumulate near degenerating, as compared to healthy, axons. Importantly, the absence of neuronal cell bodies or other glial cells in the fluidically isolated axonal compartment implies that microglia respond specifically to axon-derived signals. Such signals may either be soluble or membrane-bound,³⁰ and the identities of these molecules as well as the microglial receptors that recognize these ligands remain to be determined. We anticipate that further investigations with our microfluidic platform will allow for the dissection of the molecular mechanisms of microglial chemotaxis and accumulation in response to degenerating axons.

We employed a device configuration in which adjacent axonal compartments were merged while maintaining independent somal compartments to investigate microglial responses to axon degeneration. However, the device can be easily configured to investigate other novel aspects of axon–glia interactions. For example, one device configuration, in which neurons residing within a uniform microenvironment give rise to groups of axons that can be independently manipulated (ESI† Fig. 2B), would be particularly suited to study axon-specific responses to a variety of chemicals, toxins, and glia co-culture interactions. Alternatively, the device can be configured to yield a single large somal and axonal compartment (ESI† Fig. 2D) which can take full advantage of the large array of microchannels present in the device (1500) to collect extensive data on a specific axon–glia interaction.

Lastly, co-culturing conditions can be readily modulated depending upon the biological question in mind. We have demonstrated the utility of loading small numbers of glial cells in spatially defined areas within the device. However, if high co-culturing efficiencies are needed, for example, glia can be directly loaded along the axonal compartment–microchannel interface. The close contact of such a high percentage of co-cultured cells with axons in the device lends itself to a variety of future applications, including the study of gene expression changes in glial cells interacting with axons, as well as the development of robust in vitro CNS myelination systems.³¹

5. Conclusion

The multi-compartment co-culture platform presented here demonstrates an innovative approach to studying axon–glia interactions. By virtue of the circular configuration of the platform, large radial array of microchannels, compartment configurability, and ports within close proximity to microchannel features, many novel device characteristics were realized including (a) the use of centrifugal forces to modulate placement of both neurons and glial cells, (b) fluidic isolation of large numbers of axonal processes, (c) glial micropatterning, and (d) intimate axon–glia co-culturing. Additionally, the circular geometry requires minimal external forces to enhance cell placement within the device, thereby maximizing cell viability and providing new techniques to co-culture glia with axons. We utilized this platform to demonstrate preferential accumulation of microglia specifically to injured as compared to healthy axons, serving as a foundation to elucidate mechanisms of axon–glia interactions in neurological disease maintenance and progression. Overall, this novel multi-compartment co-culture platform enables distinct modes of axon–glia co-culture and provides experimental versatility to investigate axon-specific and axon–glia-specific cellular and molecular events implicated in neurobiological disease.

Acknowledgements

The authors would like to acknowledge Tanya Malpica-Llanos, Josh Kim, Steve Wang, Rezina Siddique, and Suraag Patel for their help. Funding was generously provided by the Johns Hopkins Institute for Nanobiotechnology (NT and AV), Maryland Technology Development Corporation Grant 104307 (NT), National Institutes of Health Grant 1F31NS066753-01 (SH), and NIDA K08DA22946, NIMH 5P30MH075673 Pilot Award, and Howard Hughes Medical Institute Early Career Physician-Scientist Award (AV).

References

1. M. Coleman, Nat. Rev. Neurosci., 2005, 6, 889–898.
2. M. P. Coleman and V. H. Perry, Trends Neurosci., 2002, 25, 532–537.
3. R. B. Campenot, Proc. Natl. Acad. Sci. U. S. A., 1977, 74, 4516–4519.
4. L. S. Zweifel, R. Kuruvilla and D. D. Ginty, Nat. Rev. Neurosci., 2005, 6, 615–625.
5. M. C. Raff, A. V. Whitmore and J. T. Finn, Science, 2002, 296, 868–871.
6. A. J. Blake, T. M. Pearce, N. S. Rao, S. M. Johnson and J. C. Williams, Lab Chip, 2007, 7, 842–849.
7. B. G. Chung, L. A. Flanagan, S. W. Rhee, P. H. Schwartz, A. P. Lee, E. S. Monuki and N. L. Jeon, Lab Chip, 2005, 5, 401–406.
8. J. El-Ali, P. K. Sorger and K. F. Jensen, Nature, 2006, 442, 403–411.
9. L. J. Millet, M. E. Stewart, J. V. Sweedler, R. G. Nuzzo and M. U. Gillette, Lab Chip, 2007, 7, 987–994.
10. T. M. Pearce and J. C. Williams, Lab Chip, 2007, 7, 30–40.
11. S. W. Rhee, A. M. Taylor, C. H. Tu, D. H. Cribbs, C. W. Cotman and N. L. Jeon, Lab Chip, 2005, 5, 102–107.
12. A. M. Taylor, M. Blurton-Jones, S. W. Rhee, D. H. Cribbs, C. W. Cotman and N. L. Jeon, Nat. Methods, 2005, 2, 599–605.
13. A. M. Taylor, S. W. Rhee and N. L. Jeon, Methods Mol. Biol., 2006, 321, 167–177.
14. E. Claverol-Tinturé, M. Ghirardi, F. Fiumara, X. Rosell and J. Cabestany, J. Neural Eng., 2005, 2, L1–7.
15. B. J. Dworak and B. C. Wheeler, Lab Chip, 2009, 9, 404–410.
16. J. Park, H. Koito, J. Li and A. Han, Biomed. Microdevices, 2009, 11, 1145–1153.
17. J. El Khoury, M. Toft, S. E. Hickman, T. K. Means, K. Terada, C. Geula and A. D. Luster, Nat. Med., 2007, 13, 432–438.
18. P. T. Nelson, L. A. Soma and E. Lavi, Ann. Med., 2002, 34, 491–500.
19. R. S. Kane, S. Takayama, E. Ostuni, D. E. Ingber and G. M. Whitesides, Biomaterials, 1999, 20, 2363–2376.
20. J. C. McDonald, D. C. Duffy, J. R. Anderson, D. T. Chiu, H. Wu, O. J. Schueller and G. M. Whitesides, Electrophoresis, 2000, 21, 27–40.
21. J. C. McDonald and G. M. Whitesides, Acc. Chem. Res., 2002, 35, 491–499.
22. S. K. Sia and G. M. Whitesides, Electrophoresis, 2003, 24, 3563–3576.
23. G. B. a. K. Goslin, Culturing Nerve Cells, 2nd edn, The MIT Press, Cambridge, MA, 1998.
24. R. C. a. J. d. Vellis, in Protocols for Neural Cell Culture, ed. S. F. a. A. Richardson, Humana Press, Inc., Totowa, NJ, 3rd edn, 2001, pp. 117–127.
25. S. W. Rhee, A. M. Taylor, D. H. Cribbs, C. W. Cotman and N. L. Jeon, Biomed. Microdevices, 2007, 9, 15–23.
26. A. Folch, B. H. Jo, O. Hurtado, D. J. Beebe and M. Toner, J. Biomed. Mater. Res., 2000, 52, 346–353.
27. G. W. Kreutzberg, Trends Neurosci., 1996, 19, 312–318.
28. G. Stollg and S. Jander, Prog. Neurobiol., 1999, 58, 233–247.
29. B. Liu and J. S. Hong, J. Pharmacol. Exp. Ther., 2003, 304, 1–7.
30. L. Cartier, O. Hartley, M. Dubois-Dauphin and K. H. Krause, Brain Res. Rev., 2005, 48, 16–42.
31. T. A. Watkins, B. Emery, S. Mulinyawe and B. A. Barres, Neuron, 2008, 60, 555–569.

---

Footnote

† Electronic supplementary information (ESI) available: Supplementary Fig. 1–3. See DOI: 10.1039/b918640a

---