Towards Neuronal Organoids: A Method for Long-Term Culturing of High-Density Hippocampal Neurons

One of the goals in neuroscience is to obtain tractable laboratory cultures that closely recapitulate in vivo systems while still providing ease of use in the lab. Because neurons can exist in the body over a lifetime, long-term culture systems are necessary so as to closely mimic the physiological conditions under laboratory culture conditions. Ideally, such a neuronal organoid culture would contain multiple cell types, be highly differentiated, and have a high density of interconnected cells. However, before these types of cultures can be created, certain problems associated with long-term neuronal culturing must be addressed. We sought to develop a new protocol which may further prolong the duration and integrity of E18 rat hippocampal cultures. We have developed a protocol that allows for culturing of E18 hippocampal neurons at high densities for more than 120 days. These cultured hippocampal neurons are (i) well differentiated with high numbers of synapses, (ii) anchored securely to their substrate, (iii) have high levels of functional connectivity, and (iv) form dense multi-layered cellular networks. We propose that our culture methodology is likely to be effective for multiple neuronal subtypes–particularly those that can be grown in Neurobasal/B27 media. This methodology presents new avenues for long-term functional studies in neurons.


Introduction
One goal that cell-biologists seek to achieve is cellular cultures that closely recapitulate physiologically relevant organ systems, commonly termed organoids. These types of cultures have advantages due to their mixed cellular populations which can form differentiated structures in vitro [1][2][3]. One of the challenges in creating these organoids is culturing cells for long periods of time (weeks to months). As a prerequisite to creating neuronal organoids, long-term culturing conditions for dense neuronal populations must be optimized. This is particularly challenging, as dense long-term cultures are often not possible, due to glial proliferation, contamination, and/or improper nutrient supplementation. Therefore, developing better culturing techniques that can maintain viable neuronal cultures for longer periods of time is necessary to explore more developed neural systems.
A series of benchmarks are used by experimentalists to ensure viable cultures including: (i) dense synapses, (ii) activity, (iii) connectivity, and in some cases (iv) protein expression (e.g. MAP2, Synapsin, Tau, etc.). The hippocampus is particularly relevant to long-term culturing experiments when considering the enhanced neurogenerative capacity associated with this area of the brain and memory formation [4,5]. However, previous studies that focused on creating long-term hippocampal cultures met with limited success [6,7]. In past years, most neuronal cultures have been prepared in the presence of Cytosine b-D-arabinofuranoside (Ara-C) that contained small numbers of glia, so as not to contaminate the neuronal population. Recent studies have demonstrated that neurons require glia for proper formation, connectivity, and activity. In particular, the long-term culture studies performed by Bertrand et al. (2011) and Majd et al. (2008) demonstrated that E18 rat hippocampal cultures could be maintained up to 45 days; however, after this span the neurons experienced significant degradation [6,7]. Considering the available data, it becomes clear that a more successful method would not only maintain the neurons for longer periods of time, but would also contain a sufficiently large, but controlled glial population, so that proper neuronal morphology and physiology may develop in vitro.
Neuronal harvesting has been reduced to common practice such that industrial resources now exist for obtaining neurons ready for use. However, plating and maintaining these cultures still proves to be problematic. Furthermore, while the Banker method [8], and other methodologies that closely resemble it, are excellent for producing low-density pure neuronal cultures, methods for producing long-term co-cultures (for months at a time) are lacking. Co-cultures have the advantage of producing tripartite synapses with glia [9]. Indeed, 50% of synapses in the mature hippocampus are in direct contact with astrocytes, the predominating cell type in glial populations of this tissue [10].
Considering the limitations mentioned above, we here present an optimized, hybrid method for maintaining long-term hippocampal cultures-one that incorporates key elements of the Banker method along with the addition of other improved methodologies which were determined empirically. This newly proposed method can produce tens of cultures from a single dissection and provides alternative solutions to the problems associated with traditional neuronal cultures. Moreover, we determined that this method Long-Term Hippocampal Neuronal Culturing PLOS ONE | www.plosone.org allows viable neurons in high densities to be grown for at least 120 days, forming large-scale neuronal circuits. We observe that cultures mature over time, and in some cases mirror functional results obtained from in vivo studies such as glutamate responses, synaptic density, and neural net integration. We provide here a detailed protocol for performing these cultures and suggest that this method may be highly effective for any neuronal culture that already performs in B27/Neurobasal media.

Materials and Methods
(Important technique tips are provided in brackets). [Slow and steady media addition rate and proper pipette position are necessary for successful plating density consistency. Position pipette tip at 45u angle along middle of 6-well interior sidewalls, dispense 1.5 mL as slowly and steadily as possible (see .] 16. Incubate cells at 37uC with 5% CO 2 and/or 9% or 20% oxygen. 17. Add Cytosine b-D-arabinofuranoside (Ara-C) to a final concentration of 5 mM, 5-6 days after plating to curb glial proliferation. [Remove 1/3 of media from each well and replace with equal volume containing final concentration of Ara-C] 18. After 4 days or longer, neurons are well differentiated. If further culture is desired, change 1/3 of medium with fresh, pre-warmed 1:1 ACM/NbActiv4 every 7-8 days.

Antibodies and Reagents
Fura-2, AM, Alexa Fluor 488 goat anti-rabbit IgG, Alexa Fluor 647 donkey anti-goat IgG, Alexa Fluor 546 goat anti-mouse IgG, Alexa Fluor 647 goat anti-chicken IgG, and ProLong Gold antifade reagent were from Molecular Probes. Chicken IgY MAP-2 antibody was from Neuromics. Rabbit polyclonal anti-GFAP and mouse monoclonal anti-b-catenin antibodies were from Abcam, Inc. Rabbit polyclonal Synapsin antibody was from Millipore. Mouse S100B and rabbit bIII-Tubulin antibodies were generous gifts from Daniel Daugherty of the Winben Deng Lab (University of California, Davis, School of Medicine). Rabbit polyclonal IP 3 R antibody was a generous gift from Dr. Darren Boehning (University of Texas Medical Branch). Wnt3a and Wnt5a ligands were obtained from R&D systems; L-glutamine was from Cellgro; goat serum was from Gibco; L-glutamic acid, bovine serum albumin, and saponin were from Sigma; NaCl was from VWR; KCl was from EMD Chemicals; MgCl 2 h?6H 2 O, glucose, and HEPES were from J.T. Baker; paraformaldehyde was from Mallinckrodt Chemicals. Lipofectamine 2000 and Opti-MEM were from Life Technologies. Papain Dissociation System was from Worthington Biochemical Corporation.

Intracellular Calcium Measurements
The methodology for culture and cytosolic Ca 2+ measurements within rat E18 hippocampal neurons using fura-2 were as described previously [11][12][13]. Briefly, cells were grown on glass coverslips as described above and fura-2-loaded by incubation with 2 mM fura-2, AM in HEPES-buffered saline solution with CaCl 2 and L-glutamine (107 mM NaCl, 7.24 mM KCl, 1.18 mM MgCl 2 ?6H 2 O, 2 mM CaCl 2 , 11.5 mM glucose, 20 mM HEPES, 2 mM L-glutamine, pH 7.2, 0.1% bovine serum albumin) for 30 min at 37uC. After de-esterification in fresh loading medium for 30 min at room temperature, coverslips were inserted into a Dvorak-Stotler Controlled-Environment Culture Chamber (DSC200, Nicholson Precision Instrument, Gaithersburg, MD). Groups of 40-100 individual fura-2-loaded cells were viewed through a Nikon Eclipse TS100 microscope and Plan Fluor 20X/ 0.50 objective. Fluorescence emission at 510 nm was monitored at room temperature using an Intracellular Imaging Inc. InCyt Standard IM Fluorescence Imaging System with excitation at 340 and 380 nm. Emission ratios were converted to [Ca 2+ nM] using a standard curve generated using Ca 2+ standards. Resting Ca 2+ levels were approximately 200-250 nM; maximal activation by glutamate resulted in up to 2400 nM Ca 2+ .

Transfection
Transfection was performed as previously described [14]. The following briefly describes the process as performed 4 days after cell seeding. 8 mL of Lipofectamine 2000 were diluted in 100 mL Opti-MEM and briefly vortexed while 1.5 mL YFP vector was combined with 100 mL Opti-MEM and briefly vortexed. After each tube was left to incubate at room temperature for 5 minutes, they were mixed, vortexed, and incubated for an additional 10 minutes. The 1.5 mL (per well) of ACM/NbActiv4 was removed and replaced with 800 mL of Opti-MEM, in which the previously incubated Lipofectamine/DNA complex was added. The removed media was saved, and mixed with fresh ACM/NbActiv4 at a 1:1 ratio. After 4-6 hours, the transfection media was replaced with the 1.5 mL of the previously prepared ACM/NbActiv4. Images were obtained using a Nikon Eclipse TS100 microscope and Plan Fluor 10X/0.25 Ph1 ADL objective. Fluorescence emission at 510 nm was observed at room temperature using an Intracellular Imaging Inc. InCyt Standard IM Fluorescence Imaging System with excitation at 485 nm. Animals P0 California mouse (Peromyscus californicus) neonates (n = 2) were euthanized by rapid decapitation. Brains were rapidly removed and dissected to remove hippocampus and cortex. Tissue was homogenized using a papain dissociation kit (Worthington) according to the manufacturer's instructions. Tissue was kept on papain at 37uC for 1.5 hr. Following dissociation, brain cell cultures were maintained as described above. Animals were treated according to procedures approved by the UC Davis Institutional Animal Care and Use Committee and which adhered to the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Substrate and Media
During the optimization of this protocol, several steps were identified as critical for achieving our desired results, including the use of ACM/NbActiv4 media, German glass coverslips as a substrate, and Ara-C for properly curbing glial proliferation. After assessing many different combinations and various ratios of media types (N2 media, Neurobasal media, ACM, NbActiv4, etc.), we determined that a 1:1 ratio of ACM and NbActiv4 was most conducive to neuronal maturation, differentiation, and longevity. In accordance with the Banker method, we performed multiple trials comparing coverslips made from different glass sources and determined empirically that coverslips manufactured from German borosilicate glass had a significantly higher rate of neuronal attachment and survivability versus coverslips obtained from domestic sources. Lastly, we determined that the addition of 5 mM Ara-C after 5-6 days of culturing was a suitable time frame for optimal retardation of glial proliferation; however, depending on the application, this time could be extended or reduced. Figure 2 depicts representative laser scanning confocal images of E18 hippocampal cultures grown for 30 (A) and 120 (B) days in vitro (DIV) using the method described above. The 120 DIV image depicts the longest culture maintained in the literature to date by more than 60 days. Under these conditions, we are able to maintain a stable glia to neuron ratio (10-20% glia) over the course of our cultures. Thus, as consistent with previously described long-term hippocampal cultures, we observe increased arborization, but not proliferation, of both glia and pyramidal cells when compared to younger cultures [6]. However, unlike earlier described long-term hippocampal cultures we do not observe the same decrease in MAP-2 expression, which has been previously reported [6]. This is likely due to the improved health of these cultures.  while 120 DIV cultures averaged 1.29 synapses per micrometer (with P,0.05). This is in strong correlation to the 1.10 total synapses per micrometer reported in previous studies of 14 DIV cultures [15]. However, unlike findings from Boyer et al. (1998) where total synapse numbers decrease after 14 DIV, our cultures show an increase in synaptic puncta per unit length of dendrite over time. Furthermore, our 120 DIV results are closer in proximity to estimates of 1.7 spine synapses per micrometer in the adult rat hippocampus [16]. Additionally, we used ImageJ software (NIH) to quantify relative synapse to dendrite area and found that 120 DIV cultures showed a 16.5% increase over 15 DIV cultures (0.424 versus 0.364 respectively).

Tripartite Synapse
Previous studies underline the importance of the contribution of glia in the development and function of synapses [9,[17][18][19][20][21]. It has been shown that individual astrocytes can contact up to 140,000 synapses [22]. In addition to diffusible secreted factors, direct astrocytic contacts have proven do induce large-scale maturation and a subsequent up-regulation of synapse formation in neurons [23]. Indeed, astrocytic membrane-bound ligands such as ephrin-A3 regulate spine morphology in pyramidal cells of the hippocampus, and as a whole, may contribute to initial synapse formation [24]. Additionally, successful labeling has revealed intricate interactions between astrocytes and dendritic arbors of CA1 neurons [25]. These interactions have been directly associated with synaptic plasticity, where astrocytes release glutamate and play a large role in regulating synaptic transmission [26][27][28][29].
For reasons mentioned above, it is important to preserve intimate contacts between astrocytes and neurons during culturing. Figure 4 illustrates a multi-layered culture with neuronal/glial contacts resembling those observed in tissue slices. This figure depicts the direct interaction between pyramidal dendrites stained with anti-MAP-2 (green), and glial processes stained with anti-GFAP (red), consistent with observations reported by Nishida et al. (2007) [24]. Careful examination reveals intricate associations where dendrites appear to entwine glial processes (see Fig. 4 and Fig. 5). In addition, Figure 5 depicts glial projections (blue), as stained by S100B, closely associated with synapse-rich dendrites (red) as stained by MAP-2. This crucial interaction cannot be achieved using the conventional Banker method due to the physical separation of neuronal and glial cultures.

Intracellular Calcium Measurements
To ensure that these cultures were comprised of viable neurons, we performed calcium imaging experiments and observed a glutamate-induced increase in intracellular calcium concentration over nearly all of the sampled cells (Fig. 6). Consistent with previous results comparing age-related calcium changes in rat hippocampal brain slices, the addition of 1 mM glutamate elicited a significantly larger calcium flux in 15 DIV cultures compared to those of 120 DIV [30]. These results further elucidate the degree in which these long-term cultures appear to mirror neuronal organoid activity.
These long-term in vivo cultures also form complete large-scale neuronal circuits that produce spontaneous, synchronous calcium oscillations (SSCOs) that can persist throughout the imaging field during both baseline and excitatory phases of intracellular calcium   (Fig. 7). It has been previously demonstrated that SSCOs are dependent on the establishment of functional synaptic contacts, and that in all cases, an extensive network of synaptic contacts were required for this oscillatory phenomenon to occur. Moreover, it has recently been shown that glial populations present in hippocampal cultures influence the total level of synchronization, and so, again, this indicates a specific interdependence between the glial and neuronal networks [31]. These intracellular calcium oscillations represent an example of typical physiological interactions in neuronal circuits, and are the result of intense synaptic activity [32]. SSCOs observed in these experiments represent yet another example of the viability and physiologically accurate nature of these long-term cultures. Figure 8 is a representative confocal image of older cultures imaged using confocal microscopy at various magnifications (10-63X). These cultures have a propensity to grow neurites that are reminiscent of parallel fibers found in the brain (Fig. 8A). We have only observed this particular type of growth to occur after 30 days in culture, and more regularly after at least 50 days in culture. In magnified views (Fig. 8B-C), we observe a high degree of order and remarkable growth patterns. These elaborate patterns and long stretches of parallel growth suggest some form of communication throughout neurite development. This idea is strengthened by examining neuron-specific immunostaining (bIII-Tubulin), where neurites appear to form complex matrix patterns (Fig. 8E). It is also worth noting that the neurite network continues to increase in overall density and field depth, as can be seen by comparing 40 DIV (Fig. 8E) to 120 DIV (Fig.8D), where the neurites have developed into an additional network layer above glial processes. As a result, we found that as cultures age, the neurite field continues to increase in overall diameter and depth.

Growth and Development
These observations suggest that these cultures are significantly different in their cellular functions at these disparate time points, which is consistent with the calcium imaging results mentioned above. Further studies monitoring SSCOs along with different levels and types of neurite development are needed to characterize these associations. In addition, future studies are needed to determine what types of co-cultures and/or organoid precursors can be created using this method. Of note, we have observed a propensity of these cultures to produce what appears to be neurospheres with a high degree of proliferation (Fig. 9,10D-E). Indeed, the neurosphere-like structures we observed from both the rat E18 hippocampal neurons (Fig. 9) and the P0 neonatal California mice hippocampal neurons (Fig. 10D-E) appear very similar to those reported in the literature [33]. Further, they accord with sizes reported and levels of protein expression. As shown in Fig. 9A, the center of the neurosphere stains positive for b-catenin (red), while the migrating cells appear to express much lower levels. This is consistent with previous studies demonstrating the requirement of b-catenin expression for obtaining developing and adherent neurospheres [34,35]. As a positive control for bcatenin staining, we stimulated our neurons with Wnt ligands, which promote b-catenin re-localization to the cytosol (Wnt5, Fig. 9B), or the nucleus (Wnt3, Fig. 9C).

Other Types of Neuronal Cultures
Given these results, we assessed whether our protocol could be used successfully in multiple neuronal types from a different, but related laboratory animal model species. The California mouse is an undomesticated species that is more closely related to rats than mice (Mus musculus). The lack of domestication makes them useful in studies that examine environmental impacts on physiology and behavior. When we applied our method to both the hippocampus and cortex from P0 neonates of this species we observed that both of these neuronal populations performed well. We observed dramatic expansion and the initiation of migration within the hippocampal neuronal populations over 21 days (Fig. 10. A-C). In addition, the cortical neurons were transfected using a standard Lipofectamine 2000 transfection protocol (see Methods). When cortical neurons are transfected 4 days post-plating we obtain 10-15% transfection efficiency of both glia and neurons (Fig. 10 F-G) as observed at day 9. The efficacy of our cell culture protocol in both a domesticated species like the rat and undomesticated California mouse suggests that the method could apply to broad areas of research that utilize specialized animal models.

Conclusion
Given the above results, as a whole, we suggest that this method provides an efficient, cost-effective, and scalable method for producing both short-and long-term hippocampal cultures that have significant advantages over current techniques, in particular, as regards their developmental progression. Furthermore, this method can be performed in almost any laboratory due to the availability of neuronal dissections from industrial sources. While the limits to which these cultures can replicate in vivo physiological responses has yet to be determined, the data presented in this study demonstrates that many of the conventional measurements taken from brain slice studies are closely recapitulated by our cultures. Ultimately, this culturing method results in a physiologically relevant model system that can be cultured for significantly longer periods of time than other long-term hippocampal studies that we are aware of [6,7]. In short, our cultures represent an ex vivo model system that more accurately portrays an in vivo hippocampal tissue with respect to glial populations while the primary neurons remain in direct contact with each other, thereby forming physiologically relevant synapses. In closing, we suggest that exploration of this method may allow for the production of true organoid neuronal cultures. Moreover, as hippocampal cultures already serve as an ongoing model for Alzheimer's disease and in other neurodegeneration studies [6,36], these cultures may provide distinct advantages when testing the effects of long-term genetic modification and/or long-term pharmacological tests. Our future research will focus on (i) exploring the time limits of growing these neurons and cell types, (ii) the physiological processes that can be measured in these cultures, and (iii) understanding how closely the development of these cultures mirrors in vivo development.