Antisense Oligodeoxynucleotide Perfusion Blocks Gene Expression of Synaptic Plasticity-related Proteins without Inducing Compensation in Hippocampal Slices

The elucidation of the molecular mechanisms of long-term synaptic plasticity has been hindered by both the compensation that can occur after chronic loss of the core plasticity molecules and by ex vivo conditions that may not reproduce in vivo plasticity. Here we describe a novel method to rapidly suppress gene expression by antisense oligodeoxynucleotides (ODNs) applied to rodent brain slices in an “Oslo-type” interface chamber. The method has three advantageous features: 1) rapid blockade of new synthesis of the targeted proteins that avoids genetic compensation, 2) efficient oxygenation of the brain slice, which is critical for reproducing in vivo conditions of long-term synaptic plasticity, and 3) a recirculation system that uses only small volumes of bath solution (< 5 ml), reducing the amount of reagents required for long-term experiments lasting many hours. The method employs a custom-made recirculation system involving piezoelectric micropumps and was first used for the acute translational blockade of protein kinase Mζ (PKMζ) synthesis during long-term potentiation (LTP) by Tsokas et al., 2016. In that study, applying antisense-ODN rapidly prevents the synthesis of PKMζ and blocks late-LTP without inducing the compensation by other protein kinase C (PKC) isoforms that occurs in PKCζ/PKMζ knockout mice. In addition, we show that in a low-oxygenation submersion-type chamber, applications of the atypical PKC inhibitor, zeta inhibitory peptide (ZIP), can result in unstable baseline synaptic transmission, but in the high-oxygenation, “Oslo-type” interface electrophysiology chamber, the drug reverses late-LTP without affecting baseline synaptic transmission. This comparison reveals that the interface chamber, but not the submersion chamber, reproduces the effects of ZIP in vivo. Therefore, the protocol combines the ability to acutely block new synthesis of specific proteins for the study of long-term synaptic plasticity, while maintaining properties of synaptic transmission that reproduce in vivo conditions relevant for long-term memory.

compound as a bolus at a high concentration that then must diffuse through the brain tissue in the intact animal prior to the slice preparation. As a result, the experimenter does not have precise control over the concentration of the compound present in the slice at the time of recording. If a high drug concentration in the bolus injection is required, this in turn can lead to questions about pharmacological specificity.
The method described in this protocol renders the perfusion of brain slices with antisense-ODN as easy as the application of any other soluble reagent or drug. Therefore, the technique may be preferable to the standard method of intracranial or intraventricular injections followed by preparing brain slices. In addition, electrophysiological recordings are in an "Oslo-type" interface brain slice recording chamber. Interface chambers are preferable to submersion chambers for long-term recordings of brain slices because they provide superior oxygenation (for detailed discussion, see Note 1). This protocol was first used in Tsokas et al. (2016) to block specifically the new synthesis of protein kinase Mζ (PKMζ) in response to strong afferent tetanic stimulation (high-frequency stimulation, HFS) without inducing genetic compensation by the other atypical protein kinase C (PKC) isoform, PKCι/ λ, as occurs in PKMζ-null mice. Indeed, in these mutant mice, long-term potentiation (LTP) and memory formation appear largely intact because the normal physiological function of PKMζ is largely compensated by PKCι/λ (Tsokas et al., 2016).
Genetic compensation for LTP and spatial long-term memory in PKMζ-null mice was revealed in Tsokas et al. (2016) by a pharmacogenetic analysis of wild-type and PKMζ-null mice using PKMζ-antisense ODN. In these experiments, the normal physiological function of PKMζ was selectively blocked by taking advantage of the specific nucleotide sequence of the PKMζ-mRNA translation start site to design a PKMζ-antisense ODN ( Figure 1A) that suppresses the activity-dependent de novo synthesis of PKMζ. Twenty μM PKMζ-antisense was bath-applied to acute wild-type mouse slices and recirculated for 1 h before tetanization and during the critical period of new protein synthesis after tetanization when PKMζ is formed (Osten et al., 1996), i.e., during the temporal window when general protein synthesis inhibitors such as anisomycin are effective in blocking late-LTP induction (Frey and Morris, 1997). The PKMζ-ODN suppressed the new synthesis of PKMζ ( Figure 1B) and late-LTP ( Figure 1C) without affecting the upregulation of PKCι/λ or the eukaryotic elongation factor 1A (eEF1A), proteins that are also rapidly synthesized in LTP (Tsokas et al., 2016; Figure 1B). Because the turnover of PKMζ is relatively slow (Osten et al., 1996), basal amounts of PKMζ in untetanized slices were unaffected (Tsokas et al., 2016). The application of PKMζ-antisense did not suppress LTP in PKMζ-null mice, in which the target of the PKMζ-antisense is absent ( Figure 1C). These results demonstrate that in the mutant mice another molecule compensates for the loss of PKMζ. Conversely, a selective PKCι/λ inhibitor reversed established LTP only in the PKMζ-null mice and not in wild-type mice (Tsokas et al., 2016, Figure 3). This double dissociation between the mechanisms of LTP in PKMζ-null and wild-type mice revealed that when PKMζ is absent there is functional compensation by PKCι/λ.
The blockade of late-LTP and the suppression of activity-dependent PKMζ synthesis with bath-applied, recirculating PKMζ-antisense was also replicated in acute rat hippocampal slices (See Figures 2A, 2B, and S4B from Hsieh et al., 2017). Taken together, these findings suggest that the crucial pool of PKMζ protein that sustains synaptic potentiation is provided by de novo synthesis in response to tetanization, rather than through the recruitment of preexisting, basal PKMζ that had been synthesized before the tetanus. Thus, the use of antisense ODN to specifically suppress new synthesis of PKMζ without affecting basal levels of the kinase is advantageous over genetic knockdown/knockout not only because it prevents genetic compensation, but because it distinguishes between basal PKMζ and the pool of PKMζ that is synthesized in response to HFS. The results reveal that only the newly synthesized pool can support late-LTP. A similar argument can be made for the use of acute translational blockade by antisense ODN rather than genetic knockdown/knockout in behavioral experiments designed to elucidate the mechanisms involved in the formation and persistence of long-term memory under physiological conditions in wild-type animals (Tsokas et al., 2016;Hsieh et al., 2017).
In our protocol, antisense ODN is delivered by a custom-built recirculation system that perfuses brain slices resting in an "Oslo-type" interface recording chamber with a small (5 ml) recirculating volume of artificial cerebrospinal fluid (ACSF). The system employs the mp6 micropump, a piezoelectric diaphragm pump developed by Bartels Mikrotechnik GmbH (Dortmund, Germany) for the transport of liquids or gases at varying flow rates and/or pressures controlled by an external electronics circuit. For liquids, each mp6 micropump supplies a maximal flow rate of 5.5 ml/min. With parallel connection of multiple mp6 units the volume flows summate. The mp6 can be controlled by the commercially available mp-x controller, or alternatively by the mp6-OEM, both of which are manufactured by Bartels. The mp6-OEM is a small (10.5 × 20.5 × 6 mm) driving circuit capable of generating up to 270 V peak-to-peak voltage at 100 Hz frequency from a 5 V power supply (therefore proper safety measures are required). The OEM controller drives the micropump at adjustable performance and can be integrated into system electronics, a PCB design, or as in the case of this protocol a breadboard.
Each mp6-OEM is intended for operating one mp6 micropump. Therefore, the six mp-x controllers that would be required for driving the mp6 arrays used in this protocol would make the cost of using the commercially available controller quite steep-as opposed to using six mp6-OEMs and building a custom-made circuit at a small fraction of the price. The mp6-OEM has a built-in interface that allows the user to adapt the adjustable parameters (frequency and amplitude) of the rectangular signal generated by the OEM by the use of external components, such as a circuit consisting of a potential divider and a capacitor, or a microcontroller capable of pulse-width modulation (PWM). For the purposes of this protocol, an Arduino Uno is used with a simple program that performs PWM to control the amplitude and/or the frequency. Using an open source Arduino library and a program provided with this protocol, the user may achieve flow rates between 0.2 ml/min and 5.5 ml/min from each mp6 micropump, appropriate for maintaining brain slices (see Figure 20). The schematic in Figure 2A represents a simplified version of the circuit that controls the flow rate of the pumps which supply ACSF to the recording chamber (inflow micropumps, controlled by the OEM circuit at the bottom via Microcontroller 2). Figure 2A also shows a simplified version of the circuit controlling the suction pumps (outflow micropumps, controlled by the OEM at the top). To achieve laminar flow when the chamber is operated in submersion mode, i.e., during ODN delivery prior to stimulation and recording, slices are perfused on two sides-at the top and the bottom of the mesh. Two micropumps driven by two separate OEMs are therefore used for controlling the inflow: one pump perfuses the top of the mesh, and the other supplies ACSF to the part of the chamber below the mesh ( Figures  15A and 15B). A "sandwich" consisting of four mp6 pumps connected in parallel fashion is used to supply the suction that completes the ACSF recirculation circuit ( Figure 15C). These four mp6 micropumps operate constantly at maximal flow rate and are therefore controlled by a simpler version of the circuit used for the two inflow micropumps, which does not perform pulse-width modulation.
A separate Arduino Uno microcontroller (Microcontroller 1) is used to turn all the mp6 micropumps on/off in a concerted fashion according to inputs received by the ADC and the software driving the electrophysiology experiment. In this manner, the recirculation system can be used in combination with a peristaltic pump via a three-way valve and activated only during drug application.
The first part of this protocol is a description of how to build the electronic circuit that controls the operation and flow rate of the mp6 micropump arrays. The second part provides a detailed hippocampal slice preparation protocol for electrophysiology in interface chambers. The protocol includes modifications of the classic method described in the scientific literature, which, based on our experience, we believe are necessary for making high-quality slices suitable for long electrophysiological recordings and subsequent biochemical and immunocytochemical analysis.
In addition, we provide evidence for the advantage of interface over submersion electrophysiology chambers for long recordings of acute brain slices required for the study of long-term synaptic plasticity (see Note 1). Whereas low-oxygen submersion chambers show unstable baseline synaptic transmission following long-term applications of the zeta inhibitory peptide (ZIP), the interface electrophysiology chambers maintain stable baseline synaptic transmission, as observed after applications of ZIP in vivo (Pastalkova et al., 2006). Thus, because brain slices maintained in interface chambers more closely preserve the in vivo physiological state, they are the preferred method for investigating the long-term synaptic plasticity thought to underlie learning and memory.

1.
To fit all the necessary components, a breadboard with 3 terminal strips and 63 rows of 10 tie points per terminal strip is required ( Figure 4). Such breadboards include the Digilent 340-002-1 or the Wisher WBU-206. Each of these has 5 power distribution buses and 3 terminal strips. Remove (unscrew) the rightmost distribution bus and the adjacent terminal strip.

2.
Attach with screws or Velcro pads the two Arduinos-ideally in a staggered fashion to allow easy access to the two USB ports, as shown in Figure 4.

3.
Connect the 5 Volt DC Power Supply to the horizontal distribution bus of the breadboard ( Figure 4, designated by the red "plus" and the black "minus" signs). Include a simple toggle SPST push button "master switch" (not shown), for switching the power supply to the circuit.

4.
Connect the power and ground of all vertical distribution buses to 5 V and ground, respectively, of the horizontal distribution bus ( Figure 4)

5.
On the middle (now rightmost) terminal strip attach the two 10K-Ohm potentiometers, and connect them to power and ground, as shown in Figure 5.
Note: The screws that rotate the wiper and vary the resistance in each potentiometer should be facing away from each other, therefore allowing enough clearance for the mini screwdriver that will be used to make adjustments in Step 6.

6.
Using a mini screwdriver, adjust voltage to get 1.3 V across the wiper of the top potentiometer and 1.8 V across the bottom Note: For safety, you should disconnect the power supply when proceeding with the next steps of the protocol.

7.
On the terminal strip that includes the two potentiometers also install the two 8channel logic level converters ( Figure 6). Connect VCCB on each logic level converter to 5 V and GND to ground ( Figure 6; also refer to Figure 2A for pinout diagram and voltage information).

8.
Connect VCCA and OE on the top logic level converter to 1.3 V and on the bottom converter to 1.8 V (see gray jumper wires in Figures 6; also refer to Figure 2A for voltage and pin information).

9.
Attach the 26-pin Assembled Pi Cobbler Breakout Board. Use the shield stacking headers to elevate the Breakout Board, and therefore allow enough clearance for the third and fourth mp6-OEM controllers that will be attached next (Step 10) and are directly adjacent to the Breakout Board (see Figure 15D).

10.
Connect the six mp6-OEM controllers to the leftmost terminal strip of the breadboard as shown in Figure 7 (also refer to Figure

11.
Short Pins 11 and 12 on all six mp6-OEMs, using U-shaped loops made out of cut jumper wires, long enough to connect two contacts on adjacent breadboard rows (see six-pointed asterisks in Figure 7; refer to Figure 2A for numbering of pins).

12.
Similarly, short Pins 2 and 3 of the top four OEMs (see five-pointed asterisks in Figure 7; refer to Figure 2A for numbering of pins).

13.
Connect the ground (Black) to Pin 1, marked with a white spot of all six mp6-OEMs ( Figure 7; refer to Figure 2A for numbering of pins).

14.
Connect 5 V (Red) to Pin 14 of all six mp6-OEMs, as shown in Figure 7 (refer to Figure 2A for numbering of pins).  Figure 2A).

24.
Connect one of the Digital outputs of the ADC to TX 1 and GDN of the top Arduino Uno (Figure 2A). This connection will turn on and off the recirculation perfusion controlled by the mp6 pumps, as directed by the winLTP Program.

25.
Connect the two Arduino Uno units to the computer, using USB connection.

26.
Connect the GDN of the bottom Arduino to the central ground of the electrophysiology rig. (Figure 2A. The top Arduino is grounded via the ADC ground-see Step 24).

27.
Copy and load into the top Arduino controlling all six micropumps (Microcontroller 1) the program shown in Figure 18.

28.
Copy and load into the bottom Arduino controlling the two inflow micropumps (Microcontroller 2) the program shown in Figure 19.
B. Assembly of the mp6 micropump array (see reference 29)

2.
Position both components as indicated in Figure 2C, the mp6/mp6-pp facing upwards (with its company logo and serial number markings visible from above), and the Molex connector with the four small openings facing down. Then slide the mp6/mp6-pp flex into the Molex connector. Clamp the Molex connector to complete the connection between both components.

3.
Manufacture three custom-made plexiglass spacers similar to the ones shown in Figures 15A and 15B. Also prepare cork pads cut to size in order to compensate for differences in the thickness of the micropump and the Molex connectors.

4.
Make an assembly consisting of two mp6 micropumps as shown in Figures 15A  (front) and 15B (side) as follows: First, glue the marked side of micropump 1 (mp6-1) on the plexiglass spacer using double-sided adhesive tape. Glue cork pads to the unmarked side of the first micropump. The order of parts in the resulting component is: plexiglass/mp6-1/cork.

5.
Repeat these steps for micropump 2 (mp6-2), but in the opposite order, i.e., with the cork pad glued to marked side and the plexiglass spacer on the unmarked. The order of parts in the resulting component is: cork/mp6-2/plexiglass.

7.
The resulting assembly can be held together either with pieces of double-sided adhesive tape, or by a clamp, or with long screws that pass through drilled holes into the plexiglass spacers. In the latter case, make sure to tap threads into the holes of only one out of the three spacers.

8.
Make mini-Faraday cages out of copper wire mesh ( Figure 15F) to allow the elimination of the noise generated by the piezoelectric micropumps (see Note 3).
Ground the mini Faraday cages on the main ground of the airtable.

9.
The completed assemblies consist of either two or four mp6 micropumps ( Figure  15F). As discussed above, four pumps connected in parallel are required for the suction, and two pumps with independent inlets and outlets perfuse the top and bottom of the hippocampal slice, from both sides of the mesh in the recording chamber.

1.
Prepare Dissection and Recording Solutions: Prepare two stock 10× ACSF solutions, one for dissection (Dissection 10× ACSF) and one for recording (Recording 10× ACSF). Maintain the two stock 10× ACSF solutions at 4 °C until the day of the experiment.

2.
On the day of the experiment, prepare 1× Dissection ACSF and 1× Recording ACSF.

4.
Reducing the temperature of the slice helps prevent ischemic damage (see Note 2). To rapidly chill room-temperature 1× Dissection ACSF Buffer to an appropriate cold temperature without freezing the buffer, place it into a −70 °C freezer for 45 min. If ice crystals form, remove them with a small spoon or sieve to avoid contact with, and possible damage to the brain tissue. After removing from the freezer continue to oxygenate the Dissection ACSF with 95% O 2 /5% CO 2 at 4 °C.

5.
Prepare the interface electrophysiology rig: Heat the water jacket of the interface chamber so that the temperature of the recording chamber is 31.5 °C. Preheat 250-500 ml of the 1× Recording ACSF by placing it in a 32 °C water bath while oxygenating with 95% O 2 /5% CO 2 . The purpose of preheating the solution to a temperature slightly above the temperature of the recording chamber is to prevent the formation of bubbles in the tubes and under the mesh when the solution is heated again after passing from the heat jacket. Start perfusing the bath with 1× Recording ACSF, at a flow rate of 500 μl/min.

7.
When the animal is deeply anesthetized, decapitate it with a small animal guillotine. Submerge the decapitated head inside a metal surgical tray sitting on ice in an icebox and filled with oxygenated ice-cold 1× Dissection ACSF Note: Following the decapitation of the animal, the preparation of the brain slices should be completed within less than 10 min.

8.
Quickly remove the brain: expose the skull by making an incision with a single edge razor blade on the scalp.

9.
Using a pair of surgical scissors with their sharp pointed blade working along the inner side of the skull, cut along the sagittal suture from the foramen magnum to the forehead.

10.
Using the surgical scissors, make one cut at the foramen magnum on each temporal side of the skull, then cut across the frontal bone along the coronal suture. Carefully pry the skull open with a rongeur and expose the brain.

11.
Holding the skull upside down, sever the cranial nerves that hold the brain to the skull using a spatula and allow the brain to fall into a 50 ml beaker containing ice-cold oxygenated 1× Dissection ACSF.

12.
Using an icebox, transfer the beaker with the brain to a 4 °C cold room.

13.
Isolate the hippocampus: use a 90 mm Pyrex Culture Petri Dish covered with a moistened filter paper as a dissecting platform.

14.
Place the chilled brain on the platform and bisect it along the longitudinal fissure. Place the right hemibrain in a 50 ml beaker containing ice-cold oxygenated 1× Dissection ACSF.

15.
To isolate the hippocampus of the left hemibrain, sever what remains of the midbrain and brain-stem using a flat spatula. Scoop out the thalamus to expose the ventral face of the hippocampus nested inside the cortex and clearly visible as a curved structure. Sever the septal and temporal connections of the hippocampus from the cortex with a spatula. Using a filled plastic Pasteur pipet, gently squeeze a few droplets of ice-cold oxygenated 1× Dissection ACSF into the pocket of the lateral ventricle to distinguish the border of the fimbria.

16.
Gently insert into the lateral ventricle a fine brush moistened with ice-cold oxygenated 1× Dissection ACSF and roll the hippocampus out gently from the surrounding cortex. Isolate the peeled-out hippocampus by cutting it free from the cortex using a flat spatula. Place the left hippocampus in a 20 ml beaker with ice-cold oxygenated 1× Dissection ACSF.

17.
Repeat previous steps with right hippocampus.

18.
Slicing: The slicing stage should be prepared in advance and can be reused multiple times. It consists of a circular thin piece of cork pad of equal diameter to the plastic disc of the McIlwain tissue chopper and glued on it with parallel pieces of double-sided adhesive tape. The clamps of the McIlwain tissue chopper dissection platform can be adjusted to match the thickness of the slicing stage ( Figure 16C).

19.
Place a piece of filter paper on the slicing stage and moisten it with chilled oxygenated 1× Dissection ACSF.

20.
Using a fine brush (or a spatula with its flat side bent at right angles) gently lift the hippocampus from the bottom of the beaker.

21.
Lay the isolated hippocampus on the slicing stage with its top side up (the dorsal face containing the alveus). It is often possible to view striations on the alvear surface of this side, with oblique light from a fiber optic. Using the brush gently straighten the bottom surface of the hippocampus so that it lies completely flush on the moistened filter paper.

22.
Only the dorsal hippocampus will be sliced from each side. To immobilize the hippocampus, one may therefore affix the ventral hippocampus (the thicker end) on the cork of the slicing stage with an insect pin ( Figure 16A).

23.
Slide the slicing stage with the affixed hippocampus through the clamps of the McIlwain tissue chopper dissection platform ( Figure 16C).

24.
Rotate the slicing stage on the platform so that the dorsal hippocampus is properly oriented with respect to the razor blade: the blade should make a 15° angle with the transverse axis of the dorsal hippocampus ( Figures 16B, 16C). Excitatory pathways are better preserved when an angle of 15 to 30° from this axis is used (Alger et al., 1984). In general, the alvear striations will tend to run parallel to the blade at this orientation. If the hippocampus on the left hemibrain is used, it will look like an upright "U" as viewed from above with the dorsal hippocampus on the right and the (pinned) ventral hippocampus on the left ( Figures 16A-16C). To achieve the proper cutting angle, rotate the slicing stage 15° counterclockwise. Similarly, if the right hippocampus is used, it will resemble an inverted "U", and the stage is rotated 15° clockwise.

25.
Wet the blade with ACSF and make the slices by lowering the blade through the hippocampus. Each slice should stick on the wet surface of the blade as it ascends. The first 3 (most dorsal) complete slices are usually discarded.
Depending on the age of the animal, 6 to 8 slices of 450 mm thickness can be obtained from rats ( Figure 16B) and 4-5 such slices from mice. The speed and the strength of impact of the blade should be adjusted so that the slices are cut gently, but quickly enough to avoid sticking of the uncut hippocampus on the ascending blade. The blade should not slam on the surface of the cutting stage; ideally it should just dimple the surface of the wet filter paper on the chopping stage.

26.
Quickly remove the slices from the blade as they are produced using a moistened fine sable brush. Immediately after making each slice, transfer it to a test tube filled with ice-cold oxygenated 1× Dissection ACSF. Take great care to avoid stretching or excessively bending the slices during handling. One method is to pick the slices from the blade with the brush premoistened with ice-cold oxygenated 1× Dissection ACSF, using a gentle rolling motion. Some of the ACSF moisture on the brush will also stick on the blade and help the next slice stick to the blade after it has been sliced off the hippocampus.

27.
Immediately after making the slices, transfer them to the interface recording chamber to recover for at least 2 h. The slices should rest on a mesh at the gasliquid interface, forming a small meniscus around them ( Figures 17A, 17B and 17C), and should be constantly superfused from both their top and bottom side with (non-recirculating) 1× Recording ACSF, at a flow rate of 500 μl/min.
Note: Please also consult the following video articles offering detailed information on acute hippocampal slice preparation for long-term recordings: Villers and Ris (2013); Shetty et al. (2015).

Part III: Application of oligodeoxynucleotides to slices in recirculation mode
In ODN experiments, after recovery from dissection in interface mode, increase the bath level to fully submerge the slices and allow the superfusate containing 20 μM of the oligodeoxynucleotide to recirculate (5 ml total volume at 5 ml/min for 30 min). To achieve this, follow these steps:

1.
Copy and load the program of Figure 18 on the top Arduino of Figure 14 ("Microcontroller 1" in Figure 2A). Change the value of "delay" to "1800000", save the new program and reload to top Arduino. Upon activation by the digital output of ADC via the winLTP program, the top Arduino will turn on all six of the micropumps for 30 min.

2.
Load the Arduino PWM library (see link) and copy and load the program of Figure 19 on the bottom Arduino of Figure 14 ("Microcontroller 2" in Figure 2). Change the "int32_t clockfrequency" value to "200", save the new program and reload to the bottom Arduino. Upon activation by the digital output of the ADC via the winLTP program, the bottom two pumps will supply a flow rate of 5 mL/min (Figure 20) to the inlets of the recording chamber (the four ouflow pumps of the suction, once on, will always operate at maximum flow rate).

3.
Turn off the peristaltic pump, and at the same time turn off the vacuum suction.

4.
Turn on the piezoelectric pump circuit using the SPST push button master switch.

5.
Use a three-way valve to manually switch between the peristaltic pump drawing solution from a main reservoir (typically a 250 ml Erlenmeyer flask held inside a water bath with a clamp) and the two-micropump assembly of the bath inlet, which should be drawing solution from a 15 ml conical tube containing the recirculating solution with the ODN (inside the same water bath as the main reservoir).

6.
Start Note: The automated perfusion control is currently only supported by National Instruments M-or X-Series ADC boards). Use the Slow0 Perfusion Change for controlling one perfusion line to one extracellular slice chamber.
Note: The digital output from the National Instruments M-or X-Series board is connected to Arduino 1, and it will activate both the assembly consisting of the two independent micropumps that control the inlets to the bath, as well as the multipump array that controls the suction. Make sure the tube at the output of the aspirator leads the solution back to the 15 ml conical tube to complete the recirculation circuit.

7.
If necessary, adjust the height of the suction tube in the suction well of the chamber to raise the bath level and fully submerge the slice.

8.
After 30 min of recirculation in the submersion mode and to begin recording, lower the bath level of the ODN-containing solution by lowering the aspirator tube to interface level and return the flow rate to 0.5 ml/min for the remainder of the experiment by changing the "int32_t clockfrequency" value from 200 to 10, and loading again the program of Figure 19 to the Bottom Arduino Uno.

9.
Place stimulating and recording electrodes in the hippocampal layers of interest and begin electrophysiological recordings.

10.
For experiments to study long-term synaptic plasticity lasting many hours, record field EPSPs (fEPSPs) with a glass extracellular recording electrode (2)(3)(4)(5) placed in the CA1 stratum radiatum, with concentric bipolar stimulating electrodes placed on either side within CA3 or CA1. Exclude from study slices if initial analysis shows fEPSP spike threshold < 2 mV. Confirm independence between the two stimulated pathways by the absence of paired-pulse facilitation between the two pathways. Set the baseline fEPSP at 25% of the spike threshold and monitor it by delivering stimuli at 0.033 Hz to each pathway. Induce LTP by strong HFS, consisting, for example, of standard two 100 Hz-1s tetanic trains, spaced 20 s apart, which is optimized to produce a relatively rapid onset of protein synthesis-dependent late-LTP (

Comparison of interface and submersion chamber: oxygenation and efficacy of aPKC inhibitor ZIP on synaptic transmission
Most studies examining long-term synaptic plasticity in hippocampal slices, employ Oslo-type interface recording chambers, rather than submerged chambers, because the former are believed to provide superior oxygenation and preservation of normal synaptic function (Khurana and Li, 2013). We tested this assumption, by using a galvanic dissolved-oxygen probe (Atlas Scientific KIT-106; zinc anode, silver cathode, 15% sodium tetraborate/15% sodium chloride electrolyte; polyethylene membrane) to measure the dissolved oxygen (DO) concentration in the Oslo-type interface recording chamber and in the same chamber used in submersion mode with flow rates ranging from 2 to 15 ml/min.
The dissolved oxygen probe was calibrated using distilled water and ambient air, as described by the manufacturer. The accuracy of the instrument was further tested by measuring dissolved atmospheric oxygen in ACSF (14,000 μS/cm) at room temperature (23.5 °C). The average value (8.1 ± 0.01 mg/L; n = 5) was in agreement with previously reported data for atmospheric dissolved oxygen equilibrium at different temperatures and salinity conditions (Radtke et al., 1998).
Air from a medical air cylinder, or 100% oxygen (both from TW Smith) was then bubbled through the gas dispersion ring (or the ceramic aerators) of the water jacket of the Oslo-type chamber at 31.5 °C, and the concentration of dissolved oxygen was measured inside the recording chamber, under interface conditions, with the probe clamped on a micromanipulator and held a few millimeters above the surface of the mesh (where the meniscus surrounding the slice is normally formed). A standard curve of oxygen saturation vs. concentration was thus generated for the tip of the probe moistened with ACSF (6.9 ± 0.3 mg/L at 21% O2; 31.1 ± 0.8 mg/L at 100% O2; n = 5; linear fit through 0, r 2 = 0.99). With 95% O 2 −5% CO 2 bubbling through the water jacket, the oxygen concentration immediately above the slice at the interface was 29.6 ± 0.7 mg/L (95.1 ± 2.3% saturation; n = 5).
The Oslo-type chamber was then converted into a submersion chamber: the suction was raised to increase the level of the superfusate inside the recording bath, the oxygenation via the aerators in the water jacket was turned off, and the flow rate was increased to obtain oxygen concentration/saturation measurements at 2, 5, 10 and 15 ml/min. Under submersion conditions, and immediately above the position where the slice would be in an actual experiment, the following oxygen concentration (mg/L) and % saturation measurements were obtained (n's = 4)-2 ml/min: 19.6 ± 0.9 (62.5 ± 2.8%); 5 ml/min: 22.7 ± 1.4 (72.4 ± 4.6%); 10 ml/min: 25.5 ± 1.7 (81.6 ± 5.7%); and 15 ml/min: 26.9 ± 1.9 (86.3 ± 6.2%). reported that ZIP had effects on both late-LTP and baseline synaptic transmission (Volk et al., 2013). This study, however, was unusual in that the late-LTP recordings and basal synaptic transmission were recorded in hippocampal slices held in a submerged chamber.
We therefore directly compared the effects of ZIP on LTP and baseline synaptic transmission in hippocampal slices recorded in an interface and in a submersion chamber. To keep all other parameters as constant as possible, the interface chamber slices recovered in the same chamber in which their recordings take place; whereas the submersion chamber slices were placed in an interface chamber, which was then immediately converted into a submersion chamber by increasing the flow rate and bath level of the ACSF solution, and the slices were then allowed to recover for another 90 min (the total recovery time of both sets of slices in the two experiments is kept equal).
As seen in Figure 21, in slices recovered and recorded in an interface chamber, ZIP specifically reverses late-LTP maintenance in the tetanized pathway without affecting basal neurotransmission in the untetanized pathway. In contrast, in slices recorded under submerged conditions, ZIP reverses LTP maintenance, eventually below baseline, and also has an effect on an independent untetanized pathway. This effect of ZIP on the untetanized/basal pathway was similar to that seen in Volk et al. (2013).
A plausible explanation for this difference is that there is a pool of atypical PKC involved in neuroprotection from hypoxia (see Tian et al., 2008) that is separate from the pool of PKMζ that maintains late-LTP. Because the oxygenation conditions in an interface chamber are superior to those for submerged slices, the effect of ZIP on basal neurotransmission in a submersion chamber may reflect the effects of the inhibitor on the neuroprotective role of atypical PKC that is induced by the relatively hypoxic conditions of the slice (Tian et al., 2008).
Alternatively, the low-oxygen conditions of submerged chambers may have induced hypoxic LTP in the untetanized synapses, rendering them sensitive to ZIP. Importantly, as mentioned above, the specificity of ZIP to potentiated and not basal synaptic transmission is seen in vivo in the hippocampal perforant path dentate gyrus input (Pastalkova et al., 2006), hippocampal CA3-CA1 (Madroñal et al., 2010), and layer 4 primary visual cortex (Cooke and Bear, 2010). Thus, high-oxygen interface chambers, but not low-oxygen submerged chambers appear to produce conditions of synaptic transmission and plasticity that more closely reflect in vivo conditions.
In addition to inferior oxygenation conditions, another factor that may have contributed to the artifactual response to ZIP of the untetanized pathway in Volk et al. (2013) is the intensity of the conditioning stimulus used to induce late-LTP. The intensity of the Volk et al. (2013) protocol, which was delivered at 75% of the maximum EPSP response, is appropriate for rat Schaffer collateral/ commissural-CA1 synaptic stimulation, but is much higher than what is normally used to induce LTP in mouse hippocampal slices, which is typically between 25% of the spike threshold, as in Tsokas et al.

2.
Protection from excitotoxicity In order to protect from excitotoxicity during the slice-making process we recommend using ice-cold high magnesium/low calcium ACSF solutions and performing the slice preparation procedure inside a 4 °C cold room.
Lowering the temperature decreases the metabolic rate of cells thus reducing their energy consumption, allowing the cells to survive ischemia during slice preparation. In addition, we recommend using a special, non-physiological ACSF during dissection (Dissection ACSF) that contains 10 mM magnesium and 0.5 mM calcium. Both these departures from physiological CSF are known to reduce excitotoxicity (Feig and Lipton, 1990;Sacktor et al., 1993; Wang and Kass, 1997; Ting et al., 2018).

mp6 micropump noise
There are two sources of noise in the system: (a) noise from the electrical signal of the OEM, and (b) noise generated in the solution because of the vibration of the ionic ACSF by the pump actuators. These sources of noise can be completely silenced by (a) making a metal enclosure for the circuit and mini-Faraday cages for the pump "sandwich" arrays, and (b) grounding the recirculating ACSF both at the inflow to the bath and at the suction, using T-shape tubing connectors with a silver chloride wire connected to the main ground of the rig.

Passive check valves mp-cv
To eliminate potential back flow of the ACSF owing to differential pressure between the pump inlet and the outlet (due to differences, for example, between the height of the recording chamber and the height of the ACSF reservoir, or the connecting tubes), use a passive check valve encased in stainless steel manufactured by Bartels Mikrotechnik (mp-cv). The valve, which will influence the volume flow of the micropump, should be placed between the micropump and the outflow reservoir. Figure 15C shows four such mp-cv check valves connected to the array of four mp6 micropumps controlling the suction of the interface recording chamber.

5.
Metal enclosure for the piezoelectric pump system For safety reasons (to protect against electric shock from the OEMs), as well as to eliminate the noise produced by the OEMs and to protect the circuit from damage, the completed circuit of Figure 14 should be placed inside a metal enclosure that is grounded. The enclosure should have an opening for the ribbon cable of the Breakout Board connecting the micropump assemblies, as well as openings for the power supply cables, the two USB cables of the Arduinos, the master switch of the circuit, and the BNC cable connecting the top Arduino to the ADC board.

6.
Hybrid persistaltic/piezoelectric pump arrangement A simplified alternative to the recirculation device described here is to use piezoelectric pumps only for the aspirating the recirculation ACSF, and a peristaltic pump to provide the inflow to the chamber. In this arrangement, a peristaltic pump with a two-channel pump head would control the inflow (alternatively, a four-channel pump head can be used, if the interface chamber supports two independent baths for slices). One piezoelectric array (or two, depending on the number of baths supported by the interface chamber) would control the aspirator, with each array consisting of four mp6 micropumps driven by the suction part of the circuit (i.e., the top four OEMs) at the maximum flow rate.

7.
Pre-flushing the dead volume of the perfusion system before switching solutions When 95% O 2 −5% CO 2 -aerated ACSF stays in polyethylene or Tygon tubing for a long time before being perfused onto slices, it degases and becomes hypoxic. The loss of carbon dioxide also alters the pH of the standard bicarbonate buffer of the ACSF. It is therefore important to "pre-flush" the dead volume in the perfusion system before switching solutions.
In the experiments described in Tsokas et al. (2016), two manual three-way valves in series allowed pre-flushing the dead volume in the tubing of the inflow mp6 micropump assemblies by switching the upstream three-way valve from the idle peristaltic pump (OFF) to the idle mp6 micropump (ON), and switching the downstream three-way valve so that the outflow from the micropump flows directly into waste (ON), rather than into the slice chamber (OFF). In this configuration, turning the micropump on for 10 seconds before switching it back off allows the dead volume to clear. Following this procedure, first turn the downstream valve's outlet to the waste to OFF, before turning the mp6 micropump back on to resume operation of the perfusion system. WinLTP 2.30 has four perfusion line controls (Slow0, Slow1, Fast0, and Fast1) that can be used for controlling piezo steppers and pinch valves for automatic changing of bath perfusion. For an excellent discussion of how to use these features with commercially available pinch valve systems for automated preflush perfusion of slices, see Chapter 10 of the WinLTP 2.30 manual (Anderson, 2018).

8.
The mp6 micropumps are sensitive and will be severely damaged if liquid is forced into them by applying pressure with a syringe. If the micropumps fail to draw liquid when initially turned on (usually due to incomplete purging of their contents during previous use), it is best to connect the inlet of the mp6 to proper tubing attached to the tip of a vertically positioned 20 ml syringe barrel (i.e., without the plunger, and with its tip pointed towards the ground). Turn the micropump on, pour a few mL of water into the barrel of the syringe, and allow gravity to assist the micropump's vibrating piezoelectric diaphragms in clearing the chambers of the pump.

1.
10× Stock Dissection ACSF Buffer (store at 4 °C) (   A. The cables connecting the different components are color-coded for easy assembly of the circuit; a step-by-step outline is provided in Figures 4-14  A and B. Two mp6 micropumps are stacked in parallel connection, which causes their volume flows to summate. Each such "sandwich" array consists of three custom-made plexiglass panels, one of which is threaded so that screws can hold the assembly together. Cork pads cut to size allow for precise contact of different surfaces. C. Two assemblies shown in (A) and (B) stacked on top of each other, constitute the pump array that controls the suction. Notice the y connectors and mp-cv valves. D. The 26-pin Assembled Pi Cobbler Breakout Board is connected to 26-pin shield stacking headers with longer pins, so as to allow enough clearance for two of the mp6-OEM controllers that are attached directly adjacent to it. E. Cable that connects six micropumps to the Pi Cobbler Breakout Board. F. Mini-Faraday cages made of copper wire mesh to allow grounding the noise from the pumps.

Author Manuscript
Author Manuscript A. The slice lies on a mesh, near the wall of a recording bath. The chamber has two such recording baths, both of which are covered by a removable convex canopy that creates a humidified, oxygen-rich atmosphere above the slice. The canopy has a triangular opening that allows recording from each slice individually (while the slice in the other bath remains covered). Each slice is constantly superfused with ACSF flowing into the bath through an inflow tube and out through an outflow conduit (across from the inflow), which leads to the suction well (the suction is provided by a yellow syringe needle). B and C. At interface, a small meniscus is formed around the slice. A larger meniscus is formed around the walls of the bath.  The program allows turning all six of the mp6 micropumps on simultaneously when prompted by one of the digital outputs of the ADC, which is in turn controlled by the "Slow0 Perfusion Change" function of the winLTP program. By changing the "delay" value (highlighted in yellow) the slices can be perfused for variable intervals. In this particular example the pumps will remain on for 1800000 ms (i.e., 30 min).  By inserting different values for "int32_t clockfrequency" (x-axis of the graph), the flow rate (y-axis) can be changed from the low rates of a few hundred microliters per min (shown in the insert) required for interface chamber recordings, to a maximal flow rate of 5.5 ml/min.  A. ZIP (5 μM) applied during the maintenance phase of late-LTP reverses potentiation without effect on the untetanized pathway (from Tsokas et al., 2016; n = 5). B. In contrast, in slices (n = 3) that have recovered in an interface chamber (for equal time as the slices in A), which was then converted into a submersion chamber by increasing the bath level and flow rate of the ACSF, 5 μM ZIP not only reverses LTP in the tetanized pathway but also decreases basal synaptic transmission in the untetanized pathway. This difference may be due to the superior oxygenation conditions in interface chambers (oxygen saturation is 95.1 ± 2.3% in an Oslo-type interface chamber, compared to 72.4 ± 4.6% in a submersion chamber with a typical flow rate of 5 ml/min).