Transcranial photoacoustic imaging of NMDA-evoked focal circuit dynamics in the rat hippocampus

Transcranial fPA imaging of neural circuit dynamics was performed during NMDA infusion into the rat hippocampus (figure 2). The representative fPA sagittal planes obtained at 790 nm presented a sufficient sensitivity at the depths of interest, i.e. 3.6 mm, at a contralateral position to the microdialysis probe (figure 3(a)). The sagittal PA imaging plane of the hippocampus revealed the cross-sections of the transverse sinus and inferior cerebral vein in the superficial depth range (figure 3(b)). From the temporal dimension, time-averaged VSD responses were reconstructed pixel-by-pixel for the pre-injection phase (−10 to 0 min) followed by the results for time bins in 10 min intervals: 0–10 min, 10–20 min and 20–30 min. The total recording duration (40 min) was limited by the internal memory of the fPA neuroimaging system. Figure 3(c) shows the representative VSD responses in the rat hippocampus collected during 0.3 mm and 3.0 mm NMDA infusion. Note that the fPA VSD responses at the regions-of-interest (ROIs blue circles in figure 3(c)) represent the NMDA-evoked electrophysiological activity within the DG circuit in the rat hippocampus before being filtered by the DG gating function. As a result, the VSD responses averaged in each NMDA acquisition phase (−10 to 0 min, 0–20 min, 20–30 min) demonstrated a positive correlation to the range of extracellular glutamate concentration with distinct doses of focal NMDA infusion, i.e. 0.3 mm (n = 6) and 3 mm (n = 3) (figure 5). In addition, in the 3.0 mm NMDA infusion group, the averaged VSD responses were consistently maximized within the first 20 min of NMDA infusion, and their intensities corresponded well to the extracellular glutamate concentration measured in the forward microdialysis (figure S1 (stacks.iop.org/JNE/17/025001/mmedia)). Further comprehensive and quantitative analysis will be given in section 2.4.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Intrahippocampal infusion of NMDA diluted into artificial cerebrospinal fluid at the required concentrations was delivered through the microdialysis probe. Concentrations of glutamate release in the dialysate were calculated from the values of three baseline samples collected at 20 min intervals before NMDA infusion. During infusion into the hippocampus (bregma −3.8 mm, lateral 3.0 mm, depth 3.6 mm; figure 4(a)), samples continued to be collected every 20 min for the remainder of the study. Glutamate release was calculated as a percentage of basal values. Figure 4(b) shows the effect of NMDA infusion on glutamate release in the hippocampus of anesthetized rats. The results demonstrate the dose-related response of NMDA infusion at 0.3 (n = 3), 1.0 (n = 1) and 3.0 mm (n = 6) into the hippocampus, where 0.3 mm NMDA caused a 17.93 ± 19.05% increase in glutamate levels that remained elevated for the duration of the infusion; the 1.0 mm NMDA infusion caused doubling of the glutamate levels in the hippocampus compared to baseline (i.e. 112.42%). The increase observed at the 1.0 mm dose remained elevated to the end of the NMDA infusion. At the 3.0 mm dose, the NMDA infusion raised the glutamate level up to 392.99 ± 250.54% above the baseline level, which peaked during the second 20 min sampling period.

Zoom In Zoom Out Reset image size

Hippocampal infusion of 3.0 mm NMDA induced a significant increase in extracellular glutamate concentration, which correlated with fPA VSD responses, suggesting a potential break in the DG gate and synchronized DG firing that may lead to acute seizure activity. We therefore utilized qEEG during the same hippocampal 3.0 mm NMDA infusion protocol to identify if the maximal VSD response and extracellular glutamate concentrations were associated with significant electrophysiological activity resulting from the breakdown of the DG gate and synchronous DG firing resulting in seizure onset. The DG is a hippocampal region specifically subjected to a barrage of excitatory inputs. However, the majority of excitatory activity does not propagate through the DG and into the hippocampus as the DG performs a gating function. Otherwise, excessive activation of the DG disrupts its gating function and induces acute seizures in naïve animals [19].

In patients with focal epilepsy, gamma and theta activity from scalp EEG are indicators of seizure onset zone and ictal onset [20, 21]. Here, the temporal specificity of qEEG recorded circuit responses in real-time, before and after focal NMDA infusion. The 3.0 mm NMDA infusion into the hippocampal circuit induced focal seizure activity recorded by qEEG (figure 4(c), black asterisk, and figure S2). When NMDA was delivered to the hippocampal circuit by the microdialysis probe (figure 4(a)) [22], a significant change in qEEG spectral power was detected, particularly in the gamma range (figure 4(c)). In the theta range, spectral power increased during the seizure (figure 4(d)), further supporting the identification of focal seizure activity. The seizure trace measured by EEG demonstrated both a gamma and theta component (figure 4(f)). The dose-related VSD and extracellular glutamate concentration response, along with the qEEG findings, suggest that the 3.0 mm NMDA infusion resulted in a break in the DG gate associated with focal seizure activity that can propagate to other cortical regions.

However, qEEG lacks spatial resolution, and recordings from deep brain structures cannot be isolated without implantation of invasive depth electrodes. Therefore, in order to maintain the tri-modal experimental paradigm, we opted not to place depth electrodes. We made the stimulation focal, according to the previous proof-of-concept results using fPA for detection of the onset of generalized seizures through an intact skull and scalp [18].

The fPA VSD responses at the contralateral side to the microdialysis probe (blue circlesin figure 3(c)) were measured and plotted as a function of the corresponding changes in extracellular glutamate concentration (figure 5). Note that the ROIs to quantify the VSD responses were in 1 × 1 mm² size in the sagittal cross-sections of rat hippocampus. The reference phase for VSD response quantification was obtained from the baseline phase, 5–10 min. In fPA imaging, 3.0 mm NMDA infusion into the hippocampus yielded significant elevation of VSD response: 0.08 ± 0.35, 2.24 ± 1.06 and 1.17 ± 1.56 for the baseline (−10 to 0 min), NMDA1 (0–20 min) and NMDA2 (20–30 min) phases, respectively (figure S1(a)). The increase in glutamate concentration was correspondingly elevated from −4.67 ± 4.48%, 369.75 ± 254.91%, to 392.99 ± 250.54%. In addition, when re-analyzing the data for −50% to 100%, 100%–500% and 500%–1000% bins of fractional glutamate concentration change, a high positive correlation was obtained with VSD responses: 0.38 ± 0.55 (n_sample = 9), 0.97 ± 1.40 (n_sample = 5) and 3.17 ± 0.32 (n_sample = 4) for 11.04 ± 29.60%, 279.85 ± 141.64% and 666.49 ± 73.63% of glutamate concentration changes (red lines in figure 5). The goodness of fit (R²) among the mean values was 0.95 with the slope and y-intercept at 0.12 and 0.00, respectively. The qEEG revealed increased bursts of gamma power during 3.0 mm NMDA infusion, reaching up to a ∼261% increase with a focal seizure, which indicates a breakdown of the DG gate with the glutamatergic excitation induced (figures 4(c) and S1). In contrast, focal 0.3 mm NMDA infusion into the hippocampus did not produce any statistically significant increase in circuit activity. The VSD responses were −0.11 ± 0.02, −0.15 ± 1.71 and −0.07 ± 0.64 in the baseline, NMDA1 and NMDA2 phases, respectively, while the corresponding glutamate concentration changes in the hippocampus were −1.65 ± 6.90%, 3.90 ± 10.41% and 17.93 ± 19.05% (figure S1(b)). However, even in this case, the VSD responses presented a strong positive correlation with fractional glutamate concentration changes when re-analyzed in −10% to 0%, 0%–10% and 10%–50% bins (blue lines in figure 5): −1.00 ± 1.00, 0.10 ± 0.35 and 0.58 ± 0.48 (n_sample = 3 for each) with −6.84 ± 3.38%, 4.43 ± 4.05% and 22.59 ± 11.64% of fractional glutamate concentration changes, respectively. R² was 0.88 with the slope and y-intercept at 0.05 and −0.46, respectively. Therefore, the fPA VSD neuroimaging revealed a positive correlation with focal glutamatergic excitation in the rat hippocampus.

Zoom In Zoom Out Reset image size

Brains were extracted after all in vivo experimental protocols. The brain was frozen-sectioned into slices of 300 µm thickness and evaluated to confirm probe placement using the scar caused by the microdialysis probe insertion for the NMDA infusion and collections as a marker. Figure 6(a) shows the bright-field images of the coronal plane of the hippocampus at −3.8 mm from the bregma, respectively. White arrows indicate the lesion caused by the microdialysis probe positioned at the following coordinates in the hippocampus: bregma −3.8 mm, lateral 3 mm, and depth 3.6 mm. The VSD staining of the brain tissue was confirmed with near-infrared fluorescence microscopy. Uniform VSD fluorescence was found in the VSD perfusion animal, while the negative control (VSD–) presented negligible fluorescence emission (figure 6(b)), which again confirms the results in our previous publication [23].

Zoom In Zoom Out Reset image size

For theproposed in vivo experiments, 8–9-week-old male Sprague Dawley rats (Charles Rivers Laboratories, Wilmington, MA, USA) weighing 275–390 g were used. The use of animals for the proposed experimental protocol was approved by the Institutional Animal Care and Use Committee of Johns Hopkins Medical Institute (RA16M225). Rats were housed in groups of three per cage with free access to food and water, and maintained on a 12 hr light/12 hr dark cycle.

On the day of the study, rats were weighed and anesthetized with urethane (1.2 mg/kg). Urethane was given incrementally with alternating IP and subcutaneous (SC) dosing. Three (3) ml of isotonic saline was given SC on each side of the body to keep the animal hydrated during the experimental procedure. Body temperature was maintained until the animal was fully anesthetized and ready for surgery. For fPA and qEEG studies, an intravenous (IV) catheter was inserted into a tail vein prior to dye administration during the studies. Once a stable plane of anesthesia was established, hair was shaved from the scalp of each rat to allow acoustic coupling for transcranial fPA recording. The rat was placed into a stereotaxic device (Stoeling Co., Wood Dale, IL, USA). This fixation procedure was required to determine the coordinates of microdialysis probe implantation and for preventing any unpredictable movements during fPA or EEG recording of neural activities. A CMA12 microdialysis probe (Harvard Apparatus, Holliston, MA, USA) was implanted into the CA₃ region of the right hippocampus (stereotaxic coordinates: 3 mm lateral and 3.8 mm posterior to the bregma, and 3.6 mm below the surface of the dura, figure 3(a)) [22]. The probe active exchange surface was 2 × 0.5 mm. The probe was secured to the skull using dental acrylic cement. The fPA and qEEG probes were placed on the contralateral side of the microdialysis probe.

In the present in vivo study, we used the fluorescence quenching-based near-infrared cyanine VSD, IR780 perchlorate (576 409, Sigma-Aldrich Chemicals, St Louis, MO, USA) as used in our previous in vivo study investigating chemo-convulsant seizure activity [18], and it has an analogous chemical structure to PAVSD800-2, our new VSD validated in our previous in vitro study [41]. This VSD yields fluorescence quenching and de-quenching depending on membrane polarization and subsequent changes in the local VSD molecule density, leading to a reciprocal change of PA contrast with non-radiative relaxation of absorbed energy.

We used real-time PA data acquisition to record electrophysiological neural activities in vivo as in our previous study [18]. The ultrasound research system consisted of an ultrasound linear array transducer connected to a real-time data acquisition system (SonixDAQ, Ultrasonix Medical Corp., Canada). To induce the PA signals, pulsed laser light generated by a second-harmonic (532 nm) Nd:YAG laser pumping an optical parametric oscillator system (Phocus Inline, Opotek Inc., USA) provided 690–900 nm of tunable wavelength range and 20 Hz of the maximum pulse repetition frequency. A bifurcated fiber optic bundle, each branch 40 mm long and 0.88 mm wide, was used for laser pulse delivery. The PA probe was situated between the outlets of the bifurcated fiber optic bundles using a customized 3D printed shell for evenly distributing laser energy density in the imaging field-of-view. The alignment of outlets was focused specifically at 20 mm depth. The PA probe was positioned in the sagittal plane contralateral to the microdialysis probe (3 mm) to cover the hippocampal cross-section. The distance between the PA probe and the rat skin surface was 20 mm filled with acoustic gel, and the resultant energy density was at ~3.5 mJ cm⁻², which is far below the maximum permissible exposure of skin to laser radiation by the ANSI safety standards [48]. A wavelength of 790 nm was used, at which sufficient absorbance can be obtained by the near-infrared VSD, i.e. IR780 perchlorate. Also, excitation at that wavelength prevented the undesired time-variant change of blood oxygen saturation, since the wavelength corresponds to the isosbestic point of the Hb and HbO₂ absorption spectra. Detailed information on neural activity reconstruction using normalized time-frequency analysis can be found in our previous publication [18].

In vivo microdialysis sampling was carried out as previously described [49, 50]. For infusion experiments, NMDA (Sigma-Aldrich Chemicals) was weighed, solubilized and diluted to the desired concentration in artificial cerebrospinal fluid (NaCl, 147 mmol l⁻¹; KCl, 2.7 mmol l⁻¹; CaCl₂, 1.2 mmol l⁻¹; MgCl₂, 0.85 mmol l⁻¹) (Harvard Apparatus) on the study day. Once the probe was inserted and secured, it was perfused with artificial cerebrospinal fluid pumped at a flow rate of 2 μl/min. Samples were collected at 20 min intervals, and immediately transferred to a −80 °C freezer until they were assayed. To allow sufficient time for the glutamate levels to equilibrate, three baseline samples were collected an hour after the initiation of infusion. Following collection of these samples, NMDA was infused into the brain directly through the dialysis probe with the same pump parameters as used for the baseline samples. Dialysate samples were assayed for glutamate by a two-step process using HPLC-ECD on an Eicom HTEC-500 system (EICOM, San Diego, CA, USA). After passing the samples through a separation column, they were processed via a column containing immobilized L-glutamate oxidase enzyme, resulting in the release of hydrogen peroxide. The hydrogen peroxide concentration was then determined using a platinum working electrode. Chromatographic data were acquired online and exported to an Envision software system (EICOM) for peak amplification, integration and analysis.

Our method of reverse microdialysis exerts effects that are largely mediated via the DG circuit in the rat hippocampus, since our results agree with those of Ludvig et al [51] who used similar methods to evoke hippocampal seizures. Our microdialysis probes have an active surface of 2.0 × 0.5 mm, and most evidence suggests a 1 mm cylinder of influence around the probe for substances that are reverse dialyzed into the neuropil [52]. It is important to note that the 3 mm concentration of NMDA in the dialysis perfusion solution is not the concentration of NMDA reaching the brain tissue. Perfusion flowrate, probe membrane characteristics and tissue properties serve to limit the penetrance of reverse dialyzed substances into the brain [53]. Ludvig et al addressed this issue previously and showed that the biophase concentration of reverse dialyzed NMDA is about 15% of the source concentration. Thus, the NMDA concentrations reaching the brain tissue in our experiments probably ranged from 0.045–0.45 mm. Another possibility to consider is that infused NMDA is taken up into cells surrounding the probe, thereby keeping biophase concentrations low. Bruhn et al demonstrated that radiolabeled D-aspartate reverse dialyzed into the rat brain is rapidly and effectively extracted from the extracellular fluid by glutamate transport mechanisms [54]. Collectively, this information suggests that the neurochemical, EEG and fPA endpoints that we measured in response to reverse dialyzed NMDA reflect changes localized to the hippocampus.

All EEG recordings utilized a three-electrode paradigm: one recording, one reference (aligned to the site of activation) and one ground over the rostrum. The electrodes (IVES EEG; Model # SWE-L25-IVES, EEG Solutions, MA, USA) were fixed with minimal cyanoacrylate adhesive (KrazyGlue), similar to previous protocols [55]. Data acquisition was performed using Sirenia software (Pinnacle Technologies Inc., Kansas, USA) with synchronous video capture. The data acquisition had a 14 bit resolution, 400 Hz sampling rate, and a band pass filter between 0.5 Hz and 50 Hz. The acquisition files were stored in an .EDF format and scored manually, using real-time annotations from the experiments. EEG power for two second epochs was achieved using an automated fast Fourier transformation module in Sirenia software [56].

The in vivo protocols were designed for simultaneous multi-modal sensing of the neural activity in the hippocampus: microdialysis—qEEG and microdialysis—fPA neuroimaging. Figure 2 shows a detailed schematic protocol for each group representing the response to the administration of NMDA, Lexiscan and VSD (i.e. IR780 perchlorate). fPA and qEEG data acquisition were performed for 40 min to correlate with three microdialysis samples collected at 20 min intervals. Graded NMDA infusion concentrations were applied to identify the dose-dependent glutamatergic excitation of hippocampal circuit: 0.3 mm (n = 3) and 3.0 mm (n = 6). VSD and Lexiscan followed the data acquisition sequence with a 3 min delay, thereby 5 min of baseline phase was guaranteed for the VSD response reconstruction in fPA neuroimaging before starting NMDA infusion. The dosing protocol for Lexiscan and VSD administration was as follows: through an IV tail vein catheter, 150 µl of Lexiscan (0.4 mg/5 ml) was injected, followed by 200 µl of VSD at 2 mg ml⁻¹ concentration, flushed immediately with 150 µl of 0.9% isotonic saline. The EEG signal was recorded separately in an identical preparation procedure as the fPA neuroimaging, including animal preparation and administration of IR780, Lexiscan and experimental duration time for all recordings.

The rats used for the above protocol were euthanized, and the whole brains immediately harvested and placed in 10% formalin. All brains were allowed to fix in fresh 10% formalin for at least 48 h with gentle agitation on a conical rotator. Subsequently, the brains were processed through a series of sucrose gradients (15%, 20%, 30% for 12–24 h each) for cryoprotection. Brains were sectioned frozen with 300 µm thickness. Tissue sections were mounted on slides in ProLong Diamond Anti-face mountant. Slides with sections were imaged using an Olympus OM-D E-M5 Mark II for bright field imaging and using LI-COR Odyssey for fluorescence visualization.