Experimental protocols utilizing animals were approved by the Stony Brook Institutional Animal Care and Use committee. F344 adult male (wild-type) were acquired from Charles River Lab, and HIV-1 Tg rats were acquired from the University of Maryland, the creator of this neuroHIV rat model (F344 background). The rats were single-housed in a 12-h light/dark cycle and were given ad libitum access to food and water. To avoid possible interference of sex hormones on cerebral vasculature and neuronal excitability, we choose to only study male rats, but future studies will evaluate female animals. Animals were divided into experimental groups, as outlined in Table 1.

The genetically encoded calcium indicator GCaMP6f was delivered into the PFC of rats to study neuronal activity. Briefly, rats were anesthetized with isoflurane and mounted onto a stereotaxic frame. An incision was made along the scalp, and a hole was drilled into the skull over the PFC area (A/P: +3 mm; M/L:+0.8 mm). A total of 1 μl of AAV5. Syn.gCaMP6f.WPRE.SV40 (Addgene) was infused at two depths (D/V: −1.4mm; D/V: −1.0 mm) at a rate of 0.2 μl/min. Following the infusion of 0.5 μl, a 20-min break occurred to allow for viral diffusion through the tissue. After the surgery, rats were returned to their home cage and observed until awake. Flunixin was given immediately post-operatively and every 12 h thereafter for 3 days. Post-operative recovery was evaluated based on the level of activity as well as food and water intake. Imaging of PFC took place 3–4 weeks after viral delivery. Ex vivo imaging was conducted at the end of experiments to confirm the location of GCaMP6f expression (Figure 1B₀ ).

Three to four weeks after viral infusion, rats were anesthetized with isoflurane and placed on a ventilator via intubation with a 16-gauge IV catheter into the trachea. Venous and arterial accesses were obtained through the insertion of catheters into the left lower limb femoral artery and vein (0.58 mm ID, 0.99 mm OD, Scientific Commodities Inc.). The arterial line was used to monitor blood pressure, whereas the venous line was used for drug delivery. The rat was then placed in a stereotaxic frame (Kopf 900), and a 4 × 6 mm² craniotomy was performed above the PFC (A/P: 1–5 mm, M/l: −3 to 3 mm). The dura was carefully removed, and 1.25% agarose gel was placed onto the brain surface. A glass coverslip was placed over the agarose gel and secured into place using superglue. During experimentation, the animal’s physiology was recorded including the mean arterial pressure (mean = 82.28, SE = 1.88), body temperature (∼37–38°C), and respiration (∼40–45 breaths per minute) (small Animal Monitoring and Gating System, model 1025L, SA Instruments Inc.).

The NIDA drug supply program provided the cocaine–HCL used in this experiment. Cocaine solutions were freshly made for each experiment by diluting the drug in saline. Memantine (MEM) was purchased from Sigma-Aldrich (St. Louis, MO) and was dissolved in saline (10 mg/ml). For vehicle, saline was utilized.

Using a custom-built multi-modality imaging platform (MIP) utilizing previously described methods by our group (Allen et al., 2019; Yuan et al., 2011; Chen et al., 2016), in vivo images of the PFC were acquired. As shown in Figure 1, the simultaneous acquisition of [Ca²⁺]_i, HbR, and HbO₂ was achieved by the time-sharing approach (Yuan et al., 2011). The light source cycled three LEDs at three different wavelengths (λ₁ = 568 nm, λ₂ = 630 nm, and λ_excitation = 488 nm; Spectra Light Engine, Lumicor; Figure 1A₀ ) at a rate of 1 Hz to illuminate the brain tissue through the cranial window implanted over the PFC (Figure 1A₁ ). Images were acquired by a cMOS camera (pixel size: 6.5 μm; Zyla 4.3, Andor) using modified Solis software (version 4.26, Andor). Figure 1B₀ demonstrates the representative multi-wavelength images obtained from the PFC of a rat in vivo. Total hemoglobin (HbT) was calculated based on the changes in HbR and HbO₂ (Eq. 1).

To determine the effects of memantine on baseline Ca²⁺ and hemodynamics, both HIV-1 Tg and control F344 were given 10 mg/kg s.c. of memantine (Table 1: experiment 1). These animals were first imaged for 5–10 min to acquire a baseline followed by SC injection of memantine (MEM) with an additional 20 min of recording.

Animals in experiments 2 and 3 underwent two sets of imaging with at least 90 min between each cocaine infusion (set a: VEH pretreatment followed by cocaine challenge; set b: memantine (MEM) pretreatment followed by cocaine challenge). Rats were first (in set a) pre-treated with VEH (saline, 1 ml/kg) followed by a 10-min baseline acquisition. Cocaine was infused (1 mg/kg i.v.) over 1 min followed by an additional 60 min of imaging. For the second set of images (in set b), rats were pre-treated with 10 mg/kg s.c. memantine. A baseline was acquired over 10 min followed by the administration of acute cocaine challenge over 1 min (1 mg/kg i.v.) with an additional 60-min imaging.

To quantify changes in Ca²⁺ and hemodynamics in response to cocaine challenge, we selected five regions of interest (ROI) for each animal (as illustrated in Figure 3: yellow circles). From full-field images, we selected the hemisphere in which GCaMP6f was injected and expressed (Supplementary Figures S2, S3). ROIs were then positioned in regions along the periphery of the GCaMP6f expression and were placed to avoid vasculature, as described in our previous publications (Allen et al., 2019; Du et al., 2020; Du et al., 2021; Park et al., 2022). This ROI positioning criterion was used to minimize the risk of GCaMP6f overexposure following cocaine administration as well as to prevent confounding effects of hemodynamic changes on the measured [Ca²⁺]_i fluorescent signal. The same ROIs are used for all three image channels to ensure that the quantification of [Ca²⁺]_i, HbT, and HbO₂ was conducted in the same location of the cortex.

Hemodynamic changes (i.e., Δ[HbR] and Δ[HbO₂]) were calculated using a modified Beer–Lambert law, as outlined in Eq. 1: [ Δ H b O 2 Δ H b R ] = [ ε H b O 2 λ 1 ε H b R λ 1 ε H b O 2 λ 2 ε H b R λ 2 ] − 1 [ ln ( R λ 1 ( 0 ) / R λ 1 ( t ) ) / L λ 1 ( t ) ln ( R λ 2 ( 0 ) / R λ 2 ( t ) ) / L λ 2 ( t ) ] (1)

Through the MIP, images from illumination with λ₁ and λ₂ were captured and the reflectance matrices were determined (R_λ1(t) and R_λ2 (t)). Molar excitation coefficients for HbO₂ and HbR at the given wavelengths are represented by ε^λ1 _HbO2, ε^λ1 _HbR, ε^λ2 _HbO2, and ε^λ2 _HbR and L_λ1 (t) and L_λ2 (t) are the path lengths of light propagation. When the sum of Δ[HbR] and Δ[HbO₂] is calculated, the changes in total hemoglobin (Δ[HbT]) can be determined and used as a correlate for total blood volume in the cerebral cortex (Yuan et al., 2011).

Changes in intracellular Ca²⁺ levels were calculated based on the relative change in fluorescent intensity when compared to those in baseline (Δ[Ca²⁺]_i). Data are represented as the percent change relative to the baseline. To correct for possible absorption in emitted light due to changes in hemodynamics, the fluorescent intensity in the ROIs was normalized to the intensity from areas of non-fluorescence.

Mean and standard error of the mean were calculated using Microsoft Excel. Statistics were carried out using SigmaStat software (Systat Software INC.). Two-way ANOVAs were used to compare the difference between groups at each time point. Either one-way or two-way ANOVAs with a Bonferroni post hoc test were used for comparison of curve parameters (i.e., peak). p<0.05 was considered statistically significant. Raw data and p-values are summarized in Supplementary Tables S1–S4.

To determine whether memantine would affect neuronal [Ca²⁺]_i and hemodynamics (e.g., Δ[HbT], Δ[HbO₂]) in the PFC, we administered memantine (10 mg/kg, s.c.) into control (F344, n = 3) and HIV-1 Tg (n = 4) animals. Neuronal Δ[Ca²⁺]_i fluorescence, tissue oxygen hemoglobin (Δ[HbO₂]), and total blood volume (Δ[HbT]) in PFC were detected for 15–20 min (Supplementary Figure S1). Figure 2 summarizes the changes in these three parameters for control and HIV- 1 Tg rats. Figure 2A shows that MEM induced a slight increase in the mean neuronal Δ[Ca²⁺]_i to 1.40 ± 0.23 (%/minute, F (1,4) = 29.391, p = 0.006) and 0.81 ± 0.16 (%/minute, F (1,6) = 19.126, p = 0.005) in PFCs of control and HIV-1 Tg rats, respectively. These changes did not differ between CTL and HIV-1 Tg animals (F (1,5) = 4.756, p = 0.081). The mean blood volumes tended to decrease, but these effects were not significant neither in HIV-1 Tg animals (∆[HbT] = -0.062 ± 0.018 (%/min), F (1,6) = 1.28, p = 0.061) nor in controls (∆[HbT] = -0.64 ± 0.33 (%/min), F (1,4) = 3.144, p = 0.15). Oxygenated hemoglobin did not change in control (∆[HbO₂] = −1.15 ± 0.37 (%/min), F (1,4) = 7.608 p = 0.051) or HIV-1 Tg (∆[HbO₂] = -2.87 ± 1.52 (%/min), F (1,6) = 3,648, p = 0.105) animals. None of the responses to memantine differed significantly between the two groups.

Panel B₀ of Figure 1 shows a representative image of GCaMP6f neuronal expression in the left PFC, where the virus was infused. Fluorescence was primarily expressed in the prelimbic cortex (PrL), which is implicated in cocaine intake (Chen et al., 2013), and reinstatement (Kalivas and McFarland, 2003). To compare cocaine’s effect on PFC with and without NMDA blockade by memantine in the same animal, we administered cocaine twice, first following the vehicle and then after MEM pretreatment (expt 2, Table 1). To minimize the effects of remaining cocaine from the first injection on the response of the second cocaine injection, we waited 2 h between infusions (Du et al., 2021).

Figure 3A _Veh and 3A_MEM include representative images of the GCaMP6f neuronal expression before and after MEM infusion, respectively. Figure 3A _veh (t = 0 min) and 3A_veh (t = 20 min) show the ratio images of GCaMP6f [Ca²⁺]_in fluorescence at baseline and 20 min after cocaine without MEM pretreatment, whereas Figure 3A _MEM(t = 0 min) and 3A_MEM(t = 20 min) are ratio images of Ca²⁺ fluorescence at baseline and 20 min after cocaine with MEM pretreatment. These images demonstrate that neuronal [Ca²⁺]_i fluorescence increases after cocaine with either vehicle or MEM pretreatment. Figure 3B shows the mean time courses of cocaine-induced neuronal [Ca²⁺]_i changes in the PFC (n = 5) with VEH or MEM-pretreatment, indicating that cocaine triggered an influx in neuronal [Ca²⁺]_i in the PFC with different recovery profiles when given after vehicle (cocaine only) or after cocaine with MEM pretreatment (10 mg/kg, s.c.). It illustrates a long-lasting [Ca²⁺]_i increase in response to cocaine in VEH animals (i.e., >60 min after cocaine challenge). However, with MEM pretreatment, cocaine-induced Δ[Ca²⁺]_i increase recovered at 45–60 min. A two-way ANOVA showed no significant difference between these two-time courses from baseline to 40 min after cocaine (F (13, 84) = 0.706, p = 0.753). However, a significant difference was observed from 40 min until the end of recording at 60 min post cocaine (p = 0.04). The average rate of Ca²⁺ influx did not differ following memantine pretreatment (F (1, 6) = 1.268, p = 0.303; Figure 3C). The peak neuronal ∆[Ca²⁺]_i fluorescence increase induced by cocaine corresponded to 4.46 ± 0.92% for vehicle and did not differ from 4.41 ± 0.78% induced with MEM pretreatment (F (1, 6) = 0.00188, p = 0.967) (Figure 3D).

Acute cocaine reduced cerebral blood volume as measured by changes in ∆[HbT] (Figure 4). Figure 4A displays grayscale images with representative images of PFC in a drug-naïve animal at baseline and after MEM pretreatment. Ratio images of PFC before (t = 0 min) and after (t = 20 min) cocaine injection in VEH (upper panels) or MEM pretreatment (lower panel) indicate a decrease in ∆[HbT] in PFC after cocaine (1 mg/kg, i.v.). Comparison between CTL-Coc and MEM-Coc with a two-way ANOVA showed no significant difference within 5 min after cocaine (p>0.05) followed by a significant difference from t = 10–60 min post cocaine (F(13, 84) = 5.47, p<0.001). Cocaine’s reduction of ∆[HbT] was persistent (i.e., >60 min) after vehicle but returned to baseline after 58.83 ± 14.83 min in the MEM group (Figure 4B). We also compared that cocaine-induced changes in ∆[HbT] integrated over 60 min, which showed significant differences between the vehicle and MEM corresponding to −7.0 ± 0.78% and −1.87 ± 0.31%, respectively (F(1, 6) = 23.075, p = 0.003). Figure 4D illustrates the mean peak changes in ∆[HbT] following cocaine, showing no differences between the vehicle and MEM pretreatment (F(1, 6) = 3.887, p = 0.096).

Cocaine decreased ∆[HbO₂] in the PFC after vehicle with MEM pretreatment, attenuating the amplitude and duration of the cocaine-induced decrease (Figure 5). Specifically, for VEH, cocaine induced a persistent ∆[HbO₂] decrease, whereas with MEM pretreatment, the decrease from cocaine returned to baseline at 38 min (t = 1–38 min; p < 0.001, t = 38 min; p = 0.909) (Figure 5B). Panels A_VEH (upper panels in Figure 5A) and A_MEM (lower panels in Figure 5A) show representative images of the surface of the PFC with ROI selections and ratio images to illustrate the percent change in [HbO₂] concentration at t = 0 min and t = 20 min following cocaine infusion (vehicle and MEM pretreatment, respectively). A two-way ANOVA of Δ[HbO₂] of VEH and MEM showed a significant interaction between pretreatment and time (F(13, 84) = 4.493, p<0.001). Point-by-point comparison between these two-time courses showed no differences within 5 min after cocaine between VEH and MEM pretreatment (p>0.05) followed by significant differences between them from t = 10–60 min (p < 0.05; Figure 5B). Cocaine-induced mean ∆[HbO₂] change showed a −6.12 ± 0.76 (%/min) reduction with vehicle that was significantly attenuated to −1.23 ± 0.49 (%/min) with MEM (F(1, 6) = 21.023, p = 0.004; Figure 5C). The peak cocaine-induced decrease in ∆[HbO₂] did not differ significantly between vehicle and MEM (−7.63 ± 0.84% with vehicle and −5.95 ± 0.64% with MEM, (F(1, 6) = 1.937, p = 0.213; Figure 5D).

Cocaine triggered significant increases in neuronal Δ[Ca²⁺]_i in PFC of HIV-1 Tg animals that differed when given after vehicle or after MEM pretreatment (Figure 6). Figure 6A _VEH-HIV and Figure 6A _MEM-HIV show the representative images of GCaMP6f neuronal expression in the PFC of a HIV-1 Tg rat with and without MEM. Figure 6A _VEH (t = 0 min) and A_VEH (t = 20 min) show the ratio images before and 20 min after cocaine (1 mg/kg, i.v.) for vehicle pretreatment and Figure 6A _MEM-HIV(t = 0 min) and A_MEM-HIV(t = 20 min) for MEM pretreatment. A two-way ANOVA showed a significant time effect ([F(13, 84) = 1.701, p = 0.072]). Comparing the responses to cocaine at each time point showed no significant difference in Δ[Ca²⁺]_i increases from baseline to 30 min (p>0.05) between VEH and MEM pretreatment. This is followed by a significant difference after 35 min (p_35-60<0.05; Figure 6B). The cocaine-induced mean Δ[Ca²⁺]_i increase was 7.96 ± 1.66 (%/min) with vehicle, which was significantly reduced to 3.01 ± 0.57 (%/min) with MEM (F(1, 7) = 6.451, p = 0.039; Figure 6C). Also, the peaks of ∆[Ca²⁺]_i in responses to cocaine differed significantly, corresponding to 6.71 ± 0.86% and 3.32 ± 0.63% for vehicle and MEM, respectively (F(1, 7) = 9.102, p = 0.019) (Figure 6D).

Cocaine decreased Δ[HbT] in the PFC of HIV-1 Tg animals with both vehicle and MEM pretreatments (Figure 7A). A two-way ANOVA for the comparison between CTL-Coc and MEM-Coc showed a significant interaction of pretreatment and time ([F(13, 84) = 1.236, p = 0.27]). Cocaine’s reduction of ∆[HbT] was persistent with vehicle but recovered at 58.78 ± 26.91 min with MEM pretreatment (Figure 7B). The mean ∆[HbT] was −15.49 ± 3.2 (%/min) and −4.04 ± 3.01 (%/min) for vehicle and MEM pretreatment, respectively (F(1, 7) = 6.418, p = 0.039). There was no significant difference in the cocaine-induced peak of ∆[HbT] between vehicle (−14.83 ± 2.41%) and MEM (−6.95 ± 2.3%) (F(1, 7) = 4.306, p = 0.077; Figure 7D).

Cocaine reduced Δ[HbO₂] in the PFC of HIV-1 Tg animals with vehicle or MEM pretreatment (Figure 8). Similar to memantine’s attenuation of the reduction of cerebral blood volume (∆[HbT]; Figure 7), oxygenated hemoglobin in PFC was also reduced with MEM pretreatment (Figure 8). With vehicle, cocaine decreased Δ[HbO₂] to −33.97 ± 2.3% at t = 4 min and with memantine to −23.48 ± 4.0% (F(1, 7) = 5.714, p = 0.048; Figure 8D). Additionally, cocaine induced a persistent decrease in ∆[HbO₂] over 60 min with vehicle, whereas with MEM pretreatment, the decrease recovered at 34.93 ± 12 min followed by an overshoot (Figure 8B). The mean changes in cocaine-induced ∆[HbO₂] were significantly reduced from −18.7 ± 1.6 (%/min) with vehicle to −1.98 ± 5.3 (%/min) with MEM pretreatment (F(1, 7) = 11.066, p = 0.013; Figure 8C).

Figure 9 summarizes the comparison of cocaine’s effects on neuronal [Ca²⁺]_i and hemodynamic changes in PFC between control and HIV-1 Tg animals with vehicle and memantine pretreatment. As shown in Figure 9A, the time course of ∆[Ca²⁺]_i in the PFC of saline pretreated HIV-1 Tg animals showed a persistent increase over 60 min, whereas in controls, it reached a peak at ∼10 min after cocaine followed by a slow recovery. HIV-1 Tg rats pretreated with memantine displayed a similar neuronal Δ[Ca²⁺]_i response to cocaine as control rats (from t = −5 min to t = 45 min, p_{-5 to 45}>0.05, green shaded area) (Figure 9D). However, the ∆[Ca²⁺]_i fluorescence in controls returned to baseline at t = 47.88 ± 10.53 min but not in HIV-1 Tg animals. A two-way ANOVA examining the interaction of drug (pretreatment with saline v memantine) and animal (control v HIV-1 Tg) for ∆[Ca²⁺]_i found no significant interaction between memantine treatment and animals (F(16,1) = 2.409, p = 0.145). The significant main effects were found for both memantine (p = 0.02) and animal (p = 0.042) on ∆[Ca²⁺]_i. Post hoc analysis revealed that the mean neuronal ∆[Ca²⁺]_i was significantly higher in HIV-1 Tg rats pretreated with saline (∆[Ca²⁺]_i = 7.96 ± 1.6 (%/min), n = 5) than in controls (∆[Ca²⁺]_i = 3.46 ± 0.8 (%/min), n = 5, p = 0.01; Figure 9G). The mean rate of accumulated ∆[Ca²⁺]_i did not differ between controls (∆[Ca²⁺]_i = 2.18 ± 0.6 (%/min)) and HIV-1 Tg rats (3.01 ± 0.57 (%/min), p = 0.654). Pretreatment of HIV-1 Tg rats with memantine significantly reduced the accumulated Δ[Ca²⁺]_i following cocaine exposure (p = 0.008; Figure 9G).

In addition, the total cerebral volume in the PFC of HIV-1 Tg animals pretreated with saline decreased to −15% within t = 5 min after cocaine and remained reduced over the time of recording (60 min; Figure 9B); whereas in controls, Δ[HbT] decreased to −5% at t = 5 min and also remained reduced until the end of recording (Figure 9B). The time courses of hemodynamic response to cocaine for Δ[HbT] for memantine pretreated control and HIV-1 Tg rats showed no significant differences (Figure 9E). A two-way ANOVA of ∆[HbT] found only a simple main effect of drug pretreatment of the total cerebral blood volume (p = 0.006). The mean amplitude of Δ[HbT] was 2-fold larger in saline pretreated HIV-1 Tg PFC (−15.49 ± 3.2 (%/min)) than in controls (−7.0 ± 0.78 (%/min), p = 0.02). Cocaine-induced mean Δ[HbT] reductions in control and HIV-1 Tg animals were −1.87 ± 0.3 (%/min) and −4.04 ± 3.01 (%/min), respectively, and were not significantly different (p = 0.584; Figure 9H). Similar to ∆[Ca²⁺]_i, a reduction in the total blood volume was attenuated in HIV-1 Tg rats pretreated with memantine when compared to saline pretreatment (p = 0.005).

HIV-1 Tg rats pretreated with saline had a rapid decrease in oxygenated hemoglobin followed by a slight recovery from −30% to −20% that remained over 60 min (Figure 9C). Cocaine-associated reduction in Δ[HbO₂] that peaked at t = 4 min (Figure 9F) was larger in HIV-1 Tg rats (−23.48 ± 4.02 (%)) than in controls (−5.95 ± 0.64 (%), p = 0.015), but the returned time to baseline was similar (control: 38.33 ± 1.67 min and HIV-1Tg: 34.93 ± 12 min, p = 0.821; Figure 9F). A two-way ANOVA of ∆[HbO₂] found no significant interaction between pretreatment and animal (F16,1) = 4.556, p = 0.052). Only the simple main effect of pretreatment was found to be significant (p = 0.002). The mean reduction in Δ[HbO₂] was greater for HIV-1 Tg rats (−18.7 ± 1.6 (%/min)) than for controls (−6.12 ± 0.76 (%/min), p = 0.003; Figure 9I). The mean reduction in Δ[HbO₂] following memantine pretreatment did not differ between the two groups (p = 0.863; Figure 9I). Memantine pretreatment of HIV-1Tg rats significantly attenuated Δ[HbO₂] reduction (p<0.001).