Novel elvitegravir nanoformulation for drug delivery across the blood-brain barrier to achieve HIV-1 suppression in the CNS macrophages

The use of antiretroviral therapy (ART) has remarkably decreased the morbidity associated with HIV-1 infection, however, the prevalence of HIV-1-associated neurocognitive disorders (HAND) is still increasing. The blood-brain barrier (BBB) is the major impediment for penetration of antiretroviral drugs, causing therapeutics to reach only suboptimal level to the brain. Conventional antiretroviral drug regimens are not sufficient to improve the treatment outcomes of HAND. In our recent report, we have developed a poloxamer-PLGA nanoformulation loaded with elvitegravir (EVG), a commonly used antiretroviral drug. The nanoformulated EVG is capable of elevating intracellular drug uptake and simultaneously enhance viral suppression in HIV-1-infected macrophages. In this work, we identified the clinical parameters including stability, biocompatibility, protein corona, cellular internalization pathway of EVG nanoformulation for its potential clinical translation. We further assessed the ability of this EVG nanoformulation to cross the in vitro BBB model and suppress the HIV-1 in macrophage cells. Compared with EVG native drug, our EVG nanoformulation demonstrated an improved BBB model penetration cross the in vitro BBB model and an enhanced HIV-1 suppression in HIV-1-infected human monocyte-derived macrophages after crossing the BBB model without altering the BBB model integrity. Overall, this is an innovative and optimized treatment strategy that has a potential for therapeutic interventions in reducing HAND.


Results formulation design and in vitro release of PLGA-EVG NPs. The PLGA-EVG NP formulation in our
study is an oil-in-water (o/w) emulsion. A hypothetical structure of PLGA-EVG NP is described in Fig. 1A. PLGA was the main polymer core to hold EVG molecules, PVA was used to stabilize the emulsion; poloxamer 188 was used as a stabilizer and to provide a better brain penetration; PLL was used to provide slightly positive charge, which helps internalize into the cells. The in vitro release of PLGA-EVG NPs in cell culture media is shown in Fig. 1B. The release of EVG from PLGA NPs followed a zero-order release profile with r 2 of 0.98 within 24 h. Only ~15% of cumulative EVG release was observed at 24 h, which suggested the potential controlled release profile of PLGA-EVG NPs. The drug release from PLGA NPs may be related to the drug/PLGA loading ratio, NP protected layer stability, and physicochemical properties of EVG.
Size and zeta potential of PLGA NPs. The size of freshly made PLGA NPs was approximately 135.7 ± 1.5 nm in PBS (Table 1). No specific change in particle size and zeta potential of NPs was observed for 7 days when stored at room temperature. PLGA NPs can be stored at 4 °C and −20 °C for at least 30 days with no change on size and zeta potential. PLGA NPs showed a stable particle size in 1 mM HEPES buffer within the pH range of 1-11 (Table 2). PLGA NPs showed a positive charge (~7 mV) at all the recorded pH values except pH 4, 5, and 11. A higher positive charge (~20 mV) of PLGA NPs was observed at the pH range of 4-5, and a lower charge (~3 mV) of PLGA NPs was observed at pH 11. Evaluation of human serum protein binding on PLGA NPs. It is expected that HS protein binding occurs when NPs are taken by human body 24 . Upon introduction of NPs into the body, NPs dynamically interact with all the serum proteins and form the NP-protein complex 25 . Understanding the interaction of NPs with HS can predict their safety and efficacy profiles when used in the clinic 24 . Therefore, a DLS assay of PLGA NPs was investigated in PBS and 1-50% HS. A slight increase in particle size was observed for PLGA NPs in 1-50% HS ( Fig. 2A). HS proteins are loosely bound to the PLGA NPs, leading to this slight increase in particle size. Zeta potential of PLGA NPs was observed to be negative in 1-50% HS (Fig. 2B). This negative charge is expected, because HS is negatively charged in pH 7.4 PBS. The negative zeta potential indicates that PLGA NPs are stable and safe for therapeutic applications. PLGA NPs did not precipitate in the presence of HS. To confirm the human serum binding on PLGA NPs, we performed a FTIR analysis of PLGA NPs, HS, and HS@NPs. The FTIR data was presented from 600 to 1800 cm −1 to observe the HS protein peaks. The peaks at 1661 cm −1 and 1541 cm −1 was found in HS and HS@NPs due to the amide bond stretch of the HS protein. The peak extent of the HS proteins in HS@NPs showed a concentration dependent increase at increasing percentage of HS (Fig. 2C). Moreover, we further confirmed the PLGA NPs-protein interaction using SDS-PAGE analysis, in which, it can be observed the HS proteins binding on HS@NPs. The identified major proteins that bind on NPs are: transferrin, human serum albumin (HSA), fibrinogen beta, fibrinogen gamma, apolipoprotein E (Apo E), and apolipoprotein A1 (Apo A1) according to the molecular weight 26 . We found that HS proteins were adsorbed to the surface of PLGA NPs when the concentration of HS increased (Fig. 2D). With an increased incubation time, an increased protein binding on the PLGA NPs was observed (Fig. 2E). HS protein binding also showed an increased pattern when the amount of PLGA NPs was increased up to 180 µg while HS protein binding decreased when incubated with more than 180 µg of NPs (Fig. 2F). This indicates the saturation of binding with NPs due to the limited HS amount and incubation time.
Hemocompatibility of PLGA NPs. Hemocompatibility tests evaluate the effects on RBCs when introduce a foreign compound to the blood. Both visual inspection (Fig. 3A) and spectrophotometric measurement of hemoglobin release (Fig. 3B) suggested that PLGA NPs do not cause hemolysis even at the highest concentration (200 µM) that was used in the assay. A dose dependent hemolysis was observed in RBCs treated with EVG native drug. At the highest tested concentration (200 µM), EVG native drug caused ~30% hemolysis while PLGA-EVG NPs did not cause hemolysis. Similarly, HS30@EVG NPs also caused low hemolysis (~4%) compared to EVG native drug (~30%). Moreover, EVG native drug caused morphology change on RBCs, while PLGA-EVG NPs and HS30@EVG NPs caused no change on the RBC morphology (Fig. 3C). The hemocompatibility of PLGA NPs could be because of the biocompatibility of the polymer and the HS adsorption on the PLGA NPs.

Toxicity evaluation of PLGA-EVG NPs in monocytes-derived macrophages (MDM).
It is important to carefully control the biocompatibility and cytotoxicity of a biomaterial for the potential use in the clinic 27 . Thus, we evaluated the toxicity profile of PLGA-EVG NPs with monocytes-derived macrophages (MDM). Over the range of 0-20 µM EVG, PLGA-EVG or 30HS@EVG NPs, ~100% cell viability was observed with all these groups when compared to control cells (Fig. 4A). No change in cell morphology was observed even after treating with the highest concentration (20 µM) of EVG free drug/PLGA-EVG NPs/HS30@PLGA-EVG (Fig. 4B).
Internalization mechanism of the PLGA NPs in MDM. In our recent report, we showed a time-and concentration-dependent uptakes of PLGA NPs in monocytes 23 . To further study the endocytosis pathways of these PLGA NPs in macrophage, various endocytosis inhibitors were utilized. The inhibitors include nocodazole, Cyto-D, CPZ, monensin, genistein, and MβCD, which suppress microtubule-related internalization, macropinocytosis, clathrin-mediated endocytosis, lysosome-involved internalization, caveolae-mediated pathways, and caveolae-/clathrin-mediated endocytosis, respectively 28,29 . To detect by flow cytometry, a fluorescence probe, C6 www.nature.com/scientificreports www.nature.com/scientificreports/ (green) was encapsulated in PLGA NPs. The inhibition effect of all the endocytosis inhibitors on the internalization of PLGA-C6 NPs in MDM are minor, indicating that the internalization of PLGA NPs in MDM may relate to a combination of endocytosis pathways (Fig. 5A,B). We further examined the sub-cellular fate of PLGA NPs using a confocal microscopic analysis (Fig. 5C). PLGA NPs were efficiently internalized in MDM after 2.5 h exposure. A co-localization of PLGA-C6 and early endosome marker were observed, but less co-localization was observed in the presence of late-endosomal marker, and no co-localization was observed in the presence of the lysosomal and mitochondria marker. This suggests that PLGA NPs can escape from endo-lysosomal compartments and deliver the therapeutics to the macrophages efficiently. In vitro BBB model and teeR assessments. The in vitro BBB model used in this study was co-cultured endothelial cells (bEnd.3) with astrocytes (C8-D1A) in a Transwell ® plate as described in the method section ( Fig. 6A). The bEnd.3 cell are a commercially available cell line and a useful BBB-mimicking system for biological and pharmacological research because it expresses efflux transporters P-glycoprotein (P-gp) and breast cancer resistance protein (BCRP) 30,31 . A bEnd.3/C8-D1A co-culture model has shown to present significantly higher tight junctions than the monoculture of bEnd.3 32 . In our study, TEER values were measured for 6 days after co-culturing. We observed that the TEER value reached a plateau on day-4 and it was stable from 4 to 6 days (Fig. 6B). A mean TEER value of 100 to 120 Ohms × cm 2 was observed in the confluent BBB model, which is consistent with the literature report 32 . All transmigration experiments were conducted after membrane integrity was stable with consistent TEER values.
Transmigration of PLGA-C6 NPs across the in vitro BBB model. The characterization of the PLGA-C6 NPs was reported previously, which demonstrated a time-and dose-dependent uptakes in monocytes 23 . To test the transmigration of PLGA-C6 NPs across the in vitro BBB model, three different concentrations of PLGA-C6 NPs were exposed in the upper chamber, and the culture media in the bottom chamber was collected after 24 hours. The penetration of PLGA-C6 NPs was examined by quantifying the fluorescence intensity of the culture media, which was collected from the bottom chamber. We observed a dose-dependent increase in fluorescence intensity from media samples, which suggests a dose-dependent penetration of PLGA-C6 NPs across the BBB model (Fig. 7A). This finding is also supported by a quantification of C6 from the same media. C6 concentrations of the media collected from the bottom chamber were calculated using a C6 calibration curve. The penetration of C6 across the BBB model also followed a dose-dependent profile, and all the PLGA-C6 groups showed significant increase in C6 fluorescence intensity compared with PLGA-Blank after crossing the BBB model (Fig. 7B, *p < 0.05 compared with PLGA-Blank, #p < 0.05 compared with the previous concentration). These results were also validated by the qualitative imaging of recipient astrocytes, which showed that PLGA-C6 NPs were taken up by cells in a dose-dependent manner after crossing the BBB model. The highest fluorescence was observed in cells treated with the highest concentration of PLGA-C6 NPs (Fig. 7C).
Increased penetration of PLGA-EVG NPs in the in vitro BBB model. To understand whether nanoparticle formulation can increase the transmigration of EVG in the in vitro BBB model, we measured EVG amount in the bottom chamber from both media and cell lysate samples. As expected, PLGA-EVG NPs showed higher penetration compared to EVG native drug. A dose-dependent penetration profiles for both EVG native drug and PLGA-EVG NPs were observed. A significant increase of EVG penetration was observed for both free EVG and PLGA-EVG NPs from 2.5 µg/mL to 5 µg/mL and from 5 µg/mL to 10 µg/mL (Fig. 8A, *p < 0.05 compared with PLGA-Blank, #p < 0.05 compared with the previous concentration). The results showed that a maximum of 27% of PLGA-EVG NPs can cross the BBB model after 24 hours. The results also showed a significant increase in penetration of PLGA-EVG NPs compared to EVG native drug at the dose of 5 µg/mL and 10 µg/mL. We also performed a time-dependent penetration study using 5 µg/mL of native EVG/PLGA-EVG NPs. Both   www.nature.com/scientificreports www.nature.com/scientificreports/ native drug and NPs showed a time-dependent penetration across the BBB model. Penetration of PLGA-EVG NPs across the BBB model showed a significant increase in the area under the curve (AUC) compared to EVG native drug (Fig. 8B, *p < 0.05). These results demonstrate that PLGA-EVG NPs improve the ability of EVG to cross the BBB model in the in vitro model. Moreover, no changes were observed for TEER values (Fig. 8C) and endothelial cell morphology (Fig. 8D) before and after the treatment, suggesting that PLGA-EVG transmigration did not change the monolayer integrity of the BBB model. Additionally, we identified the mechanistic contribution of P-gp in interfering the penetration of EVG in the in vitro BBB model and the capability of PLGA NPs to bypass the BBB efflux transportation. For this, we performed a penetration study in the presence and absence of a P-gp inhibitor elacridar for 24 hours (Fig. 8E). The AUC (0-t) of in vitro BBB penetration for EVG was significantly increased in the presence of the P-gp inhibitor, suggesting that the low EVG level in the CNS is, at least in part, mediated by P-gp efflux transportation (Fig. 8F). In contrast, the penetration of PLGA-EVG NPs showed no change in the presence and absence of the P-gp inhibitor, suggesting the ability of NP to bypass the efflux transporter P-gp in the in vitro BBB.   www.nature.com/scientificreports www.nature.com/scientificreports/ on viral replication on HIV-1-infected MDM in a modified in vitro BBB model following a one-month treatment paradigm as described in the treatment scheme (Fig. 9A). HIV-1-infected MDM were cultured on the bottom chamber, which will not be exposed to EVG directly. Blank PLGA NPs, EVG, and PLGA-EVG NPs were given on the upper chamber, which contains BBB monolayer. TEER values of the BBB model were measured for all the groups every day. Both EVG and PLGA-EVG NPs did not compromise the physiological integrity of the in vitro BBB model (Fig. 9B). HIV p24 protein level was measured from the bottom chamber daily. The effect of viral suppression in the MDM will be caused by the EVG/PLGA-EVG upon crossing the BBB monolayer to the bottom chamber. Our results showed that PLGA-Blank NPs have no effect on viral replication. However, both native EVG and PLGA-EVG NPs suppressed the viral replication to 70-100% of control (Fig. 9C). We calculated the p24 area under the curve (AUC) to compare the viral suppression efficacy of native EVG and PLGA-EVG NPs from day-1 to day-7 (Fig. 9D, *p < 0.5). The results showed that the HIV-1-infected MDM, when exposed to PLGA-EVG NPs have a significantly lower viral load, indicating that PLGA-EVG NPs have better efficacy on viral suppression in HIV-1-infected MDM after crossing the in vitro BBB model. We also measured EVG levels in MDM at the end of the treatment. EVG levels in MDM was significantly increased when exposed to PLGA-EVG NPs, compared with EVG native drug (Fig. 9F, *p < 0.5). This result suggests that the enhanced viral suppression efficacy, achieved by PLGA-EVG NPs, is correlated with increased EVG intracellular uptake in MDM.

Discussion
Although HIV-1-associated dementia (HAD), the most severe form of HAND, is rare in developed countries 34 , there is no specific treatment option available for HAND 2 . Currently, ART is the best option for delaying or preventing HIV-1 disease progression 35 . However, due to the inability of ARVs to cross the BBB following systemic administration, ART is still not clinically applicable for the treatment of HAND 2 . HIV-1 infection in CNS cells affects not only cognitive functioning but also the opportunity of HIV-1 eradication 2 . Individuals with HAND commonly show characteristics of dysfunctional and impaired judgment, memory, multitasking, and attention 36 . The CNS can serve as a viral reservoir for production of HIV-1, and limit the opportunity for viral eradication 37 . Thus, there is a need for novel treatment regimens that cross the BBB and deliver therapeutic into the CNS to suppress the viral replication in the CNS cells, especially viral reservoirs. The Nair group has explored a field-controlled magnetic-NP drug formulation with the capability of magnetic field-triggered release across the BBB, for delivery of ARVs or siRNAs to the brain [38][39][40] . The same group has also collaborated with the El-Hage group and demonstrated an intranasal delivery of siRNA in mouse brain to achieve HIV-1 attenuation 41 . The Gendelman group also has extensive experience with ARV nanoformulation where they have developed several long-acting nanoformulated ARV prodrugs that show slow release profiles and improved antiretroviral activity [42][43][44] . Although these studies have demonstrated the use of nano-carriers to deliver ARVs to the brain [38][39][40][41]45 , none of these strategies are in clinic yet. Therefore, it is necessary to study and develop alternative, safe, and clinically applicable ARV delivery systems for the treatment of HAND 9,19 .
We selected EVG as our model drug, which is a FDA-approved integrase strand transfer inhibitor, and is part of the first line ART regimen for the majority of HIV-1-infected populations in the USA 46 . EVG in this www.nature.com/scientificreports www.nature.com/scientificreports/ nanoformulation provides an effective antiretroviral ability and a favorable safety profile. In the last three years, researchers started formulating EVG into NPs to combat HIV-1 infection 47,48 . For example, EVG has been formulated into a bioadhensive PLA NPs with a hyperbranched polyglycerols (HPG) coating on the surface 47 . These bioadhensive PLA NPs were distributed widely throughout the reproductive tract, and showed a prolonged intravaginal delivery of EVG. Moreover, EVG has been used together with tenofovir alafenamide (TFV) in a PLGA-based NPs coated with poloxamer 407 as a prevention strategy to against HIV-1 48 . This EVG NPs can extend the t 1/2 of EVG from 10.8 h to 3.3 days and protect mice with 100% and 60% uninfected rate on day 4 and day 14 post-nanoparticle injection after a HIV-1 challenge. In contrast to the published EVG nanoparticle studies, our goal is to study the potential use of EVG NPs in CNS drug delivery to suppress the HIV-1 replication in the brain. Like other EVG NPs, we also developed a polymeric nanoparticle of EVG for a better stability and safety profile. An FDA-approved polymer, PLGA, was used to incorporate EVG and formulate the core of the NP. PLGA is a biocompatible polyester, which can be converted into water and carbon dioxide via tricarboxylic acid cycle and eventually eliminated from the body 49 . In addition, human body has shown to have low inflammatory response to PLGA suggesting its applicability for systemic administration clinically 50 . In this work, we identified that the safety and stability profile of the PLGA NPs, and both the safety and the stability profile of the PLGA NPs support their use as a potential CNS delivery strategy.
Understanding the mechanistic internalization of NPs by the targeted cells can increase our knowledge of how these particles are taken up and are transported within the cells, and to which extent they are secreted 51 . In general, most of the NPs are taken up into cells via endocytosis into endosomes and subsequently trafficked into lysosomal compartments 52 . Our data showed that a combination of different endocytosis pathways were involved in the uptake of PLGA NPs in the macrophages, which is consistent with previous studies [53][54][55] . Although initially present in early endosomes, PLGA NPs can still bypass the endo-lysosomal compartments in macrophages with less and no co-localization found in the presence of late endosome and lysosome marker, respectively. Our design allows the PLGA NPs escape the endo-lysosomal degradation/secretion, which helps the therapeutic to work effectively inside the cells 52 .
NPs have been demonstrated as a promising strategy to deliver drugs at the site of action, bypass transporters and drug metabolic enzymes, and cross the BBB 19,21 . Thus, we hypothesize that, EVG in PLGA NPs can bypass transporters on the BBB to perform a higher drug transmigration and can also bypass the transporters and drug metabolic enzymes on macrophages after crossing the BBB, showing improved viral suppression on the HIV-1-infected macrophages. In order to have an efficient brain delivery, we used poloxamer 188 as the surfactant in our PLGA NP formulation. Poloxamer 188 is known to have low toxicity and able to decrease inflammation in the injured brain of experimental rats 56 . Poloxamer 188 has been used as an effective surfactant for brain delivery in cancer 57,58 . Literatures showed that poloxamer 188 coating can help PLGA NPs become to an efficient brain delivery method 57 . They demonstrated that poloxamer 188-coated PLGA NPs have a high anti-tumor effect against an intracranial 101/8 glioblastoma in rats, compared with the non-coated NPs. In another study, poloxamer 188-coated PLGA NPs demonstrated the greatest cellular uptake in the brain tissues over the NPs coated with other commonly used surfactants such as polysorbate 80 and poloxamer 407 58 . Similarly, the poloxamer 188-coated EVG NPs demonstrated an effective brain delivery profile in our study.
Currently, the magnetic nanoformulation is the most commonly studied nanoformulation for NeuroAIDS 38,40 . Magnetic NPs of azidothymidine showed a 3-fold higher transmigration in an in vitro BBB model, compared to the native drug 38 . This magnetic nanoformulation of azidothymidine also showed a slightly lower p24 antigen production when treating with HIV-1-infected PBMCs. The same group also developed a magnetically layer-by-layer NPs with a ~37.95% BBB transmigration and ~33% p24 suppression on astrocytes in vitro 40 . Compared with those magnetic NPs, our PLGA-EVG NPs showed a comparable percentage of transmigration and an efficient viral suppression in HIV-1-infected macrophages. To the best of our knowledge, this is the first report of using EVG nanoformulation as a potential delivery method to cross the BBB model and suppress the HIV-1 in the CNS macrophages. We understand that the use of mouse model of the BBB with human macrophages is a limitation of this study. However, a combined model with mouse BBB cells and human macrophages fulfils the basic requirement for an in vitro BBB model that provides a comparison of both drug penetration and viral suppression of PLGA-EVG NPs and EVG native drugs in our study. In this model, the mouse BBB model can be used to study BBB transmigration because of the tight junction and the expression of efflux transporters P-gp and BCRP [30][31][32] . The validity of the BBB model in this study is further strengthen, because we are able to study the efficacy of the viral suppression in HIV-1-infected human primary macrophages 59 . Overall, compare to commonly used magnetic NPs, our PLGA-EVG NPs is easy to prepare, is FDA-approved non-toxic NPs, and does not require an application of an external magnetic field. We expect that this PLGA-EVG NPs have the potential to move forward to the clinical use.

conclusion
Our current study demonstrated a poloxamer-PLGA based nanocarrier is an efficient delivery approach for EVG. The PLGA-EVG NPs demonstrated a favorable stability and safety profile. The PLGA-EVG NPs demonstrated significant transmigration across an in vitro BBB model. Most importantly, PLGA-EVG NPs showed an improved viral suppression in HIV-1-infected macrophages after crossing the BBB model. Ongoing study in our laboratory is to determine its safety and efficacy in animal model. This is the first step towards achieving our goal to create a safe and efficient drug delivery method to target HIV reservoirs in the CNS for potential clinical.  (14-826-49) and BD 1 mL TB syringe (14-826-88) were obtained from Fisher Scientific (Hampton, NH). Sterile phosphate-buffered saline was obtained from Gibco (Dublin, Ireland). Ethylenediaminetetraacetic acid (EDTA) was brought from Boston Bio Products (Ashland, MA).

Materials
The immortalized mouse brain endothelial cells (bEnd.3, CRL-2299) and mouse astrocytes (C8-D1A, CRL-2541) were purchased from American Type Culture Collection (Manassas, VA, USA). Human monocytes-derived macrophages (MDM) were differentiated from the de-identified human blood, which was obtained from Interstate Blood Bank Inc. (Memphis, TN)  All methods were carried out in accordance with relevant guidelines and regulations. All experimental protocols with cell lines and primary cells were approved by the Institutional Biosafety Committee (IBC) at the University of Tennessee Health Science (UTHSC). The protocols for analysis of blood samples from de-identified subjects obtained from Interstate Blood Bank (Memphis, TN) were approved by Institutional Review Board (IRB), UTHSC. The use of blood samples from de-identified subjects, bought from a blood bank, does not require informed consent from subjects and is waived from the IRB approval.
Preparation of PLGA-EVG/PLGA-C6 NPs. EVG loaded PLGA NPs were prepared by nano-precipitation technique as described in our previous report 23 . Cells were treated with 5 µg/mL of coumarin-6 equivalent PLGA nanoparticles at different Briefly, 45 mg of PLGA was dissolved in 4 mL of acetone (organic phase). EVG (4 mg) was added to the above organic phase, under constant magnetic stirring. The PLGA-EVG organic phase was then added dropwise into 10 mL of 1% PVA-aqueous solution on a magnetic stir plate at 400 rpm. After 3 hours, aqueous solution of 10 mg of PLL and 50 mg of F-68 were added to the nanoparticle suspension and stirred at room temperature to allow acetone evaporation. Finally, the larger aggregates were removed by centrifugation at 1000 rpm for 10 min. The finer and uniform PLGA-EVG NPs were stored at 4 °C/−20 °C for further study. Coumarin-6 (C6) loaded NPs were also prepared in a similar manner as EVG loaded PLGA NPs. C6 was used as a fluorescent model to track, locate, and measure the green fluorescence for semi-quantitative and qualitative cellular uptake. We have characterized both PLGA-EVG NPs and PLGA-C6 NPs in our previous report 23 . PLGA-EVG NPs and PLGA-C6 NPs showed similar particle sizes, and PLGA-C6 NPs were used to study subcellular localization, cellular uptake and internalization, and BBB model transmigration in this study.
In vitro drug release. The in vitro drug release study was performed using a Float-A-Lyzer ® dialysis device with molecular weight cut-off of 8-10 kDa (Sigma-Aldrich Co. Z726508). A volume of 1 mL PLGA-EVG NPs formulation was enclosed in a dialysis device and incubated in 30 mL cell culture media at 37 °C under mild agitation. At 15, 30 min, and 1, 3, 6, 9, 12, 24 h, 500 µL incubation media was withdrawn and analyzed for EVG level using LC-MS/MS. The same volume of fresh media was added to the incubation media after each sampling.
Human serum protein binding to PLGA NPs. Three hundred micrograms of PLGA NPs were incubated with 1-50% HS for 3 hours at 37 °C. The HS bound PLGA NPs were obtained by centrifugation of this NPs-HS mixed solution at 10,000 rpm for 20 min. The HS bound PLGA NPs 60 in the pellet were denoted as % of HS@ NPs. For example, the 10% HS bound PLGA NPs were recognized as 10HS@NPs. To examine the HS protein adsorption on PLGA NPs, the HS@NPs samples were either resuspended in PBS for a dynamic light scattering (DLS) analysis or lyophilized using the Labconco Freeze Dry System (−53 °C, 133 × 10 −3 mBar; Labconco, Kansas City, MO) to perform a Fourier Transform Infrared (FTIR) spectroscopy. The HS protein adsorption on PLGA NPs was determined by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) analysis as described 61 . Briefly, the adsorption capacity of HS proteins was studied based on 3 protocols: (1) equal amount (300 µg) of PLGA NPs were incubated with various concentration (1-50%) of HS for a fixed incubation period (3 h); (2) equal amount (300 µg) of PLGA NPs were incubated with a fixed concentration (10%) of HS for a various incubation period (0.5, 1, 2, 3 and 24 h); and (3) various amount (10-300 µg) of PLGA NPs were incubated with a fixed concentration (10%) of HS for a fixed incubation period (3 h). The HS@NPs were obtained by centrifugation at 10,000 rpm for 20 min. The HS@NPs samples were separated by size in a 10% SDS-PAGE analysis using the constant voltage of 150 V for 75 min. The SDS-PAGE gel was processed for a Bio-Safe ™ Coomassie Stain (BioRad, Hercules, CA), and scanned using a BioRad Gel Doc.

Dynamic light scattering.
A DLS analysis was performed to measure the particle size and zeta potential of PLGA NPs or HS@NPs using a Zetasizer instrument (Nano ZS, Malvern Instruments Ltd., Worcestershire, UK). Both particle size and zeta potential of NPs/HS@NPs were measured in PBS at 25 °C. The particle size of each Scientific RepoRtS | (2020) 10:3835 | https://doi.org/10.1038/s41598-020-60684-1 www.nature.com/scientificreports www.nature.com/scientificreports/ sample was measured for 3 min (~15 runs), and the zeta potential of each sample was measured for 90 runs. The average particle size and zeta potential of 3 readings were calculated and reported for each sample.
Fourier transform infrared spectroscopy. FTIR spectroscopy was employed to confirm the extent of binding or presence of HS proteins on HS@NPs. The FTIR spectra of lyophilized HS@NPs were acquired using PerkinElmer Spectrum 100 FTIR spectrometer (Waltham, MA) equipped with a Diamond/ZnSe Attenuated Total Reflection crystal plate. PLGA NPs were used as a negative control, and the pure HS was used as a positive control. An average of 32 scans was performed on all the samples. The spectra were scanned from 4000 to 600 to cm −1 with a resolution of 4 cm −1 .
Stability assay. We determined the particle size and zeta potential of PLGA NPs using DLS. PLGA NPs were stored at room temperature, 4 °C, or −20 °C for a stability test. DLS analysis was performed on the room temperature sample for 7 days. The size and zeta potential of PLGA NPs from 4 °C and −20 °C were measured on day 7, and an additional measurement was performed on day 30 for the −20 °C sample. The particle size and zeta potential of PLGA NPs in 1 mM HEPES buffer from pH 1 to pH 11 were measured to investigate the alteration of PLGA NPs with this pH range.
Hemocompatibility. The red blood cells (RBCs) were collected from the de-identified human blood. The freshly collected RBCs were resuspended in RPMI phenol red free media for a hemolysis assay as described in our previous report 61 . Briefly, the RBC suspension (600 µL) was incubated with 5-200 µg/mL EVG or equivalent concentrations of PLGA-EVG, 30HS@PLGA-EVG, and PLGA-blank, 30HS@PLGA-blank in Eppendorf tubes for 3 hours at 37 °C. After incubation, the experimental tubes were centrifuged at 1000 rpm for 10 min and the supernatant was transferred to a 96-well plate to measure the release of hemoglobin using a plate reader (Cytation 5 imaging, BioTek, Winooski, VT) at 570 nm. Ten microliter of RBCs from the pellet were imaged on a glass slide using the AMF4300 EVOS ® FL Imaging System (Life Technologies, Carlsbad, CA). RBCs dissolved in MilliQ water was used as a positive control #1 (100% of hemolysis). SDS was used as positive control #2 (hemo-toxicity), and PBS was used as a negative control (0% of hemolysis) in this experiment.

Generation of HIV-1-infected primary macrophages. Peripheral blood mononuclear cells (PBMCs)
was used to generate MDM according to previously published reports 62,63 . Briefly, the buffy coats were diluted and layered over the lymphocyte separation medium and centrifuged for 20 min at 1200 g. PBMCs for MDM differentiation were collected and cultured in RPMI media containing 10% HS, 1% L-glutamine, and 5% penicillin-streptomycin. After overnight incubation, non-adherent cells were washed by PBS. Cellular toxicity assay. Cellular toxicity assay was performed on MDM in the presence of EVG native drug, PLGA-EVG NPs, and HS30@PLGA-EVG using an XTT cell viability kit (Cell signaling, Danvers, MA). In this study, MDM were seeded into 96-well plates (2 × 10 4 in each well). Cells were incubated with 0-20 µM of EVG or equivalent concentrations of PLGA-EVG, and HS30@PLGA-EVG for 24 hours. The cells were washed twice with PBS and replaced with 100 µL fresh media. Subsequently, 50 µL of XTT detection solution, which contains electron coupling solution, was added to each well of the 96-well plate. MDM were incubated with XTT detection solution for 3 hours and absorbance was measured using a plate reader at 450 nm. The percentage of viable cells from treatment groups was calculated by comparing the absorbance with the untreated cells. Data presented are from five replicates.
Sub-cellular localization. In this study, PLGA-C6 NPs were used to visualize the sub-cellular localization of the NPs. MDM were seeded on 4-well cell culture chamber glass slides (CellTreat, Pepperell, MA). Cells were treated with 2.5 µg/mL of PLGA-C6 NPs for 2 hours. Subsequently, cells were incubated with markers for mitochondria (MitoTracker TM Deep Red), early endosome (Texas Red TM Conjugate), late endosome (CellLight TM Late Endosomes-RFP), and lysosome (LysoTrackerR Red DND-99, Life Technologies) for 30 minutes. After the incubation, cells were fixed using 4% paraformaldehyde for 20 minutes and permeabilized with 0.2% TritonX-100 for 5 minutes. All the cells were stained with DAPI (Life Technologies) and mounted in Vectashield Mounting Medium (Vector Labs, Burlingame, CA) to visualize nuclei. All the images were taken using a laser confocal microscope (Carl Zeiss LSM 710, Oberkochen, Germany) under 400 × magnification using oil immersion objective.
Cellular uptake and internalization mechanism. MDM were seeded in 6-well plates at a density of 8 × 10 5 cells/well. The cells were pretreated with various endocytosis inhibitors for 30 min at 37 °C. The inhibitors were: nocodazole (10 µg/mL), Cyto-D (10 µg/mL), CPZ (10 µg/mL), monensin (200 nM), genistein (200 µM), and MβCD (1 mM). The cells were incubated with 2.5 µg/mL PLGA-C6 NPs for 2 hours at 37 °C. After the treatment, cells were washed twice with PBS, trypsinized, and collected to measure the cellular uptake. Semi-quantitative PLGA-C6 uptake in MDM were measured using Accuri C6 Flow Cytometer (Accuri Cytometer Inc, Ann Arbor, MI). (2020) 10:3835 | https://doi.org/10.1038/s41598-020-60684-1 www.nature.com/scientificreports www.nature.com/scientificreports/ Images were taken with a laser confocal microscope (Carl Zeiss LSM 710) under 400× magnification using oil immersion objective. In vitro BBB model. Mouse brain endothelial cells bEnd.3 and mouse astrocytes C8-D1A were cultured in complete media (DMEM media with 10% fetal bovine serum and 1% penicillin-streptomycin solution) in T75 flasks at 37 °C in a humidified incubator with 5% CO 2 before building in vitro BBB model. Transwell ® -COL collagen-coated 0.4 μm pore polytetrafluoroethylene membrane insert (Sigma-Aldrich) was used to build an in vitro BBB model as previous published protocols 32,64 . Mouse astrocytes were seeded on the bottom of 12-well plates at a density of 2 × 10 5 cells/well. After 24 hours of adhesion, mouse endothelial cells were seeded onto the upper side of the Transwell ® -COL inserts at a density of 2 × 10 5 cells/well, and the inserts were placed in a 12-well plate containing astrocytes. The in vitro BBB model was grown for 5 days to achieve ~90% confluency for transmigration assay.
Trans endothelial electrical resistance (TEER) measurements. The integrity of the tight junction dynamics of the BBB model was measured in terms of transendothelial electrical resistance (TEER) using EVOM2 Epithelial Voltohmmeter (World Precision Instruments, Sarasota, FL) 65 . TEER values were measured regularly every day for 6 days from the day of cell seeding. The TEER values were also measured during the experiment period to assess the integrity of the endothelial cell layer. TEER of the blank insert with media alone was subtracted from the TEER final reading to obtain the TEER value for the endothelial monolayer. The TEER value was presented as the value of resistance (current, Ohm) multiplied by the area (cm 2 ) of the endothelial monolayer (Ohm⋅cm 2 ).
Transmigration study of PLGA NPs. The transmigration of PLGA NPs was visualized and measured using fluorescence dye C6 in a confluent in vitro BBB model. PLGA-C6 NPs at concentrations of 2.5, 5, and 10 µg/mL were added to the upper inserts and incubated at 37 °C for 24 hours. After incubation, astrocytes in the bottom chamber were washed twice with PBS. The media of the bottom chamber was measured using a fluorescence plate reader (SpectraMAX M2e; Molecular Devices, San Jose, CA) with excitation at 488 nm and emission between 400 and 650 nm. The transmigration of PLGA-C6 NPs was presented in terms of fluorescence intensity. C6 concentration in the media was also calculated using a C6 calibration curve with the matrix of media. The images of receiving astrocytes were taken by EVOS ® FL Imaging System. Penetration of PLGA-EVG NPs in the in vitro BBB model. The in vitro BBB model was used to assess the penetration of the PLGA-EVG NPs. We performed a 24-hour EVG native drug/PLGA-EVG NPs treatment on the BBB cells. EVG native drug/PLGA-EVG NPs at concentrations of 2.5, 5, and 10 µg/mL were added to the upper inserts and incubated at 37 °C for 24 hours. A sample of 50 µL of receiver media was collected at 1, 3, 6, 9, and 24-hour time point, and an equal volume of fresh media was added. At the end of the experiment, the cells from the bottom chambers were collected in RIPA buffer as the cell lysate. The amount of EVG flux to the bottom chambers from both media and cell lysate was measured using our developed LC/MS method 66 . Quantification of EVG using LC-MS/MS. A 50 µL of media or cell lysate samples were mixed with 3-volumes of cold acetonitrile containing internal standard (IS) ritonavir (50 ng/mL). The mixed solution was vortexed and centrifuged at 8000 rpm for 10 min for protein precipitation. A clear supernatant was collected and analyzed using our established LC-MS/MS method 66 . Briefly, a Shimadzu liquid chromatographic system (Kyoto, Japan) coupled with an AB SCIEX Triple Quad 5500 tandem mass spectrometer (Framingham, MA) was used for the analysis. Chromatographic separation was performed on an Xterra ® MS C18 column (125 Å, 3.5 μm, 4.6 mm × 50 mm; Waters, Milford, MA). The mobile phase used for EVG separation consisted of (A) water with 0.1% formic acid and (B) acetonitrile with 0.1% formic acid (v/v) at a flow rate of 1 mL/min. The gradient elution was as follows: 0-1.5 min, 50% B; 1.5-5.1 min, 60% B (v/v). The quantification of the validated assay of EVG was 1 to 500 ng/mL. EVG and IS were eluted separately at approximately 3.27 and 2.72 min, respectively. The multiple reactions monitoring (MRM) transitions (m/z) Q1/Q3 selected for quantitative analyses were 447.9/343.8 for EVG and 721.3/296.1 for the IS. In order to reduce matrix effects, calibration curves were prepared with blank media or cell lysate based on the sample types.
Mechanistic contribution of P-gp in interfering the penetration of EVG in the in vitro BBB model. The mechanistic contribution of P-gp in interfering with the penetration of EVG was assessed in the presence and absence of the P-gp inhibitor elacridar. The in vitro BBB layer was preincubated with/without elacridar (500 nM) for 30 min before the penetration study. After preincubation, 5 µg/mL of EVG native drug/PLGA EVG NPs were added to the upper inserts and incubated at 37 °C for 24 hours. A sample of 100 µL of media from the bottom chamber was collected at 1, 3, 6, 9, and 24-hour time point, and an equal volume of fresh media was added. The amount of EVG flux to the bottom chambers was measured using LC-MS/MS method.

Viral suppression of PLGA-EVG NPs in HIV-1-infected MDM after crossing in vitro BBB model.
To determine the efficacy of the viral suppression of PLGA-EVG NPs, we used HIV-1-infected primary MDM to create a modified in vitro BBB model in a Transwell ® plate. The confluent brain endothelial monolayer was obtained as described in the previous section. The upper insert of the in vitro BBB model containing confluent brain endothelial cells was transferred to the 12-well plate containing HIV-1 infected MDM, and co-cultured with HIV-1-infected MDM. In this modified in vitro BBB model, HIV-1-infected MDM were cultured on the bottom of the 12-well plate. Endothelial monolayers in the upper inserts were exposed to control, blank PLGA NPs, EVG native drug (5 µg/mL), and PLGA-EVG NPs (5 µg/mL) for 7 days. HIV-1 viral loads were measured every day using a p24 ELISA kit from the culture media from the bottom chamber. The upper inserts