Commercial cell lines THP-1 and MOLT-4/CCR5 obtained from ATCC are provided with certification that they are Mycoplasma-free. Stock cultures are grown up, and aliquots are frozen down to enable the use of the same cell lines throughout the study. The cell line was maintained in the respective cell-culture media at 37°C/5% CO₂/125 rpm shaker.

To produce the R5 tropic HIV-1/eGFP pBR43IeG-nef+/EGFP (AIDS Research and Reagents Program #ARP-11100) was used as a parental plasmid construct. The pBR43IeG-nef+-EGFP V3-Env region was removed by digestion with StuI/BsaBI restriction enzymes (NEB, Ipswich, MA) following the treatment with Calf Intestinal (CIP) Alkaline Phosphatase (NEB, Ipswich, MA) to prevent self-ligation. Next, gel purification was performed to remove the pBR43IeG-nef+-EGFP V3 Env region. DNA synthesis (Blue Heron Bio, Bothell, WA) was used to recreate the excised pBR43IeG-nef+-EGFP V3-Env sequence containing a BaL V3 DNA sequence. Synthesized DNA was digested with StuI/BsaBI (NEB) and then gel purified. DNA fragments were ligated with T4 DNA ligase (NEB) and used to transform STBL2 cells (Invitrogen, Waltham, MA). Clones were screened by restriction mapping, (EcoRI/StuI, NEB) with two correct restriction digest pattern clones selected for sequence confirmation (CIBR, Baltimore, MD).

HIV-1 JRFL pseudoviruses were generated by co-transfection of HEK293T cells with an Env-deficient HIV-1 backbone plasmid pNL4-3-ΔE-EGFP along with Env-expression plasmids (35, 36) and pCAGGS-JRFL (kindly provided by J. Binley, Torrey Pines Institute of Molecular Studies, San Diego, CA). Transfections were accomplished using FuGENE 6 (Roche, Indianapolis, IN) transfection reagent at a 3:1 reagent-DNA ratio. To produce the infectious molecular clone of HIV-1 BaL or transmitted/founder (T/F) HIV-1 AD17 virus (37) or HIV-1/eGFP, HEK293T cells were transfected with the BaL (obtained through the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID) or AD17 plasmid (kindly provided by B. Hahn, University of Pennsylvania) or pSF345-02-gfp plasmid at a FuGENE-to-DNA ratio of 3:1. Virions-containing supernatant was harvested after three days, and concentrated about 10-fold by incubating with PEG-it™ virus precipitation solution (System Biosciences, Mountain View, CA) for 18 hours at 4°C as recommended by vendor. The antigen content of all virion preparations was quantified using p24 and gp120 antigen capture ELISAs. Infectivity was established using standardized procedures (38) and quantified as a function of TCID₅₀ in TZM-bl cells. HIV-1 BaL and HIV-1 JRFL pseudoviruses with gp120 to p24 ratio of 1:10-1:50, and 200,000 – 500,000 TCID50/mL; HIV-1 AD17 T/F with gp120 to p24 ratio of 1:200, and 600,000 to 1,000,000 TCID50/mL; were used.

MOLT4/CCR5 cells were infected using MOI 1 of HIV-1/eGFP via spinoculation at 2000 rpm for 2 hours at 7°C. Cells were washed and seeded at a concentration of 2×10⁶/mL. After 48 hours, 10 µL of cell suspension was incubated in poly-lysine coated slides for 30 mins at room temperature and washed two times before imaging under a 2p-FLIM microscope.

To parallel the results of the in vitro studies in the previous section and complement the outcome allowing cross-comparisons of in vivo and in vitro data and the generation of a comprehensive picture of NADH metabolism in tissues, we have performed NADH fluorescence in HIV-1 infected humanized mouse tissues. Furthermore, HIV-infected tissues allow us to survey the host-pathogen interaction and determine the tissue level response to infection. Accordingly, the 2p-FLIM method investigated NADH metabolism in cryo-sectioned p24 confirmed HIV-1 infected tissues. NADH levels in HIV-infected mouse tissues (specifically spleen and lymph nodes) were quantified using 2p-FLIM and compared with the uninfected tissues. NSG mice reconstituted with human PBMCs (hu-PBMC-NSG-SGM3) (39) were used for the current study investigating HIV-1 infection in a humanized mouse model. These strains/models do not suffer significantly from graft versus host (GVH) effects. This is a robust humanized mouse model, which unavoidably enhances HIV replication via developing GVH immune activation.

Fixed, frozen 7µm thick tissue sections of lymph-node and spleen from HIV-1 BaL-infected humanized NSG mice or controlled humanized NSG mice (39) were made on a Leica CM1860 UV - Cryostat. Cryo-sectioned tissues were mounted on a glass slide with Invitrogen prolong antifade reagent. HIV-1 p24 monoclonal antibody D45F tagged with FITC (Invitrogen catalog # MA1-7378) was used to stain the 7µm thick lymph node tissue section from HIV-1 infected mice.

A customized confocal microscope (based on ISS Q2 laser scanning nanoscope) with single-molecule detection sensitivity was used for performing 2p-FLIM. The excitation source is a pulsed femtosecond laser (Calmar, 780nm, 90 fsec pulse width, and 50MHz repetition rate) equipped with an ISS excitation power control unit. An incident wavelength of 780 nm was used for exciting NADH in cells or tissue samples (27). The excitation light was reflected by a dichroic mirror to a high-numerical-aperture (NA) water objective (60X; 1.2 NA) and focused onto the sample. The fluorescence was collected by single photon counting avalanche photodiodes (SPAD) through a dichroic beam splitter, Chroma short pass (750SP), and 460/55 bandpass filters, thus eliminating the scattered excitation light and collecting fluorescence from the NADH in the region of interest. Using the 780nm two-photon excitation, eGFP fluorescence intensity and lifetime were recorded in the 500 to 529nm spectral window with a narrow bandpass filter (514/30, Chroma). We used one excitation wavelength for simultaneous detection of both NADH and eGFP in two separate detection channels that included high quality bandpass emission filters without spectral bleed through. NAD(P)H fluorescence detection has been reported in the literature using 780nm excitation (3, 27). The imaging in our Q2 setup was performed with Galvo-controlled mirrors with related electronics and optics controlled through the 3X-DAC control card. The software module in ISS VistaVision for data acquisition and processing and the time-correlated single photon counting (TCSPC) module from Becker & Hickl (SPC-150) facilitate FLIM measurements and analyses. Each image included 512x512 pixels with a dwell time of 200 µsec for each pixel translating to an image acquisition time of ~53sec. The decay sampling resolution was 19.5 psec which was determined from the number of the decay time bins and the total time length of the decay. The FWHM of the measured IRF by directly recording the laser scatter light was ~ 300psec. The intensity decays were fitted with a bi-exponential model: I(t) = α₁exp(-t/τ₁) + α₂exp(-t/τ₂), where τ₁ and τ₂ are the short and long decay times; α₁ and α₂ are the amplitudes for the short and long decay times respectively ( Figure 1 ). The sum of the amplitude fractions (α₁ + α₂₎ in the observed biexponential are normalized to unity. The fractional contributions of short and long components to the steady-state intensity are denoted as f₁=α₁τ₁/(α₁τ₁+α₂τ₂) and f₂=α₂τ₂/(α₁τ₁+α₂τ₂). The average (intensity-weighted) lifetime is described by f₁τ₁+f₂τ₂. The values of FLTs (τ₁ and τ₂) were obtained using the ISS vistavision software with the deconvolution of instrument response function and nonlinear least-squares fitting. For the decay analyses we used the estimated IRF generated by the ISS VistaVision software. Three different fields of view per independent experiment were used to compute the average of different metrics. For the HIV-1 infected cell samples at least 20 cells were used for calculating various metrics.

All time-resolved fluorescence data were analyzed using ISS VistaVision software. TCSPC histograms of individual pixels are depicted as usual as the logarithm of the time resolved intensity as a function of time. TCSPC histograms were mostly fitted by a standard method to a bi-exponential decay model using the Marquardt algorithm, by minimizing the sum of the squares of differences between the measured and calculated values of time-dependent intensity and calculating a reduced χ². Two components were usually chosen for NADH decay analyses because the raw data indicate there is more than one component (the decay lines aren’t linear), with multiple fluorescent lifetimes present. For determining the FLTs (τ_1, τ₂, and average lifetime), pre-exponential factors α₁, α₂ and fractional contributions from each measurement, the regional average of the time-resolved fluorescence decays were fitted with a bi-exponential model as described before. However, the intensity decay for the eGFP channel could be fitted using a single-exponential decay model. The difference (residuals) between the actual measured intensity and the calculated value from the fit at that time point were evaluated. The standard deviations of the residuals were evaluated to assess the quality of the fit. A good fit is characterized by residuals that are mostly small (90% are less than 2 standard deviations) and randomly distributed around zero. The derived χ² value is expected to be close to 1.0.

We have avoided the crosstalk between NADH and eGFP channels using high quality narrow bandpass emission filters (Chroma) which minimize the spectral overlap in the spectral region of 435-485nm for the detection of NADH signal and 500-529nm for the detection of fluorescence from eGFP. The quantum yield of eGFP is significantly (over 10-fold) higher than NADH (4, 40). Furthermore, fluorescence recorded in the detection channel using a high quality 514/30 bandpass emission filter showed a single exponential decay with a fluorescence lifetime of 2.6 nsec confirming that the fluorescence emanated from eGFP only in 500 to 529 nm spectral range (15). Moreover, detection of NADH fluorescence with a narrow bandpass filter in the spectral region of interest alleviated the issue of fluorescence contributions from other structural proteins (collagen or elastin) (4). We have avoided any emission spectral overlap between flavins and NADH using high quality narrow bandpass filters in the spectral region of 435-485nm for detection of NADH signal. For each experimental FLIM measurement with the HIV-infected tissue samples, we included control mouse tissue samples detected under identical settings (same excitation wavelength, incident laser power, detection emission filter, image acquisition parameters). We have used tissues from three HIV-1 BaL infected mice for FLIM studies, correspondingly tissues were collected from uninfected controlled mice. In all HIV infected tissue section, either from spleen or lymph nodes, we consistently observed higher NADH signals in infected tissues compared with the control mouse tissues. Furthermore, the quantum yield of collagen and elastin fluorescence is lower than NADH either free or protein bound form. The emission spectra of collagen (emission maximum at 400nm) and elastin (emission maximum at 415nm) are ~50nm blue-shifted compared to NADH emission spectra (4).

Statistical analysis of data was performed using GraphPad Prism 8.0 and Origin 2021 software. A two-tailed test evaluated the statistical differences between control and infected cells or tissues. A p-value of less than 0.05 was considered statistically significant.

The test panel includes HIV-1 virus phenotypes and neutralization sensitivities (tiers) (41, 42). We examined pseudoviruses expressing neutralization-resistant tier-2 CCR5-tropic HIV-1 JRFL, tier-1 HIV-1 BaL infectious molecular clones (IMC) (43–46), and transmitted/founder (T/F) (47, 48) HIV-1 subtype B AD17 viruses, which are replication-competent IMC and express unmodified, native trimers (45, 46). The pseudoviruses have only a single round of infection, whereas the T/F or IMC virions are replication-competent and have multiple rounds of infection. All virion preparations are characterized for Env protein content, p24 protein content, and viral RNA copies by in-house quantitative PCR. We emphasize that there is no further genetic or chemical manipulation of the envelopes on these viruses.

The monocytic THP-1 cell line has been used to investigate the mechanisms by which HIV-1 infects and replicates in monocytes and macrophages, as well as the factors that influence the latency of HIV-1 in these cells (49, 50). Label-free live 2p-FLIM imaging of NADH in THP-1 cells upon infection with HIV-1 JRFL pseudovirions after 90 mins at different virion concentrations are illustrated in Figure 2 (A-C: NADH intensity, D-F: average FLT and G-I: τ₂, bound form of NADH FLT). Figure 2 display the time-resolved intensity decays of NADH from control THP-1 (panels J) and HIV JRFL pseudovirus (p24 values of 6µg/ml and 12µg/ml) infected THP-1 cells (panels K and L). A significant change in NADH intensity and FLT in the bound form of NADH in THP-1 cells upon incubation with JRFL compared to THP-1 control is observed and quantified in bar graphs ( Figures 2M, N ). It is important to note that the FLT is independent of probe (NADH in this case) concentration, so any FLT changes observed in the bound form of NADH (τ₂) represent the inherent changes in molecular properties. To further investigate if the changes in NADH levels in infected cells are associated with multiple rounds of infections, we have used T/F AD17 and BaL-IMC infected THP-1 cells as shown in ( Figures 3A–C : NADH intensity, D-F: average FLT and G-I: τ₂ bound form of NADH FLT). Using a bi-exponential fit (see methods), the intensity decay of NADH from uninfected THP-1 cells yielded a τ₁ value of 0.8 nsec (f₁ = 0.58) and τ₂ value of 3.2 nsec (f₂ = 0.42) with a χ² value of 1.19. In contrast, the bi-exponential fit with the intensity decay of NADH from HIV-1 JRFL infected THP-1 cells exhibited a τ₁ value of 1.0 nsec (f₁ = 0.27) and τ₂ value of 4.25 nsec (f₂ = 0.73) with a χ² value of 1.16. Likewise, the fit to the intensity decay of NADH from HIV-1 AD17 infected THP-1 cells exhibited a τ₁ value of 0.95 nsec (f₁ = 0.32) and τ₂ value of 4.1 nsec (f₂ = 0.68) with a χ² value of 0.96. Similarly, the two-component fit to the intensity decay of NADH from HIV-1 BaL-IMC infected THP-1 cells displayed a τ₁ value of 0.98 nsec (f₁ = 0.33) and τ₂ value of 4.25 nsec (f₂ = 0.67) with a χ² value of 1.10. The χ² values for all the fits to the different intensity decay data sets were in the range of 0.96 to 1.19, indicating overall good fit to the experimental data. Interestingly, more HIV-1 infected THP-1 cells were attached to the poly-lysine coated coverslips compared to the uninfected THP-1 cells. Similar to JRFL-infected THP-1 cells, the enhanced fluorescence intensity of NADH and long FLT associated with the bound form of NADH was observed for both BaL and AD17 virion-infected THP-1 cells compared to uninfected THP-1 cells as displayed in Figure 3 . The two-tailed t-test showed a significant difference in both intensity and FLT of NADH between infected and uninfected cells ( Figures 3J–L ). Similarly, we observed enhanced fluorescence intensity of NADH and longer FLT signaling bound form of NADH on HIV-1 BaL infected T lymphoblast MOLT-4/CCR5 cells (not shown). These experiments revealed how the virus phenotypes and doses affect cellular NADH levels.

The MOLT-4/CCR5 cells are highly permissive for infection by HIV-1. A recent study (51) reported that HIV-1 selectively infects metabolically active CD4+ T cells with high oxidative phosphorylation and glycolysis. Blocking metabolic pathways in HIV-1 infected cells can stop the virus replication process early on and kill the cells. This suggests that HIV-1 is vulnerable to metabolic disruptions, which could be exploited for therapeutic purposes. Accordingly, we have set out to identify and evaluate the relationship between HIV-1 infection and altered metabolism using an indicator of metabolic activity (NADH levels) and MOLT-4/CCR5 cells during HIV-1 infection. Toward this objective, eGFP expressing HIV-1 BaL was incorporated to assess HIV-1 infection correlated with NADH levels. The MOLT-4/CCR5 cells showed a positive eGFP signal upon being infected with eGFP expressing HIV-1 BaL pseudovirus ( Figure 4B ). Figures 4D, F, H showed a noticeable increase in the level of NADH intensity, average FLT of NADH, and longer FLT corresponding to the bound form of NADH for the HIV-1/eGFP infected MOLT-4/CCR5 cells at 48 hours post-infection. Next, HIV-1/eGFP infected MOLT-4/CCR5 cells expressing eGFP fluorescence were analyzed using ImageJ and compared with the corresponding NADH signal. ( Figures 4I, J ). A positive correlation between eGFP and NADH fluorescence is observed, indicating that higher infection of MOLT-4/CCR5 cells leads to an increased level of NADH signal ( Figure 4K ). The intensity decay of NADH from HIV infected cells is shown in Figure 4L . A bi-exponential fit was used to analyze the intensity decay, which yielded two decay times: 0.9 nsec (32%) and 4.1 nsec (68%). The fit had a χ² value of 1.06, indicating a good fit to the data. In contrast, a mono-exponential fit to the intensity decay of eGFP signal from the HIV infected cell recorded using a separate detection channel (see methods) exhibited a fluorescent lifetime of 2.6 nsec with a χ² value of 1.29. A single fluorescence lifetime of 2.6 nsec observed for eGFP expressed THP-1 cells detected in the 500 to 529 nm spectral window (which is the optimal emission spectral region for eGFP) confirms that the fluorescence signals emanated from eGFP molecules (15). The changes in NADH’s average FLT and NADH’s bound form for the HIV-1/eGFP infected MOLT-4/CCR5 cells compared to uninfected cells are quantitated ( Figures 4M, N ).

HIV-1 utilizes metabolically active cellular environments for establishing productive and latent HIV-1 infection in CD4+ T cells (51). Consequently, it is important to identify infected cells and tissues to study the relationship between metabolic activity and T cells in different tissues where HIV-1 replicates. Towards this goal, 2p-FLIM was applied to extract NADH binding profiles within infected tissues from spleen and lymph nodes of humanized NSG mice. 2p-FLIM studies with cryo-sectioned tissues from HIV-1 BaL-infected humanized NSG mice (39) are presented ( Figures 5 , 6 ). These experiments also include control spleen and lymph node sections. For each FLIM measurement with the HIV-infected tissue samples, we used a corresponding control uninfected mouse tissue section sample detected under identical settings (same excitation wavelength, incident laser power, detection emission filter, and image acquisition parameters). Label-free 2p-FLIM measurements with humanized NSG mouse spleen tissue display a substantially lower NADH level than the HIV-1 BaL-infected NSG mouse ( Figures 5A–C ). The average FLT and bound form of NADH FLT showed a longer lifetime with a higher fractional amount for the infected tissues, implying a consistent qualitative and quantitative approach to distinguish between uninfected versus HIV-1 infected tissues. The significant difference in NADH signal level is more evident for the HIV-1 BaL infected lymph node tissue section ( Figures 6B, G ). Quantitative analyses evaluating changes in NADH intensity and FLT are presented for infected versus control tissue samples. The two-tailed t-test showed a significant difference in both intensity and FLT of NADH between infected and control tissue sections ( Figures 6G–I ). A likely explanation of the significantly higher levels of NADH including the protein bound form of NADH observed in the infected lymph nodes compared to uninfected control is due to the fact that HIV-1 virions is harbored at these sites. This is a noteworthy observation from this study. A two-tailed test was performed for statistical analysis in panels (G, H & I) of Figure 6 yielded P values of * <0.05, and *** <0.001.

FLIM can be used to determine whether there are differences in NADH signal distribution in tissues (lymph node and spleen) in control versus HIV-1 BaL infected mice. The method is also useful to establish the correlation between infection sites versus metabolic distribution around the infectious sites. For determining the infection sites in the tissue, the antigen in the infected cryo-sectioned tissues is detected by a FITC-labeled p24 detection antibody. Our 2p-FLIM setup allows for the simultaneous detection of FITC and NADH in two different detection channels with a single excitation wavelength at 780nm and detection with an appropriate dichroic, short pass, and band pass filters. An example of two-photon imaging of p24-FITC and NADH with HIV-1 BaL infected NSG mouse lymph node (cells in the mesenteric lymph nodes of CD34+ hematopoietic stem cell reconstituted mice) tissue is shown in Figure 7 . Consequently, it creates a microscopic framework for contextualizing the “snapshot” tissue images of localized infection with metabolic information, as can be visualized in Figures 7A (p24-FITC), 7B (NADH intensity), and 7C (merged intensity images of p24-FITC and NADH). Thus, 2p-FLIM can be used to determine the correlation between the infection locus and the overall change in metabolic signature via NADH. Figures 7D, E represent NADH’s average FLT and NADH’s longer FLT (bound form) in the HIV-1 infected mouse lymph node tissue. The substantial increase and fraction of longer FLT of NADH correspond with an enzyme-bound fraction is consistent with an increase in oxidative phosphorylation (OxPhos) (4, 52, 53). Indeed, CD4+ T cells with elevated oxidative phosphorylation are selectively infected by HIV-1 (51). Furthermore, the label-free quantitation of endogenous NADH fluorescence by high-spatial and temporal imaging of virally infected systems at single organism, sub-cellular, cellular, and tissue levels provides a mechanistic view of the microbial-host interface involving perturbations in cellular metabolism for future studies ( Figure 8 ). The ability to determine discrete and quantitative changes of NADH lifetimes, label-free and with minimal invasion in cells and tissues, presents an excellent opportunity to evaluate virally induced shifts in metabolism throughout the infection life cycle.