In this study, our goal is to test if the LC-MS/MS technology can monitor HIV-1 induced cellular metabolic alterations in CD4+ T cells and macrophages. To test this we compared HIV-1 infected versus non-infected primary human CD4+ T cells from the same donor (). CD4+ T cells were infected with NLEGN-1 HIV-1 virus, which is the infectious HIV-1 NL4-3 stain that has all viral genes expressed in addition to GFP (Kutsch et al., 2002
). CD4+ T cells were FACS-sorted into GFP+ (HIV-1 producing) and GFP− cells from the same donor, achieving 7–20% infection at the time of sorting. All CD4+ T cell experiments compared infected versus uninfected T cells at the same time for individual donors.
Figure 1 Diagram of methods. A) Primary CD4+ T cells were expanded in culture in the presence of 5 μg/ml PHA and 10 ng/ml IL-2. After five days the cells were infected with NLEGN-1 HIV-1 virus for 24h before FACS sorting the cells into GFP+ (HIV-1 producing (more ...)
Due to the nature of macrophages being difficult to infect completely with HIV-1 and their tendency not to re-adhere once detached, we utilized a cell line model to mimic human HIV-1 infected macrophages (). The human promonocytic U937 cells are the most commonly used HIV-1 monocyte/macrophage model system (Cassol et al., 2006
). U1 cells are chronically HIV-1 infected U937 cells (Folks et al., 1987
) and are widely used as a model system to study HIV-1 biology in macrophages (Bristow et al., 2008
; Cassol et al., 2006
; Fernandez Larrosa et al., 2008
; Olivares et al., 2009
; Spector and Zhou, 2008
). U937 and U1 cells were differentiated into macrophages using Vitamin D3 and PMA treatment (). U1 cells produced virus within 2 days as measured by intracellular expression and secreted p24 (Supplemental Data 2a and b
). In addition, U1 cells remained alive and have comparable cell death to U937 cells (Supplemental Data 2c and d
), indicating that HIV-1 production does not induce apoptosis for U1 cells, which is similar to what we observe with primary human macrophages.
A basic overview of metabolic pathways is illustrated in showing the many different pathways that glucose can travel once entering the cell (Wamelink, Struys, and Jakobs, 2008
). First, we used tritiated 2-deoxyglucose (2-DOG) to monitor the cellular capacity for glucose uptake (Reusch et al., 1991
). As shown in and Supplemental Data 3a
, HIV-1 infected CD4+ T cells (GFP+) from three different human donors have a significant increase in 2-DOG uptake as compared to GFP− CD4+ T cells (control) from the same donor under the same culture conditions. In contrast, when we examined glucose uptake of the U937 and U1 macrophage model system, and observed that U1 macrophages have a significant reduction in 2-DOG uptake as compared to U937 macrophages ( and Supplemental Data 3b
). Since U1 and U937 cells have been passaged for many years, there may be additional, subtle changes that have occurred to both cell lines. However, both U1 and U937 cells have comparable 2-DOG uptake before differentiation (Supplemental Data 3c
), suggesting no predisposed gross changes in glucose uptake in latently infected U1 cells is observed. Overall, the absence of glucose uptake enhancement in U1 macrophages (HIV-1 producing) versus U937 macrophages implies that HIV-1 production does not induce additional glucose uptake during virus production. This finding is in contrast to our HIV-1 infected CD4+ T cell data () and the observations of another lytic viral infection (Munger et al., 2006
; Munger et al., 2008
Figure 2 Diagram of metabolic pathways. FBP is the major regulatory metabolite for glycolysis. It influences its own production and pyruvate, to stimulation more glucose consumption. Pyruvate can be converted into lactate or further be processed to enter the TCA (more ...)
Next we used LC-MS/MS to examine steady state levels of metabolic intermediates during viral infection. For HIV-1 infected CD4+ T cells and macrophage model system, a total of 66 metabolic small molecules were identified during steady state analysis (Supplemental Table 1
: complete list). Our data for HIV-1 infected CD4+ T cells (analyzed as ratio of GFP+/GFP) indicate that several glycolytic key metabolites: hexose-P, fructose 1,6-bisphosphate (FBP), glyceraldehyde-3P (G3P) and 3-phosphoglycerate (3PG) have significantly increased pool sizes by T-test analysis (; shown as Log2
converted data). This data together with the increased glucose uptake (), suggest an increased rate of glycolysis for HIV-infected CD4+ T cells. These observations parallel previous findings in which increases in glycolytic metabolite concentrations correlated with increased glycolytic flux (Munger et al., 2006
; Munger et al., 2008
Figure 3 LC-MS/MS analysis of glycolysis and TCA cycle metabolic pools. A) LC-MS/MS analysis of glycolytic intermediates for CD4+ T cells (red) and macrophages (blue) are shown as fold change in data. Metabolites that have a significant difference are denoted (more ...)
Next, we examined metabolic pools in macrophage model system (analyzed as the ratio of U1/U937) and found significant reductions in glycolytic intermediates: hexose-P, FBP and G3P (). Coupled with the decrease in glucose uptake (), these results suggest an HIV-induced suppression of glycolysis in macrophages. One surprising aspect of these data is the increase in pyruvate pool size despite the reduction of glucose uptake and decreased FBP levels. This could reflect either decreased pyruvate consumption by the TCA cycle in macrophages or perhaps increased pyruvate production from other carbon sources such as amino acid oxidation. Regardless, these data indicate that HIV-1 infection in CD4+ T cells and macrophages has very different metabolic outcomes.
Next, we examined metabolite pool sizes for the TCA cycle. As shown in , aconitate and isocitrate are significantly increased for HIV-1 infected CD4+ T cells, while HIV-1 producing macrophages have a significantly larger malate pool. Collectively, the lack of detectable changes in multiple metabolite pool sizes suggests that the viral impact on the TCA cycle is not analogous to its impact on glycolysis. Ringrose et al.
reported that several TCA cycle proteins are upregulated during late HIV-1 infection (Ringrose et al., 2008
). While we do detect substantial changes to a few TCA metabolite pool sizes, we cannot clearly delineate this impact on the TCA cycle. Moreover, our data does not address HIV-modulation of TCA cycle flux or changes in metabolite pool sizes that feed into and out of the TCA cycle for other biosynthesis pathways. These data are in contrast to earlier findings for HCMV infection which showed increases in TCA cycle intermediates (Munger et al., 2006
; Munger et al., 2008
), suggesting that evolutionarily divergent viruses interact differently with the host-cell metabolic network. Alternatively, activation of glycolysis is sufficient to support HIV-1 production in CD4+ T cells, while HCMV, having a larger DNA genome (230 kb) as compared to the small RNA genome (9.5 knt) of HIV-1, may require greater metabolic resources thus promoting the activation of both glycolysis and the TCA cycle.
Next we examined metabolite pool sizes. For the PPP, HIV-1 infected CD4+ T cells have a significant increase in sedoheptulose 7-P (S7P; non-oxidative PPP) and ribose-P, whereas macrophages have a significant decrease in 6-P-gluconate (6PG; oxidative PPP) and no significant change in S7P and ribose-P (). These data suggest that infected CD4+ T cells may use the non-oxidative PPP to generate ribose at this late time point after infection. In addition, the PPP metabolite 6PG level is not significantly different (Log2
= −0.09). These data are consistent with Chan et al
.; they showed no difference at the enzyme level for this metabolite at a late time point of infection (Chan et al., 2009
). HIV-1 has an RNA genome; therefore we also examined nucleotide triphosphates (NTP) pool sizes in CD4+ T cells and macrophages. We recently reported macrophages have slightly lower NTP pools than activated PBMC, and the concentrations of pyrimidines are lower than purines (Kennedy et al., 2010
). As shown in , no significant differences in NTP pool sizes were detected for CD4+ T cells. Yet for the macrophage model system, the pyrimidines: CTP and UTP pool sizes were significantly increased. We detected a decrease in GTP, yet this did not reach the level of significant difference using paired T-test.
Figure 4 LC-MS/MS analysis of metabolites used in viral production. A) PPP pathway was examined for CD4+ T cells (red) and macrophages (blue). Significantly different metabolites are indicated by asterisks. B) NTP pools were examined and found to change more for (more ...)
We next examined the ATP/ADP, NAD+/NADH and NADP+/NADPH ratios ( and Supplemental Data 4
). For both cell populations, the ATP/ADP and NAD+/NADH ratios were not significantly different. These data supported our conclusion of very little change in the TCA cycle by HIV-1 infection. It has been well established for HIV-1 and other viruses that the redox levels: glutathione and NADPH, change within the infected cells (Gu et al., 2001
; Jana and Pahan, 2004
; Korenaga et al., 2005
; Sagi and Fluhr, 2001
). The NADP+/NADPH ratio for macrophages was significantly different, but not for CD4+ T cells (). This data and the 6PG PPP metabolite data suggest an overall reduction in the ability to generate NADPH. Furthermore, these data suggest that HIV-1 producing U1 macrophages may be compromised in their ability to generate de novo
nucleotides and fatty acids. This can be addressed in future studies using more tradition biochemical assays.
Finally, we examined the UDP-sugar pools within the different cell types. UDP-sugars are metabolic intermediates used in glycosylation of proteins. HIV-1 gp120 and gp41proteins are highly glycosylated, are key targets for antibody binding to prevent infection, and are critical for HIV-1 pathogenesis (Montefiori, Robinson, and Mitchell, 1988
; Pal, Hoke, and Sarngadharan, 1989
; Raska et al., 2010
). We detect a significant increase in UDP-glucose pool size for CD4+ T cells (). UDP-D-glucurononate pool size was not significantly changed in either CD4+ T cells or macrophages. UDP-N-acetyl-glucosamine pool size was significantly increased in both CD4+ T cells and macrophages, whereas N-acetyl-glucosamine-1P was significantly increased in macrophages only (). These data indicate that UDP-sugar pathways may be potential targets for metabolic chemical inhibitors, since they are increased in both cell types. Future experiments will need to be done in order to confirm that reducing UDP-sugar synthesis has an affect on HIV-1 viral replication.
Clearly our data indicates that LC-MS/MS technology is attractive when measurement of multiple metabolic intermediates is desired. Our strongest data using LC-MS/MS comes from analysis of glycolytic intermediates (), which will require additional future studies to examine both RNA and protein levels under these conditions. The limitations of these results are 1) the relative low detection sensitivity presumably due to the limited biomass sizes, compared to the previous HCMV data, particularly in PPP and TCA pathways (Munger et al., 2006
; Munger et al., 2008
), 2) the potential delay of sample preparation for suspension cells, which requires centrifugation before lysis, which may be problematic for metabolic kinetic readouts (i.e. flux analysis), as compared to adherent cells which can be directly methanol lysed on the plates, and finally, 3) the difficult nature of primary human macrophage infection from donor to donor. These technical limitations partially attribute cell types used during this study, and require further work.