A Link Between Methylglyoxal and Heart Failure During HIV-1 Infection

Early-onset heart failure (HF) continues to be a major cause of morbidity and mortality in people living with human immunodeficiency virus type one (HIV-1) infection (PLWH), yet the molecular causes for this remain poorly understood. Herein NOD.Cg-PrkdcscidIl2rgtm1Wjl/SzJ humanized mice (Hu-mice), plasma from PLWH, and autopsied cardiac tissues from deceased HIV seropositive individuals were used to assess if there is a link between the glycolysis byproduct methylglyoxal (MG) and HF in the setting of HIV-1 infection. At five weeks post HIV infection, Hu-mice developed grade III-IV diastolic dysfunction (DD) with an associated two-fold increase in plasma MG. At sixteen-seventeen weeks post infection, cardiac ejection fraction and fractional shortening also declined by 26 and 35%, and plasma MG increased to four-fold higher than uninfected controls. Histopathological and biochemical analyses of cardiac tissues from Hu-mice 17 weeks post-infection affirmed MG increase with a concomitant decrease in expression of the MG-degrading enzyme glyoxalase-1 (Glo1). The endothelial cell marker CD31 was found to be lower, and coronary microvascular leakage and myocardial fibrosis were prominent. Increasing expression of Glo1 in Hu-mice five weeks post-infection using a single dose of an engineered AAV2/9 (1.7 × 1012 virion particles/kg), attenuated the increases in plasma and cardiac MG levels. Increasing Glo1 also blunted microvascular leakage, fibrosis, and HF seen at sixteen weeks post-infection, without changes in plasma viral loads. In plasma from virally suppressed PLWH, MG was also 3.7-fold higher. In autopsied cardiac tissues from seropositive, HIV individuals with low viral log, MG was 4.2-fold higher and Glo1 was 50% lower compared to uninfected controls. These data show for the first time a causal link between accumulation of MG and HF in the setting of HIV infection.

Early-onset heart failure (HF) continues to be a major cause of morbidity and mortality in people living with human immunodeficiency virus type one (HIV-1) infection (PLWH), yet the molecular causes for this remain poorly understood. Herein NOD.Cg-Prkdc scid Il2rg tm1Wjl /SzJ humanized mice (Hu-mice), plasma from PLWH, and autopsied cardiac tissues from deceased HIV seropositive individuals were used to assess if there is a link between the glycolysis byproduct methylglyoxal (MG) and HF in the setting of HIV-1 infection. At five weeks post HIV infection, Hu-mice developed grade III-IV diastolic dysfunction (DD) with an associated two-fold increase in plasma MG. At sixteenseventeen weeks post infection, cardiac ejection fraction and fractional shortening also declined by 26 and 35%, and plasma MG increased to four-fold higher than uninfected controls. Histopathological and biochemical analyses of cardiac tissues from Hu-mice 17 weeks post-infection affirmed MG increase with a concomitant decrease in expression of the MG-degrading enzyme glyoxalase-1 (Glo1). The endothelial cell marker CD31 was found to be lower, and coronary microvascular leakage and myocardial fibrosis were prominent. Increasing expression of Glo1 in Hu-mice five weeks post-infection using a single dose of an engineered AAV2/9 (1.7 × 10 12 virion particles/kg), attenuated the increases in plasma and cardiac MG levels. Increasing Glo1 also blunted microvascular leakage, fibrosis, and HF seen at sixteen weeks post-infection, without changes in plasma viral loads. In plasma from virally suppressed PLWH, MG was also 3.7-fold higher. In autopsied cardiac tissues from seropositive, HIV individuals with low viral log, MG was 4.2-fold higher and Glo1 was 50% lower compared to uninfected controls. These data show for the first time a causal link between accumulation of MG and HF in the setting of HIV infection.
Keywords: HIV-1, heart failure, humanized mice, methylglyoxal, echocardiography INTRODUCTION Modern antiretroviral drug therapies (ART) have profoundly reduced morbidities and mortality in human immunodeficiency virus type one (HIV-1) infected individuals (1,2). However, by contemporary estimates more than 40% of persons living with HIV-1 infection (PLWH) on long-term combination ART therapies are developing early-onset heart failure (HF) (3)(4)(5)(6)(7). This disease which starts at least a decade earlier in PLWH compared to uninfected individuals and is independent of arteriosclerosis, and/or myocardial infarction. This HF also starts earlier in women than in men (6,(8)(9)(10)(11)(12)(13). To date, pharmacological strategies to blunt/slow the development HF in PLWH remain virtually non-existent, in part because of an incomplete understanding of the mechanisms involved.
Available data suggest that the pathophysiology of early-onset HF in PLWH is multifactorial, arising from persistent systemic immune activation, elevation in inflammation and oxidative stress, off-target effects of antiretroviral drugs, alcohol, aging, illicit drug use and the composition of the gut microbiome (5,(14)(15)(16)(17)(18)(19)(20)(21)(22). However, specific molecular pathways by which these cues negatively impact cardiac function are not well-defined. Some investigators using transgenic rodents have suggested that the HIV-1 auxiliary proteins Nef, gp120 and Tat contribute to the early-onset HF by impairing mitochondria and contractile functions of myocytes (23)(24)(25)(26)(27)(28). However, to the best of our knowledge, plasma levels of HIV viral proteins in PLWH on ART are significantly lower than that in the transgenic rodent models (29)(30)(31). Also, the low-grade systemic inflammation seen in PLWH is minimally observed in these transgenic rodents, raising concerns about the disease relevance of the latter. Whether EcoHIV mice, another model in which gp120 from HIV-1 is replaced with murine leukemia virus gp80 for cell entry, also develops HF remains unclear (32). Others have suggested that off-target effects of antiretroviral drugs, alcohol and illicit drug use, and accelerated aging could exacerbate traditional risk factors of cardiovascular diseases and HF potentiating dyslipidemia, hyperglycemia, and endothelial dysfunction (5,(14)(15)(16)(17)(18)(19)(20)(21)33). However, the effects of antiretroviral drugs, alcohol, and illicit drugs, and aging on cardiac function in the setting of HIV-1 infection are limited, making delineation of mechanisms that trigger HF development in HIV-1 setting challenging.
MG is a potent activator of the inflammation transcription factor, nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB), the NLR family pyrin domain containing 3 (NRLP3) inflammasome and mitochondria production of reactive oxygen species (ROS) (52,53). Glo1 expression is also negatively regulated by inflammation and oxidative stress (49,(54)(55)(56)(57). Thus, as MG levels increase it can activate NF-kB and the NRLP3 inflammasome, increase oxidative stress and downregulate Glo1, resulting in accumulation of MG. At high levels MG can also diffuse from infected immunocytes into the microenvironment. Earlier we showed that acute exposure of vascular endothelial cells and cardiac myocytes to supraphysiologic levels of MG, perturb their intracellular Ca 2+ homeostasis and increase ROS production (58,59). Long-term exposure of vascular endothelial cells to MG also diminished their responses to vasodilators and expression of tight junction proteins, established causes of microvascular leakage, decreased microvascular perfusion, fibrosis, and HF (58)(59)(60).
Earlier we showed that NOD.Cg-Prkdc scid Il2rg tm1Wjl /SzJ mice reconstituted with human CD34+ hematopoietic stem cells obtained from umbilical cord blood (Hu-mice) can be productively infected with the HIV-1 virus (61)(62)(63). These Humice also develop a progressive HF with microvascular leakage and ischemia, akin to that reported in PLWH (64)(65)(66)(67), suggesting commonalities in the pathogenesis of HF in Hu-mice and PLWH. Herein, we investigated if the microvascular leakage, fibrosis, ischemia, and HF seen in HIV-infected Hu-mice could be arising from accumulation of MG. Plasma from PLWH and cardiac tissues from deceased HIV-seropositive individuals with HF were also used to confirm elevation in MG.

General Characteristics of Hu-Mice
The general characteristics of the animals used in this study are shown in Table 1. Intraperitoneal injection of HIV-1 into Hu-mice led to productive infection; the plasma HIV viral loads four weeks post-infection was 1.5-1.64 × 10 6 RNA copies/mL and remained elevated during the 16-week period. The percentage of CD4+ T cells in blood declined in HIV-1 infected Hu-mice from 76.6 ± 0.4% to 55.2 ± 2.1% during the course of the study (Supplementary Figure S1A). CD4+ T cells in blood of HIV-infected Hu-mice treated with AAV2/9-Endo Glo1 also declined to 50.2 ± 3.1%. CD8+ T cells increased in blood of HIV-1 infected Hu-mice and HIVinfected Hu-mice treated with AAV2/9-Endo-Glo1 after 16 weeks of infection (Supplementary Figure S1B). The gating strategy used was CD45→ CD3→ CD4/CD8, and the total CD45 and CD3+ T cells did not notably change during the study (Supplementary Figures S1C,D). None of the animals used for this study had to be sacrificed prematurely due to weight loss or graft-vs. host disease.
AAV2/9-Endo-Glo1 Treatment Blunt Lefted Ventricular Function in HIV-1 Infected Hu-Mice Non-invasive, multi-modal echocardiography (pulsed-wave, tissue Doppler, M-mode, and speckle tracking) did not reveal any impairments in diastolic or systolic functions of the uninfected humanized mice at the start of the study. At five weeks post-infection, all hu-mice (males and females) developed DD characterized by a reduction in peak latediastolic transmitral velocity (A-wave), an increase in E:A ratio and an increase in E wave deceleration time [ Figure 1B (Figure 3A), and a decrease in reverse longitudinal strain (Pk, %) using the reverse peak algorithm ( Figure 3B) (63). After five weeks of infection, Mmode Doppler recordings did not show any significant changes in left ventricular systolic functions (percent fractional shortening (FS), percent ejection fraction (EF), left ventricular end diameterdiastole (LVED-diastole), left ventricular end diameter-systole (LVED-systole), and cardiac output (Figure 4).
After sixteen weeks of infection, pulsed wave and tissue Doppler revealed worsened DD in HIV-1-infected Hu-mice (males and females) with significant declines in E-wave, Awave and E-wave deceleration time (Figures 1, 2). L-waves were also pronounced in 4/6 HIV-1 infected Hu-mice [ Figure 1B(ii), middle right panel, yellow arrow]. E:A ratio, E:e' and IVRT also increased sixteen weeks post-infection (Figure 2) as did global longitudinal strain and global circumferential strain (Figures 3A,C). Reverse global longitudinal strain decreased further ( Figure 3B). Strain analyses of long and short axes B-mode images during systole revealed dyskinesis (expansion of a wall segment during systole Figure 4 left panels, red arrows) and dyssynchrony (opposite walls moving in counter directions, Figure 4, right panels blue arrows). After 16 weeks of infection, M-mode echocardiography revealed small but significant declines in FS, EF and cardiac output, and an increase in LVED-diastole and LVED-systole ( Figure 5).
A single intravenous injection of AAV2/9-Endo-Glo1 to express the Glo1 five weeks after HIV infection, attenuated impairments in myocardial diastolic and systolic dysfunctions that developed 16 weeks post-infection (Figures 2, 3, 5). Administration of AAV2/9-Endo-Glo1 also attenuated the dyskinesia and the dyssynchrony (Figure 4). Video recordings of parasternal long-and short-axis loops showing direction and magnitude of endocardial deformation between uninfected control, HIV-infected Hu-mice and HIV-infected mice treated with AAV2/9-Endo-1, are shown in Supplementary Videos Files, Videos 1-6. The non-specific virus AAV2/9-Endo-eGFP had no effect on diastolic and systolic parameters (data not shown). In this study, uninfected Hu-NSG mice also did not develop diastolic and systolic deficits during the 16-week study period. Animals used in this study also did not show physical signs of graft-vs-host disease including hair loss, hunch back or reduced mobility.
AAV2/9-Endo-Glo1 Treatment Blunted Plasma Elevation of MG and Semicarbazide-Sensitive Amine Oxidase (SSAO) in HIV-1 Infected Hu-Mice After five weeks of HIV-1 infection, MG level was 100.0 ± 6.8% higher in plasma of HIV-1 infected Hu-mice compared to uninfected controls. The activity of the non-selective inflammation enzyme SSAO (the soluble form of vascular adhesion protein-1, VAP-1) was also 38.7 ± 2.3% higher in plasma of HIV-1 infected Hu-mice compared to uninfected controls. After 16 weeks of infection, plasma MG increased further to 250.0 ± 10.3% and SSAO activity increased 84.2 ± 4.3%. A single intravenous injection of AAV2/9-Endo-Glo1 five weeks after HIV-1 infection, attenuated the increases in plasma MG and SSAO as observed at 16 weeks of infection with minimal change in HIV-1 viremia. Uninfected Hu mice showed minimal change in plasma MG and SSAO activity during the study period ( Table 1).

AAV2/9-Endo-Glo1 Treatment Blunted Leakage of Microvessels and Restored Vascular Perfusion in HIV-1 Infected Hu-Mice
In uninfected control animals, the green fluorescence of injected FITC-labeled BSA (FITC-BSA) appeared throughout the vascular network of mice, indicative of perfusion of the coronary microvessels. Larger diameter vessels which typically contain more blood had more FITC-BSA fluorescence compared to the smaller capillaries ( Figure 6A, yellow arrow). There was also minimal leakage of the FITC-BSA from the vasculature into the myocardium of uninfected mice. Interestingly, in ventricular tissues in HIV-1 infected animals, FITC-BSA fluorescent was present as "blobs" in some regions, indicative of leakage from the confines of the microvessels into the myocardium ( Figure 6A, white arrows). Intravenous administration of AAV2/9-Endo-Glo1 to Hu-mice five weeks after HIV-1 infection blunted the impairment in microvascular permeability and microvascular leakage seen in heart at seventeen weeks post-infection ( Figure 6A, lower right panel). Quantitation of microvascular leakage is shown in Figure 6B.
Myocardial tissues from hu-mice seventeen weeks following viral infection contained extensive perivascular and interstitial fibrosis. This was indicated by blue Masson's Trichrome

Duration of infection (weeks)
NA 16 16 Body weight (g) staining ( Figure 6C, middle panels, blue staining, yellow arrows), as compared to uninfected controls which exhibited minimal Masson's Trichrome blue staining ( Figure

AAV2/9-Endo-Glo1 Treatment Blunted Endothelial Cell Dysfunction in HIV-1 Infected Hu-Mice
Immunofluorescence staining was also conducted for the endothelial cell marker CD31, to assess the integrity of endothelial cells (ECs) in the coronary microvasculature. After seventeen weeks of infection, CD31 immunofluorescence was reduced by 27.8 ± 2.1% in ventricular sections from HIVinfected animals compared to uninfected controls ( Figure 8A, and graphs on right, and Figure 8B). Administration of AAV2/-Endo-Glo1 blunted the upregulation of the loss of CD31 seen in the myocardium of Hu-mice 16 weeks post-HIV-1 infection ( Figure 8A).

AAV2/9-Endo-Glo1 Treatment Restored Glo1 and Decreased VAP-1 Protein in Ventricular Homogenates From HIV-Infected Hu-Mice
Western blot analyses conducted to validate changes in ventricular levels of Glo1 and VAP-1 seen in immunofluorescence assays. In this study we found that after seventeen weeks of HIV-1 infection, cardiac Glo1 level was 40.2 ± 3.5% lower than that in control animals ( Figure 8C). Myocardial VAP-1 level also increased by 58.6 ± 6.5%. A single intravenous injection of AAV2/9-Endo-Glo1 to hu-mice five weeks after HIV-1 infection, increased myocardial Glo1 protein and attenuated the increase in VAP-1 as observed at seventeen weeks following viral infection ( Figure 8D).

Increased MG and SSAO in PLWH Plasma and Increased MG, Glo1, and VAP-1 in Autopsied Cardiac Tissues in HIV+ Individuals
The amount of MG and SSAO activity in plasma of PLWH, and MG-H1, Glo1 and VAP-1 in autopsied cardiac tissues from deceased HIV-1 sero-positive individuals with HF were assessed to validate the clinical significance of our pre-clinical findings. In plasma of PLWH (age 48.9 ± 3.5 years, duration of infection, 8.55 ± 2.12 years, plasma viral load < 20 RNA copies/ml, Supplementary Excel File, patient plasma), MG was 3.2-fold higher when compared to uninfected controls (1607.2 FIGURE 2 | Longitudinal pulse-wave and tissue doppler echocardiography before and after administration of AAV2/9-Endo-Glo1 to Hu-mice after HIV-1 infection reversed/attenuated the development of diastolic dysfunction. The panels below show (A) E:wave velocity, (B) A:wave velocity, (C) E:A ratio, (D) E:wave deceleration time, (E) E:e ratio and (F) isovolumetric relaxation time (IVRT) in uninfected control animals ( ) before (0), 5 weeks after and sixteen weeks after injection of saline; in HIV-infected mice ( ) before, 5 weeks and sixteen weeks after HIV-1 infection, and in AAV2/9-Endo-Glo1-treated HIV-infected Hu-mice ( ) before, 5 weeks after infection with HIV-1, and eleven weeks after administration of AAV2/9-Endo Glo1. Data shown on graphs are mean ± SEM from n ≥ 6 mice per group. * Denotes significantly different from uninfected Hu-NSG mice (p < 0.05). # denotes significantly different from HIV-1 infected Hu-mice (p < 0.05).
± 89.3 HSA-MG eq (µg/mL) compared to 502.6 ± 61.8 HSA-MG eq (µg/mL)). Mean SSAO activity was also 60% higher compared to uninfected controls (15.5 ± 1.6 units/mL/2 vs. 5.0 ± 0.5 units/mL/2 h, Supplementary Excel File, patient plasma). In autopsied ventricular tissues from HIV+ patients with HF MG-H1 was 4.2 ± 0.3-fold higher in vascular smooth muscle cells than control HIV-1 seronegative samples with no HF and 2.0 ± 0.2-fold higher than HIV-patients with atherosclerosis and HF (Figure 9, upper panels and graph bottom left, also see Supplementary Excel File, autopsied tissues). Glo1 levels were about 50% lower in cardiac tissues from HIV+ with HF as compared to HIV-with no HF and 35 ± 5.1% lower than HIV-group with atherosclerosis and HF (Figure 9, middle panels and middle graph). VAP-1 was 5.0 ± 0.3-fold higher in HIV-1 seropositive group with HF as compared to HIV-1 seronegative with no HF and 3.1 ± 0.2fold higher in HIV-1 seronegative patients with atherosclerosis and HF.

DISCUSSION
HF remains a major cause of morbidity and mortality in >40% of PLWH (4-13). Drugs to blunt or slow the progression of HF in PLWH are not available in part to an incomplete understanding of its underlying cause(s). Herein, we demonstrate for the first time that elevation of the cytotoxic glycolysis byproduct MG is an initiating cause for HF during HIV-1 infection. This conclusion is based on new findings obtained from infected Hu-mice, plasma from PLWH and autopsied cardiac tissues from deceased HIV-1 seropositive individuals with HF. MG has been previously linked to HF in other inflammatory diseases (60,68).
In this study we confirmed our HIV-infected Hu-mice developed a progressive HF using longitudinal echocardiography and histopathological analyses (63). At five weeks postinfection, HIV-infected Hu-mice developed diastolic dysfunction with/without mitral regurgitation that progressed to systolic dysfunction as the duration of infection increased (63). We also found for the first time a progressive increase in plasma MG with worsening HF. Prior studies have reported that following HIV-1 infection, immunocytes upregulate glucose transporter 1 (GLUT1) and increase aerobic glycolysis to generate the necessary substrates needed for HIV-1 replication (43)(44)(45)(46). Since 0.1% of glucotriose flux is converted to MG (49,50), an increase in glycolysis in immunocytes could account in part for the increase in MG. However, additional work is needed to confirm this. Accumulation of MG in plasma and cardiac tissues could also arise from a reduction in its degradation. Using immunofluorescence and Western blot assays, in this study we found for the first time a reduction in Glo1 protein in hearts of Hu-mice seventeen weeks post-infection.
These data indicate that the accumulation of MG during HIV-1 infection is also arising in part from a reduction in its degradation.
Next, we assessed if increasing expression of Glo1 would blunt accumulation of MG in HIV-infected hu-mice. Under non-stressed conditions, the nuclear factor erythroid 2-related factor 2 (Nrf2) binds to the antioxidant response element (ARE) on the promoter region of the Glo1 gene induces expression of Glo1 (54,69). However, under inflammatory as is the case in HIV-1 infection, activated NF-κB would compete with Nrf2 to suppress Glo1 expression (49,56,57). As such, to induce expression of Glo1 under inflammatory conditions, we replaced the endogenous CMV promoter of AAV2/9 with the promoter of inflammation-induced protein endothelin-1 (58,60). Using this strategy, we show for the first time that increasing expression of Glo1 in hearts of HIVinfected Hu-mice decreased MG and blunted the HF seen 16 weeks post-infection, establishing that elevated MG is an underlying cause of HF in HIV-1-infected Hu-mice. It should be mentioned that in this study AAV2/9-Endo-Glo1 was not given to uninfected Hu-mice as these mice have low systemic and tissue inflammation, and as such this viral construct would not express Glo1 (58,60).
Studies were then conducted to delineate mechanisms by which elevated MG elicited HF during HIV-1 infection. First, we found that increasing Glo1 expression blunted the loss of CD31, and microvascular leakage seen in HIV-1 infected Humice, indicating that Glo1 was protected vascular endothelial cells (ECs) in the heart from MG insults. Earlier we showed that microvascular ECs are especially susceptible to elevation in plasma MG due to their low expression of Glo1 (58). We also showed that chronic exposure of microvascular ECs to MG decreased expression of their tight junction proteins (58). Second, we found that increasing Glo1 in the heart of HIV-1infected hu-mice attenuated perivascular and interstitial fibrosis (60) indicating that the myocardial fibrosis that developed in hearts of HIV-infected Hu-mice were linked to elevated MG. Although the specific molecular pathways by which elevated MG induce fibrosis is not clear, we posit that MG-induced reduction in tight junction proteins would increase extravasation of blood substances and immunocytes into the cardiac interstitium, triggering inflammation, activating matrix metalloproteinases and the deposition of collagen fibers (70). Third, in this study we found that increasing Glo1 expression in hearts of HIV-1 infected mice blunted plasma SSAO and cardiac VAP-1 expression, consistent with the notion that elevation in MG is contributing to systemic and cardiac inflammation (71-73). However, additional studies are needed to determine the causeeffect relationships between MG, cardiac inflammation, NF-κB and NRLP3 inflammasome activation (52,53).
Plasma from PLWH and autopsied cardiac tissues were assayed for MG, Glo1, SSAO, and VAP-1 to define the clinical relevance of our findings. In plasma from PLWH with low HIVviremia and autopsied tissues from deceased HIV-1 seropositive individuals, MG levels were 3.7-fold and 4.2-fold higher than  that in uninfected controls, respectively. SSAO was also 60% higher in plasma of PLWH compared to uninfected controls. In autopsied cardiac tissues from deceased HIV+ individuals, VAP-1 was 5.0-fold higher and Glo1 was 50% lower as compared to uninfected controls without HF. Thus, the elevation in MG and the reduction in Glo1 observed in Hu-mice is of clinical relevance to HF for PLWH.
There are some limitations with the present study. First, although our data show that MG (measured as its surrogate MG-H1) is elevated in plasma and cardiac tissues from HIVinfected Hu-mice, PLWH and autopsied cardiac tissues from HIV-seropositive deceased individuals, the underlying cause for this is not well-delineated. In the present study, we focused on expression of Glo1. However, Glo-1 degrades the hemiacetal formed between MG and reduced glutathione, GSH (49,50). Thus, a reduction in GSH could result in MG accumulation. As such, additional studies will be needed to determine total glutathione, the ratio of reduced and oxidized glutathione, and the activities of the two enzymes involved in the synthesis of GSH, namely γ-glutamylcysteine ligase (ligates L-glutamate and L-cysteine), and glutathione synthetase (adds glycine to γglutamylcysteine) in hearts of HIV-infected Hu-mice. Second, ischemic regions were observed in hearts of HIV-1 infected Hu-mice as well as in PLWH (63,66,67). Under normoxia, the heterodimeric transcription factor that regulates cellular and systemic adaptive responses hypoxia-inducible factor 1α (HIF-1α) is targeted for degradation by the proteasome via hydroxylation of proline residues, mediated by the oxygendependent prolyl hydroxylase domain (PHD) family of enzymes (74,75). When oxygen delivery is compromised, as is the case with ischemia, the PHD enzymes are inhibited, and HIF-1α escapes hydroxylation, allowing it to migrate to the nucleus and induce transcription of HIF-1 target genes, including those involved in glycolysis and erythropoiesis. HIF-1α also binds to the antioxidant response element (ARE) on the promoter region of the GLO1 gene to inhibit Glo1 expression (49,56,57). Thus, ischemia and activation of HIF-1α will also inadvertently lead to an increase in MG. Additional work will also be needed to investigate if inhibitors of HIF-1α are cardio-protective in the setting of HIV-1 infection. Third, the present study focused on EC dysfunction and microvascular leakage, vascular changes that are known mediators of inflammation, fibrosis, and HF. However, supraphysiologic level of MG can also perturb intracellular Ca 2+ , induce reactive oxygen species (ROS), and form adducts on accessible basic moieties of proteins that could negatively impact the function of cardiac myocytes (59). In the future, we will assess if myocytes from HIV-1 infected Hu-mice have impaired Ca 2+ homeostasis, increased ROS production, and diminished contractile properties.
In conclusion, the present study shows for the first time that early-onset HF seen in HIV-1-infected Hu-mice is arising in part from accumulation of the cytotoxic glycolysis metabolite MG. We also showed that this elevation in MG is arising in part from a decrease in its degradation and precipitating dysregulation of coronary microvascular ECs, microvascular leakage, and fibrosis (Figure 10). We posit that an elevation in MG could also be an underlying cause increase in systemic and cardiac inflammation seen during HIV-1 infection by activating NF-κB and NRLP3 (4,5,(19)(20)(21). These data also suggest that therapeutic strategies to lower MG levels may be useful in reducing inflammation and HF during HIV-1 infection.

Ethics Statement
All experimental protocols involving the use of laboratory animals were approved by the University of Nebraska Medical Center (UNMC) Institutional Animal Care and Use Committee (IACUC) ensuring the ethical care and use of laboratory animals in experimental research. All animal studies were performed in compliance with UNMC institutional policies and NIH guidelines for laboratory animal housing and care. Human CD34+ Hu-NSG were isolated from umbilical cord blood obtained from UNMC labor and delivery department at UNMC with written consents from adult parents to use the remaining or discarded biological material for research. Samples were collected without identifiers under UNMC Institutional Review Board (IRB) exempt. The UNMC institutional IRB determined that these studies using anonymized cord blood samples doesn't constitute human subject research as defined at 45CFR46.102(f). We regularly collect the cord blood samples and isolate CD34+ hematopoietic cells which are either injected immediately into mice or stored in liquid nitrogen for future human reconstitution. NOD.Cg-PrkdcscidIl2rgtm1Wjl/SzJ, NSG mice were obtained from the Jackson Laboratories (Bar Harbor, Maine, USA; stock number 005557), and a breeding colony was developed at the University of Nebraska Medical Center. All animal procedures are approved under the IACUC protocols 18-110-08 and 10-107-01 for ECHO procedures (61)(62)(63).

Autopsied Ventricular Tissues From Deceased HIV+ Individuals and Plasma From PLWH
De-identified, autopsied left ventricular tissues on glass slides (consecutive sections) from seven de-identified HIV+ patients with myocardial dysfunction (4 males and 3 females) were obtained from the National NeuroAIDS Tissue Consortium (NNTC), under approved protocol # R605. Age, HIV-1 infection duration, plasma viral load and ARDs taken are in Supplementary Excel File-a. Deidentified autopsied cardiac tissues from seven patients with cardiac hypertrophy and atherosclerosis (4 males and 3 females) and from seven uninfected individuals who died in accidents with no reported history of heart diseases were obtained from the tissue bank at the UNMC. De-identified plasma from ten HIV-infected and uninfected "control" patients were also obtained from the UNMC tissue bank. Plasma viral load and ART taken are in Supplementary Excel File-b.

Generation of Humanized Mice
Humanized mice (Hu-mice) were prepared as described in prior publications (61)(62)(63). Shortly after birth, NOD.Cg-Prkdc scid Il2rgt m1Wjl /SzJ mice were briefly irradiated with a sub-lethal dose of radiation (1Gy) using a RS-2000 X-Ray Irradiator (Rad Source Technologies). CD34+ cells (50,000 cells/mouse) enriched from human cord blood (>90%) were then injected intra-hepatically and left for humanization. At monthly intervals mice were bled via a submandibular vein into EDTA-coated tubes and screened for human immune cells using flow cytometry (LSR-II FACS analyzer, BD Biosciences, Mountain View, CA, USA). The CD45 percentage in humanized mice used in this study ranged from 25 to 50%. About 5% of our Hu mice develop graft vs.-host disease (GVHD) and these mice were excluded from our study.

Infection of Hu-Mice With HIV-1
Twenty weeks after humanization, 12 Hu-mice were infected intraperitoneally (IP) with 2 × 10 4 tissue culture infectious dose 50 (TCID 50 ) of HIV-1 ADA (a macrophage tropic viral strain) (61)(62)(63). An additional six Hu-mice served as uninfected agedmatched controls. Peripheral blood samples were collected every four weeks via submandibular vein bleeding to assess HIV-1 viral RNA and to assess the dynamics of human immune cell markers by flow cytometric analysis. Plasma HIV-1 RNA levels were measured using an automated COBAS Ampliprep V2.0/Taqman-48 system (Roche Molecular Diagnostics, Basel, Switzerland) as per the manufacturer's instructions. The detection limit after dilution factor adjustment was 200 viral RNA copies/mL.

Glo1 Gene Transfer to HIV-1 Infected Hu-Mice
Five weeks post-HIV-1 infection, Hu mice were divided into two groups. Animals in Group-1 received a single intravenous injection of AAV2/9-Endo-Glo1 (1.7 × 10 12 virion particles/kg in sterile physiologic saline solution), while animals in Group-2 received saline (58,60). Uninfected animals were injected with saline and were kept for the whole duration of the study (4 months total). The dose (multiplicity of infection) of AAV used in this study was selected from our earlier (60) and other previous studies (76,77). AAV2/9 was selected because of its high tropism for cardiac myocytes and endothelial cells (76,77). The endothelin-1 promoter was used to induce expression of Glo1 in these cells under inflammatory conditions.

Assessment of Longitudinal Cardiac Function Using Echocardiography
Transthoracic conventional echocardiography was performed using a Fujifilm VisualSonics Vevo 3100 system (Fujifilm VisualSonics, Toronto, ON, CAN), employing a MX550D transducer with a center frequency of 40 Hz and an axial resolution of 40 µM, prior to, 5 and 16 weeks after infection with HIV-1 or saline injection (63). For this, hair on chests of mice were removed (Nair, Church & Dwight Co., Inc. NJ, USA). Twenty-four hours later, mice were anesthetized with 1-2% isoflurane (Cardinal Health, Dublin OH, USA) and taped in the supine position on a heated 37 • C pad. Anesthesia was maintained with 0.5-3% isoflurane via a nose cone. Feet of mice were connected to ECG leads, and pulsedwave Doppler images were acquired in the apical four chamber view with appropriate stage tilt and probe tilt to acquire maximum flow and digitally stored in cine loops. The offline Program Vevo LAB 3.1.1 was then used to assess peak earlyand late-diastolic transmitral velocities (E and A waves), E- Early diastolic tissue relaxation velocity (E') was measured using tissue Doppler, and E:e ′ ratio was calculated. All ultrasound imaging and analyses were done in a blinded manner then decoded for statistical evaluation.

Speckle Tracking
Parasternal long-axis and short axis B-mode echocardiographic images were obtained at a rate of >300 frames/second using the Fuji VisualSonics Vevo 3100 system and digitally stored in cine loops (60). Vevo LAB 3.1.1 was used to determine global longitudinal, radial, and circumferential strain using three to four consecutive cardiac cycles. The Vevo Strain Software was used to determine longitudinal, radial strain/strain rates, dyskinesis, and dyssynchrony during systole using six segment (anterior base, AB; anterior middle, AM; anterior apex, AP, posterior base, PB; posterior middle, PM; and posterior apex PA) analyses. All analyses were done in a blinded manner but decoded for statistical analyses.

Microvessels Perfusion and Permeability in Cardiac Tissue
One week after the last echocardiographic measurement (17 weeks post-infection), half of mice from each group were injected with fluorescein isothiocyanate-labeled bovine serum albumin (FITC-BSA 40 mg/kg in sterile 1X PBS buffer, 50 µL) via tail vein (58,60) and was allowed to circulate for 10 min, after which animals were anesthetized with 5% isoflurane. Chest cavities were opened and hearts were quickly removed and immersed in 4% paraformaldehyde for 24 h at 4 • C. Hearts were then cut longitudinally into three sections, and the right 1/3 was transferred to 4% paraformaldehyde/15% sucrose solution for 24 h, followed by 4% paraformaldehyde/30% sucrose solution for 24 h, and then 30% sucrose solution for 24 h. Cryoprotected hearts were cut into 20 µm thick longitudinal/coronal sections on a microtome (Leica EM-UC 6, Leica Microsystems, Wien, Austria) and mounted onto precleaned glass slides. Cardiac sections were then washed three times with 1X PBS to remove cutting medium. Vectashield TM mounting medium containing DAPI was added to the sections, and slides were cover slipped and dried overnight. Next day slides were placed on the head stage of a Nikon TE2000 microscope attached to a Coolsnap HQ2 CCD camera (Photometrics, Tuscon AZ, USA) and images were collected to assess the density of microvessels perfused with BSA-FITC and microvascular leakage. Analyses were done in a blinded manner but decoded for statistical evaluation.

Masson-Trichrome and Immunofluorescence Staining in Cardiac Tissue
The left longitudinal 1/3rd of hearts from Hu-mice were also placed in 4% paraformaldehyde for 24 h and then processed and embedded in paraffin as described earlier (63). Five micrometer sections were then cut and placed onto glass slides. Slides were de-paraffinized with xylene (3 changes, ten minutes each) and rehydrated in decreasing concentrations of ethanol (100, 95, 70% and distilled water, three minutes each) followed by a saline wash. Masson Trichrome staining was conducted using rehydrated sections without modification to assess fibrosis (Sigma-Aldrich, St Louis, MO, USA). Sections were then cover slipped with Prolong Gold Anti-fade reagent. Images were then taken with a Nikon inverted fluorescence microscope (TE 2000) equipped with a CoolSNAP HQ2 CCD Camera (Photometrics, Tucson, AZ, USA). Image analysis software quantitated changes. This was completed in a blinded manner then decoded for statistical evaluation. Immunofluorescence staining was performed on cardiac tissues from Hu-mice (uninfected, HIV-1-infected and HIV-1-infected and treated with AAV2/9-Endo-Glo1) and from autopsied cardiac tissues obtained from uninfected and HIV-1 seropositive persons to determine the levels of the MG (hydroimidazolone isomer 1), Glo1, CD31 (measure of endothelial cell) and VAP-1 (a non-selective, inflammation-induced protein). Calponin-related protein (SM22α), a marker of contractile smooth muscle cells, served as reference to define microvessels. Primary antibodies were used at concentrations of 1:100 to 1:200 while secondary antibodies concentrations were 1:250 to 1:500. Horse serum (10%) was used as the blocking agents to reduce non-specific interactions. Images were taken with a Nikon inverted fluorescence microscope (TE 2000). Nikon Elements image analysis software was used to quantitate changes of VAP-1, MG-H1 and Glo1 immunoreactivities using 20× frames.

MG and Semicarbazide-Sensitive Amine Oxidase (SSAO) in Plasma
MG levels (as surrogate MG-H1, OxiSelect TM Methylglyoxal Competitive ELISA, Cell Biolabs Inc, San Diego CA) in plasma were measured from Hu-mice (6 and 16 weeks, uninfected, HIV-infected and AAV2/9-Endo-Glo1-treated), and from uninfected healthy controls and PLWH as per manufacturers' instruction inside a BSL2+ Facility at UNMC. The activity of the non-selective inflammation marker, semicarbazide-sensitive amine oxidase (the soluble form of VAP-1) in plasma from Hu-mice was measured from (uninfected, HIV-infected and AAV2/9-Endo-Glo1-treated), and from uninfected healthy controls and PLWH using Fluoro-SSAO TM (Cell Technology, Mountain View CA) as per manufacture's instruction.

Glo1 and VAP-1
Glo1 and VAP-1 levels in ventricular homogenates were also determined using Western blot assays (59). For this, cardiac tissues (50 mg) were chopped into small pieces, placed into 200 µL of cell lysis buffer (MicroRotofor Cell Lysis Kit (mammals), BioRad Inc., Burlingame CA) and sonicated 3 × 3 s with 10 s intervals on ice in between. Samples were then centrifuged at 3,000 × g for 5 min and the supernatants were collected and protein concentration was determined using Bradford Protein Assay Kit (BioRad Inc., Burlingame CA). Western blots were then carried out as described in earlier publications. Primary antibody concentrations were used at 1:1,000 dilutions and incubated for 16 h at 4 • C and secondary antibody were used at 1:2,00 and incubated for 2 h at room temperature. β-actin served as the internal control to correct for variations in sample loading.

Statistical Analyses
Data were analyzed using GraphPad Prism 7.0 software (La Jolla, CA) and presented in text as the mean ± the standard error of the mean. All experiments listed in this manuscript were performed using a minimum of three biologically distinct replicates. One-way ANOVA with Bonferroni correction for multiple comparisons were used. For studies with multiple time points, twoway factorial ANOVA and Bonferroni's post-hoc tests for multiple comparisons were performed. Studies were from six animals per group. Significant differences were determined at p < 0.05.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Human CD34+ Hu-NSG were isolated from umbilical cord blood obtained from UNMC labor and delivery department at UNMC with written consents from adult parents to use the remaining or discarded biological material for research. Samples were collected without identifiers under UNMC Institutional Review Board (IRB) exempt. The UNMC institutional IRB determined that these studies using anonymized cord blood samples doesn't constitute human subject research as defined at 45CFR46.102(f). We regularly collect the cord blood samples and isolate CD34+ hematopoietic cells which are either injected immediately into mice or stored in liquid nitrogen for future human reconstitution. The patients/participants provided their written informed consent to participate in this study. The animal study was reviewed and approved by University of Nebraska Medical Center (UNMC) Institutional Animal Care and Use Committee (IACUC). All animal procedures are approved under the IACUC protocols 18-110-08 and 10-107-01 for ECHO procedures. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.

AUTHOR CONTRIBUTIONS
KB conceived the experiments and planned along with all other authors. SG created humanized mice. PD, FA, BH, and KB conducted echocardiography and histopathological assays and interpreted the data sets for the humanized mice. PD and KB performed molecular, virologic and immunological studies, and related data analyses. JC, HF, and BM assisted with obtaining the de-identified human plasma and tissues. KB, EM, and SG performed and analyzed the microvascular permeability and fibrosis tests and analyses. FA, PD, JM, JC, BM, HG, HF, and SG interpreted the data and wrote the manuscript with editing. All authors contributed to the article and approved the submitted version.

FUNDING
This work was supported in part by a pilot project from UNMC Center for Chronic HIV infection and Aging in NeuroAIDS (NIH P30 MH062261) and R56 HL151602-01A1.

ACKNOWLEDGMENTS
The authors are thankful to Hang Su for technical assistance, Emiko Waight for proof-reading the article and Lili Guo for generation of humanized mice.

SUPPLEMENTARY MATERIAL
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcvm.

2021.792180/full#supplementary-material
Supplementary Figure S1 | Time frame of observation of immune status of animals used in the study.
Supplementary XLSX | Showing (Sheet-1)-Patient data details from whom Plasma is analyzed. (Sheet-2)-Patient data details from whom autopsied tissues were used for the study.
Video 1 | Uninfected control Hu-mice Video Short axis.
Video 2 | Uninfected control Hu-mice Video Long axis.