Role of myeloid cells in system-level immunometabolic dysregulation during prolonged successful HIV-1 treatment

Objective: Why people with HIV-1 on ART (PWHART) display convoluted metabolism and immune cell functions during prolonged suppressive therapy is not well evaluated. In this study, we aimed to address this question using multiomics methodologies to investigate immunological and metabolic differences between PWHART and HIV-1 negative individuals (HC). Design: Cross-sectional study Methods: Untargeted and targeted metabolomics was performed using gas and liquid chromatography/mass spectrometry, and targeted proteomics using Olink inflammation panel on plasma samples. The cellular metabolic state was further investigated using flow cytometry and intracellular metabolic measurement in single-cell populations isolated by EasySep cell isolation. Finally, flow cytometry was performed for deep-immunophenotyping of mononuclear phagocytes. Results: We detected increased levels of glutamate, lactate, and pyruvate by plasma metabolomics and increased inflammatory markers (e.g. CCL20 and CCL7) in PWHART compared to HC. The metabolite transporter detection by flow cytometry in T cells and monocytes indicated an increased expression of glucose transporter 1 (Glut1) and monocarboxylate transporter 1 (MCT-1) in PWHART. Single cell-type metabolite measurement identified decreased glucose, glutamate, and lactate in monocytic cell populations in PWHART. Deep-immunophenotyping of myeloid cell lineages subpopulations showed no difference in cell frequency, but expression levels of CCR5 were increased on classical monocytes and some dendritic cells. Conclusions: Our data thus suggest that the myeloid cell populations potentially contribute significantly to the modulated metabolic environment during suppressive HIV-1 infection.


Introduction
Successful long-term antiretroviral therapy (ART) inhibits human immunodeficiency virus type-1 (HIV-1) replication for a prolonged period in people with HIV-1 (PWH). We and others have earlier reported metabolic modifications involving dysregulation of the amino acid (AA) and central carbon metabolism in PWH on successful long-term ART (PWH ART ) [1][2][3][4][5][6][7][8][9]. Our recent study also reported a system-level upregulation of oxidative phosphorylation (OXPHOS) and glycolysis in PWH ART compared to HIV-1 positive elite controllers who naturally control viral replication [10]. These alterations can play a role in latent reservoir dynamics and immunosenescence in PWH ART . In our earlier studies from India [1], Denmark [2], and Cameroon [3], we reported persistent metabolic reprogramming in patients with successful therapy. A subset of these PWH ART also showed an altered metabolic profile linked with the development of age-related comorbidities that can potentiate accentuated aging or age-related diseases.
In this study, we performed a system-level plasma metabolomics profiling using untargeted and targeted metabolomics in a well defined Swedish cohort of PWH ART , together with matched HIV-negative controls (HC). We also performed immune phenotyping of metabolite transporters for some critical metabolites altered in PWH ART and intercellular measurements of the key metabolites in different cell populations. Finally, we phenotyped myeloid cell lineages, focusing on the expression levels of chemokine receptors that play a significant role in cellular trafficking during inflammation. We observed that intercellular metabolites and metabolite transporters are altered in monocytes but not T cells, potentially linked with the systemic metabolic environment. Our study thus provides a comprehensive understanding of immuno-metabolic regulation during long-term successful therapy.
The study was approved by the Regional Ethics Committee of Stockholm and performed according to the Declaration of Helsinki. All participants gave informed consent before inclusion in the study.

Metabolomics
Plasma un-targeted metabolomics was performed using the HD4 Platform (Metabolon, Morrisville, North Carolina, USA) in a cohort including HC (n ¼ 22), PWH ART (n ¼ 29), and PWH VP (n ¼ 11). The samples were selected randomly. The method was performed as previously described [11]. Plasma targeted metabolomics was performed towards central carbon metabolism (CCM) and sugars using gas chromatography-mass spectrometry (GC-MS) and amino acids by liquid chromatography (LC)-MS/MS in a cohort of HC (n ¼ 37), PWH ART (n ¼ 55), and PWH VP (n ¼ 24) at the Swedish Metabolomics Centre (Umeå, Sweden).

Proteomics
Targeted plasma proteomics was performed using the Olink inflammation panel (Olink, Sweden) as previously described [12]. In brief, a proximity extension assay was used to evaluate the plasma levels of a set panel of proteins. Cycle threshold (Ct) values were normalized to an extension control and inter-plate control, and all proteins were reported as normalized protein expression levels (NPX).

Flow cytometry
Purity evaluation of isolated cell populations was acquired on BD Fortessa (BDBiosciences, Franklin Lakes, New Jersey, USA). Expression levels of metabolite transporters was acquired on BD Symphony (BdBiosciences). The myeloid lineage panel was acquired on BD Symphony (BdBiosciences). Staining was complemented with Dead cell (Near-IR or Aqua) viability stain (Invitrogen, ThermoFisher, Scientific). Specifics about antibodies can be found in Table 4, Supplemental Digital Content, http://links.lww.com/QAD/C814. All flow cytometry analysis was performed using FlowJo 10.8.1 (TreeStar Inc., Ashland, Oregon, USA).

Results
Increased levels of glutamate, pyruvate and lactate in people with HIV-1 on antiretroviral therapy compared to HIV-negative controls We performed untargeted metabolomic analysis using the HD4 Platform (Metabolon) on plasma samples obtained from PWH ART (n ¼ 29), untreated PWH with viremia (PWH VP , n ¼ 11), and matched HIV-negative controls (HC, n ¼ 22). A total of 841 metabolites were detected, of which 143 (17%), belonging to xenobiotics, were excluded from further analysis as these are not naturally produced (Fig. 1a). After adjusting for gender, body mass index (BMI), and age, known to influence the plasma metabolite profile, 14 metabolites were significantly different (adjusted P < 0.1) between PWH ART and HC ( Fig. 1a and Supplementary File 1, Supplemental Digital Content, http://links.lww.com/QAD/C813). We also detected 36 (adjusted P < 0.1) metabolites that differed between PWH VP and PWH ART and 70 (adjusted P < 0.1) between PWH VP and HC. Therefore, a pronounced host metabolic alteration was detected during the initial viraemic phase. Since our primary aim was to identify the metabolic dysregulation in PWH ART compared to matched HC, we restricted the analysis to these two groups. Heatmap visualization of significantly different metabolites between PWH ART and HC showed a distinct enrichment pattern of metabolites between the groups (Fig. 1b). This included a significant enrichment of glutamate, pyruvate, and lactate in PWH ART (Fig. 1c). Uniform manifold approximation and projection (UMAP) could clearly distinguish clusters for PWH ART and HC based on these three metabolites (Fig. 1d). For 11 individuals, we had paired samples from the initial viraemic phase without therapy and after a median duration of eight years of treatment. In five of these patients, the median levels of glutamate, both before and during treatment, were higher than in HC, while glutamate levels increased following treatment initiation. In most of the patients, lactate levels decreased (9 of 11) while pyruvate levels increased (7 of 11) following therapy (Fig. 1e). Next, we tried to identify biomarkers that could differentiate HC from successfully treated PWH ART using multivariate methods with unbiased variable selection in R (MUVR). This analysis identified glutamate and g-carboxyglutamate as key biomarkers (Fig. 1f). We further performed weighted co-expression analysis using the Leiden algorithm, where coordination between pyruvate, lactate, and glutamate was observed (Fig. 1g). Therefore, our data indicate a dysregulation of glutamate metabolism in PWH ART with long-term successful HIV-1 treatment. This modulation could play a significant coordinating role in pyruvate and lactate metabolism.
Dysregulated central carbon and amino acid metabolism together with elevated levels of inflammatory markers in people with HIV-1 on antiretroviral therapy Next, to validate our findings in a larger cohort of PWH ART (n ¼ 55), PWH VP (n ¼ 24), and HC (n ¼ 37), we performed targeted metabolomic analysis for AA and metabolites of the central carbon metabolism (CCM) and sugars. Among the quantified AA (n ¼ 26), CCM metabolites, and sugars (n ¼ 15), 19 metabolites were significantly different (adjusted P < 0.05) between PWH ART and HC (Fig. 2a). Of these, the majority of AA (tryptophan, methionine, lysine, glutamine, glycine, arginine, threonine, and kynurenine) decreased in PWH ART while glutamate was increased (Fig. 2a). Amongst the CMM metabolites and sugars, we detected a decrease in glucose while its isoform beta-glucose was increased in PWH. Furthermore, a similar trend with an increase in lactate and pyruvate was seen for the untargeted metabolomics, strengthening the result (Fig. 2a). As the proportion of metabolites can be a consequence of clinical parameters or confounders, we next employed UMAP to evaluate if gender, HIV-positive status, BMI, or length of treatment could separate PWH ART from HC. The UMAP analysis identified that alterations in CCM metabolites and sugars were associated with the duration of therapy (Fig. 2b). Finally, to evaluate the inflammation level, we performed secretome analysis in plasma using the Olink  (adjusted P < 0.001) in PWH ART compared to HC (Fig. 2c). Another 12 inflammatory markers were significantly higher in PWH ART than HC, using a significance cut-off adjusted P < 0.05. These markers Change High in ART  , sugar metabolites and amino acids are shown from the cohorts. Column annotation denotes the study cohorts and right annotation shows the metabolite levels in PWH ART compared to HC. Significantly changed metabolites in PWH ART compared to HC are labelled. (b) Sample distribution based on targeted metabolomics data of CCM, sugar metabolites and amino acids by UMAP dimensionality reduction. Individual UMAPs are coloured in respective category for cohort, gender, duration of treatment, and BMI. (c) Boxplot of significantly regulated proteins in PWH ART compared to HC (adj.P < 0.05). Statistical significance was determined using MannÀWhitney U-test (adjusted P < 0.05). HC, HIV-1 negative controls; PWH ART , people with HIV on antiretroviral therapy; UMAP, uniform manifold approximation and projection. and propagation [11,16,17]. As the significant systemic alteration of the metabolites was glucose, pyruvate/ lactate, and glutamate, next we performed flow cytometry analysis of metabolite transporters, glucose transporter-1 (Glut1), pyruvate and lactate transporter monocarboxylate transporter 1 (MCT-1), and cysteine/ glutamate antiporter (xCT) in a cohort of HC (n ¼ 9) and PWH ART (n ¼ 27) (Fig. 1A, Supplemental Digital Content, http://links.lww.com/QAD/C814). The CD4 þ T cells were decreased while CD8 þ T cells were increased in PWH ART compared to HC (Fig. 3a). No significant differences were observed in classical (CM), intermediate (IM), or nonclassical monocytes (NCM). Receptor expression analysis showed that the percentage of CM þ Glut1 þ increased in PWH ART compared to HC (Fig. 3b and c). Analysis of the median fluorescence intensity (MFI) showed no difference in CD4 þ or CD8 þ T cells (Fig. 3d). Even though there was no significant difference in MCT-1 þ CM, the MFI of MCT-1 was increased on CM in PWH ART compared to HC ( Fig. 3c and d). To further evaluate the intracellular metabolite levels in blood cell populations, a cohort of HC (n ¼ 10) and PWH ART (n ¼ 29) was used to isolate CD4 þ T cells and monocytes ( Fig. 3e and Fig. 1B, Supplemental Digital Content, http://links.lww.com/ QAD/C814). One PWH ART sample was excluded due to low cell viability. After CD4 þ T-cell isolation, two HC samples were excluded from the analysis due to low cell count. The purity of isolated CD4 þ T cells had a median of 97% (IQR 93.9-97.575) (Fig. 1C, Supplemental Digital Content, http://links.lww.com/QAD/C814). Intracellular measurement of glucose, lactate, and glutamate showed no significant difference within the CD4 þ T-cell subset (Fig. 3f). After monocyte isolation, four HC samples and one PWH ART sample were excluded from the analysis due to low cell count. The purity of the monocytes had a median of 93.15% [interquartile range (IQR) 95.55-90.55] (Fig. 1D, Supplemental Digital Content, http://links.lww.com/ QAD/C814). Intracellular measurement of glucose, lactate, and glutamate showed a significant decrease of all three metabolites in PWH ART compared to HC (Fig. 3g). These data indicate that the main differences in the metabolic profile occur in the monocytes during long-term suppressive therapy.
Characterization of monocytic subpopulations in people with HIV-1 on antiretroviral therapy vs. HIV-1 negative controls As we identified the main differences in intracellular metabolite and transporter expression in monocytes, we wanted to evaluate the distribution in monocytic cell populations. Therefore, we ran a panel characterizing myeloid cells in HC (n ¼ 10) and PWH ART (n ¼ 29) (Fig.  2A, Supplemental Digital Content, http://links.lww. com/QAD/C814). The panel identified subsets of monocytes, dendritic cells (DCs), and myeloid-derived suppressor cells (MDSCs) (Fig. 4a). The distribution of cell populations can be viewed in Fig. 4b. Even as the proportion of DC2/DC3 showed a slight reduction in PWH ART (P ¼ 0.085), no significant differences were detected in the frequency of cell populations except for the frequency of CD11b þ CD33 þ low density granulocytes (LDGs) (Fig. 2B, Supplemental Digital Content, http://links.lww.com/QAD/C814). Additionally, we looked at chemokine receptor expression CCR2, CCR5, and CX3CR1 as they are known markers of migration and activation of myeloid lineages (Fig. 4c). CM expressing CCR5 were increased in frequency, while CCR5 þ granulocytic (G)-MDSC were decreased and CX3CR1 þ G-MDSC were increased in PWH ART compared to HC (Fig. 4d). In DC lineages, the frequency of CX3CR1 þ plasmacytoid (p)DC and CCR2 þ DC1 was decreased, while the frequency of CX3CR1 þ DC1 and CCR5 þ DC2/DC3 was increased in PWH ART compared to HC (Fig. 4d). No other significant differences in the proportion of cells expressing any of the receptors were identified between the groups (Fig.  3A, Supplemental Digital Content, http://links.lww. com/QAD/C814). Within the cell populations, the MFI was decreased for CCR2 on DC1, pDC, and G-MDSC in PWH ART compared to HC (Fig. 4E). Furthermore, an increase in MFI for CCR5 on DC2/DC3 and CX3CR1 on G-MDSC was observed in PWH ART compared to HC (Fig. 4e and Fig. 3B, Supplemental Digital Content, http://links.lww.com/QAD/C814). Collectively, this data shows the relevance of chemokine receptors on myeloid lineages as the key regulators of myeloid cell trafficking during HIV-1 infection, rather than the distribution of cell populations during suppressive therapy.

Discussion
Our comprehensive immuno-metabolic study in PWH ART identified system-level alterations in amino acid metabolism during long-term treatment. We identified a coordinated role of pyruvate, glutamate, and lactate in HIV-1 infection, indicating central carbon and energy metabolism dysregulation. Furthermore, the plasma inflammatory markers MIP-3a and MCP-3 increased in PWH ART . Macrophages and monocytes mainly produce the MIP-3a, while MCP-3 is a chemoattractant that plays a significant role in monocyte mobilization and trafficking to the inflammation sites. Therefore, our data indicate that monocytes play a potential role in the modulation of the inflammatory profile during suppressive ART. This could also be linked with the metabolic alterations in the monocytic subsets for transporter expression and intracellular levels.  Schematic representation of CD4 þ and monocyte isolation for intracellular metabolite measurement. (f) Intracellular measurement of glucose, lactate, and glutamate in a cohort of HC (n ¼ 8) and PWH ART (n ¼ 28) in CD4 þ T cells. (g) Intracellular measurement of glucose, lactate, and glutamate in a cohort of HC (n ¼ 6) and PWH ART (n ¼ 27) in monocytes. Statistical significance was evaluated using MannÀWhitney U-test (P < 0.05) and represented as median with interquartile range (IQR). See also Figure S1. HC, HIV-1 negative controls; IQR, interquartile range; MUVR, multivariate methods with unbiased variable selection in R; PWH ART , people with HIV on antiretroviral therapy; UMAP, uniform manifold approximation and projection.   The three main pathways of CCM and energy metabolism are compartmentalized between the cytoplasm and mitochondria, namely, glycolysis/gluconeogenesis, glutamine metabolism (glutaminolysis), and tricarboxylic acid cycle (TCA cycle) [providing electrons for OXPHOS]. The host dependencies for HIV-1 include glycolysis and TCA [18]. Susceptibility to HIV-1 is also regulated by a cell's metabolic activity and activation stage, where CD4 þ T cells with elevated OXPHOS and glycolysis are more permissive to infection [9,19]. In this study, we detected decreased Glut1 expression on CD8 þ T cells while the percentage of CM expressing the receptor was increased. The uptake of glucose through Glut1 is essential for viral production [20]. In our recent study, we detected reduced Glut1 expression on T-lymphocytes in HIV-positive elite controllers compared to the HIV-negative controls, which could be one mechanism restricting HIV-1 infection [21]. A recent study also indicated that latent HIV-1 reservoirs use glutaminolysis as an alternative fuel source for energy generation [22]. Our cohort detected decreased glutamine and increased glutamate in plasma from PWH ART . Glutaminolysis is the primary pathway charging the TCA-cycle and OXPHOS in naive and memory T-cell subsets. Higher HIV-1 infections have been shown in T cells selected for OXPHOS activity [19] and compromised metabolic steps preceding OXPHOS result in lipid accumulation [22]. Furthermore, we recently observed how modulation of glutaminolysis significantly affects the reactivation of the latent virus [3]. This hints toward a substantial role of glutaminolysis in HIV-1 reactivation.
A recent seminal study also showed that senescent cells relied on glutaminolysis and proposed that inhibiting glutaminolysis in the aging body could prevent ageassociated disorders and even prolong the lifespan [23]. It is known that in glutamate toxicities, pyruvate plays a significant role in quenching glutamate to keep the level of inflammation low [24]. Our longitudinal data thus indicates that the increase of pyruvate could result from elevated glutamate levels. In our earlier cross-sectional studies, we observed higher glutamate levels in PWH ART in cohorts from India [1], Cameroon [3], and Denmark [2]. These cohorts, collectively including more than 500 PWH, indicated a prevalent disruption of glutaminolysis, that is, lysis of glutamine to glutamate, in PWH ART . We also saw that glutaminolysis was central to comorbidities such as metabolic syndrome (MetS) [2,3]. Interestingly, the level of pyruvate was high in the Danish cohort but not in India or Cameroon. Additionally, the level of inflammation was high in the Indian cohort [25], while our earlier Swedish study on long-term treated individuals (nearly two decades) indicated near normalization of the inflammatory profile [26]. Based on these observations, we posit that the increased level of pyruvate in cohorts from high-income countries could be linked with the low level of persistent inflammation. This alteration could be associated with using newer antiretrovirals with a better toxicity profile while indicating an enhanced metabolic profile and improved quality of life.
Herein, we also show how intracellular metabolic modulation mainly occurs in the monocytic cell population in PWH ART . We did not see any differences in the frequency of myeloid cell populations, indicative of maintained cellular frequencies during suppressive ART. However, there was a variation in the receptor expression of CCR2, CCR5, and CX3CR1 in some sub-populations. Activation of these chemokines' receptors mediates immune cell trafficking. Even as the general classification of CM is CCR2 high and CX3CR1 low , our cohort exhibited an increase in CCR5 in PWH ART [27]. CCR5 is one of the main co-receptors used for HIV-1 entry into cells, and in monocytes, the dependency of CCR5 for the M-tropic virus is well described [28]. Earlier studies have also shown how activated monocytes and macrophages shift to a glycolytic metabolism by increasing the expression of Glut1, one of the primary glucose transporters [29]. As described here by us and others, these activated Glut1 þ monocytes were enriched during HIV-1 infection [30]. Therefore, the elevated levels of CCR5 on CM could indicate an increased susceptibility towards HIV-1 or, on a phenotype level, an increased transition into a proinflammatory phenotype to elicit heightened inflammatory responses. Simultaneously, the evaluated immune activation seen in PWH ART could support an environment of increased glucose metabolism and higher activation-induced differentiation of monocytes, as previously reported in T cells [31,32]. studies have shown the depletion of DCs during acute viral infections (e.g. COVID-19 [33] and Hantavirus [34]). Possibly, similar events occur during initial HIV-1 viremia but normalize during suppressive ART. This would be in accordance with our data, where we did not see any differences in the frequency of DC subsets. However, our data show that the main differences in receptor expression occur in the DCs with a decrease of CCR2 and CX3CR1 in some subpopulations, while CCR5 increases in DC2/DC3. The DCs belong to the mononuclear phagocytes (MNPs) together with monocytes and macrophages, which are crucial for linking the innate and adaptive immune systems [35]. In HIV-1, DCs can transfer the virus to T cells by delivery of membranebound virus through a viral synapse or de novo synthesis of virus from infected cells [35]. After capturing the virus and antigen presentation, DCs migrate to the lymph nodes and spleen. This can facilitate HIV-1 spread to compartments rich in target cells for infection. However, our data do not indicate increased dissemination of DC subsets to distant body compartments as the frequencies of DC subsets are maintained during suppressive ART [36]. Instead, the altered chemokine receptor expression could hint at the function of immune cells in PWH ART . For the CX3CR1 receptor, also known as the fractalkine receptor, the receptor-ligand interaction is strongly linked to the survival of immune cells [37]. Thus, the reduced expression in some DC subsets in PWH ART could indicate a decreased capacity to counteract cell death mechanisms. This could result from the hypothesized rapid DC decline during peak viremia. MNPs are potent producers of CCL2, the ligand for CCR2, mediating a proinflammatory function [38]. A reduced expression of CCR2 could affect the responsiveness of DC recruitment to the site of inflammation. Our results show a dysregulated phenotype on some DC subsets in PWH ART . However, the implication on immune cell function, metabolic reprogramming, and immunological aging during suppressive ART remain to be elucidated.
Though our study used a relatively large number of samples, including the longitudinal samples, our study has limitations that merit comments. First, the study's crosssectional nature limits the observations to associations that do not infer causation. Second, the ART regimens were heterogeneous. Third, though we have a more homogenous group, CMV status, diet and lifestyle can influence the inflammatory status. We have initiated a large study to understand the myeloid cell dysfunction in PWH ART. Finally, our analysis is more steady-state in the absence of functional assays.
In conclusion, our cross-sectional study provides systemlevel snapshots of the metabolic reprogramming in the PWH on successful therapy, dynamics of the metabolic alterations in the longitudinal study, and its association with systemic inflammation and myeloid lineage cells. This metabolic reprogramming and altered chemokine signaling may subsequently affect monocyte trafficking and polarization during persistent inflammation. In turn, these aberrations might influence adaptive immunity. A better understating of immune cell trafficking and polarization of myeloid cell lineage in PWH on successful therapy, together with strategies to modulate the activation of macrophage phenotypes, could provide adjunctive therapeutic targets to improve metabolic health while mediating the control of viral replication in PWH.