Role of IL-6 in Mycobacterium avium–Associated Immune Reconstitution Inflammatory Syndrome Free

Human immunodeficiency virus infection results in defects in CD4 T cell numbers and function leading to susceptibility to microbial infections, and control of viral replication with ART allows CD4 T cell responses to recover and restores normal host resistance. In most patients, this process leads to an improvement of clinical symptoms. However, some individuals, who harbor a microbial coinfection and have extremely low numbers of CD4 T cells at the time of antiretroviral therapy (ART) initiation (1), experience a rapid deterioration within the first few weeks of treatment. This worsening of disease in ART-treated patients is referred to as immune reconstitution inflammatory syndrome (IRIS) and occurs despite successful control of HIV viremia and recovery of circulating CD4 T cell numbers. Indeed, IRIS is thought to be mediated by dysregulated immune responses mounted against an underlying opportunistic infection. Many different pathogens have been associated with IRIS, but mycobacterial infections such as Mycobacterium tuberculosis and Mycobacterium avium complex (MAC) are frequent culprits. Although the specific symptoms of IRIS depend on the particular microbial coinfection and the affected tissue, the manifestation of IRIS can be broadly categorized as either a paradoxical or unmasking presentation. Paradoxical IRIS occurs in patients already receiving treatment for an opportunistic infection who then develop an exacerbation of inflammation associated with that coinfection when ART is initiated. Unmasking IRIS describes the appearance of pathology associated with an infection that only becomes recognized following ART because of the IRIS event itself.

The mechanisms of IRIS are poorly understood, and no targeted therapies exist for the treatment of IRIS. Although it may seem counterintuitive to broadly immunosuppress an HIV patient with a disease causing microbial coinfection, corticosteroid therapy is currently the best approach for reducing inflammation and pathology during IRIS (2–6). Therefore, a better understanding of the immunopathology of IRIS is needed for the development of targeted therapies. Many clinical studies have examined inflammatory cytokine profiles that are associated with IRIS. IFN-γ levels (5–8) and IFN-γ–secreting T cells (9–13) have been found to be increased in individuals who develop IRIS following ART, suggesting that this cytokine may play an important role in IRIS pathology. There have also been indications that the IL-6 pathway may be engaged during IRIS. C-reactive protein (CRP), an acute-phase reactant know to be downstream of IL-6 signaling, is one of the most often observed biomarkers of IRIS. Although a few reports have failed to find an association between IL-6 and IRIS susceptibility (5, 14, 15), other studies covering diverse manifestations of IRIS have found associations between the risk of IRIS and increased levels of IL-6 preceding ART (16), at the time of the IRIS event (6, 16–19), and sometimes even for years following recovery from IRIS (16, 20, 21). It also has been shown that effective corticosteroid treatment of IRIS decreases IL-6 concentrations in the serum (4). Moreover, a single nucleotide polymorphism in the IL6 locus has been shown to associate with the susceptibility to MAC/IRIS (22). Thus, a large amount of clinical evidence acquired over the past decade has indicated that increased IL-6 production correlates with IRIS, but its role in directly driving the pathology of IRIS has not been addressed.

We have previously described a model of experimentally inducible M. avium IRIS in mice (23, 24). In the system used, T cell–deficient mice harboring a disseminated M. avium infection are injected with CD4 T cells to mimic the ART-induced recovery of Th cells in a mycobacterial coinfected, T cell–deficient AIDS patient. In this robust model of severe IRIS, adoptive transfer of CD4 T cells into chronically infected TCRαKO mice leads to a rapid, IFN-γ–dependent wasting disease, and most of the mice succumb 2–4 wk following T cell transfer. In this study, we perform both patient and animal model studies to examine the role of IL-6 in M. avium–associated IRIS. We show that IL-6 levels rise sharply in HIV/MAC–coinfected patients at the time of an unmasking IRIS event and use the experimental animal model to demonstrate that IL-6 blockade greatly reduces IRIS disease. Moreover, coblockade of IFN-γ and IL-6 in the experimental model further reduced IRIS associated disease and was effective even when therapy was initiated after the onset of wasting. These data strongly support the hypothesis that IL-6 is a major mediator of disease during IRIS in M. avium–coinfected HIV⁺ individuals.

Plasma samples were obtained from seven HIV⁺ patients (five male/two female; median age of 45 y; range, 27–50 y) with a median CD4 T cell count of 24 cells/μl (range, 1–86 cells/μl) recruited as part of a cohort study at the Clinical Center of the National Institutes of Health and who developed unmasking MAC IRIS after ART initiation. All participants provided written informed consent at the Clinical Center of the National Institute of Allergy and Infectious Diseases, National Institutes of Health, under an Institutional Review Board–approved protocol (NCT00286767). All patients were ART naive or had interrupted ART for >6 mo after receiving <3 mo of therapy (n = 1) and had less than or equal to 100 CD4⁺ T cells/μl at baseline. IRIS was defined according to the AIDS Clinical Trials Group criteria. The median time to unmasking MAC IRIS was 19 d (range, 7–99 d).

C57BL/6 and athymic nude mice were obtained from Taconic Farms, and IL-6^−/− mice were purchased from The Jackson Laboratory. Ag85b-specific TCR transgenic (Tg) P25 mice (25) were bred in our animal facility. To generate bone marrow (BM) chimeric mice, animals were lethally irradiated with 950 rad and reconstituted with 8–10 × 10⁶ BM cells isolated from wild-type (WT) or IL-6^−/− mice the same day after irradiation. Thy1.2-depleting Ab (500 μg) was added to the BM cell inoculum to ensure depletion of contaminating mature CD4 T cells. Mice were allowed to reconstitute for 10–12 wk before they were exposed to M. avium. For bacterial infections, the mice were inoculated i.v. in the lateral tail vein with 1 × 10⁶ CFU M. avium SmT 2151. All Kaplan–Meier graphs of mouse survival represent mice that were found dead or reached moribundity and were euthanized. All animals were bred and maintained in an American Association for the Accreditation of Laboratory Animal Care–approved facility at the National Institute of Allergy and Infectious Diseases, National Institutes of Health, according to the Guide for Care and Use of Laboratory Animals, and all procedures were conducted according to an animal study proposal that was approved by the National Institute of Allergy and Infectious Diseases Animal Care and Use Committee.

Spleen and lymph node CD4 T cells were isolated from naive WT, IL-6^−/−, or P25 TCR Tg mice as indicated using MACS magnetic beads, according to the manufacturer’s instructions (Miltenyi Biotec). Cells were routinely >90% pure CD4⁺ cells after column separation. For adoptive transfer, cells were suspended in saline, and 2 × 10⁶ T cells were injected i.v. in a volume of 300 μl. Control Ab (clone HRPN), anti–IL-6 (clone MP5-20F3), anti–IL-6R (clone 15A7), anti–IFN-γ (clone XMG1.2), and anti-Thy1.2 (clone 30H12) mAbs were purchased from BioXCell. For in vivo blockade experiments, 200–500 μg Ab was administered every third day for the indicated amount of time. No difference was observed between mice given PBS or control Ab.

CD4 T cell frequencies in blood were determined by flow cytometry. Blood was collected in EDTA tubes, and RBCs were lysed with ACK lysing buffer. In different experiments, cells were stained with various combinations of CD4, TCRβ, CD3ε, Vβ11, and fixable viability dye eFlour780. All Abs and live/dead dyes were purchased from eBioscience and BioLegend. All samples were acquired on a BD LSR Fortessa and analyzed with FlowJo version 9.5.

Cytokine measurements were performed on cryopreserved human plasma samples or mouse serum samples stored at −80°C. Mouse CRP, mouse IL-6, mouse TNF-α, and human IL-6 was measured in with ELISA kits purchased from R&D Systems. Human CRP and mouse IFN-γ was measured with a high-sensitivity ELISA purchased from eBioscience.

Weight loss was represented as a percentage of initial weight before T cell transfer as follows: percentage of weight change = ([current weight/pretransfer weight] × 100) − 100. Core body temperature was measure by rectal thermometry on nonanesthetized mice with a murine probe thinly coated with a petroleum-based lubricant.

The statistical differences between groups were analyzed by Student t test, and the difference between survival curves was calculated with a log–rank Mantel–Cox test in GraphPad Prism software version 6. Groups were considered significantly different when p ≤ 0.05.

Increases in the plasma concentrations of IL-6 and CRP after ART are frequently found to distinguish HIV patients who develop IRIS from those who do not. To characterize the dynamics of IL-6 and CRP expression during IRIS, we performed a detailed kinetic analysis of these proteins in a cohort of seven HIV-infected individuals who developed unmasking MAC/IRIS. The patients developed IRIS at time points ranging from 2 to 10 wk following ART. In patients 1, 2, 3, 5, and 6, CRP was at or close to normal concentrations (< 3 mg/l) before ART and spiked at the time of the IRIS event (Fig. 1A). This peak in CRP was then followed by a rapid decline afterward. In patient 4, the plasma CRP concentration was high even before ART but increased further at the time of the IRIS episode.

We next measured IL-6 in these patient samples. In patient 1, the concentration of plasma IL-6 was within the range found in healthy donors but then rapidly increased at the time of the IRIS event at week 2 and then dropped back to normal by week 4 (Fig. 1B). Moreover, in patients 3, 4, 5, 6, and 7, the IL-6 concentrations increased between the time point immediately preceding and the IRIS event. Interestingly, in patients 3, 4, 5, and 7, the concentration of IL-6 initially dropped in the first weeks following ART but then rose sharply during IRIS. In patient 2, the concentration of IL-6 was high before ART and remained elevated until after the patient developed IRIS at week 2. These data indicate that CRP and IL-6 increase in HIV-infected individuals at the time of unmasking MAC/IRIS.

To model IRIS in mice, we developed a system that recapitulates the immunological scenario underlying IRIS in HIV-infected individuals: T cell reconstitution of a lymphopenic host harboring a microbial infection. TCRαKO mice are infected with M. avium, and after at least 60 d of infection, the mice are injected with purified CD4 T cells (Fig. 2A). Although CD4 T cells are normally critical for control of mycobacterial infections, in this setting of immune reconstitution, CD4 T cell transfer triggers a rapid wasting disease that results in the death of most of the animals (Fig. 3) (24). We found that CD4 T cell transfer into M. avium–infected TCRαKO mice induces increases in the serum concentration of IL-6 (Fig. 2B). IFN-γ production by the CD4 T cells was not required for IL-6 induction because transfer of IFN-γ knockout (KO) CD4 T cells induced similar amounts of IL-6 compared with WT T cells (Fig. 2B). Serum CRP was elevated in M. avium–infected TCRαKO mice compared with naive animals, and during IRIS following adoptive transfer of CD4 T cells, CRP increased further (Fig. 2C). CRP expression can be induced by IL-6, and CRP and IL-6 concentrations positively correlated in the serum (Fig. 2D). Therefore, we next examined the role of IL-6 in inducing CRP expression during IRIS. IL-6 neutralization with mAbs decreased serum IL-6 back down to the concentrations seen in uninfected mice (Fig. 2C). IFN-γ–deficient CD4 T cells also failed to drive increases in CRP above what is found in TCRαKO mice that did not receive T cells (Fig. 2C). Collectively, these data indicate that IL-6 and CRP increase during experimentally induced M. avium–IRIS. Furthermore, these commonly observed biomarkers of IRIS are directly linked as elevated CRP expression requires IL-6 and IFN-γ.

Given the association of IL-6 with disease during M. avium–IRIS in both humans and mice, we sought to directly evaluate the role of IL-6 in the pathogenesis of IRIS. M. avium–infected TCRαKO mice were reconstituted with purified CD4 T cells and treated with anti–IL-6–neutralizing mAb starting on the day of T cell transfer. Mice that received CD4 T cells alone developed a severe wasting disease starting around day 7 posttransfer and lost >30% of their body weight by day 12 (Fig. 3A). In contrast, mice that received CD4 T cells and anti–IL-6 first began to show signs of weight loss around day 10 posttransfer, and by day 20 when blockade was stopped, the mice had lost ∼20% of their body weight (Fig. 3A). CD4 T cell transfer also induced a dramatic drop in core body temperature. TCRαKO mice harboring a chronic M. avium infection were ∼35°C, and on day 12 posttransfer of CD4 T cells, untreated control mice had dropped to ∼32°C, whereas mice receiving CD4 T cells and anti–IL-6 had dropped to ∼34°C (Fig. 3B). Moreover, IL-6 blockade significantly extended the survival of mice following CD4 T cell transfer (Fig. 3C). The protective effect of IL-6 blockade was not due to differences in the degree of the expansion of the donor CD4 T cells as untreated, and anti–IL-6–treated mice had similar percentages of CD4 T cells in their PBMC (Fig. 3D). TNF-α can sometimes also be a major mediator of wasting disease, and we have previously shown that it partially contributes to pathology in our model of murine IRIS (23). IL-6 blockade also had no effect the induction of TNF following CD4 T cell reconstitution (Fig. 3E), and conversely TNF blockade had no effect on IL-6 induction (data not shown). Taken together, these data indicate that IL-6 production not only correlates with IRIS in humans and mice but also directly contributes to the pathology seen in M. avium–infected T cell–deficient mice following CD4 T cell reconstitution. Interestingly, the protective effect of IL-6 blockade in experimentally induced IRIS is even greater than what we previously reported with TNF neutralization.

Cells of both hematopoietic and non-hematopoietic origin can produce IL-6, so we next examined the cell types involved in mediating IL-6–dependent disease in experimentally induced IRIS. To test the role of CD4 T cell–derived IL-6 in IRIS pathogenesis, we transferred either WT or IL-6 KO CD4 T cells into M. avium–infected TCRαKO mice. WT and IL-6 KO CD4 T cells induced identical courses of wasting diseases, indicating that CD4 T cells are not a significant source of IL-6 in murine IRIS (Fig. 4A).

We next tested the role of hematopoietic cell–derived IL-6 in IRIS. Athymic nude mice were lethally irradiated and then reconstituted with BM from either WT or IL-6 KO mice (Fig. 4A). Thy1.2-depleting mAb was added to the BM cell inoculum to ensure that any mature CD4 T cells contaminating the BM donor cells were removed. Ten weeks after BM reconstitution, the mice were injected with M. avium, and the infection was allowed to progress for an additional 2 mo. Although the injected BM was inherently capable of producing T cells, the use of athymic nude mice recipients prevented the development of T cells allowing us to later study the impact of CD4 T cell reconstitution. This approach yielded T cell–deficient mice that harbor a chronic M. avium infection and selectively lack IL-6 in BM-derived cells. Last, TCR Tg CD4 T cells specific for Ag85b (P25 cells) were adoptively transferred to induce IRIS. Interestingly, CD4 T cell reconstitution induced identical wasting disease in both WT and IL-6 KO BM recipient nude mice (Fig. 4C). Although there was a slight trend toward a 2- to 3-d delay in mortality of the IL-6 KO BM recipients, the difference in the survival of WT or IL-6 KO BM recipients during IRIS did not reach statistical significance, and all mice eventually succumbed by day 31 posttransfer (Fig. 4D). Moreover, serum IL-6 concentrations were similar between WT and IL-6 KO BM–recipient mice undergoing IRIS (data not shown). These data indicate that radio-resistant cells are the predominant source of IL-6 that drives wasting disease and death during experimentally induced IRIS in M. avium–infected T cell–deficient mice.

We previously showed that IFN-γ produced by the reconstituting CD4 T cells is a major mediator of disease in the murine model of M. avium–IRIS (23), so we next asked whether IL-6 blockade affects the production of IFNγ in this system. M. avium–infected TCRαKO mice were injected with CD4 T cells and treated with anti–IL-6 blocking mAb, and IFN-γ was measured in the serum. CD4 T cell reconstitution induced elevations in serum concentrations of IFN-γ equally in mice treated or not with IL-6 blockade (Fig. 5A). Thus, IL-6 and IFN-γ both contribute to IRIS pathology but do not affect the expression of each other in this system.

These observations indicated that IL-6 and IFN-γ are independent mediators of wasting and mortality during M. avium–IRIS, so we next tested the therapeutic efficacy of combined IL-6 and IFN-γ blockade. M. avium–infected TCRαKO mice were injected with CD4 T cells and treated with anti–IL-6/anti–IL-6R, anti–IFN-γ or a mixture of anti–IL-6/anti–IL-6R and anti–IFN-γ starting on the day of T cell reconstitution, and weight loss was measured on day 14 posttransfer. As expected, individual blockade of each pathway with anti–IL-6/ anti–IL-6R, or anti–IFN-γ alleviated the extent of wasting (Fig. 5B), but IL-6 blockade was slightly more effective than IFN-γ blockade. When anti–IL-6/ anti–IL-6R and anti–IFNγ were administered simultaneously at the beginning of T cell transfer, only a slight decrease in weight loss was detected on day 14 posttransfer (Fig. 5B) compared with mice that did not receive T cells, indicating that IL-6 and IFN-γ represent two major mediators of IRIS pathology. Moreover, coblockade of IL-6 and IFN-γ had no effect on serum concentrations of TNF, indicating that the protective effects of this treatment is not likely to be due to suppression of TNF (data not shown).

Finally, we tested the ability of combined IL-6 and IFN-γ blockade to treat IRIS after the onset of wasting disease. To do so, IRIS was induced as before and T cell recipients were treated with a mixture of anti–IL-6+ anti–IL-6R+ anti–IFN-γ mAbs from day 0 to 16 or from day 10 to 16. Survival after withdrawal of the blocking Abs was monitored. Untreated CD4 T cell recipients showed a median survival time (MST) of 15 d, whereas combined cytokine blockade administered day 0–16 of T cell transfer extended the MST to 43 d (Fig. 5C). Importantly, mice that began cytokine blockade on day 10 had already undergone ∼20% weight loss (data not shown) but still showed an increased MST of 38 d (Fig. 5C). These data indicate that even after the onset of a rapidly progressing wasting disease, blockade of these two effector pathways can dramatically extend survival of this severe M. avium–associated IRIS.

Many studies of HIV patients have sought to find correlations between inflammatory mediators and susceptibility to IRIS. Although several pathways have been associated with mycobacterial IRIS, the lack of a preclinical model has made it difficult to ascribe a role of these molecules to the immunopathology of IRIS. This is the first study, to our knowledge to combine human data and a preclinical model to evaluate the potential therapeutic efficacy of targeting an inflammatory mediator for the treatment of IRIS. We show that elevations in serum IL-6 concentrations are associated with disease onset in HIV/MAC–coinfected individuals who develop unmasking IRIS following ART and use the mouse model of experimentally induced M. avium–IRIS to directly test the role of IL-6 in driving IRIS pathology. In fact, the system of experimentally induced IRIS used in this study can even be considered a form of unmasking IRIS similar to the patients described, because the mice are not treated with antimycobacterial drugs and display no overt symptoms resulting from the M. avium infection until after CD4 T cell reconstitution. We find that IL-6 not only correlates with disease but also is a major mediator of IRIS pathology in M. avium–infected T cell–deficient mice. Moreover, we find that IL-6 also likely contributes to the increased levels of the acute-phase reactant CRP, which are so often observed in IRIS patients, indicating that these two biomarkers of IRIS are likely indicative of the same pathway. Although the studies performed here were limited to M. avium–associated IRIS, the observation of increased IL-6 and CRP in IRIS associated with M. tuberculosis and Cryptococcus neoformans infections suggests that the mechanisms delineated here may be of general relevance to multiple manifestations of the syndrome.

Numerous studies have examined many different inflammatory pathways in HIV/tuberculosis-coinfected individuals during ART, and we have experimentally addressed the role of several key cytokines that have been associated with IRIS susceptibility. We showed that IFN-γ and to a lesser extent TNF drive IRIS pathology in M. avium–infected mice, whereas IL-4 and IL-17 have no role (23). Moreover, we have found that IL-7R, IL-1R1, GM-CSF, M-CSF, and G-CSF blockades have no therapeutic effect in this murine model of IRIS, indicating that these pathways may play little role in the induction of mycobacterial immune reconstitution disease (data not shown). Therefore, although cytokine biomarkers may be found to associate with IRIS, it is critical to functionally examine the role of each molecule in experimental models. Moreover, the accumulating negative data on other inflammatory pathways we are observing emphasizes the major role of IL-6 in IRIS pathology.

Given that IL-6 and IFN-γ represent two major pathways that drive IRIS, we also examined the interplay between these two cytokines. In this study, we found that single neutralization of IL-6 or IFN-γ does not impact the expression of the other, and simultaneous blockade of both cytokines synergistically alleviates wasting disease. Therefore, IL-6 and IFN-γ represent two major and independent pathways of IRIS pathology. Indeed, the pathogenic sources of these two cytokines are different cell types. During experimentally induced IRIS the pathogenic IFN-γ is derived from CD4 T cells. It is possible that many cell types produce IL-6 during IRIS, but our nude mouse BM chimera experiments show that the pathogenic IL-6, which contributes to IRIS, is largely produced by radio-resistant cells. Although we cannot exclude the possibility that IL-6 is derived from a long-lived myeloid cell population, our T cell transfers were performed 18–20 wk following BM reconstitution of the nude mice, so it is more likely that nonhematopoietic cell types are key producers of the pathogenic IL-6 in experimentally induced IRIS. Indeed, multiple nonhematopoietic cell types such as adipocytes, myocytes, fibroblasts, and endothelial cells have all been shown to produce IL-6 in response to a variety of stimuli. Because CD4 T cell transfer induces IL-6 expression, this raises the possibility that the communication between the repopulating CD4 T cells and nonhematopoietic tissues may be a key factor driving the systemic disease during mycobacterial IRIS. Further experiments will be required, however, to determine the relevant nonhematopoietic sources of IL-6 and the mechanisms through which CD4 T cells drive its expression. Likewise, the cell types that are targeted by IL-6 and the downstream activities of IL-6R signaling that lead to pathology will require further investigation.

Although the data indicate that IL-6 is a key determinant of IRIS pathology associated with M. avium infection, the clinical feasibility of blocking IL-6 to treat IRIS HIV patients undergoing ART is not clear. It is possible that IL-6 blockade could exacerbate pathogen replication in HIV patients harboring MAC coinfections, because IL-6 blocking Abs have been shown to impair control of M. avium infections in mice (26, 27). There are, however, indications that IL-6 blockade can be beneficial for the treatment of other inflammatory disorders. For example, the mAb against the IL-6 receptor tocilizumab is prescribed for the treatment of rheumatoid arthritis (28). Our data indicate that IL-6 is a major mediator of the cachexia associated with M. avium–associated IRIS in mice, and IL-6 also has been shown to play a major role in cancer cachexia (29). In a recent case report, a patient suffering from cachexia with an IL-6–overexpressing lung cancer was treated with tocilizumab (30). The elevated CRP in the patient quickly returned to normal concentrations and the patient’s weight loss and other symptoms were greatly alleviated. Cachexia is also frequently observed in HIV patients experiencing mycobacterial IRIS during ART, and it is likely that treatments to prevent wasting would greatly reduce morbidity and mortality in individuals that experience a severe episode of IRIS.

In conclusion, we have shown that CRP/IL-6 elevations occur both in HIV⁺ persons who develop unmasking MAC-IRIS after ART initiation and in M. avium–infected TCRα KO mice after CD4 T cell transfer. In the mouse model we showed that both CD4 T cell–derived IFN-γ and IL-6 produced by radio-resistant cells have independent pathogenic roles and that blocking both cytokines was synergistically beneficial. These data suggest that the IL-6 pathway may be a reasonable target in therapeutic interventions in HIV⁺ patients with mycobacterial IRIS.

We thank Sandra Oland, Sara Heiny, and Keith Kauffman for technical assistance.

The authors have no financial conflicts of interest.