Kidney transplant has become the key therapeutic strategy for organ failure, especially in the context of chronic diseases such as diabetes, lupus, polycystic kidney disease, and hypertension (1–3). Since its beginning in the 20th century, advances in surgical and conservation techniques allowed an increase in its numbers worldwide: from 2000 to 2020, the total sum of kidney transplants has drastically increased (from 23,084 to above 100,000), as well as its rate per population (from 9.8 to 14.01 per million) (4). These values, however, are below the general demand (5, 6), which prompted a continuous search for strategies that allowed a greater acquisition and offer of organs—such as the use of expanded criteria donor and a central distribution strategy—as well as interventions that reduce organ rejection and dysfunction (7–9).

During transplantation, the kidney is subjected to ischemia (10). This phenomenon leads to graft inflammation and necrosis, resulting in the release of damage-associated molecular patterns (DAMPs) (11). After graft reperfusion, the presence of DAMPs triggers an immune response from the host, leading to immune cell migration and adaptive response, which could contribute to acute rejection (AR) (12, 13).

AR can manifest as an early (<12 months) or late (>12 months) complication of kidney transplant (14). It can be divided according to Banff criteria histopathological findings in antibody-mediated rejection (ABMR), T cell-mediated rejection (TCMR), and borderline for TCMR. TCMR is initiated via the presentation of donor alloantigens to recipient T lymphocytes, leading to T-cell (CD4+ and CD8+) and natural killer recruitment to the allograft, accumulating in kidney tubules, the interstitium, and vascular areas. This process leads to histologic features that allow its classification via Banff criteria in interstitial inflammation, tubulitis, or intimal arteritis (15). The activation of T helper cells leads to the release of soluble factors that support the activation of B cells that could trigger ABMR. ABMR is a severe form of acute rejection and can occur with or without TCMR. After recognition and presentation of recipient HLA and DAMPs, T helper cells interact with B cells, leading to antibody affinity maturation and plasma cell activation. In the graft, antibodies formed to activate the complement pathway and trigger neutrophil recruitment, leading to the endothelial lesion and cell death (16, 17). In the Banff criteria, ABMR is characterized for vascular inflammation—utilized as evidence of acute tissue injury—current antibody and endothelial interaction and serologic markers of donor-specific antibodies (18).

A common mechanism that connects both TCMR and ABMR is the innate inflammatory response triggered upon recognition of DAMPs and mismatched HLA. In this sense, one potential component of innate immunity that has been recently put in perspective in the context of transplant rejection is the inflammasomes (19). Inflammasomes are multimeric protein complexes that are activated via PRR located in the cytosol, recruiting and activating caspase precursors, leading to IL-1β and IL-18 production and consequent induction of proinflammatory immune responses (20).

Another current therapy target is immunomodulation via metabolism reprogramming. Studies regarding T-cell metabolism showed that glycolysis is linked to an increase in inflammation, with Th1 and Th17 mostly relying on glycolytic activity and the mammalian target of rapamycin complex I (MTORC1) pathway to shape an effector response (21, 22). However, OXPHOS and lipid metabolism are linked to immunosuppressive response, with the products of these metabolic pathways being used to produce immunosuppressive cytokines as well as the persistence of memory T cells and generation of T regulatory cells (Tregs) (21–24). For graft rejection, animal studies have shown that the inhibition of glycolysis could prevent or delay rejection by blocking T-cell polarization and increasing relative Treg frequency (25).

Here we performed a transcriptomics analysis, where we aimed to evaluate non-canonical genes and pathways linked to AR in the context of human kidney transplant in comparison to a non-rejecting group. Our analysis indicated possible associations related to the regulation of the inflammatory response and metabolism that could drive graft rejection, which might pave the way for new therapeutic approaches to the patients.

Acute rejection is a life-threatening condition affecting nearly 10% of kidney transplants of which 25% do not recover previous graft function (44, 45). Despite the improvement of immunosuppression, the currently available drugs show limitations in reducing these statistics, possibly due to targeting canonical pathways of lymphocyte activation, such as the mammalian target of rapamycin (mTOR), antigen presentation, and IL-2 transcription (46, 47). In this sense, the investigation of new processes involved in pathogenesis could act as adjuvant therapies, for either acute rejection prevention or treatment.

Here, using a transcriptomics-based approach, we innovatively suggest that inflammasomes could act as a complementary branch, reporting for the first time their enhancement and reproducibility on acute rejection, mainly for TCMR. Indeed, inflammasomes are known to be involved in renal homeostasis and inflammatory response (48, 49). Curiously, while NLRP3 has been the one most described in kidney acute and chronic injuries (48), we showed that its upregulation is less remarkable in acute rejection when compared to NLRC4 and AIM2. We hypothesize that this may reflect either the relevance of particular signals, for example, an NLRC4 activation by interferon-regulated genes (50), since interferon processes were enriched in our analyzed datasets, or an AIM2 stimuli secondary to dsDNA release, possibly donor-free DNA molecules (51–54) since apoptotic and necrotic cells are proven to be present in these grafts (55, 56).

Few studies have explored the association between inflammasome activation and the regulation of adaptive immunity. In this sense, it has been demonstrated that activated T cells display a complement C5 protein-dependent NLRP3 inflammasome assembly, which induces T helper 1 (Th1) responses (57), which are typically associated with autoimmune and infectious diseases. However, our group demonstrated that a gain of function in NLRP3 specifically in CD4+ T cells improved the clinical course of disease in a model of experimental autoimmune encephalomyelitis (EAE), which was associated with a reduction of Th1 and Th17 inflammatory responses and a shift toward a Th2 pattern (58). It is important to point out that inflammasome receptors can also modulate adaptive immune responses regardless of their canonical roles in inducing the assembly of the complexes. Accordingly, NLRP3 has been shown to translocate to the nucleus of Tregs and negatively regulate this population’s master transcriptional regulator FOXP3, impairing their differentiation efficiency (59). AIM2, in turn, has been shown to enhance the stability of Tregs through metabolic reprogramming toward beta-oxidation in the mitochondria, restraining autoimmune responses in an inflammasome-independent fashion (60). Altogether, these data indicate that inflammasome receptors are important players in the regulation of T-cell responses and suggest that these processes are context-dependent. Our results here provide evidence that AIM2 expression correlates positively with T cells and is linked to allograft rejection, highlighting a detrimental role for this molecule in the context of kidney transplantation, which could be better explored and targeted in further studies.

Additionally, T cells typically undergo metabolic rewiring after stimulation. The interplay between cell metabolism and effector immune function sets an emerging field termed “immunometabolism” (23), which is unraveling mechanisms that may propose promising therapeutic targets to dampen allograft rejection. Th1 and Th17 cells mostly rely on glycolytic activity in order to shape an effector response, and the MTORC1 and HIF-1α transcription factor are upregulated in both of these cells. However, Tregs usually activate the AMPK pathway, leading to lipid oxidation, OXPHOS, and suppression of immune responses (21, 22). In our results, the enrichment analysis showed that OXPHOS was negatively correlated to AIM2 expression and rejected grafts. We also observed that OXPHOS-related genes were negatively correlated to lymphocyte markers such as CD3 and CD4. Studies show that pharmacological glycolysis ablation can enhance levels of CD4+Foxp3+ cells and reduce CD8+IFN-γ+ cells in transplanted mice (25). Also, in graft-versus-host disease (GvDH), when T cells encounter alloantigens, they preferentially upregulate glycolytic metabolism to produce pro-inflammatory responses. Ablation of glycolysis activator MTORC1 alleviated GvDH progression in mouse models (61). Although our results do not directly provide evidence for different T lymphocyte subset metabolic profiles, they suggest a correlation with these studies in a way that T cells present in rejected kidney allografts may have reduced levels of OXPHOS and lipid oxidation, which may enhance their inflammatory phenotype and, therefore, increase transplant rejection. A proposed integrative mechanism for AIM2 based on our findings is demonstrated in Figure 8 .

Finally, meta-analysis is a strategy to quantify a combined effect size of different studies (62), in the case of bioinformatics, usually employing the standardized mean difference (SMD) variable, which was also applied to remove batch effects, in order to reduce heterogeneity (63–65). In our data, AIM2 attained a significant SMD in the TCMR subgroup, in accordance with our individual-dataset heatmap, suggesting a probable confident non-random finding. Since demographic information is usually incompletely available either in the GEO platform or in published transcriptome papers, our analyzed microarrays could have embraced grafts from different donor types, distinct ischemia periods, and immunosuppressive schemes, which are conditions that affect acute rejection development, potentially modifying our measured effect (66–68). In spite of diagnostic heterogeneity, expression profiling by array has intrinsic variations that may also contribute to this observation. It has been noted that microarrays carried out at different time points but performed in the same laboratory still show discrepancies among them (69). Since the datasets analyzed in our work were retrieved from different centers, we standardized the type of annotation that the microarray was provided, which aids in reducing batch effects (69, 70). In this line, further clinical studies are desired to overcome the intrinsic heterogeneity of in silico studies (65), supporting and refining information concerning which grafts are more susceptible to AIM2 signaling and, thus, identifying the patients who could benefit from drug trials targeting the inflammasomes as adjuvant therapies (71–73).

Currently, acute rejection is suspected with parameters for acute kidney injury—rise in 48 h of serum creatinine >0.3 mg/dl or >50% or oliguria < 0.5 ml/kg/h for more than 6 h, with the diagnosis being made by the realization of biopsy to directly identify allograft rejection (74, 75). These criteria, however, present some clinical challenges: serum creatinine is not specific for allograft rejection and can be elevated in both acute rejection and progression of the disease or new pathologic processes in the kidney (76). Furthermore, the confirmation via kidney biopsy brings financial limitations—since it is necessary for the patient to be hospitalized—as well as the bias of sampling error or variability between professionals (77). This context led to the search for non-invasive biomarkers that could predict acute rejection, specifically in blood and urine samples that were linked to TCMR, like CXCL10, TNF-alpha, and miR-155-5p (74, 75). Our findings for AIM2 expression in the tissue and accuracy for graft failure could prompt new analysis for its expression in patient biosamples.

Even though AIM2 has been in recent discussions as a potential target to understand and regulate the progression of inflammatory conditions (78), we were unable to find references that directly linked the molecule to acute rejection both inside and outside the context of a kidney transplant. Going further from the potential mechanism linked to AIM2 and acute rejection previously discussed, markers that permit identifying and predicting acute rejection in kidney transplants are still being researched. In summary, potential non-invasive markers bring a new perspective to the diagnosis of acute rejection as well as a potential way of introducing different interventions according to the findings. In this sense, AIM2 could be a potential new biomarker, especially for TCMR. Even though the previously cited markers showed a good overall diagnostic accuracy with a more practical clinical approach since they are collected from the blood and urine; for acute rejection, here we also include the potential use of AIM2 as a predictor for overall graft survival and not only acute rejection. This could prompt the evaluation of AIM2 as a marker in blood and urine as well.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/ Supplementary Material .

NF and JV selected datasets and wrote the first draft of the manuscript. LE and BG contributed writing sections of the manuscript. NF and LE performed computational analysis. NF, JV and BG designed correlations. NF, JV, LE, IZ, GR and BG collaborate in data interpretation and visualization. NO conceived and supervised the study. All authors contributed to edition, critical revision and approved the submitted version.

The research was supported financially by São Paulo Research Foundation (FAPESP) (18/21844-6, 2019/22409-4, 17/05264-7) and National Council for Scientific and Technological Development (CNPq) and CAPES (financial code 001) grants.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.