The pathogenic epitopes in vaccines stimulate adaptive immune cells, resulting in cellular and humoral immune responses that can be sustained for years by maintaining memory cells in the body (Nicholson, 2016). In addition to inducing adaptive immune memory, another mechanism, called trained immunity, is triggered by vaccines such as Bacillus Calmette–Guérin (BCG) and the yellow fever vaccine (Netea et al., 2011; Saeed et al., 2014; Bekkering et al., 2016). This mechanism utilizes the ability of monocytes and natural killer (NK) cells to respond better to a second, non-specific heterologous stimulus. It has been associated with epigenetic modifications in regions that promote the immune response, immunometabolism, and remodeling of cellular energy metabolism towards aerobic glycolysis (Cheng et al., 2014), which may increase oxidative phosphorylation (Arts et al., 2016; Netea et al., 2016). Innate immune cells can be trained by ligands of NOD2 or dectin-1 receptors (van der Meer et al., 2015), which may result in a trained cell with a pro-inflammatory (Quintin et al., 2012; Kleinnijenhuis et al., 2014b) or regulatory (Quinn et al., 2019; Debisarun et al., 2021) phenotype, depending on the stimulus.

Pro-inflammatory-trained cells are characterized by increased production of pro-inflammatory cytokines such as interleukin (IL)-6, IL-18, tumor necrosis factor alpha (TNF-α) (Kleinnijenhuis et al., 2012; Quintin et al., 2012; Rizzetto et al., 2016), and, in some cases, IL-1β, and an improved ability to kill pathogens such as Candida albicans, Staphylococcus aureus, and Escherichia coli (Kleinnijenhuis et al., 2014a; Rizzetto et al., 2016; Arts et al., 2018). In addition, pro-inflammatory trained cells show increased expression of SET7 protein, which causes an increase in the expression of the enzymes MDH2 and SDHB, both of which are involved in producing cellular energy in the Krebs cycle, promoting the accumulation of metabolites that promote oxidative phosphorylation and, consequently, the production of pro-inflammatory cytokines (Keating et al., 2020). Alternatively, regulatory trained cells are characterized by long-lasting enhanced anti-inflammatory responsiveness (Cauchi and Locht, 2018), such as increased production of IL-10 and IL-1Ra, both immune regulatory cytokines (Quinn et al., 2019; Guha et al., 2021), and reduced production of IL-6, TNF-α, and IL-1β (Guha et al., 2021; Wimmers et al., 2021). In addition, human monocytes trained in the regulatory profile expressed more CD163 and CD206, which promoted inflammation resolution (Guha et al., 2021).

Considering the ability of vaccines to provoke the training phenotype of monocytes, which initiate an immune response to a vaccine antigen via cytokine and chemokine production and antigen presentation, the underlying mechanisms are important to improve vaccine technology and evaluate the potential benefits of immunization. In this context, several vaccines for coronavirus disease 2019 (COVID-19) were developed due to recent health emergency, and the protective role of the innate immune system in vaccination remains to be explored. Here, we showed that vaccination against COVID-19 with a non-replicating viral vector vaccine (nRVVac) and with an inactivated vaccine (InVac) may trigger a training phenotype during the first months following immunization. Both vaccines evaluated in this study induced innate training, and we suggest that the training was characterized by a regulatory phenotype that could be crucial for controlling the exacerbated inflammatory response in further COVID-19 infection. Our results support the widespread vaccination and reinforce its value, particularly considering its effectiveness in modulating epigenetically the innate immune response.

This study was conducted with aim of analyzing whether COVID-19 vaccines have the ability to induce trained immunity, as has been observed for other vaccines for human diseases. We assessed the results of phenotype characterization of monocytes isolated from vaccinated subjects who were subjected to full immunization protocols with an inactivated vaccine or a non-replicating viral vector vaccine. Both groups were defined based on currently available vaccines in Brazil, where the study was conducted. COVID-19 has been characterized as a disease that is markedly inflammatory, with a cytokine storm that is deleterious for patients with severe clinical presentation (Fara et al., 2020; Kim et al., 2020; Manjili et al., 2020; Wang and Perlman, 2022).

The vaccine became globally available at the end of 2020 and in Brazil from January 2021, when the population could receive an inactivated vaccine (CoronaVac—Sinovac/Butantan) and a non-replicating viral vector vaccine (ChAdOx1 nCoV-19—Oxford/AstraZeneca/FioCruz). Recently, other RNA vaccines have become available; however, in this study, our cohort was comprised of the first two vaccines. Vaccination was advantageous primarily to control the development of severe forms of disease and death, which was true for both inactivated and viral vector vaccines (Bayhan and Guner, 2022; Cerqueira-Silva et al., 2022; Sevinc et al., 2022; Promlek et al., 2023; Shi et al., 2023; Wei et al., 2023). The adaptive immune response induced by these vaccines, production of neutralizing antibodies, and cellular response have been extensively studied (Escobar et al., 2022; Kim et al., 2022; Sheng et al., 2022; Wang et al., 2022; Zhao et al., 2022). Furthermore, the understanding of COVID-19 pathology is evolving since its emergence in 2019 (Mazza et al., 2022; Shafqat et al., 2022; Davis et al., 2023). Here, we sought to understand whether vaccination could interfere with innate immunity and consequently regulate the exacerbated cytokine storm caused by the virus. Then, our initial objective was to analyze whether the cells from immunized donors were already trained due to the vaccination or not.

In vitro training of human monocytes is well stablished and consists in a first stimulus with β-glucan, the Bacillus Calmette–Guérin (BCG) vaccine, or oxidized low-density lipoprotein (oxLDL). In vivo, vaccines could serve as a first stimulus. After the first stimuli, a challenge with LPS is used as an unrelated secondary bacterial stimulus (Bekkering et al., 2016). To evaluate the extension of training, we analyzed if monocytes from unvaccinated and vaccinated donors presented altered phenotype under the challenge with LPS. Moreover, the training phenotype is not limited to just one feature but actually to a set of markers that we showed throughout this study.

We observed an increase in the frequency of classical monocytes in InVac-immunized individuals following stimulation with LPS. The same result was observed in the literature using another inactivated vaccine against COVID-19 (Liu et al., 2021). Classical monocytes are the main cells involved in phagocytosis, production of reactive oxygen species, high production of IL-10, and low production of TNF-α in response to LPS. The increase in this population in the InVac group could favor the control of the inflammatory response (Yang et al., 2014). Moreover, CD163 is a scavenger receptor involved in regulating the immune response and resolution of inflammation (Onofre et al., 2009) and is also a marker that characterizes trained cells (Quintin et al., 2012; Guha et al., 2021). We observed an increase in classical and intermediate monocytes from the nRVVac-vaccinated group, suggesting that monocytes vaccinated with nRVVac are trained for the regulatory profile. It is important to note that LPS challenge activates and stimulates classical monocyte survival and causes a reduction in CD163 expression (Onofre et al., 2009). Therefore, we observed that the frequency of CD163 expression in classic monocyte is increased, accompanying the increase in the frequency of this population; however, the MFI of this marker showed a reduction after stimulation with LPS. In addition, after the stimulus with LPS, there was a reduction in the frequency of intermediate monocytes; consequently, there was a reduction in the frequency of CD163 and in the MFI due to the stimulus with LPS in CTRL group, indicating that both vaccines are playing a role in monocytes profile differentiation.

To concatenate all the immunophenotyping data and provide an overall view of our findings, we used tSNE as an integrative data analysis tool. For this, we considered CD163 as a marker that was increased in trained cells and therefore considered the starting point; monocytes from vaccinated individuals, after stimulation with LPS, presented a down-modulation of clusters that did not express CD163. More clusters, specifically Pop1 and Pop2, expressed this marker. In addition, cells that express more CD163 also expressed the highest levels of CD16, CD86, and HLA-DR, demonstrating that they are monocytes with an antigen-presenting function, a typical function of intermediate monocytes (Jakubzick et al., 2017), which could be advantageous post-immunization. We also observed that before stimulation with LPS, all individuals had the same MFI and frequency of expression of surface markers, except for CD163 in classic monocytes, suggesting that the basal profile of monocytes from different experimental groups was similar, but they responded differently when stimulated, suggesting that monocytes from vaccinated individuals were less inflammatory than those from non-immunized individuals, showing a regulatory phenotype. In the context of COVID-19, the balance of cytokines determines the course of the disease (Xu et al., 2020). Vaccines could help regulate the exacerbated production of cytokines by monocytes, thus avoiding high levels of inflammation when the vaccinated individual encounters SARS-CoV-2.

Another evidence that reinforces the regulatory phenotype found in monocytes from vaccinated individuals is the reduction in IL-6 production in the InVac group and reduction in TNF-α in both the vaccinated groups. Although the nRVVac group did not show a reduction in IL-6, it was the only group that showed an increased expression of IL-10, an important regulatory cytokine (Ip et al., 2017). Indeed, IL-6 and TNF-α are pro-inflammatory cytokines involved in the immunopathology of COVID-19; IL-6, in particular, can lead to macrophage activation syndrome and cytokine storm, the most serious complications of COVID-19 (Qin et al., 2020; Tan et al., 2020). Between other inflammatory cytokines, IL-6 and TNF-α are downregulated and IL-10 can be upregulated in monocytes trained for the regulatory profile (Cauchi and Locht, 2018; Quinn et al., 2019; Guha et al., 2021).

Moreover, all groups produced the same levels of IL-1β, and only the InVac group did not increase IL-1Ra production after stimulation with LPS. Therefore, our results differ from others in the literature that demonstrated that cells trained with a regulatory profile release less IL-1β when trained by natural infection with Plasmodium falciparum (Guha et al., 2021) and more IL-1Ra when trained with helminth extract (Quinn et al., 2019). Considering that we were comparing with a vaccine and that the supposed training took place in vivo, the type of stimulus likely influenced which genes are affected during training-induced chromatin remodeling, which will determine how the cell will respond when challenged.

In addition to cytokine production, we also evaluated the gene expression of proteins related to immunometabolism and the tolerance phenotype, such as IRG1 (Li et al., 2013), to confirm whether the reduction in IL-6 and TNF-α production was not associated with the tolerance phenotype (Li et al., 2013). Also associated with epigenetic changes and contrary to training, cells of the innate immune system can be suppressed after prolonged exposure to an infectious agent, persistent pro-inflammatory signals, or specific antigens such as LPS. In monocytes, this phenomenon is called immune tolerance (Li et al., 2013). Although in the trained regulator profile, there is also a reduction in the production of pro-inflammatory cytokines, it differs from the tolerance phenotype, as tolerant cells increase the expression of immunoresponsive gene 1 (IRG1) (Li et al., 2013), which causes immunometabolic changes associated with itaconate pathway that leads the monocyte to the state of immunoparalysis (Domínguez-Andrés et al., 2019). This phenotype is characterized by reduced production of pro-inflammatory cytokines but without increasing the production of regulatory cytokines and the expression of membrane proteins that characterize the trained cell (Li et al., 2013). We observed a reduction in IRG1 expression in the nRVVac group and maintenance of expression levels in the InVac group, demonstrating that the reduction in pro-inflammatory cytokine production was not caused by tolerization. Additionally, an increase in the aerobic glycolysis pathway was associated with the training phenotype (Cheng et al., 2014; van der Meer et al., 2015). Trained monocytes from individuals vaccinated with BCG showed increased expression of GLUT1, LDHA, and HK2 (Kühtreiber et al., 2020). Here, we observed an increased expression of glycolysis-related proteins in the nRVVac group. Moreover, we did not observe alterations in the expression of SET7 and SDHB proteins, which are related to the induction of the pro-inflammatory training phenotype (Keating et al., 2020), reinforcing the idea that monocytes from vaccinated groups may be trained to the regulatory phenotype. Altogether, the immunophenotype, cytokine production, and immunometabolic control could indicate the polarization of the training phenotype to a regulatory milieu.

Chromatin remodeling of cytokine gene promoters or transcription factors is a hallmark of innate training that suppresses the antigen receptor somatic recombination seen in the adaptive immune response (Van Der Heijden et al., 2018; Kar and Joosten, 2020). Hence, we performed a chromatin immunoprecipitation assay followed by specific gene amplification to evaluate which gene promoters or regions of promoters were under epigenetic control in monocytes from vaccinated subjects. Here, we evaluated three parameters of epigenetic modulation: pan acetylation and H3K4me, which are often associated with gene activation and increased DNA accessibility (Van Der Heijden et al., 2018), and H3K27me3, which is associated with gene repression and DNA inaccessibility (Van Der Heijden et al., 2018). A reduction in pan acetylation was observed for the IL6 gene in both vaccinated groups, with unchanged H3K4me3 levels. These data are consistent with the reduction in IL-6 release observed in vaccinated individuals. Guha et al. (2021) trained monocytes for the regulatory profile using Plasmodium falciparum-infected red blood cells as a stimulus and observed an increase in chromatin accessibility for IL-6 transcription through an increase in H3K4me3. However, 5 days after challenge with P. falciparum-infected red blood cells, they observed a reduction in H3K4me3 and, consequently, a decrease in chromatin accessibility in the IL6 gene (Guha et al., 2021). In our results, despite the absence of alterations in H3K4me3 of IL6, we observed chromatin closure due to a reduction in pan acetylation in vaccinated individuals, corroborating the regulatory phenotype.

Interestingly, we observed that although TNF-α was suppressed in vaccinated individuals, the TNF gene was more accessible for transcription in nRVVac-vaccinated individuals. Using qPCR of the TNF genes and IL6, we evaluated if the genes were transcribed more or less and verified that the transcription of IL6 did not differ between the groups, both basal and stimulated. However, when we analyzed the TNF gene, we detected an increase in the transcription of this cytokine when stimulated with LPS. These data suggest that although TNF-α transcription is increased, its translation or release may be regulated by a post-transcriptional mechanism.

Regarding IL1B, the promoter regions evaluated here were not under epigenetic control, which is in line with IL-1β production by monocytes, although other studies using trained cells in the regulatory profile showed a reduction in IL-1β release (Quinn et al., 2019; Murphy et al., 2023). The IL10 gene showed an increase in chromatin accessibility in both vaccinated groups with reduced H3K27me3, although only the nRVVac group showed an increase in IL10 cytokine release, probably due to its delayed release kinetics, which increased in a time-dependent manner, with a release peak between 20 and 48 h after stimulation with LPS (Malefyt et al., 1991; Amoroso et al., 2012). Finally, we observed enhanced chromatin accessibility in the CCL2 gene, mediated by a reduction in H3K27me3. The increased release of this chemokine has been reported in monocytes from individuals trained with the adenoviral vector COVID-19 vaccine to a regulatory profile (Murphy et al., 2023) and in trained monocytes with lipoprotein to a pro-inflammatory phenotype (Groh et al., 2018). The increased production of this chemokine enhances the LPS-induced production of IL-10 and causes increased migration of immune cells to a site with an infectious agent, which may favor a rapid response and resolution of the infection (Sierra-Filardi et al., 2014; Gschwandtner et al., 2019).

In our cohort, we compared two types of vaccines and demonstrated that inactivated COVID-19 vaccines trained human monocytes to a regulatory phenotype, mediated by histone modifications in the IL6 and IL10 genes, whereas the non-replicating viral vector COVID-19 vaccine drove human monocytes to a regulatory phenotype, mediated by histone modifications in the IL6, IL10, TNF, and CCL2 genes. Our data collectively suggest that both vaccines may be associated with the training of monocytes, leading them to adopt a regulatory phenotype. We demonstrated that each vaccine activates distinct chromatin remodeling regions, resulting in different immune responses and metabolic regulation. These differences could potentially contribute to the effectiveness of vaccines in controlling excessive inflammation upon subsequent SARS-CoV-2 infection in patients.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding author.

This study was approved by the Research Ethics Committee (CEP) of the School of Pharmaceutical Sciences of Ribeirão Preto (CAAE 30736120.7.0000.5403). The patients/participants provided their written informed consent to participate in this study.

MA: conceptualization, methodology, validation, formal analysis, investigation, data curation, writing—original draft, writing—review and editing, and visualization. FS, HG and RC: conceptualization, methodology, validation, formal analysis, investigation, and writing—review and editing. FM, JL and CF: methodology, validation, and writing—review and editing. FF: conceptualization, methodology, resources, writing—review and editing, supervision, project administration, and funding acquisition. All authors contributed to the article and approved the submitted version.