A new study by Swedish researchers from Linköping University – Sweden and Ryhov County Hospital, Jönköping – Sweden has that SARS-CoV-2 infections especially in those that were hospitalized, caused major alterations to key components of the immune system such as the monocytes and dendritic cells for up to six months after recovery.
The study findings were published in the peer reviewed journal: Frontiers in Immunology.
To address the lack of knowledge concerning myeloid cells in COVID-19 we have investigated the effects that SARS-CoV-2 infection exerts, both initially and long-term, on monocyte and DC subsets. We found both initial and long-term alterations to the frequencies and phenotypes of these cells, indicating an immune dysregulation and dysfunction.
Monocytes are today divided into three functionally different subpopulations distinguished primarily by their expression of CD14 and CD16 (9). In mouse models of acute lung injury and in acute respiratory distress syndrome, as seen in severe COVID-19, circulating monocytes have been shown to play a pivotal role in driving inflammation (42).
Monocyte frequencies have been found to be lower during acute COVID-19 compared to convalescence (43). In line with this, the levels of overall monocytes in the patients in our study were significantly decreased at inclusion, when compared to the later time points.
A factor affecting the levels of monocyte subsets could be virus-mediated cell death during COVID-19, since they all express the SARS-CoV-2 receptor, ACE2 (43), and there is evidence of viral antigens present within ncMo for over a year after initial infection (44).
In addition, CD16+ monocytes can also be infected in an ACE2-independent manner, leading to inflammatory cell death (45). Our observed sustained increase in cMo in COVID-19 patients in comparison to healthy controls is consistent with previously published data (26).
The elevated levels of these cells at 6-7 months could be part of the host resolving long-term effects of the SARS-CoV-2 infection as the key roles of cMo include tissue repair and anti-apoptotic functions (46, 47). We saw an initial drastic drop in the level of ncMo, which recovered in convalescence, consistent with other reports (26, 48).
This initial drop could be due to the recruitment of CD16+ myeloid cells to inflamed lung tissue during infection (27, 48, 49). Different patterns have been seen in severe compared to mild COVID-19 in the levels of circulating iMo, with decreased levels in severe disease when measured during active disease (27). In our patient cohort we found decreased levels of iMo compared to healthy controls, which failed to recover, indicating a long-lasting effect.
Recent studies, through single-cell resolution methods, point to the expansion of suppressive myeloid cells in the blood as a hallmark of severe COVID-19 (17, 26), as confirmed by our data. These cells (MDSC) are a heterogeneous population of myeloid cells that exist during normal conditions in tissues (50).
The levels of MDSC have potent immune suppressive properties and have been found to correlate to impaired T cell responses (51, 52). This sustained elevation suggests a long-lasting suppressive state, together with the suppressive T cell phenotypes defined in our COVID-19 cohort (3) indicate a long lasting impaired immune response after COVID-19.
In line with other studies, we found that COVID-19 had a major impact on circulating DC subpopulations (26), with major reductions during active disease for pDC (17), cDC1 (27), and combined cDC2/cDC3 (29). The reduction in DC subsets at inclusion could be due to several factors such as redirection of circulating pDC and cDC1 to lymphoid tissue (28, 53), and the recruitment of cDC2 subtypes into the lung tissue, though this is not seen for pDC and cDC1 (27, 54).
In addition to relocation from circulation to tissue, the decrease in DC might depend on their destruction by direct or indirect viral effects, since SARS-CoV-2 can bind and enter DC through CD147 and DC-SIGN (55, 56). The combined cDC2/cDC3 population increased to levels higher than in healthy controls and these elevated levels persisted long-term.
A lasting effect with higher levels of HLA-DR+CD11c+ DC has been noted in individuals needing hospitalization, whereas normal levels of CD141+ DC, i.e., cDC1 were observed (32). This elevated level of cDC2/cDC3 could play a role in ongoing, long-term, systemic inflammation due to the damage inflicted by COVID-19 on the host.
Almost all DC functions are affected during COVID-19 (28) with probable consequences for the progression and outcome of the illness. Initial type I and III IFN responses are essential to the resolution of viral infection and are found to be impaired in severe COVID-19 (57), which could be due to initial DC depletion.
Given the importance of DC for the initial activation of T cell responses, their depletion or tissue relocation could hamper the subsequent T cell response (58). Overall, the long-lasting alteration in the DC compartment could explain the sustained T cell dysfunction seen following COVID-19 (3, 59).
In different diseases such as cancer and severe infections, the phenotype and functionality of circulating monocyte and DC subsets are altered (60, 61). One of the immune checkpoint molecules affected is CD86, which is known to be upregulated in settings with ongoing immune activation and chronic inflammation (62–64).
Our phenotypic analysis showed little to no decrease in CD86 expression during acute disease, on monocyte subsets and MDSC, and this was followed by elevated CD86 levels on all these cells at 6-7 months. Previous studies have shown that monocyte subsets, and especially iMo, have decreased CD86 during acute infection (35, 65).
We found an increased level of expression on pDC and cDC1 at inclusion but not cDC2/cDC3. The elevated CD86 levels on pDC have been documented during acute SARS-CoV-2 infection (66). However, other studies found no effect on pDC, or on the myeloid DC subsets in acute disease (4, 10, 29, 67).
The expression of CD86 returned to levels comparable to healthy controls quickly for pDC but remained elevated on cDC1, cDC2, and cDC3 subsets at 6-7 months. This contrasts with a previously published study, which showed long-term reduction in CD86 expression on pDC and HLA-DR+CD11c+ DC subsets in hospitalized COVID-19 patients (32).
Elevated CD86 expression was found on ki67+ cDC2 and cDC3 subsets, i.e., a similar pattern as we have seen for these DC subsets, highlighting that the level seems to be connected to the time that they have been in circulation (29).
We suspect that the ongoing, low-level, systemic inflammation from lung repair following COVID-19 (68) could be the cause of the increased CD86 on monocyte and DC populations since CD86 expression is elevated in chronic HIV-1 and HCV, both infections that cause inflammation (62–64).
The elevated CD86 expression we found on the myeloid cells could play a role in hyperactivation or T cell impairment. The differences between our and other results could be due to the method of defining DC subsets, with many studies assessing CD14-HLA-DR+CD11c+ DC and not the different cDC subsets we have used in our study.
Previous studies have shown decreased HLA-DR levels in acute COVID-19 for cMo and ncMo (17, 35, 48, 67, 69), which is in accordance to our results. The iMo have been found to have downregulated HLA-DR during severe acute COVID-19 (17, 48, 69), which differs to our observation of no effect on HLA-DR at inclusion.
Regarding the DC subsets, the HLA-DR expression in severe COVID-19 has previously been found to be reduced across all circulating DC besides cDC1 (26). In our study HLA-DR was only significantly altered on cDC2/3 where it was reduced at early time points before returning to levels found in controls, which is in agreement with Marongiu et al. who showed a reduction of HLA-DR in cDC2 and cDC3 subsets (70). The observed initial low expression of HLA-DR on cDC1, cDC2/cDC3, cMo, and ncMo (29, 69, 71, 72) could be a sign of impaired functionality and part of the immunosuppression seen in severe disease.
PD-L1 has been shown to be dysregulated in COVID-19 patients (73). In our study we observed a consistently lower PD-L1 surface expression across all myeloid cells. For pDC and cDC subsets this aligns with gene and protein expression levels in hospitalized patients during acute disease (32, 73, 74). In contrast, an increase in PD-L1 expression has been seen in other studies on bulk monocytes (67, 75), and circulating DC subsets (66, 67).
In our data this reduced expression of PD-L1 persisted for the entire duration of the study, as was seen also in a previous study at 7 months post-COVID-19 (32). The loss of PD-L1 on monocyte and DC subsets might be due to shedding of soluble PD-L1, which is found to be elevated in the serum of COVID-19 patients (73) and to be one feature of critical COVID-19 (76). The lasting reduction that we observe in PD-L1 expression across all myeloid cell subsets requires further study to explore if it plays any role in COVID-19 pathogenesis and if it is due to these cells shedding PD-L1.
Concerning the migratory receptor CCR7, there was little change in the DC and monocyte subsets during acute disease compared to healthy controls, while long-term effects included decreased CCR7 on cMo, pDC and the two cDC3 populations.
This decrease is in accordance with previous findings for cDC at 7 months (32), but not for pDC, which had long-term increased CCR7 (32). We did not find any major alterations in the surface expression of CD83 across DC, which concurs with Venet et al. (74).
Overall, we identified marked alterations to the expression of surface markers across myeloid cell types following COVID-19, which could play a part in the resulting immune suppression. The long-term changes to the surface expression of these proteins on monocytes and DC may be due to ongoing inflammation or epigenetic changes resulting from severe COVID-19 (77), possibly altering progenitors in the bone marrow.
There are many reports that demonstrate that there is higher morbidity and mortality in males than females. Sex hormones have been implicated in the ability of females to cope with the infection better. In addition, factors such as the innate response, cytokines, T cells, and monocytes differ between males and females (78–80).
Takahashi et al. found that male patients had higher levels of ncMo than female during acute disease (79). We could not find this difference in ncMo, or any other differences in the myeloid compartment, between males and females.
COVID-19 severity has, besides age and biological sex, been linked to an array of clinical parameters such as the level of soluble urokinase plasminogen activator receptor (suPAR), CRP, LDH, and viral load (81–86). When exploring correlations between the clinical parameters measured in this study, the viral load correlated positively with the CRP levels and negatively with anti-spike IgG and neutralizing SARS-CoV-2 antibody levels, at inclusion.
In addition, the COVID-19 patients with high CRP levels had lower levels of anti-spike IgG and neutralizing antibodies at inclusion. We have previously shown that viral load correlates to disease severity (87), that is in turn connected to suboptimal development of germinal centers (88) that in turn might be the reason for the delayed IgG response. However, our correlations do not definitely mean causality and so their underlying mechanisms warrant further investigation.
This negative association between CRP and antibody levels early on during SARS-CoV-2 infection has, to our knowledge, not yet been made. It has previously been shown that higher levels of SARS-CoV2 specific antibodies during convalescence correlated with initial higher CRP levels (89), though we did not see this.
During acute disease in hospitalized patients, the levels of pDC, cDC2, and CD163+CD14- cDC3 are negatively correlated to CRP levels (29, 69). We confirm this to be the case for all circulating DC subsets at acute and six weeks post-inclusion, i.e., a faster DC recovery was predicted by lower initial levels of CRP.
This negative effect on DC levels could be due to CRP impairment of DC development (90) and the slow recovery of DCs could affect the ability to respond to new infections and activate T cell responses.
The levels of ncMo and iMo have been previously found to negatively correlate to CRP in acute disease (91), while we found a positive correlation for iMo levels in our study. Raised CRP is part of a highly inflammatory environment in COVID-19 with prolonged high viral load due to the lack of a strong type I IFN response (92). This in turn might have long-term effects on monocyte and DC subsets, which our data strongly supports.
A potential limitation of our study is the imperfect age and sex matching of the controls to the patients. This is of particular importance for some DC subsets such as pDC, which are decreased in older (>40 years) healthy individuals, whereas there are no significant effects on cDC (93, 94). Given that the levels of pDC returned to the levels of healthy controls we do not believe this to be an issue.
While biological sex does not seem to play a role for the levels of MDSC in general, their levels could be influenced during disease. A study of MDSC in mild to severe COVID-19 found higher levels of monocytic MDSC in males than in females (80).
The increase we found of MDSC in blood from hospitalized COVID-19 patients did not correlate with biological sex or age, however, our cohort was relatively small and did not contain cases of mild disease, so a larger cohort may be needed to observe these associations.
Another effect of our relatively small cohort is that with large number of parameters explored we have comparatively low statistical power. The MDSC levels increase with age and highly elevated levels are seen in severe infections and cancers (95–101). A general problem when comparing data from different COVID-19 studies is the definition of disease severity, that differs depending on country.
We defined severity according to the NIH guidelines, which are based largely on supplemental oxygen requirements (33), as opposed to the WHO scale (102).
In conclusion, given the long-lasting changes in the monocyte, DC and MDSC compartments, as seen in the altered frequencies of cell populations and expression levels of various surface markers, it is evident that COVID-19 impacts the development and functionality of these cells.
Further studies will be required to determine for exactly how long these alterations persist after severe COVID-19, and if they affect the type and quality of immune responses elicited against future infections.