Systemic lupus erythematosus (SLE) is characterized by the dysregulation of immunocompetent cells that can generate pathogenic autoantibodies after self-antigen exposure to attack systemic organs [1]. The pathogenesis of SLE are multifactorial [2, 3]. However, few studies have discussed the potential roles of plasma cell-free DNA (DNA${}^{\text{pcf}}$) [4].

Under normal circumstances, DNA molecules are located intracellularly and embedded compactly in the nucleus as nuclear DNA (nDNA) or in the mitochondrial nucleoid as mitochondrial DNA (mtDNA) [5]. On the contrary, cell-free DNA (DNA${}^{\text{cf}}$) are the DNA molecules that are released into the tissue fluid from cells all over the whole body [6]. They have been detected in cerebrospinal fluid [7], pleural fluid [8], saliva [9], urine [10] and plasma [11] for different purposes. The DNA${}^{\text{cf}}$ present in blood circulation is called “plasma cell-free DNA (DNA${}^{\text{pcf}}$)”, and most of them are originated from circulating hematopoietic cells [12]. Analysis of DNA${}^{\text{pcf}}$ has been employed for the diagnosis of acute myeloid leukemia as a non-invasive tool, which is as accurate as the invasive bone marrow aspiration or biopsy [13]. Human DNA${}^{\text{pcf}}$ was first described by Mandel and Metais in 1948 [11]. From then on, the clinical usefulness of DNA${}^{\text{pcf}}$ has been emphasized in predicting the outcomes of critical patients with myocardial infarction [14], sepsis [15] or stroke [16]. High levels of DNA${}^{\text{pcf}}$ were first found in SLE patients in 1966 [17]. Later on, some studies suggested that DNA${}^{\text{pcf}}$ might serve as an exposed self-antigen to trigger the pathogenesis of SLE [18]. Because the DNA${}^{\text{pcf}}$ is mainly derived from intracellular mtDNA or nDNA, it is of great interest and clinical relevance to explore the roles of mitochondrial DNA${}^{\text{pcf}}$ (mtDNA${}^{\text{pcf}}$) and nuclear DNA${}^{\text{pcf}}$ (nDNA${}^{\text{pcf}}$) simultaneously in SLE patients.

Dysregulated antioxidant capacity with imbalanced production of reactive oxygen species (ROS) has been speculated in the development of SLE [19, 20]. Abnormal ROS levels would cause oxidative damages or oxidative modifications in the intracellular or extracellular components of cells in situ, nearby, or remotely. These components may include DNA, lipid, proteins, and phospholipids. The events would trigger subsequent cellular dysfunction or autoimmune reactions in SLE patients [19, 20, 21]. Elevated plasma 8-hydroxy-2’-deoxyguanosine (8-OHdG) representing oxidative DNA damage [21, 22, 23] and plasma malondialdehyde (MDA) representing lipid peroxidation [21, 22, 24] have been reported in SLE patients with different clinical implications. The levels of accumulated 8-OHdG and/or MDA in the plasma could be effective surrogates to represent the oxidative stress in SLE patients [25].

Viral infection causing incidental self-antigen presentation has been proposed as a possible mechanism underlying the autoimmune reaction in SLE. Similar to the immune interaction implicating high levels of interferon (INF) observed between invading virus and host pattern/pathogen recognition receptors (PRRs), a phenomenon of type I INF signature has also been recognized as a prominent feature of SLE [26]. A clinical trial revealed that the treatment of anti-INF-alpha monoclonal antibody, sifalimumab, could achieve a significant improvement of symptoms/signs in moderate to severe SLE patients [27]. Recently, the importance of human C-type lectin domain family 5 member A (CLEC5A, a member of PRRs) in interacting with the invading virus was reported [28]. Teng et al. [29] reported that through toll-like receptor 3 (TLR, a subtype of PRRs), influenza virus infection in bone marrow-derived macrophages from CLEC5A${}^{-/-}$ mice would result in an increase of IFN-alpha. Based on these findings, we have speculated that decreased or dysregulated leukocyte CLEC5A expression might be involved in SLE pathogenesis.

In this study, we focused on the measurement of copies of mtDNA${}^{\text{pcf}}$ and nDNA${}^{\text{pcf}}$ to explore their association with the levels of plasma 8-OHdG and plasma MDA as well as the leukocyte CLEC5A mRNA transcripts in SLE patients. We aimed to elucidate the role of DNA${}^{\text{pcf}}$ in the pathogenesis of SLE.

In summary, we have demonstrated that: (1) SLE patients tended to have lower mtDNA${}^{\text{pcf}}$ copies, higher nDNA${}^{\text{pcf}}$ copies, higher levels of plasma 8-OHdG, and lower levels of plasma MDA than did HCs, but they had similar leukocyte CLEC5A mRNA levels (Table 2). (2) SLE patients with higher SLEDAI tended to have lower mtDNA${}^{\text{pcf}}$ copies, and SLE patients with nephritis had higher nDNA${}^{\text{pcf}}$ copies but lower mtDNA${}^{\text{pcf}}$ copies (Table 3). (3) In SLE patients, higher nDNA${}^{\text{pcf}}$ copies were correlated with higher levels of plasma 8-OHdG but lower levels of plasma MDA (Table 4). (4) In HCs, higher nDNA${}^{\text{pcf}}$ or lower mtDNA${}^{\text{pcf}}$ copies were associated with higher leukocyte CLEC5A mRNA expression levels, which showed no associations in SLE patients (Table 4) and (5) In active SLE patients, increases in mtDNA${}^{\text{pcf}}$ copies and leukocyte CLEC5A mRNA levels were found after rituximab treatment (Table 5). We have thus speculated that lower mtDNA${}^{\text{pcf}}$ copies, higher nDNA${}^{\text{pcf}}$ copies, oxidative damages/stress as well as dysregulated leukocyte CLEC5A expression might be implicated in the pathogenesis of SLE. A proposed mechanism is illustrated in Fig. 1.

Fig. 1.

Illustration is a proposed mechanism of DNA${}^{\text{𝐩𝐜𝐟}}$ released from immunocytes/neutrophils under pathogen invasion or stress stimulations, which is involved in the self-antigen presentation in SLE pathogenesis. (a) Lytic NETosis, DNA molecules, including nDNA and mtDNA, as well as other proteins in extruded NET can entangle the pathogens and trigger immune reactions to result in a subsequent lysis of their own cell membrane and the death of invading microbes. (b) Vital NETosis, different from the cell death during lytic NETosis, some immune cells secrete only mtDNA, harboring high immunogenicity, into NET, which enable them to maintain alive even if they become anucleate. (c) Both lytic and vital NETosis could cause the release of DNA into blood circulation and involve the presentation of self-antigen. (d) ROS plays an important regulator in adjusting NETois, and the degrees of ROS-elicited modifications could be reflected by the abundance of 8-OHdG and MDA. (e) CLEC5A can drive human immune response to defend viral infections and is a critical receptor in innate immune system causing NETosis.

Higher levels of DNA${}^{\text{pcf}}$ in SLE patients was first described by Tan et al. in 1966 [17]. Up to now, DNA${}^{\text{pcf}}$ has been regarded as an important biomarker in autoimmune rheumatic diseases [40]. Nevertheless, DNA${}^{\text{pcf}}$ including mtDNA${}^{\text{pcf}}$ and nDNA${}^{\text{pcf}}$ and their potential implications in disease development were less explored in SLE. Giaglis et al. [41] reported that a high level of DNA${}^{\text{pcf}}$ is resulted from mtDNA${}^{\text{pcf}}$ rather than from nDNA${}^{\text{pcf}}$ in SLE. However, Truszewska et al. [42] found that high DNA${}^{\text{pcf}}$ is due to high nDNA${}^{\text{pcf}}$ in SLE. They also failed to find differences in mtDNA${}^{\text{pcf}}$ copies between SLE patients and HCs. The present investigation has revealed higher nDNA${}^{\text{pcf}}$ but lower mtDNA${}^{\text{pcf}}$ copies in SLE patients compared with those of HCs (Table 2). Similar to the Q-PCR protocol described by Giaglis and Truszewska and those reported in the literature [41, 42, 43], we measured the mtDNA${}^{\text{pcf}}$ and nDNA${}^{\text{pcf}}$ copies in a given volume through the established standard curves. There are small differences between the absolute copies reported by Giaglis and Truszewska and the relative copies detected in the present investigation. Although the data by Giaglis and Truszewska were generated in more precise manner, the results of the present study could still provide useful information. So to speak, our patients exhibited higher SLEDAI of 8 in median than the cohort in Truszewska’s of 7.42 in average or in Giaglis’ of 2 in median, indicating more severe clinical manifestations in our patients. These similarities and disparities suggest that the mechanism underlying the release of intracellular DNA into plasma to become nDNA${}^{\text{pcf}}$ or mtDNA${}^{\text{pcf}}$ in SLE patients needs re-appraisal. Through cell death, degradation or damage and extrusion of intracellular molecules, partial or complete genomes from in situ or nearby tissue cells that may or may not contain invading viruses or microbes, would be continuously shed into human body fluids to form DNA${}^{\text{cf}}$ [44]. Different from the other types of DNA${}^{\text{cf}}$, most DNA${}^{\text{pcf}}$ are originated from hemopoietic cells [12]. As a result, necrosis, apoptosis, degradation or damages of dysregulated SLE-immunocytes might be the major contributors to the release of SLE-DNA${}^{\text{pcf}}$ [40, 45]. Among the different types of immunocytes, neutrophil is the most abundant one with plenty of DNA molecules and has been implicated in the trigger and perpetuation of SLE [46]. DNA${}^{\text{pcf}}$ from dysregulated neutrophils might be involved in the pathogenesis of SLE.

Similar to necrosis, apoptosis or exocytosis of general cells, neutrophils are able to undergo a unique mechanism of metamorphosis resulting in neutrophil extracellular trap (NET), which is accordingly called “NETosis”, to deal with the invading pathogens or sterile inflammation (Fig. 1). Stimulated by ROS and extruding web-like NET, which is composed of a nucleus-derived decondensed DNA coated with histones, granular proteins, and cytoplasmic proteins, into the extracellular space, some neutrophils can entangle the pathogens and trigger immune reactions to result in a lysis of their own cell membrane and the death of invading microbes, termed lytic NETosis (Fig. 1a) [47]. Different from the lytic NETosis, some “immortal neutrophils” may undergo vital NETosis to keep intact cell membranes, in which they only extrude a small amount of mtDNA into NET, allowing themselves to remain alive and continue to exert antimicrobial activity, even if they become anucleated (Fig. 1b) [48, 49]. Through the release of neutrophil nDNA${}^{\text{pcf}}$ and mtDNA${}^{\text{pcf}}$, NETosis might lead to a self-antigen exposure and participate in the pathogenesis of SLE (Fig. 1c) [40, 50, 51].

As shown in Fig. 1, both lytic and vital NTEosis in neutrophils could contribute to the release of nDNA molecules, which could account for our results of higher nDNA${}^{\text{pcf}}$ in SLE patients than in HCs (Table 2). In our cohort, SLE patients tended to present lower levels of mtDNA${}^{\text{pcf}}$ (Table 2) and such a decrease was highly correlated with the high SLEDAI (Table 3). This implies that mtDNA are captured in the NET without releasing into the plasma, which is a possible result of vital NETosis. The mtDNA-enriched NET may triggered a vicious cycle of immune response because the high CpG motif in mtDNA that would propagate immunogenicity cascade in innate immunity [52]. Therefore, based on the findings of low mtDNA${}^{\text{pcf}}$ copies and higher median SLEDAI of 8 as demonstrated in our cohort, similar mtDNA${}^{\text{pcf}}$ copies and less higher mean SLEDAI of 7.42 in Truszewska’s cohort [42] or high mtDNA${}^{\text{pcf}}$ copies but low median SLEDAI of 2 in Giaglis’ cohort [41], we speculated that the different results of mtDNA${}^{\text{pcf}}$ or nDNA${}^{\text{pcf}}$ might be hybrid results from diverse proportions of lethal (lytic) and vital NETosis (Fig. 1a–c) regulated by oxidative stress (Fig. 1d). To validate this speculation, we need again to evaluate the NETosis activity and the differences among intracellular DNA, cell membrane-bound DNA and DNA${}^{\text{pcf}}$ of SLE patients in the future.

Lupus nephritis is an important clinical manifestation in SLE, and the role of oxidative DNA damage has been evaluated [53]. However, the role of DNA${}^{\text{pcf}}$ remained unclear. Because 8-OHdG is a stabilized product after oxidative DNA damage, it is an optimal marker to reflect the ROS levels. Our preliminary results showed that SLE patients harbored higher levels of plasma 8-OHdG and higher nDNA${}^{\text{pcf}}$ copies than did HCs’ (Table 2), and the high plasma 8-OHdG levels were related to high nDNA${}^{\text{pcf}}$ copies among SLE patients (Table 4). It is suggested that high oxidative stress might cause an abundant release of nDNA from impaired leukocytes during NETosis. We also found that SLE patients with nephritis had higher nDNA${}^{\text{pcf}}$ copies (Table 3), suggesting that damaged kidney might be related to impaied clearance of NET remnant or nDNA${}^{\text{pcf}}$. As a result, higher nDNA${}^{\text{pcf}}$ copies were detected and the vicious cycle sustained. On the contrary, we did not find any association between the levels of plasma 8-OHdG and mtDNA${}^{\text{pcf}}$ copies in SLE patients, but SLE patients with nephritis tended to have a lower mtDNA${}^{\text{pcf}}$ copies (p = 0.082) (Table 3). However, Fernandez et al. [54], reported that an elevation in mtDNA${}^{\text{pcf}}$ would define a subgroup of SLE patients with membranous lupus nephritis. Due to the dynamic changes of intracellular and extracellular mtDNA molecules and the inconsistency in reported literature, Truszewska normalized their ratios between intra to extra cellular mtDNA copy number to correlate the occurrence of nephritis, and it showed good association [42]. Besides the mtDNA${}^{\text{pcf}}$, the high levels of urine mtDNA contents from damaged kidney (i.e., a kind of mtDNA${}^{\text{cf}}$ in urine) have been advocated to correlate severity of nephritis [55, 56]. The precise role of mtDNA either intracellular or extracellular to reflect lupus nephritis needs further investigation.

On the other hand, we failed to observe a higher level of plasma MDA in SLE patients (Table 2) and the levels of plasma MDA were inversely correlated with that of nDNA${}^{\text{pcf}}$ (Table 4). Unlike the stabilized 8-OHdG, MDA is the peroxidized product of lipid with highly reactive activity, the unstable free form could conjugate with proteins through covalent alterations of amide and amine groups of peptides to form MAD adducts at a post-translational level, giving rise to neo-epitopes that can elicit autoantibody responses [52]. The literature has demonstrated that high levels of MDA adducts would trigger obvious autoimmune reactivity in SLE patients [57]. Similar to oxidized low density lipoprotein, MDA adducts could induce the formation of NET and propagate immune reaction [58]. More precisely speaking, MDA could exist in free and conjugated (adduct) forms, and a free form MDA could positively extrapolate total MDA to predict oxidative stress in exhaled breath condensate as well as urine. But the relevant role of free MDA in plasma/serum, nasal fluid, or saliva to predict total MDA remained controversial and dismal. Different proportions of free MDA and conjugated MDA (MDA-adducts) in different clinical samples might account for these discrepancies [59, 60]. In the present study, what we measured was the free plasma MDA rather than MDA adducts and lower levels of plasma MDA were inversely correlated with high nDNA${}^{\text{pcf}}$ copies in SLE patients (Table 4). It is suggested that large amounts of MDA were underestimated because of the presence of high proportion of MDA-protein adducts in SLE patients.

CLEC5A can drive human immune response to defense viral infections [29, 37]. Recent studies showed that CLEC5A is a critical receptor in innate immune system, causing NETosis to deal with Dangue virus [28, 37] or Japanese encephalitis virus [61]. Among the 24 HCs, we found that higher levels of leukocyte CLEC5A mRNA level were associated with higher nDNA${}^{\text{pcf}}$ copies or a lower mtDNA${}^{\text{pcf}}$ copies, suggesting a possible role of DNA${}^{\text{pcf}}$ in CLEC5A related immune response (Table 4). High nDNA in the NET is required for neutrophils to trigger suicidal NETosis and to defend against invading pathogens [62]. The present results have indicated that the leukocytes in healthy subjects hold functional CLEC5A to trigger proper immune reaction. In contrast, such function may vanish or be dampened in SLE patients, because the above association becomes absent among SLE leukocyte CLEC5A mRNA transcripts and nDNA${}^{\text{pcf}}$ copies, suggesting an impaired leukocytes/neutrophils in SLE [46]. Influenza infection in bone marrow-derived macrophages of CLEC5A${}^{-/-}$ mice (dampened CLEC5A) would induce an increase in the secretion of IFN-$\alpha$ [29], which is similar to the IFN signature in SLE [26]. Of note, although the added cohort of 3 SLE patients was small in number, the present preliminary results showed an increase of leukocyte CLEC5A mRNA level after the treatment with rituximab (Table 5). Taken together, we propose that an aberrant expression and impaired function of CLEC5A may be implicated in the pathogenesis of SLE.

Similar to CLEC5A, cyclic guanosine monophosphate (GMP) - adenosine monophosphate (AMP) synthase (cGAS) is also a member of PRR to act as an additional first-line host immune defense that can induce type I IFN response through its interaction with downstream stimulator of interferon genes (STING), i.e., the cGAS-STING pathway [63]. Recently, many immunological studies have focused on the cGAS-STING pathway as another player in the pathogenesis of autoimmune diseases, including SLE [64]. It is speculated that cGAS could sense and interact with free plasma intrinsic or extrinsic DNA fragments derived from engulfed NETs to trigger immune cascade [65]. This process would be more potent if the cGAS is stimulated by free mtDNA fragments [66, 67]. Intriguingly, in the present investigation, we observed that SLE patients with higher SLEDAI tended to have lower mtDNA${}^{\text{pcf}}$ copies extracellularly. But we have also demonstrated a lower mtDNA copy number intracellularly in our previous report [35]. In addition, the mtDNA${}^{\text{pcf}}$ copies was increased after rituximab treatment in severe SLE patients (Table 5). These dynamic alterations of mtDNA strengthen the possibility of the interaction among mtDNA${}^{\text{pcf}}$ molecules and consumption of mtDNA fragments through PRR, cGAS or CLEC5A, which might contribute to the pathogenesis of SLE.

Viral infection has been proposed to play a role in the SLE pathogenesis. A meta-analysis demonstrated that SLE patients express higher IgG titers to EBV viral capsid antigen (VCA) and early antigen (EA), and exhibit higher positive rates for EBV DNA [68]. Furthermore, thanks to the new computational methods to account for the genetics of human diseases, it has been found that nearly half of SLE risk loci are occupied by the Epstein-Barr nuclear antigen 2 (EBNA2) protein and many co-cluster with human transcription factors [69]. As to the lupus nephritis, some studies have revealed that a higher serum EBV load is associated with renal damage, and the involved kidneys can express higher EBV-latent membrane protein‑1 (EBV-LMP1) and EBV-encoded RNA 1 (EBER-1) [70, 71]. Notably, higher EBV loads were correlated with higher DNA${}^{\text{cf}}$ levels in SLE patients [70]. The measurement of DNA${}^{\text{pcf}}$ copies to dissect the interrelations between EBV infection and self-antigen presentation in SLE requires further investigations.

In conclusion, we have demonstrated that higher nDNA${}^{\text{pcf}}$ copies, lower mtDNA${}^{\text{pcf}}$ copies, increased ROS-elicited oxidative DNA damages and dysregulated leukocyte CLEC5A expression might participate in the pathogenesis of SLE. Further evaluation of the underlying mechanisms in this regard is necessary in the future.

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

HTL, CSL, CYT and YHW designed the study. HTL, CSL and SCP carried out the experiments. HTL, CSL, SCP and WSC wrote the original draft. CYT and YHW reviewed and edited the manuscript. All authors have read and approved the final version of the manuscript.

This study was approved by the Institutional Review Board of Taipei Veterans General Hospital (IRB No: 2013-04-043B).

We acknowledge the support of this study by Mackay Memorial Hospital and Taipei Veterans General Hospital.

This study was partly supported by a research grant from the Ministry of Science and Technology (MOST 111-2320-B-371-001), an intramural grant (110-CCH-MST-127) from Changhua Christian Hospital, and a grant from National Science Council (NSC102-2314-B-195-020), Taiwan.

The authors declare no conflict of interest. Given his role as Guest Editor and Editorial Board Member, Yau-Huei Wei had no involvement in the peer-review of this article and has no access to information regarding its peer-review. Full responsibility for the editorial process for this article was delegated to Pietro Gentile.