Adenosine deaminase 2 regulates the activation of the toll-like receptor 9 in response to nucleic acids

Abstract

Human cells contain two types of adenosine deaminases (ADA) each with unique properties: ADA1, which is present in all cells where it modulates intracellular functions and extracellular signaling, and ADA2, which is secreted by immune cells. The exact intracellular functions of ADA2 remain undetermined and less defined than those of ADA1. ADA2 has distinct characteristics, such as low adenosine affinity, heparin-binding ability, and putative lysosomal entry. Here, we confirm that ADA2 is a lysosomal protein that binds toll-like receptor 9 (TLR9) agonists, specifically CpG oligodeoxynucleotides (CpG ODNs). We show that interferon-alpha (IFN-α) is secreted in response to TLR9 activation by CpG ODNs and natural DNA and markedly increases when ADA2 expression is downregulated in plasmacytoid dendritic cells (pDCs). Additionally, the pretreatment of pDCs with RNA further stimulates IFN-α secretion by pDCs after activation with CpG ODNs. Our findings indicate that ADA2 regulates TLR9 responses to DNA in activated pDCs. In conclusion, decreasing ADA2 expression or blocking it with specific oligonucleotides can enhance IFN-α secretion from pDCs, improving immune responses against intracellular infections and cancer.

Keywords

adenosine deaminase 2 / plasmacytoid dendritic cell / TLR9 / IL-3 / IFN-α / CpG ODN

1 Introduction

Adenosine is a crucial signaling molecule and the primary nucleoside metabolite of the purine salvage pathway. In the extracellular environment, adenosine binds to type-1 purinergic receptors and modulates the response of cells to activate signals [1]. Adenosine concentration is regulated by adenosine deaminases (ADAs). Two types of ADAs are found in humans: ADA1 and ADA2 [2]. ADA1 converts (deoxy-)adenosine to (deoxy-)inosine within the cytoplasm, essential to cell survival. Genetic deficiency states influencing ADA1 expression cause lymphocytotoxicity and a dramatic decrease in B and T cells, manifesting as severe combined immunodeficiency [3].

In contrast to ADA1, ADA2 has unclear functions. ADA2 is a secreted protein that binds to glycosaminoglycans (GAGs) and has much lower ADA catalytic activity than ADA1 at the physiologic concentrations of adenosine [4, 5]. This observation and the lower rate of deoxyadenosine deamination suggest that ADA2 has another function in addition to its ADA activity. Genetic expression analysis shows that ADA2 is highly expressed in monocytes, myeloid, and plasmacytoid dendritic cells (pDCs). NK, T, and B cells also express ADA2 but at lower levels (Fig. S1) [6]. Although macrophages may be the primary cells expressing extracellular ADA2, T cells infected with the HIV-1 virus also release ADA2 as an ectoenzyme [7, 8]. The concentrations of ADA2 in several biological fluids increase in response to immune cell activation. Several studies have shown ADA2 to be a valuable and robust biomarker for diagnosis or monitoring treatment responses in, for example, pleural tuberculosis, HIV, autoimmune disorders, and several forms of cancer [9–13].

Biallelic pathogenic mutations in ADA2 genes cause loss of function in patients with ADA2 deficiency (DADA2), characterized by systemic inflammation, vasculitis, early-onset stroke, cytopenia, and immunodeficiency [14, 15]. These changes are associated with increased levels of TNF-α and type 1 interferon in the plasma of patients with DADA2 [16, 17]. These observations suggested that extracellular ADA2 regulates myeloid M1/M2 polarization, TNF-α secretion from macrophages, neutrophil activation, NET formation, and IFN-β secretion by epithelial cells [6, 14, 18]. Patients with DADA2 can display bone marrow failure and immunodeficiency, indicating involvement in myeloid cell differentiation and NKT, T, and B cell development [6, 14, 18]. Biological inhibitors targeting TNF-α remain the primary strategy for treating patients with DADA2 [19].

Some studies have focused on the potential extracellular role of ADA2 but have been unable to fully explain the protean phenotypic characteristics of DADA2 [20]. Any functionality ascribed to intracellular or organelle-associated ADA2 remains speculative. These unexplored pathways can contribute to pathological manifestations in DADA2. Toll-like receptor TLR9 is one of the intracellular TLR receptors activated by dsDNA in response to bacterial and viral infections, autoimmune diseases, and cancer [21]. In endosomes and lysosomes, human TLR9 is activated by single-strand oligodeoxynucleotides (ODNs) containing at least one unmethylated CpG motif and more than 20 nucleotides. In addition, shorter DNA fragments were found to bind to augment the activation of TLR9 with CpG ODNs [22, 23]. Human TLR9 is expressed and active in pDCs, B cells, and NK cells.

ODNs are often modified with a phosphorothioate (PTO) backbone that protects the intrinsic natural phosphodiester oligodeoxynucleotides (PD ODNs) from cleavage by DNase. Three classes of CpG ODNs that activate human TLR9 in different immune cells have been developed (Table S1) [24]. Class A CpG ODNs comprise a PD CpG–containing palindromic motif and a PTO-modified poly G string at the 3′ or 5′ end or both. These CpG ODNs induce high IFN-α production from pDCs but are weak stimulators of TLR9-dependent NF-κB signaling and proinflammatory cytokine (e.g., IL-6) production. Class B CpG ODNs contain an entire PTO backbone with several CpG motifs. In contrast to class A CpG ODNs, these ODNs strongly activate B cells and TLR9-dependent NF-κB signaling but weakly stimulate IFN-α secretion in pDCs. Class C CpG ODNs have a complete PTO backbone and a CpG-containing palindromic motif and induce strong IFN-α production from pDCs and B cell stimulation [25].

pDCs are the myeloid cells of the immune system and are characterized by the secretion of high levels of IFN-α/β upon TLR7/9 stimulation, which are crucial cytokines in viral infections, including COVID-19 [26, 27]. Along with IFN-α/β, pDCs produce the proinflammatory cytokines IL-6 and TNF-α, which regulate T, B, and NK cells and conventional DC (cDC) responses [28]. TLR9 in pDCs can be activated with class A CpG ODN in early endosomes, producing IFN-α through the IRF-7 signaling pathway. TLR9 activation with class B CpG in lysosomes activates NF-κB signaling pathways after the release of proinflammatory cytokines and chemokines, such as IL-8, from the cells (Fig. S2) [25, 29–31]. However, why classes A and B CpG ODNs activate pDCs differently is unclear.

In this report, we describe an important and novel intracellular functionality of ADA2, which serves as a DNA-binding protein and a molecular switch that controls TLR9 activation in pDC. These findings provide a rationale for three classes of CpG ODNs in humans. Classes B and C CpG ODNs preferentially bind to ADA2, whereas class A CpG ODNs do not compete for the binding of ADA2. Our studies provide the fundamental basis for improving existing drugs and developing novel drugs for combating viral infections and unleashing immune responses in cancer.

2 Materials and methods

2.1 Cell isolation and culture

Human peripheral blood mononuclear cells (PBMCs) were isolated from the fresh blood of healthy donors according to the Institutional Review Board-approved protocol. The isolation was performed using 50 mL Leucosep tubes (Greiner Bio-One). After two washes with 50 mL of PBS buffer, CD14, and monocytes were purified using anti-CD14-conjugated magnetic microbeads (Miltenyi). The corresponding kits from Miltenyi were used to purify pDCs either by negative or positive selection. The purity of the cells was confirmed to be greater than 90%, as determined by flow cytometry (Fig. S3). In the cell cultures, RPMI 1640 was used as a complete medium supplemented with 1% nonessential amino acids, 1% sodium pyruvate, 100 U/mL penicillin, 100 μg/mL streptomycin, 2 mmol/LM L-glutamine, and 10% FBS. Monocytes and pDCs were cultured in 96-well plates with 200 µL of RPMI medium containing 40 ng/mL GM-CSF (PeproTech) or 10 ng/mL IL-3 (PeproTech). The medium with 40 ng/mL GM-CSF was replaced every three days. Monocytes were differentiated for 6 days and then incubated overnight with 0.5 µmol/L CpG ODN with PTO (Invivogen), and the binding of ADA2 to CpG ODN with PTO inside the cells was analyzed. The cells were treated with 0.5 µM CpG ODN 2006 PTO labeled with FITC (Invivogen) for microscopic analysis.

2.2 Knock-down of ADA2 expression with siRNA in THP-1 cells and pDCs

THP-1 cells (0.3 × 10⁶ cells/mL) were activated with 50 nM phorbol-12-myristate-13-acetate (PMA, Sigma-Aldrich) and cultured in 96-well plates in 200 µL of RPMI medium. The medium was replaced daily, and the cells were transfected with siRNA (1 pmol/well) with Lipofectamine RNAiMax reagent according to the manufacturer’s protocol (Invitrogen). The sequences of the control siRNA and ADA2-specific siRNAs (all from Origene) were rCrGrUrUrArArUrCrGrCrGrUrArUrArArUrArCrGrCrGrUAT (siRNA control), rGrGrArUrArArGrUrUrCrArUrArGrCrArGrArUrGrUrGrGCT (siRNA A), rGrGrCrArUrArCrArGrCrArUrCrCrGrArUrUrUrArArUrCTG (siRNA B), and rCrCrUrCrUrArArUrCrArCrArGrCrUrUrArUrArArUrCrGGA (siRNA C). The cell-culture medium was replaced on the next day, and the concentration of ADA2 in the medium was analyzed on the following day by ELISA [11]. pDCs were transfected with 1 pmol siRNAs, 0.2 µg of poly U (Invivogen), or poly U21 (Synbio Technologies) and cultured in RPMI medium with 10 ng/mL IL-3 for 1 day. On the second day, the cells were either activated with 0.5 µmol/L CpG ODN with PTO (Invivogen) or transfected with 0.1 µg of THP1-1 or Escherichia coli DNA with a Lipofectamine 3000 transfection reagent (Invitrogen). The levels of IL-8 and IFN-α secreted by the cells were analyzed the next day with ELISA (Biolegend).

2.3 Immunostaining and confocal microscopy

Monocytes and pDCs were cultured in eight-well chamber slides (Thermofisher) with 400 µL of RPMI medium. Cells were fixed for 10 min in 10% formaldehyde or 4% PFA, washed three times with PBS, permeabilized for 5 min in PBS supplemented with 0.3% Triton X-100 (PBS-T), and washed three times with PBS. The cells were incubated in the primary antibody overnight, diluted in PBS containing 1% BSA at 4 °C, washed three times with PBS, and incubated with a secondary antibody in PBS containing 1% BSA for 0.5 h at room temperature. The cells were washed again three times and mounted in 80% glycerol. Dilutions for the antibodies against LAMP2 (Sino Biological, 13555-MM05) and ADA1 (Abcam, ab34677) were 1:100. Dilutions for the antibodies against ADA2, TLR9 (Abcam, ab259651), and CD303-FITC (Biolegend, 354208) were 1:50. Fluorescent stains were captured on a Leica SP8 confocal microscope. Polyclonal anti-ADA2 antibodies were purified from the sera of rabbits immunized with recombinant ADA2 (the antiserum was generated by Sino Biological) with a previously described method [11]. To validate the specificity of the antibodies, we included polyclonal rabbit antibodies as a negative control, which did not result in any observable staining. Additionally, nuclear and endoplasmic reticulum (ER) staining was conducted using DAPI (Cell Signaling Technology) and ER-tracker (Invitrogen) according to the manufacturer’s instructions.

2.4 Analysis of ADA2 binding to DNA and CpG ODNs

Recombinant ADA2 and its mutant ADA2 H88G were overexpressed in HEK 293T cells and purified as previously described [17]. A gel-shift experiment was performed, in which ADA2 was incubated with ΔR 8.2 packaging plasmid DNA in the presence of different concentrations of salt, DNase I (NovoProtein), or CpG ODN 2006 PTO (Invivogen) for 5 min at room temperature (Fig.1). Then, 15 µL of the mix was analyzed using 1% agarose gel (GenStar) containing Gel Red (Biosharp). In another assay, 96-well ELISA plates were coated with 1 mg/mL E. coli DNA (Invivogen) in 100 µL of TBS with 0.02% NaN₃ overnight at 37 °C. Then, the plates were washed three times with 200 mL of 0.05% Tween 20 in 1× PBS and blocked with 200 mL of 2% BSA in PBS and 0.02% NaN₃ at room temperature for 1 h. In the next step, 100 mL of ADA2 dilutions in 50 mM Tris pH 6.8, 150 mM NaCl, 10 µM ZnCl₂, and 0.02% NaN₃ (buffer A) was added to the plate with DNA, and the plate was incubated for 1 h at room temperature on a shaker. After three washes with buffer A, 2 mmol/L adenosine was added to the same buffer, and the amount of ADA2 bound to DNA was determined after 20 h of incubation at 37 °C by comparing the ADA activity of ADA2 retained on DNA with the ADA activity of ADA2 standards [11].

Fig.1 ADA2 binds to dsDNA. (A) Gel-shift analysis of ADA2 (μg/mL) binding to plasmid DNA (μg/mL) in the presence of increasing salt concentration (mM). (B) Gel-shift analysis of ADA2 (μg/mL) binding to plasmid DNA (μg/mL) in the presence of different concentrations of DNase I (IU/mL). (C) Gel-shift analysis of ADA2 H88G and ADA2 (μg/mL) binding to plasmid DNA (μg/mL) in the presence or absence of different concentrations of ODN 2006 PTO (µM). (D) ADA2 binding to E. coli DNA was analyzed by a DNA ELISA assay. Each dot on the graph represents the mean value and standard deviation of three independent measurements.

Full size|PPT slide

The binding of ADA2 to CpG ODNs (Invivogen) was analyzed similarly, but a different DNA capture strategy was used (Fig. S4A). ELISA plates were coated with 2 mg/mL streptavidin (New England Biolabs) in 200 µL of PBS and 0.02% NaN₃ at 4 °C overnight. After three washes with 200 mL of 0.05% Tween 20 in PBS and blocking with 200 mL of 2% BSA in PBS, 20 nmol/L biotin DNA in 100 mL of PBS was added to the wells, and the plate was incubated for 30 min at room temperature on a shaker. The binding of ADA2 was analyzed in the plates with CpG ODN biotin bound to streptavidin. To study the inhibition of ADA2 from binding to CpG ODN biotin, 50 mL of DNA/RNA dilutions (Synbio Technologies) in buffer A were added to the wells after the addition of 50 mL of 80 nmol/L ADA2 in the same buffer. The plate was incubated for 1 h at room temperature on a shaker and washed three times with buffer A. The amount of ADA2 remaining in the wells was determined as described above. The pH dependence of the binding between ADA2 and CpG ODN 2006 PD was analyzed, and 10 mmol/L sodium citrate buffer with different pH values and 150 mmol/L NaCl was used as a binding and washing buffer, respectively.

In the experiments with CpG ODNs modified with PTO, the binding of ADA2 to ODNs was investigated with a modified standard ELISA (Fig. S4B). ELISA plates (Greiner Bio-One) were coated overnight at 4 °C with 100 µL of 5 µg/mL rabbit anti-ADA2 polyclonal antibodies in PBS with 0.02% NaN₃. After the plates were washed three times with 200 µL of PBS-Tween 20 buffer and blocked with 200 µL of 2% BSA in PBS for 1 h, 100 µL of 100 ng/mL ADA2 in PBS containing 10% FBS and 0.02% NaN₃ were added to the wells. The plates were incubated for 1 h at room temperature on a shaker. Subsequently, the plates were washed three times with 200 µL of PBS-Tween 20, and 100 µL of the mixture of 4 nmol/L CpG ODN 2006 PTO biotin (Invivogen) and buffer A with different concentrations of inhibitory CpG ODNs with PTO was added to the wells. The plate was incubated for 30 min at room temperature on a shaker and washed three times with buffer A. After 100 µL of a solution containing 1:1000 streptavidin–horseradish peroxidase (HRP) solution (Abbkine), 20 mmol/L Tris HCl (pH 6.8), 50 mmol/L NaCl, 10 µmol/L ZnCl₂, and 10% FBS (buffer B) was added to the wells, the plates were washed four times with the same buffer and incubated for 30 min at room temperature on a shaker. The color reaction was initiated by adding 100 µL of TMB substrate (BioLegend). The reaction was stopped by adding 100 µL of 2 mol/L HCl, and the plates were read at 450 nmol/L in a plate reader (Thermofisher).

3 Results

3.1 ADA2 binds to dsDNA

Structural analyses have demonstrated that glycosaminoglycans (GAGs) bind to the positively charged surfaces of enzymes and stabilize the dimer of ADA2 [5]. Hence, ADA2 is a heparin-binding protein [4]. This observation suggests that ADA2 interacts with other negatively charged molecules, such as DNA.

First, we used a gel-shift assay to study ADA2 interaction with ΔR 8.2 packaging plasmid DNA [32]. The plasmids migrated in the agarose gel in linear and supercoiled forms (lane 1; Fig.1). However, adding ADA2 resulted in a complex formation between the enzyme and plasmid (lane 2). Furthermore, the ADA2 and plasmid complex was stable at physiologic salt concentrations (Fig.1), and ADA2 bound to the plasmid protected plasmid DNA from cleavage with DNAse I (Fig.1). The binding of ADA2 to the plasmid was independent of ADA activity because the mutant forms of ADA2, ADA2, and H88G exhibit 1% of ADA activity and form complexes with plasmids (Fig.1) [17]. Complexing with plasmid DNA can be competitively disrupted by adding CpG ODN 2006 PTO, which binds to ADA2 (Fig.1). In addition, ADA2 formed a complex with E. coli DNA adsorbed on an ELISA plate (Fig.1).

These results suggest that ADA2 can interact with dsDNA and ODN PTO under physiologic conditions, and this interaction is independent of ADA2’s ADA activity.

3.2 ADA2 binds to ssDNA

The toll-like receptor TLR9 is an intracellular dsDNA sensor activated by unmethylated ODNs containing CG motifs [24]. PD ODNs are modified with a PTO backbone, which protects ODNs from degradation by DNases. We studied the binding of ADA2 to three classes of CpG ODNs with PTO, which were used to activate the TLR9 receptor in different experimental systems. The enzyme was captured on an ELISA plate with anti-ADA2 antibodies (Fig. S4B). Then, class B CpG ODN 2006 PTO biotin was added to ADA2 in the presence of increasing concentrations of different classes of CpG ODNs with PTO (Fig.2). The amount of biotinylated CpG ODN 2006 PTO bound to ADA2 was visualized by adding HRP conjugated with streptavidin, followed by the color reaction with HRP substrate TMB in a standard ELISA. The ODN sequences and the inhibition constants (IC₅₀) obtained in this experiment are shown in Table S1. The binding and washing buffer contained a near-physiologic 150 mM salt concentration with a pH of 6.8, similar to early endosomes. CpG ODN 2006 PTO and its control GpC ODN 2006 PTO (ODN 2006 GC PTO) had the highest affinity for ADA2. They efficiently inhibited CpG ODN 2006 PTO biotin binding to the enzyme (Fig.2, Table S1). By contrast, class A CpG ODN with PTO (2216 and 2336) did not bind well to ADA2 (Fig.2). Classes B and C CpG ODNs with PTO were bound to ADA2 but had a lower affinity for ADA2 than CpG ODN 2006 PTO (Fig.2). Most CpG ODNs with PTO, except CpG ODN 2006 with PTO, formed a secondary structure in the solution, lowering affinity for ADA2 [24]. In addition, the backbone of class A CpG ODN with PTO is not fully modified with PTO, which decreases their electrostatic binding to ADA2. In a similar experiment, mouse TLR9 showed a higher affinity for CpG ODN 2006 PTO than other CpG ODNs with PTO and an extremely low affinity for class A CpG ODN with PTO (Fig. S5A). ADA2 added to mouse TLR9 competed for binding to CpG ODN 2006 PTO biotin and completely abolished the binding of mouse TLR9 to CpG ODN 2006 PTO biotin at high ADA2 concentrations (Fig. S5B). ADA2 had a higher relative affinity for CpG ODN 2006 PTO than for TLR9 (Fig. S5C).

Fig.2 Inhibition of the binding of ADA2 to ODN 2006 PTO biotin. (A) Inhibition of the binding of ADA2 to ODN 2006 PTO biotin with class B ODN 2006 PTO, ODN 2006 GC PTO, class A ODN 2216 PTO, and ODN 2336 PTO. (B) The inhibition of binding of ADA2 to ODN 2006 PTO biotin with class B ODN BW006 PTO, ODN D-SL01 GC PTO, class C ODN 2395 PTO, ODN M362 PTO, and ODN D-SL03 PTO. (C) ADA2 binding to ODNs PTO (ODN 2006 PTO and ODN 2216 PTO) and ODNs PD (ODN 2006 PD and ODN 2006-2006 PD). (D) ADA2 binding to ODN 2006 PD in a buffer with different pH values. Each dot on the graph represents the mean value and standard deviation of three independent measurements.

Full size|PPT slide

In the next experiment, we compared the binding of ADA2 to PTO-modified and unmodified natural PD ODNs. In this experiment, biotinylated ODNs were first bound to streptavidin and adsorbed to the ELISA plate (Fig. S4A). In the second step, ADA2 was added to the wells on the plate, and ADA2 bound to ODNs was detected based on its ADA activity. As shown in Fig.2, ADA2 was bound to CpG ODN 2006 PD, but the complex had lower stability than CpG ODN 2006 PTO, with apparent Kd values of 13 and 4 nmol/L, respectively. The doubling of the length of CpG ODN 2006 PD (ODN 2006-2006 PD) did not change the apparent binding constant (Kd = 14 nmol/L) but increased the number of binding sites threefold. These results indicate that more than one molecule of ADA2 binds to ODN 2006-2006 PD. The binding of ADA2 to CpG ODN 2006 PD was promoted at a low pH, like the pH found in endosomes and lysosomes (Fig.2) [33].

To further explore the interaction of ADA2 with natural ODN PD, we monitored the binding of ADA2 to CpG ODN 2006 PD in the presence of increasing concentrations of competitor ODNs (Fig.3). Like modified ODN PTOs, ADA2 had higher affinity for class B CpG ODN 2006 PD than class C CpG ODN 2395 and class A CpG ODNs 2216 and 2336. The binding of ADA2 to class A CpG ODN 2336 PD was the weakest (Fig.3), suggesting that the structure of ODNs is crucial for the binding of ADA2 to natural ODNs.

Fig.3 Inhibition of binding between ADA2 and ODN 2006 PD biotin. (A) Inhibition of binding between ADA2 and ODN 2006 PD biotin with class B (ODN 2006 PD), class A (ODN 2216 PD, ODN 2336 PD), or C (ODN 2395 PD) ODNs. (B) Inhibition of binding between ADA2 and ODN 2006 PD biotin with ODN 2006 PD, ODN GC2006 PD, ODN 2006 C-Met PD, poly T [24], or ODN 2006 Reverse PD, ODN 2006 (T changed to C) PD. (C) Inhibition of ADA2 binding to ODN 2006 PD biotin with modified class A ODN 2336 PD. (D) Inhibition of binding between ADA2 and ODN 2006 PD biotin with poly T of varying length. (E) Inhibition of binding between ADA2 and ODN 2006 PD biotin with RNA PTO and RNA PD. (F) The sequences of oligonucleotides. Each dot on the graph represents the mean value and standard deviation of three independent measurements.

Full size|PPT slide

ODNs are composed of poly T, which do not have a secondary structure, and bind to ADA2 equally well as CpG ODN 2006 PD (Fig.3). However, the conversion of thymidine into cytosine in CpG ODN 2006 PD results in the formation of a secondary structure and reduction in affinity. The methylation of cytidine and the conversion of CG into GC in the CpG ODN 2006 PD sequence decreased the binding constant of modified ODNs to ADA2 threefold. Notably, the addition of complementary ODN 2006 Reverse PD, which forms a double strand with CpG ODN 2006 PD, completely abolished the binding of ADA2 to biotinylated CpG ODN 2006 PD.

This result indicates that ADA2 binds much better to ssDNA than to dsDNA. However, genomic dsDNA may have a specific sequence that binds to ADA2. To further demonstrate that the secondary structure affects the binding of ODN PD to ADA2, we modified the sequence of CpG ODN 2336 PD, which poorly binds to ADA2 (Fig.3). The exchange of nucleotides that form a palindrome to thymidine facilitated the binding of modified ODNs to ADA2, and additional changes in the stretches of guanosine to thymidine from the 3′ end of ODN did not affect the binding of ODN to ADA2. The conversion of four guanosines from the 5′ part of the ODN into thymidines decreased the binding of ADA2 to ODN, suggesting that the stretches of guanosines at 5′ of ODNs may improve the binding of ADA2 to ODNs. The binding of ODN PD to ADA2 depended on the length of ODN. The binding of poly T to ADA2 revealed that ADA2 binds to ODN PD, and the length of poly T was longer than 12. Affinity increased until the length of ODN reached 24 nucleotides. ODNs with a length above 24 nucleotides bound to the several molecules of ADA2 (Fig.3).

TLR9 does not recognize RNA [34]. In contrast to TLR9, ADA2 binds to RNA (Fig.3). Thus, the enzyme had a higher affinity for RNA with a PTO backbone (PTO modified) than for natural RNAs (RNA PD). The interaction between ADA2 and an RNA PD depends on the secondary structure of the RNA PD. A linear poly U is a better ligand than RNA 2006 PD and RNA 40 PD. Hence, ADA2 can bind to ssDNA and RNA, and these interactions depend on the secondary structure of the nucleic acid.

3.3 ADA2 binds ODN PTO in the lysosomes of macrophages

According to the results above, ADA2 binds to ODN PTO at salt concentrations and pH, which were reported for endosomes and lysosomes [33]. Therefore, whether ADA2 can bind to CpG ODNs PTO inside cells was investigated. Monocyte-derived macrophages were incubated with CpG ODN with PTO, and the concentration of ADA2 was analyzed in the cell lysates and cell culture medium. Consequently, the incubation of ADA2 and macrophages with classes B and C CpG ODNs with PTO increased the concentration of ADA2 inside the cells. By contrast, intracellular ADA2 remained unchanged in the presence of class A CpG ODN with PTO (Fig.4). Furthermore, the amount of ADA2 secreted from the cells was not affected by CpG ODN with PTO (Fig.4), suggesting that ADA2 interacted with ODNs inside the endosomal compartment. Therefore, we analyzed macrophages with confocal microscopy to determine the localization of ADA2.

Fig.4 ADA2 binds to ODN 2006 PTO in macrophage lysosomes. (A, B) ADA2 levels in cell lysates (A) and culture medium (B) of macrophages incubated with 0.5 μmol/L ODN PTO for 24 h. Monocytes were isolated from PBMCs and differentiated into macrophages with 40 ng/mL GM-CSF for 6 days. Macrophages were washed twice with PBS before lysis to remove the secreted ADA2. ADA2 levels were measured by ELISA. Error bars represent the standard deviation from four independent measurements. (C–F) Confocal microscopy of macrophages differentiated with 40 ng/mL GM-CSF for 6 days and treated with 0.5 μmol/L ODNs for 24 h. Cells were fixed and stained for ADA2 (blue), lysosomal marker LAMP2 (red), nuclear marker DAPI (white), and endoplasmic reticulum marker ER-tracker (red). ODNs were ODN 2006 PTO (D), ODN 2006 G5 PD (E), and ODN 2006 PTO FITC (F). Untreated cells are shown in (C).

Full size|PPT slide

Abundant ADA2 expression was observed inside the cells (Fig.4), and the enzyme was partially colocalized with the lysosomal marker LAMP2. However, adding CpG ODN 2006 PTO to macrophages increased the density of ADA2 inside the lysosomes (Fig.4 and S6). The treatment of the cells with CpG ODN 2006 PD modified with a 3′ poly G string (ODN 2006 G5 PD) did not increase ADA2 density in the lysosomes (Fig.4) and the intracellular enzyme’s concentration (Fig. S7A). Moreover, ADA2 colocalized with a fluorescence-labeled CpG ODN 2006 PTO FITC inside the lysosomes (Fig.4). These observations explain the ELISA results in Fig.4 and suggest that ADA2 bound to CpG ODN 2006 PTO is transported to lysosomes. In addition, CpG ODN 2006 PTO may protect ADA2 from digestion induced by lysosomal proteases, potentially explaining why the enzyme concentration increased inside the cells. In contrast to ADA2, the TLR9 receptor showed no change in density in the lysosomes after the incubation of macrophages with CpG ODN 2006 PTO (Fig. S8).

HEK 293T cells overexpressing ADA2 exhibited the partial colocalization of ADA2 with the lysosomal marker LAMP2 (Fig. S8A). ADA1 was not found in the lysosomes of the cells overexpressing ADA1 (Fig. S8C), suggesting that ADA2 was the only ADA found in the lysosomes.

pDCs respond differently to three classes of CG ODNs PTO after activation with IL-3In humans, pDCs produce large quantities of IFN-α in response to TLR9 activation with unmethylated ODN PTO containing CG motifs. Class A CpG ODN with PTO, which has the lowest affinity for TLR9 (Fig. S5A), can still induce a significant amount of IFN-α secretion from freshly isolated cells. By contrast, the same cells produced IL-8 in response to class B CpG ODN with PTO [29]. TLR9 activation in early endosomes promotes IFN-α secretion via the IRF-7 pathway, and TLR9 binding to the PTO of CpG ODNs in lysosomes induces the NF-κB pathway and IL-8 secretion (Fig. S2) [30]. The activation of TLR9 in the endosomes and lysosomes appears to be differentially regulated.

The results of our experiments were consistent with previously reported observations showing that the addition of classes A and C CpG ODNs with PTO to freshly isolated pDCs results in IFN-α secretion. The addition of class B CpG ODN with PTO did not induce the release of a significant amount of IFN-α (Fig.5). In the same experiment, cells produced more amounts of IL-8 in response to TLR9 activation with classes B and C CpG ODNs with PTO than with class A ODN with PTO (Fig.5). However, responses to different classes of CpG ODNs with PTO changed when the CpG ODNs with PTO was added to pDCs after 1 day of culture in the medium containing IL-3 [30]. Here, cells started to secrete IFN-α in response to class B activation (Fig.5), whereas IL-8 secretion decreased considerably (Fig.5). Furthermore, when the number of pDCs increased, the amount of IFN-α released from the cells activated with classes B and C CpG ODNs with PTO on the second day was much higher than that on the first day of activation (Fig.5 and 5F). This pattern was similar to the pattern of IL-8 secretion after TLR9 activation on day 1 (Fig.5). These results indicate the intracellular regulation of IFN-α and IL-8 secretion from pDCs after TLR9 activation by the three classes of CpG ODNs with PTO.

Fig.5 Activation of pDCs by three classes of ODNs with PTO: (A) 1 μmol/L ODN with PTO was added to pDCs (0.0125 × 106 cells/mL) on day 1, and the amount of IFN-α (A) and IL-8 (B) released from the cells was analyzed by ELISA in 48 h. Class A ODNs are shown in blue, class B in red, and class C in green. (C, D) Class A ODN 2216 with PTO and class B ODN 2006 with PTO (1 μM) were added to pDCs (0.0125 × 106 cells/mL) on day 1 or day 2, and the amount of IFN-α (C) or IL-8 (D) released from the cells was analyzed by ELISA on the following day. (E, F) ODN with PTO (1 μM) was added to pDCs (0.05 × 106 cells/mL) on day 1 (E) or day 2 (F), and the amount of IFN-α released from the cells was analyzed by ELISA on the following day. Class A ODNs are shown in blue, class B in red, and class C in green. The error bars represent the standard deviation, and the results were obtained from four independent replicates.

Full size|PPT slide

3.4 ADA2 regulates the activity of TLR9 in pDC

In the next experiment, pDCs isolated from PBMCs were analyzed using confocal microscopy. These cells express ADA2 (Fig.6). However, no colocalization of the enzyme with the lysosomal marker LAMP2 was observed. After the pDCs were incubated with IL-3 and CpG ODN 2006 with PTO and FITC for 1 day, a significant portion of ADA2 in lysosomes colocalized with CpG ODN 2006 with FITC (Fig.6 and 6C). This finding is like the previous observation made with monocyte-derived macrophages, suggesting that ADA2 bound to CpG ODN 2006 in the endosomes of pDCs is transported to lysosomes.

Fig.6 (A) Human pDCs were isolated from PBMCs and cultured with 10 ng/mL IL-3 in an eight-well chamber. After 24 h, pDCs that were attached to the plastic were washed, fixed, and stained with anti-ADA2 antibodies (blue), ER-tracker (red) for endoplasmic reticulum staining, and DAPI (white) for nucleus staining. (B) Human pDCs were cultured with 10 ng/mL IL-3 in the presence of 0.5 μM ODN 2006 PTO FITC in an 8-well chamber. After 24 h, pDCs that were attached to the plastic were washed, fixed, and stained with anti-LAMP2 antibodies (red), anti-ADA2 antibodies (blue), and DAPI (white). (C) Colocalization of PTO-modified ODN 2006 with FITC, ADA2, and the lysosomal marker LAMP2.

Full size|PPT slide

Notably, when pDCs were treated with CpG ODN 2006 with PTO and FITC on the second day of culture with IL-3, ADA2 did not colocalize with CpG ODN (Fig. S10). This observation can be explained by a decrease in ADA2 concentration in endosomes after the treatment of pDCs with IL-3.

ADA2 can bind to CpG ODNs with PTO within cells (Fig.4), suggesting that it competes with TLR9 for binding to class B CpG ODN 2006 with PTO in endosomes. To investigate this characteristic further, we used siRNA to knock down ADA2 expression in pDCs and then activated TLR9 with dsDNA or CpG ODNs with PTO (Fig.7).

Fig.7 Effects of knocking down or blocking of ADA2 during TLR9 activation in pDCs. In all experiments, human pDCs were isolated from PBMCs and cultured in the presence of 10 ng/mL IL-3. (A–D) pDCs (0.0125 × 106 cells/mL) were transfected with siRNAs using Lipofectamine RNAiMax transfection reagent on day 1. On day 2, PTO-modified ODNs (1 μmol/L) were added to the cells. The amount of IFN-α (A, C) and IL-8 (B, D) in the cell culture medium was analyzed on day 3 using ELISA. (E, F) The knockdown of ADA2 expression with siRNA in pDC promoted TLR9 activation by dsDNA. The cells (0.0125 × 106 cells/mL) were transfected with control (siRNA control) and specific (siRNA ADA2) siRNA. After 24 h, the cells were transfected with (E) E. coli genomic DNA and (F) genomic DNA from human THP1 cells using Lipofectamine 3000. (G) Analysis of ADA2 in the culture medium 2 days after the siRNA transfection of PMA-activated THP1 cells with ELISA. (H) Blocking ADA2 with poly U promoted TLR9 activation by PTO-modified ODN 2006. Human pDCs (0.02 × 106 cells/mL) were transfected with 0.2 µg Poly U or Poly U21 using Lipofectamine RNAiMax transfection reagent on day 1. On day 2, PTO-modified ODN 2006 (0.5 μmol/L) was added to the cells. The amount of IFN-α in the cell culture medium was analyzed on day 3 using ELISA [11]. The error bars represent the standard deviation, and the results were obtained from four independent replicates. In all experiments, the difference between the control siRNA and siRNA ADA2 was statistically significant (P < 0.008).

Full size|PPT slide

Given that pDCs are rare and difficult to transfect, we validated the efficiency of ADA2 knockdown with three different siRNAs and a scrambled siRNA control by using monocytic THP1 cells expressing ADA2. PMA-activated THP1 cells transfected with siRNAs A and C had a 70%–80% reduction in the amount of ADA2 secreted from the cells, whereas siRNA control and siRNA B did not greatly affect the release of ADA2 from the cells (Fig.7).

Using the same siRNAs, we transfected freshly isolated pDCs and activated them with CpG ODN 2006 with PTO 1 day after transfection. We found that the knockdown of ADA2 expression with siRNAs A and C led to an eightfold increase in IFN-α secretion from pDCs (Fig.7). Similarly, the knockdown of ADA2 expression increased IL-8 secretion from pDCs activated with three classes of CpG ODNs with PTO (Fig.7).

In addition, the knockdown of ADA2 expression increased IFN-α and IL-8 secretion from pDCs activated with the three classes of CpG ODNs with PTO (Fig.7 and 7D). However, no cytokine production was observed with the control ODN GpC 2006 in the corresponding experiment (data not shown). However, classes B and C CpG ODNs with PTO activated TLR9 more efficiently than Class A ODNs in the cells transfected with siRNA A. These results suggest that ADA2 regulates the activity of TLR9 by competing for binding to CpG ODNs with PTO. However, whether the removal of ADA2 increases TLR9 activation in response to pDC transfection with dsDNA remains unclear. In the next experiment, pDCs were treated with siRNA after transfection with DNA isolated from E. coli and human THP1 cells. The knockdown of ADA2 expression resulted in a substantial increase in the production of IFN-α by pDC transfected with bacterial and eukaryotic dsDNA (Fig.7 and 7F), suggesting that ADA2 regulates the activation of TLR9 with PTO-modified CpG ODNs and natural dsDNA.

In contrast to TLR9, ADA2 can bind RNA [34] (Fig.3). Thus, RNA bound to ADA2 can block the enzyme, facilitating the activation of TLR9 with PTO-modified ODNs. To prove this hypothesis, we used long poly U and poly U containing 21 nucleotides (poly U21) to transfect pDCs before cell activation with PTO-modified CpG ODN 2006. As shown in Fig.7, the cells transfected with RNA did not secrete IFN-α. However, when PTO-modified CpG ODN 2006 was added to the cells transfected with poly U or poly U21, the amount of IFN-α released from the cells was threefold that released from cells not treated with RNAs. These results suggest that the knockdown of ADA2 expression with siRNA and blocking ADA2 binding to PTO-modified CpG ODNs with RNA activates TLR9, promoting the secretion of IFN-α from pDCs.

4 Discussion

In this study, we show that ADA2 is a DNA-binding protein, and our data indicate a competition between ADA2 and TLR9 for binding to ssDNA inside the endosomes of pDCs. These pathways differentially regulate IFN-α and proinflammatory cytokine secretion in response to viral infections and cancer. ADA2 binds to heparin and other acidic GAGs, suggesting that the enzyme can interact with DNA [4, 5]. The gel-shift analysis of the binding between ADA2 and plasmid DNA reveals that ADA2 formed a complex with DNA, and this complex did not enter the gel (Fig.1). This complex was stable at physiologic salt concentrations and independent of the ADA activity of ADA2 (Fig.1 and 1C). Moreover, ADA2 complexed with a plasmid that protects other DNA from the cleavage by DNase I (Fig.1) and interacted with genomic DNA from E. coli at physiologic salt concentrations (Fig.1). Notably, PTO-modified CpG ODN 2006, which is a ligand and activator of TLR9, binds to ADA2 and prevents ADA2 from complexing with plasmid DNA (Fig.1).

Differential binding of PTO-modified CpG ODN 2006 biotin to ADA2 in the presence of the three classes of PTO-modified CpG ODNs was demonstrated by a modified ELISA assay (Fig. S4). Classes B and C CpG ODNs bound to ADA2 and inhibited the enzyme from binding to PTO-modified CpG ODN 2006 biotin (Fig.2 and 2B, Table S1). By contrast, class A CpG ODN did not compete with ADA2 for binding to PTO-modified CpG ODN 2006 biotin. These results show that ADA2 has a higher affinity for ODNs with a wholly modified PTO backbone. In addition, a linear structure of PTO-modified CpG ODN 2006 may explain why this form has a higher affinity for ADA2 than other CpG ODNs, which can form a secondary structure [24].

Similar to ADA2, TLR9 binds to classes B and C CpG ODNs modified with PTO better than to class A CpG ODNs (Fig. S5A). Interestingly, ADA2 can outcompete TLR9 for binding to CpG ODN 2006 biotin (Fig. S5B and S5C). To further analyze the binding of ADA2 to ssDNA, we used ODNs with a natural PD backbone. Although ADA2 had a lower affinity for PD ODNs than PTO ODNs, the dissociation constant for ADA2-PD ODNs was in the nanomolar concentration range (Fig.2). The complex between ADA2 and CpG ODN 2006 PD was stable at pH 5.0 and pH levels reported for lysosomes (Fig.2) [33]. Like the results obtained with CpG ODNs PTO, ADA2 has a higher affinity for classes B and C CpG ODNs with PD than class A (Fig.3). Although the methylation of cytosine or changes of CG motif to GC markedly decreased the binding to ADA2 (Fig.3), these differences might not be adequate to support the postulation that ADA2 distinguishes between self and foreign DNA [24]. The addition of a reverse complementary ODN, which forms dsDNA with PD-modified CpG ODN 2006, completely inhibited the binding of ADA2 to PD-modified CpG ODN 2006 biotin. This result suggests that either ADA2 binds less to dsDNA than it does to ssDNA or that the binding of ADA2 to dsDNA requires specific intrinsic sequences.

The exact length of poly T ODN has a similar affinity to CpG ODN 2006 PD. The conversion of thymidines into cytidines in the PD-modified CpG ODN 2006 sequence inhibited the binding of ODN to ADA2. This observation suggests that ADA2 recognizes ODNs with linear structures. Change in the palindrome sequence in CpG ODN 2236, which forms a secondary structure, to a linear stretch of thymidines increased the binding of ADA2 to the modified ODN (Fig.3). The affinity of ADA2 for poly T increased with length (from 12 nucleotides to 24 nucleotides; Fig.3), suggesting that the minimal length of ODNs required for ADA2 binding is 20 nucleotides. The additional extension in the length of poly T increased the number of ADA2 molecules bound to the ODN without changing its affinity for ODNs (Fig.2 and Fig.3). ADA2 binds to RNA. The results of the binding experiment suggest that the enzyme prefers RNA molecules with a linear structure (Fig.3). Other reports have suggested that TLR9 does not bind to RNA [34]. The reason for this apparent discordance is unclear.

The binding of ADA2 to PTO-modified CpG ODNs was confirmed intracellularly. The cell lysates from the macrophages treated with PTO-modified CpG ODNs had increased concentrations of ADA2. The analysis of three classes of ODNs showed that the concentration of ADA2 increased inside macrophages in response to the treatment of cells with classes B and C CpG ODNs modified with PTO (Fig.4 and S7A). The concentration of ADA2 in the cell lysates did not considerably change after the cells were incubated with class A CpG ODN modified with PTO. These results suggest that ADA2 preferentially binds to classes B and C CpG ODNs modified with PTO inside cells. The concentration of ADA2 secreted from the cells did not change after the cells were exposed to PTO-modified CpG ODNs (Fig.4 and S7B), suggesting that ADA2 bound to CpG ODN and accumulated inside the endosomes or lysosomes.

We analyzed macrophages with confocal microscopy to reveal ADA2 localization inside the cells before and after treatment with class B CpG ODN 2006 modified with PTO. ADA2 was well expressed in untreated macrophages but was only partially colocalized with a lysosomal marker LAMP2 (Fig.4). However, when the cells were incubated with PTO-modified CpG ODN 2006, a substantial increase in ADA2 density was observed in the lysosomes (Fig.4 and S6). When the cells were treated with a fluorescent-labeled CpG ODN 2006 modified with PTO, the density of ADA2 colocalized with PTO-modified CpG ODN 2006 (Fig.4), suggesting that ADA2 bound to PTO-modified CpG ODN 2006 translocated and then accumulated inside the lysosomes. Such increases in ADA2 concentration in the lysosomes of macrophages can be explained by a decreased rate of degradation of CpG ODN bound to ADA2 by proteases. Further analysis of purified pDCs revealed ADA2 expression inside cells (Fig.6), which was like that observed in the macrophages (Fig.4). Moreover, ADA2 and CpG ODN 2006 colocalized inside lysosomes in the pDCs after incubation with the fluorescent-labeled CpG ODN 2006 modified with PTO (Fig.6 and 6C), suggesting that ADA2 in pDCs binds to CpG ODN 2006 and is transported to lysosomes.

Functional analyses show that pDCs treated with classes A and C CpG ODNs modified with PTO secrete high levels of IFN-α. The cells released additional amounts of IL-8 in response to TLR9 activation by classes B and C CpG ODNs (Fig.5 and 5B). These observations are similar to the results of previous studies, which suggested that activation of TLR9 in early endosomes results in IFN-α secretion from pDCs and the activation of TLR9 in lysosomes activates the NF-κB pathway and releases IL-8 from cells (Fig. S2) [29, 30]. Interestingly, pDCs cultured with IL-3 for 1 day responded quite differently to three classes of CpG ODNs, consistent with the results of a published report [35]. Accordingly, cells cultured with IL-3 for 1 day after incubation with PTO-modified CpG ODNs generate IFN-α in response to the treatment with class B CpG ODN 2006 modified with PTO (Fig.5, 5E, and 5F). The level of IL-8 released from these activated cells decreased (Fig.5).

These observations suggest that putative intracellular signals control the activation of TLR9 in response to CpG ODNs. ADA2 binds to classes B and C CpG ODNs modified with PTO and thus can compete with TLR9 for binding (Fig.2, S5B and S5C). In addition, ADA2 bound to PTO-modified CpG ODN 2006 intracellularly (Fig.4) and accumulated in the lysosomes of macrophages and pDCs treated with PTO-modified CpG ODN 2006 (Fig.4, 4F, 6B and 6C). These results suggest that the direct binding of ADA2 to class B CpG ODNs in the endolysosomes of pDCs can limit the activation of TLR9.

Therefore, experiments that knock down the expression of ADA2 with siRNA were performed, followed by activation with CpG ODNs. For the validation of siRNAs, THP1 cells were transfected with siRNA, and the amount of ADA2 secreted from the cells was analyzed after 2 days (Fig.7). The validated siRNAs were used for the experiment with pDCs. The cells transfected with ADA2-specific siRNA and activated with class B CpG 2006 modified with PTO exhibited 8- and 2.5-fold increase in IFN-α and IL-8 levels, respectively (Fig.7 and 7B). These increases in IFN-α and IL-8 secretion were observed after the pDCs were treated with the other classes of CpG ODNs (Fig.7 and 7D). In another experiment, pDCs were first pretreated with the control and ADA2-specific siRNA after the transfection with genomic DNA extracted from bacterial and mammalian cells. In both cases, the amount of IFN-α released from pDCs treated with the specific ADA2 siRNA was higher (Fig.7 and 7F) than the control siRNA treatment. These results demonstrated that decreases in ADA2 expression in the cells potentiate the responses of TLR9 to PTO-modified CpG ODNs and natural DNA ligands.

Prior experiments demonstrating that ADA2 binds RNA molecules (Fig.3) have suggested that RNA bound to ADA2 inside endosomes can block the binding of CpG ODNs to ADA2. Indeed, the pretreatment of pDCs with poly U after the activation of the cells with class B CpG ODNs resulted in a fourfold increase in IFN-α secretion from the cells compared with the activation with CpG ODN 2006 PTO alone (Fig.7). The transfection of pDCs with poly U did not induce the release of IFN-α from the cells, consistent with the published data demonstrating that TLR7, which is an ssRNA sensor, is poorly activated by poly U [36]. Therefore, blocking ADA2 with natural or chemically modified RNA might be a therapeutic strategy for activating pDCs in response to CpG ODNs, which are currently used in trials involving cancer immunotherapy and immunostimulation against pathogens and as adjuvants in vaccines [21].

In contrast to ADA2, ADA1 was not found in the same location as the lysosomal marker LAMP2 (Fig. S9), indicating that ADA1 is responsible for regulating adenosine levels in the cytoplasm. After ADA2 was modified with Man-6-P, it bound to the cation-dependent mannose 6-phosphate receptor (CD-M6PR) in the trans-Golgi apparatus and subsequently transported to endosomes (Fig.8). Interestingly, the Man-6-P modification site ¹⁷⁴NVT in ADA2 molecule [37] is located next to the putative receptor binding (PRB) domain [5], suggesting that CD-M6PR could be the “missing ADA2 receptor” that binds to the PRB domain of the enzyme. This work indicates that ADA2 competes with TLR9 for binding to CpG ODNs in endosomes, hence blocking TLR9 activation and IRF-7-dependent IFN-α expression and secretion from pDCs (Fig.8, left side). However, these cells can still produce NF-κB-induced cytokines and chemokines, such as IL-8, in response to TLR9 activation with class B CpG ODNs in lysosomes, where the concentration of ADA2 is low (Fig.5 and 6A). IL-3 is a growth factor produced by activated T cells in response to antigen recognition triggered by interaction with conventional DCs and macrophages [38, 39]. In patients with COVID-19, low levels of IL-3 correspond to reduced concentrations of IFN-α and a small population of pDCs [40]. To promote the maturation and survival of pDCs, IL-3 is added to cell cultures. We hypothesized that the binding of IL-3 to the IL-3 receptor may either trigger the translocation of ADA2 from endosomes to lysosomes (Fig.6 and Fig.8) or decrease ADA2 expression in activated pDCs. pDCs incubated with IL-3 for 24 h and treated with CpG 2006 ODN FITC did not show the colocalization of ADA2 with ODN 2006 (Fig. S10C), indicating that the concentration of ADA2 was reduced in the activated pDCs (Fig.8). Additional experiments are required to confirm this hypothesis. This proposed mechanism can explain why pDCs cultured with IL-3 for a day begin to produce IFN-α in response to treatment with class B CpG ODN while concurrently releasing low quantities of IL-8 (Fig.5 and 5D). According to the proposed mechanism, ADA2 regulates TLR9 activation in response to IL-3 binding to pDCs (Fig.8). This additional amplification signal requires adequate control of the response of pDC to self-DNA. In the absence of ADA2, the cells may be activated by self-DNA, resulting in IFN-α, TNF-α, and IL-6 secretion from pDCs and an autoimmune response. These events may explain why most patients with DADA2 present with an inflammatory phenotype [20]. The activation of PBMCs from patients with DADA2 and PTO-modified CpG ODN 2006 increases CXCL1 secretion from the cells [14]. The absence of ADA2 expression in HEK 293 cells (Fig. S4B) can explain why mouse and human TLR9 expressed in these cells are activated by class B CpG ODNs rather than class A CpG ODNs [41]. Similarly, B cells expressing low levels of ADA2 (Fig. S1) respond better to class B CpG ODN [42]. Interestingly, the high concentration of adenosine in lysosomes (0.2–2 mmol/L) and acidic pH matches the high Km value of ADA2 for adenosine (2 mmol/L), and the acidic pH maximizes ADA activity [4, 43]. Thus, ADA2 can regulate adenosine levels in lysosomes and can influence cell activation, particularly the polarization of macrophages into M1/M2 phenotypes [14]. In addition, ADA2 may have multiple functions in immune cells (Fig. S1), possibly explaining the various phenotypes observed in patients with DADA2 [20]. The intracellular role of ADA2 in the regulation of macrophages, cDCs, NK cells, T cells, and B cells’ functions in response to the activation signals must be explored further. We discovered a significant evolutionary change in this characteristic by analyzing the electrostatic surface distribution of ADA2 molecules from different species, ranging from flies to humans (Fig. S11). The ability of ADA2 to bind to DNA may have been acquired during evolution and is a relatively recent function. Our modeling predicts that single-stranded DNA binds to the positively charged bottom of ADA2, consistent with our previous findings that proteoglycans bound to ADA2 stabilize the dimer [5] (Fig. S12).

Fig.8 Scheme illustrates a proposed regulation mechanism of TLR9 activation by ADA2 in human pDCs.

Full size|PPT slide

5 Conclusions

In summary, the findings of this work suggest a new role for ADA2 in pDC activation and the regulation of innate immune responses by the adaptive immune system. We propose new functions for ADA2 in controlling TLR9 responses to ssDNA inside pDCs and lysosome-focused adenosine deamination. These results may explain the functional differences among the three classes of CpG ODNs and allow the design of CpG ODNs with high specificity and potency. The synergistic activation of TLR9 with RNA and CpG ODNs and blocking of ADA2 can provide novel therapeutic strategies for inducing IFN-α secretion from pDCs and modulating immune responses to intracellular infections and cancer.

References

[1] Eltzschig HK, Sitkovsky MV, Robson SC. Purinergic signaling during inflammation. N Engl J Med 2012; 367(24): 2322–2333

[2] Antonioli L, Colucci R, La Motta C, Tuccori M, Awwad O, Da Settimo F, Blandizzi C, Fornai M. Adenosine deaminase in the modulation of immune system and its potential as a novel target for treatment of inflammatory disorders. Curr Drug Targets 2012; 13(6): 842–862

[3] Kohn DB, Booth C, Shaw KL, Xu-Bayford J, Garabedian E, Trevisan V, Carbonaro-Sarracino DA, Soni K, Terrazas D, Snell K, Ikeda A, Leon-Rico D, Moore TB, Buckland KF, Shah AJ, Gilmour KC, De Oliveira S, Rivat C, Crooks GM, Izotova N, Tse J, Adams S, Shupien S, Ricketts H, Davila A, Uzowuru C, Icreverzi A, Barman P, Campo Fernandez B, Hollis RP, Coronel M, Yu A, Chun KM, Casas CE, Zhang R, Arduini S, Lynn F, Kudari M, Spezzi A, Zahn M, Heimke R, Labik I, Parrott R, Buckley RH, Reeves L, Cornetta K, Sokolic R, Hershfield M, Schmidt M, Candotti F, Malech HL, Thrasher AJ, Gaspar HB. Autologous ex vivo lentiviral gene therapy for adenosine deaminase deficiency. N Engl J Med 2021; 384(21): 2002–2013

[4] Zavialov AV, Engström A. Human ADA2 belongs to a new family of growth factors with adenosine deaminase activity. Biochem J 2005; 391(1): 51–57

[5] Zavialov AV, Yu X, Spillmann D, Lauvau G, Zavialov AV. Structural basis for the growth factor activity of human adenosine deaminase ADA2. J Biol Chem 2010; 285(16): 12367–12377

[6] Dhanwani R, Takahashi M, Mathews IT, Lenzi C, Romanov A, Watrous JD, Pieters B, Hedrick CC, Benedict CA, Linden J, Nilsson R, Jain M, Sharma S. Cellular sensing of extracellular purine nucleosides triggers an innate IFN-β response. Sci Adv 2020; 6(30): eaba3688

[7] Zavialov AV, Gracia E, Glaichenhaus N, Franco R, Zavialov AV, Lauvau G. Human adenosine deaminase 2 induces differentiation of monocytes into macrophages and stimulates proliferation of T helper cells and macrophages. J Leukoc Biol 2010; 88(2): 279–290

[8] Tsuboi I, Sagawa K, Shichijo S, Yokoyama MM, Ou DW, Wiederhold MD. Adenosine deaminase isoenzyme levels in patients with human T-cell lymphotropic virus type 1 and human immunodeficiency virus type 1 infections. Clin Diagn Lab Immunol 1995; 2(5): 626–630

[9] Porcel JM, Esquerda A, Bielsa S. Diagnostic performance of adenosine deaminase activity in pleural fluid: a single-center experience with over 2100 consecutive patients. Eur J Intern Med 2010; 21(5): 419–423

[10] Abdi M, Rahbari R, Khatooni Z, Naseri N, Najafi A, Khodadadi I. Serum adenosine deaminase (ADA) activity: a novel screening test to differentiate HIV monoinfection from HIV-HBV and HIV-HCV coinfections. J Clin Lab Anal 2016; 30(3): 200–203

[11] Luo W, Dong L, Chen F, Lei W, He L, Zhou Q, Lamy T, Zavialov AV. ELISA based assays to measure adenosine deaminases concentration in serum and saliva for the diagnosis of ADA2 deficiency and cancer. Front Immunol 2022; 13: 928438

[12] Gao ZW, Yang L, Liu C, Wang X, Guo WT, Zhang HZ, Dong K. Distinct roles of adenosine deaminase isoenzymes ADA1 and ADA2: a pan-cancer analysis. Front Immunol 2022; 13: 903461

[13] Lee PY, Schulert GS, Canna SW, Huang Y, Sundel J, Li Y, Hoyt KJ, Blaustein RB, Wactor A, Do T, Halyabar O, Chang MH, Dedeoglu F, Case SM, Meidan E, Lo MS, Sundel RP, Richardson ET, Newburger JW, Hershfield MS, Son MB, Henderson LA, Nigrovic PA. Adenosine deaminase 2 as a biomarker of macrophage activation syndrome in systemic juvenile idiopathic arthritis. Ann Rheum Dis 2020; 79(2): 225–231

[14] Zhou Q, Yang D, Ombrello AK, Zavialov AV, Toro C, Zavialov AV, Stone DL, Chae JJ, Rosenzweig SD, Bishop K, Barron KS, Kuehn HS, Hoffmann P, Negro A, Tsai WL, Cowen EW, Pei W, Milner JD, Silvin C, Heller T, Chin DT, Patronas NJ, Barber JS, Lee CC, Wood GM, Ling A, Kelly SJ, Kleiner DE, Mullikin JC, Ganson NJ, Kong HH, Hambleton S, Candotti F, Quezado MM, Calvo KR, Alao H, Barham BK, Jones A, Meschia JF, Worrall BB, Kasner SE, Rich SS, Goldbach-Mansky R, Abinun M, Chalom E, Gotte AC, Punaro M, Pascual V, Verbsky JW, Torgerson TR, Singer NG, Gershon TR, Ozen S, Karadag O, Fleisher TA, Remmers EF, Burgess SM, Moir SL, Gadina M, Sood R, Hershfield MS, Boehm M, Kastner DL, Aksentijevich I. Early-onset stroke and vasculopathy associated with mutations in ADA2. N Engl J Med 2014; 370(10): 911–920

[15] Navon Elkan P, Pierce SB, Segel R, Walsh T, Barash J, Padeh S, Zlotogorski A, Berkun Y, Press JJ, Mukamel M, Voth I, Hashkes PJ, Harel L, Hoffer V, Ling E, Yalcinkaya F, Kasapcopur O, Lee MK, Klevit RE, Renbaum P, Weinberg-Shukron A, Sener EF, Schormair B, Zeligson S, Marek-Yagel D, Strom TM, Shohat M, Singer A, Rubinow A, Pras E, Winkelmann J, Tekin M, Anikster Y, King MC, Levy-Lahad E. Mutant adenosine deaminase 2 in a polyarteritis nodosa vasculopathy. N Engl J Med 2014; 370(10): 921–931

[16] Signa S, Bertoni A, Penco F, Caorsi R, Cafaro A, Cangemi G, Volpi S, Gattorno M, Schena F. Adenosine deaminase 2 deficiency (DADA2): a crosstalk between innate and adaptive immunity. Front Immunol 2022; 13: 935957

[17] Kaljas Y, Liu C, Skaldin M, Wu C, Zhou Q, Lu Y, Aksentijevich I, Zavialov AV. Human adenosine deaminases ADA1 and ADA2 bind to different subsets of immune cells. Cell Mol Life Sci 2017; 74(3): 555–570

[18] Carmona-Rivera C, Khaznadar SS, Shwin KW, Irizarry-Caro JA, O’Neil LJ, Liu Y, Jacobson KA, Ombrello AK, Stone DL, Tsai WL, Kastner DL, Aksentijevich I, Kaplan MJ, Grayson PC. Deficiency of adenosine deaminase 2 triggers adenosine-mediated NETosis and TNF production in patients with DADA2. Blood 2019; 134(4): 395–406

[19] Deuitch NT, Yang D, Lee PY, Yu X, Moura NS, Schnappauf O, Ombrello AK, Stone D, Kuehn HS, Rosenzweig SD, Hoffmann P, Cudrici C, Levy DM, Kessler E, Soep JB, Hay AD, Dalrymple A, Zhang Y, Sun L, Zhang Q, Tang X, Wu Y, Rao K, Li H, Luo H, Zhang Y, Burnham JM, Boehm M, Barron K, Kastner DL, Aksentijevich I, Zhou Q. TNF inhibition in vasculitis management in adenosine deaminase 2 deficiency (DADA2). J Allergy Clin Immunol 2022; 149(5): 1812–1816.e6

[20] Lee PY, Aksentijevich I, Zhou Q. Mechanisms of vascular inflammation in deficiency of adenosine deaminase 2 (DADA2). Semin Immunopathol 2022; 44(3): 269–280

[21] Karapetyan L, Luke JJ, Davar D. Toll-like receptor 9 agonists in cancer. OncoTargets Ther 2020; 13: 10039–10060

[22] Pohar J, Lainšček D, Ivičak-Kocjan K, Cajnko MM, Jerala R, Benčina M. Short single-stranded DNA degradation products augment the activation of Toll-like receptor 9. Nat Commun 2017; 8(1): 15363

[23] Ohto U, Ishida H, Shibata T, Sato R, Miyake K, Shimizu T. Toll-like receptor 9 contains two DNA binding sites that function cooperatively to promote receptor dimerization and activation. Immunity 2018; 48(4): 649–658.e4

[24] Krieg AM. CpG motifs in bacterial DNA and their immune effects. Annu Rev Immunol 2002; 20(1): 709–760

[25] Verthelyi D, Zeuner RA. Differential signaling by CpG DNA in DCs and B cells: not just TLR9. Trends Immunol 2003; 24(10): 519–522

[26] Van der Sluis RM, Holm CK, Jakobsen MR. Plasmacytoid dendritic cells during COVID-19: ally or adversary. Cell Rep 2022; 40(4): 111148

[27] Swiecki M, Colonna M. The multifaceted biology of plasmacytoid dendritic cells. Nat Rev Immunol 2015; 15(8): 471–485

[28] Bencze D, Fekete T, Pázmándi K. Type I interferon production of plasmacytoid dendritic cells under control. Int J Mol Sci 2021; 22(8): 4190

[29] Kerkmann M, Rothenfusser S, Hornung V, Towarowski A, Wagner M, Sarris A, Giese T, Endres S, Hartmann G. Activation with CpG-A and CpG-B oligonucleotides reveals two distinct regulatory pathways of type I IFN synthesis in human plasmacytoid dendritic cells. J Immunol 2003; 170(9): 4465–4474

[30] Combes A, Camosseto V, N’Guessan P, Argüello RJ, Mussard J, Caux C, Bendriss-Vermare N, Pierre P, Gatti E. BAD-LAMP controls TLR9 trafficking and signalling in human plasmacytoid dendritic cells. Nat Commun 2017; 8(1): 913

[31] Marongiu L, Gornati L, Artuso I, Zanoni I, Granucci F. Below the surface: the inner lives of TLR4 and TLR9. J Leukoc Biol 2019; 106(1): 147–160

[32] Liu C, Skaldin M, Wu C, Lu Y, Zavialov AV. Application of ADA1 as a new marker enzyme in sandwich ELISA to study the effect of adenosine on activated monocytes. Sci Rep 2016; 6(1): 31370

[33] Trivedi PC, Bartlett JJ, Pulinilkunnil T. Lysosomal biology and function: modern view of cellular debris bin. Cells 2020; 9(5): 1131

[34] Haas T, Metzger J, Schmitz F, Heit A, Müller T, Latz E, Wagner H. The DNA sugar backbone 2′ deoxyribose determines toll-like receptor 9 activation. Immunity 2008; 28(3): 315–323

[35] Kim S, Kaiser V, Beier E, Bechheim M, Guenthner-Biller M, Ablasser A, Berger M, Endres S, Hartmann G, Hornung V. Self-priming determines high type I IFN production by plasmacytoid dendritic cells. Eur J Immunol 2014; 44(3): 807–818

[36] Hornung V, Guenthner-Biller M, Bourquin C, Ablasser A, Schlee M, Uematsu S, Noronha A, Manoharan M, Akira S, de Fougerolles A, Endres S, Hartmann G. Sequence-specific potent induction of IFN-alpha by short interfering RNA in plasmacytoid dendritic cells through TLR7. Nat Med 2005; 11(3): 263–270

[37] Sleat DE, Zheng H, Qian M, Lobel P. Identification of sites of mannose 6-phosphorylation on lysosomal proteins. Mol Cell Proteomics 2006; 5(4): 686–701

[38] Niemeyer CM, Sieff CA, Mathey-Prevot B, Wimperis JZ, Bierer BE, Clark SC, Nathan DG. Expression of human interleukin-3 (multi-CSF) is restricted to human lymphocytes and T-cell tumor lines. Blood 1989; 73(4): 945–951

[39] Janke M, Witsch EJ, Mages HW, Hutloff A, Kroczek RA. Eminent role of ICOS costimulation for T cells interacting with plasmacytoid dendritic cells. Immunology 2006; 118(3): 353–360

[40] Bénard A, Jacobsen A, Brunner M, Krautz C, Klösch B, Swierzy I, Naschberger E, Podolska MJ, Kouhestani D, David P, Birkholz T, Castellanos I, Trufa D, Sirbu H, Vetter M, Kremer AE, Hildner K, Hecker A, Edinger F, Tenbusch M, Mühl-Zürbes P, Steinkasserer A, Richter E, Streeck H, Berger MM, Brenner T, Weigand MA, Swirski FK, Schett G, Grützmann R, Weber GF. Interleukin-3 is a predictive marker for severity and outcome during SARS-CoV-2 infections. Nat Commun 2021; 12(1): 1112

[41] Vollmer J, Weeratna R, Payette P, Jurk M, Schetter C, Laucht M, Wader T, Tluk S, Liu M, Davis HL, Krieg AM. Characterization of three CpG oligodeoxynucleotide classes with distinct immunostimulatory activities. Eur J Immunol 2004; 34(1): 251–262

[42] Hartmann G, Battiany J, Poeck H, Wagner M, Kerkmann M, Lubenow N, Rothenfusser S, Endres S. Rational design of new CpG oligonucleotides that combine B cell activation with high IFN-alpha induction in plasmacytoid dendritic cells. Eur J Immunol 2003; 33(6): 1633–1641

[43] Zhong XZ, Zou Y, Sun X, Dong G, Cao Q, Pandey A, Rainey JK, Zhu X, Dong XP. Inhibition of transient receptor potential channel mucolipin-1 (TRPML1) by lysosomal adenosine involved in severe combined immunodeficiency diseases. J Biol Chem 2017; 292(8): 3445–3455

Acknowledgments

The authors wish to thank the staff of the Institute of Pediatrics for participating in the study.

Compliance with ethics guidelines

The authors declare that the research was conducted without commercial or financial relationships that can be considered a conflict of interest. Simon C. Robson is a scientific founder of Purinomia Biotech Inc. and consults for eGenesis, AbbVie, and SynLogic Inc.; his interests are reviewed and managed by HMFP at Beth Israel Deaconess Medical Center in accordance with the conflict-of-interest policies.

Electronic Supplementary Material

Supplementary material is available in the online version of this article at https://doi.org/10.1007/s11684-024-1067-5 and is accessible to authorized users.

RIGHTS & PERMISSIONS

2024 Higher Education Press