Streptavidin Promotes DNA Binding and Activation of cGAS to Enhance Innate Immunity.

cGAS/STING signaling plays an essential role in sensing cytosolic DNA. cGAS activity is regulated by posttranslational modifications and binding partners. cGAS interactome largely includes mammalian or viral proteins. Whether and how bacterial proteins bind cGAS to modulate innate immunity remain elusive. Here, we found streptavidin, a secreted bacterial protein, selectively bound cGAS to promote DNA-induced cGAS activation and interferon-β production. Mechanistically, streptavidin enhanced DNA binding and cGAS phase separation, therefore facilitating cGAS activation. Using an HSV-1-infected mouse model, we found streptavidin nanoparticles facilitated HSV-1 clearance through improving innate immunity. Considering the clinical usage of streptavidin as an immune stimulant and drug delivery vehicle and its biotechnological usage for biotin-labeled protein purification and detection, our studies not only provide an example for a bacterial protein regulating cGAS activity but also suggest caution needs to be taken when using streptavidin in various applications given to its ability to induce innate immunity.

Introduction

In mammals, the immune system contains both specific (adaptive immunity) and nonspecific immunity (innate immunity) to defend against pathogens. As the first line of host defense, the innate immune system utilizes germLine-encoded receptors named pattern recognition receptors (PRRs) to detect invading pathogens (Brubaker et al., 2015). As DNA or RNA is indispensable for propagation of pathogens, sensing of invading foreign RNA and DNA serves as a fundamental mechanism for host defense. Human cells respond to accumulated cytosolic DNA through activation of cGAS/STING signaling (Sun et al., 2013). Mechanistically, accumulated cytosolic DNA, from viral (Goubau et al., 2013) or bacterial (Patrick et al., 2016) infections, or intrinsic damaged mitochondria (Sliter et al., 2018) and genomic DNA (Bakhoum et al., 2018; Dou et al., 2017; Harding et al., 2017; Mackenzie et al., 2017), binds and activates mammalian cGAS (cGAMP synthase) to generate 2′3′-cGAMP that subsequently activates STING to promote type I interferon (IFN) production and expression of antiviral and immune modulatory genes (Diner et al., 2013; Li et al., 2013; Sun et al., 2013; Zhang et al., 2013). Mouse models suggest that loss of cGAS sensitizes mice to viral infection (Gao et al., 2013; Li et al., 2013), whereas hyperactivation of the cGAS/STING pathway results in autoimmune diseases (Gao et al., 2015) and radioresistance (Deng et al., 2014). Consistent with its role in innate immune recognition of DNA, cGAS/STING signaling also provides an anticancer function by which cGAS activation or administration of cGAMP synergistically functions with immune check blockades to suppress melanoma growth in murine models (Wang et al., 2017) (reviewed in (Ablasser and Chen, 2019) and (Barber, 2015)).

Beyond innate immunity, cGAS also regulates tumorigenic phenotypes. For example, cGAS is essential for cellular senescence (Yang et al., 2017), and chromosomal-instability-induced cGAS activation facilitates breast cancer metastasis (Bakhoum et al., 2018). In addition, nuclear cGAS binds PARP1 to suppress HR and promote lung cancer growth in a cGAS-enzyme-independent manner (Liu et al., 2018). Given its pivotal roles in both innate immunity and tumorigenesis, cGAS activity is tightly controlled. Various posttranslational modifications have been reported to fine-tune cGAS activity and function, including phosphorylation (Liu et al., 2018; Seo et al., 2015), acetylation (Dai et al., 2019), glutamylation (Xia et al., 2016), sumoylation (Cui et al., 2017; Hu et al., 2016), ubiquitination (Chen and Chen, 2019; Seo et al., 2018), and others (Ablasser and Gulen, 2016). In addition, cellular localization of cGAS provides an additional layer of regulation governing spatial cGAS activation (Barnett et al., 2019). cGAS also recognizes HIV DNA in the nucleus (Lahaye et al., 2018), and the nuclear cGAS binds centromeric DNA (Gentili et al., 2019) and chromatin that suppresses DNA repair (Jiang et al., 2019). On the other hand, chromatin binding in the nucleus suppressed cGAS activity (Volkman et al., 2019). Mechanistically, cGAS is activated by DNA binding to its N-terminus or enzymatic domain that triggers cGAS phase separation (Du and Chen, 2018). Intriguingly, multiple cGAS-binding proteins have been identified with roles in either suppressing cGAS activity (such as Beclin-1 (Liang et al., 2014), OASL (Ghosh et al., 2019), a herpesvirus virion protein ORF52 (Wu et al., 2015), and a cytomegalovirus tegument protein pp65 (Biolatti et al., 2018)) or promoting cGAS activity (such as PQBP1 (Yoh et al., 2015), G3BP1 (Liu et al., 2019), and ZCCHC3 (Lian et al., 2018)). Notably, most of the identified cGAS-binding proteins are either mammalian proteins or viral proteins, and there are a limited number of bacterial proteins observed to facilitate cGAS activation, such as HU from Listeria monocytogenes that bends DNA to create favorable DNA conformations for cGAS recognition (Andreeva et al., 2017). Whether and how any bacterial proteins bind and regulate cGAS activation and innate immunity remain elusive.

Streptavidin is a secreted bacterial protein produced by the soil bacterium Streptomyces avidinii with a high affinity for biotin (vitamin B7). Due to its nature as one of the strongest noncovalent interactions, streptavidin has been extensively used in molecular biology and biotechnology for purification and detection of biotin-labeled proteins or nucleotides (Dundas et al., 2013) or as a tool to identify new drug targets (Bykhovski et al., 2013), as well as in clinics as immune stimulants (Weir et al., 2014) or as a drug delivery vehicle (Jain and Cheng, 2017) for biotin-labeled biomaterials. Whether streptavidin affects innate immunity remains unknown.

Results

Streptavidin Binds cGAS In Vitro and in Cells

Surprisingly, we found that streptavidin-conjugated agarose beads displayed a strong ability to pull down bacterially purified full-length His-hcGAS (human cGAS) proteins in vitro (Figure 1A). This was not true for biotin-interacting neutravidin-conjugated agarose beads, nor commonly used glutathione, Flag-antibody-, HA-antibody-, or Myc-antibody-coupled agarose beads (Figures 1A, S1A, and S1B). Streptavidin binding to hcGAS was further confirmed by using streptavidin magnetic beads to exclude the possibility that agarose beads contribute to hcGAS interaction with streptavidin in vitro (Figures 1B, S1C, and S1D). In addition, benzonase nuclease treatment that digest nucleotides (both DNA and RNA) did not affect hcGAS binding to streptavidin in vitro (Figure S1E and S1F), suggesting that observed cGAS binding to streptavidin is not mediated by DNA or RNA. The binding affinity of His-hcGAS with streptavidin was ∼1.3 μM in vitro determined by ITC analysis (Figure 1C). Considering the tetramer form of streptavidin and allowable errors in protein concentration calculation, roughly streptavidin binds hcGAS in a 1:1 ratio (Figure 1C). Moreover, streptavidin, but not neutravidin beads, were able to pull down ectopically expressed hcGAS from HEK293 cells (Figures 1D and S1G) in a cGAS-enzyme-activity-independent manner (Figures 1E and S1H). More importantly, streptavidin beads were able to pull down endogenous hcGAS but not STING nor TBK1 in MDA-MB-231 cells (Figure 1F). Ectopically expressed streptavidin in HEK293 cells co-migrated with (Figure 1G) and interacted with transfected hcGAS in cells (Figure 1H). Moreover, streptavidin, but not neutravidin beads, was able to pull down bacterially purified mcGAS (mouse cGAS) proteins in vitro (Figure S1I) and endogenous mcGAS from either 4T1 (Figure S1J) or B16 cells (Figure S1K). Together, these data suggest that streptavidin binds both hcGAS and mcGAS in vitro and in cells.

Figure 1

Streptavidin Directly Binds cGAS In Vitro and in Cells

(A and B) In vitro pull-down assays using indicated agarose beads (A) or magnetic beads (B) with recombinant His-hcGAS proteins demonstrate that streptavidin strongly interacts with cGAS. M280, T1, C1, and M270 are streptavidin magnetic beads from Invitrogen. Data represent results from two independent experiments.

(C) ITC analysis using bacterially purified His-hcGAS proteins and streptavidin proteins suggests their binding affinity of 1.2 μM in vitro.

(D and E) Streptavidin beads pull down hcGAS expressed in cells. Immunoblot (IB) analysis of whole cell lysates (WCL) and pulldowns by indicated beads derived from HEK293 cells transiently transfected with indicated hcGAS constructs. Notably, KKEA (K173E/R176E/K407E/K411A) is a cGAS mutant deficient in binding DNA. Data represent results from two independent experiments.

(F) IB analysis of streptavidin beads or neutravidin beads pulldowns using WCL derived from MDA-MB-231 cells. Forty-eight hours posttransfection, pull-down assays were performed. Data represent results from two independent experiments.

(G) A gel filtration experiment using cell lysates derived from HEK293 cells transfected with indicated DNA constructs.

(H) WCL of (G).

(I) IB analysis of IPs and WCL derived from HEK293T cells transfected with indicated DNA constructs. Cells were collected 48 h posttransfection. Data represent results from two independent experiments. N23A/S27D/S45A-streptavidin is a biotin binding deficient mutant.

(J) (Left panel) A cartoon illustration of the experimental procedure: streptavidin agarose beads were preincubated with excess amount of biotin at room temperature for 2 h and nonbinding biotin was washed off. Biotin-saturated and nonsaturated streptavidin beads were used to pull down biotin-H4 (1-23) peptides (top right panel) or recombinant GST-hcGAS proteins (top bottom panel). Data represent results from two independent experiments.

Compared with biotin, the larger size of cGAS may prevent it from fitting into the biotin-binding pocket of streptavidin (Figure S1L). Interestingly, a streptavidin mutant (N23A/S27D/S45A) deficient in binding biotin (Howarth et al., 2006) largely retained its binding with cGAS (Figure 1I). In addition, saturating streptavidin beads with biotin, although reduced binding with biotin-labeled H4 peptides, did not significantly affect streptavidin binding with cGAS in vitro (Figure 1J). Titrating in increasing doses of biotin also did not compete with cGAS to bind streptavidin beads in vitro (Figures S1M and S1N). These data suggest that cGAS may not compete with biotin to recognize the same motif of streptavidin.

Streptavidin Binding Promotes cGAS Activation

Next, we examined if streptavidin binding to cGAS modulates cGAS activity. To this end, we generated EA.hy926 cells stably expressing either a GFP control or streptavidin (Figure S2A). We found that compared with GFP expressing EA.hy926 cells, streptavidin expression enhanced cellular TBK1-pS172 and IRF3-pS396 levels upon stimulation by ISD90 (Figure 2A) or ISD45 (Figure S2B) but not by the cGAS product, 2′3′-cGAMP (Figure 2B), and minimally affected these proteins upon RNA challenge with polyI:C (Figure S2C). As a result, streptavidin expression led to increased levels of IFNβ mRNA transcription (Figure 2C) and secreted IFNβ (Figures 2D and S2F). Notably, neither cGAS (Figure S2D) nor STING (Figure S2E) mRNA expression changes were observed under these treatments. These data suggest that streptavidin promotes DNA-induced cGAS activation and subsequent IFNβ production in cells. In addition, streptavidin was able to promote cGAS activation by either shorter ISD45 or longer ISD90 stimulation (Figure 2E). Given that streptavidin promotes cGAS activation at early ISD treatment time points, rather than sustains ISD90-induced TBK1 and IRF3 phosphorylation (Figure 2A), we further examined if streptavidin-enhanced innate immune activation depended on ISD90 concentrations. To this end, we found that under both 3 μg/mL and 6 μg/mL ISD90 stimulation conditions, expression of streptavidin enhanced IRF3-pS386 phosphorylation at early time points (Figures S2G and S2H). Moreover, to exclude possible effects of GFP as a negative control in modulating innate immune signaling, we engineered an empty vector (EV) expressing EA.hy926 cell line as an additional control. We found that compared with EV-expressing cells, streptavidin expression promoted TBK1/IRF3 phosphorylation and subsequent IFNβ production upon ISD90 (Figures S2I–S2K) but not under 2′3′cGAMP (Figures S2L–S2N) stimulation conditions, further supporting our conclusion that streptavidin binding to cGAS promotes cGAS activation. Interestingly, streptavidin itself was not able to activate STING in the absence of cGAS, and streptavidin also facilitated cGAS activation in HEK293T cells (Figure 2F), supporting the notion that streptavidin may function through cGAS to enhance DNA-induced innate immune responses. Notably, stable expression of streptavidin in EA.hy926 cells did not significantly affect cell growth (Figure S2O), revealing streptavidin-mediated cellular sensitivity to DNA insults is unlikely due to growth changes.

Figure 2

Streptavidin Binds cGAS to Facilitate cGAS Activation

(A and B) IB analysis of WCL derived from EA.hy926 cells stably expressing either GFP or streptavidin treated with 2 μg/mL ISD90 (A) or 5 μg/mL 2′3′-cGAMP (B) for indicated time periods. pTBK1 and pIRF3 signals were quantified by ImageJ and presented below each western blot. Data represent results from three independent experiments.

(C) RT-PCR analysis of IFNβ mRNA levels in EA.hy926 cells stably expressing either GFP or streptavidin treated with ISD90 or 2′3′-cGAMP for 6 h ∗p < 0.05 from Student's t tests. Data were obtained from biological triplicates.

(D) ELISA assays using cell culture media from EA.hy926 cells stably expressing either GFP or streptavidin treated with ISD90 for 16 h ∗p < 0.05 from Student's t tests. Data were obtained from biological triplicates.

(E) IB analysis of WCL derived from EA.hy926 cells stably expressing either GFP or streptavidin treated with ISD45 or ISD90 for indicated time periods. pTBK1 and pIRF3 signals were quantified by ImageJ and presented below each western blot. Data represent results from two independent experiments.

(F) IB analysis of WCL derived from HEK293T cells transfected with indicated DNA constructs. Cells were collected 48 h posttransfection. Data represent results from two independent experiments.

Streptavidin Enhances DNA Binding to cGAS to Facilitate cGAS Activation

Mechanistically, we found that both the N-terminus of cGAS (Figure S3A), a domain important for cGAS binding to DNA (Tao et al., 2017) and possible localization to plasma membrane (Barnett et al., 2019), and the C-terminal catalytic domain (ΔN) (Figures S3B–S3E) were necessary for binding streptavidin proteins in vitro. As a result, addition of recombinant streptavidin proteins only promoted DNA binding to the full-length hcGAS proteins (Figures 3A, 3B and S3F–S3G) but not N- (Figure 3C) nor ΔN-hcGAS proteins (Figure 3D) in vitro. Consistent with previous reports (Luecke et al., 2017), full-length or ΔN, but not the N-terminus of cGAS, was able to induce TBK1/IRF3 activation in cells (Figure S3H). Consistent with full-length cGAS being required for streptavidin binding, streptavidin only specifically promoted cGAS activation in cells expressing full-length, but not ΔN-cGAS (Figure S3I).

Figure 3

Streptavidin Binds cGAS to Enhance DNA Binding

(A) A cartoon illustration of the experimental design: recombinant GST-hcGAS proteins were immobilized onto glutathione (GSH) beads and incubated with linear pcDNA3.0 DNA in the presence or absence of streptavidin proteins.

(B–D) GST-hcGAS recombination proteins (including FL, full length in B, N-hcGAS in C, and ΔN-hcGAS in D) pull-down assays as described in (A) indicating streptavidin promotes full-length but not truncated cGAS binding to DNA in vitro. Where indicated, 15 cycles of PCR were used to determine DNA abundance from GST-hcGAS pulldowns. Data represent results from two independent experiments.

(E) His-hcGAS recombinant protein pull-down assays indicating hcGAS specifically binds streptavidin but not avidin in vitro. Data represent results from two independent experiments.

(F and G) Representative streptavidin protein structure (PDB: 3ry2) and avidin protein structure (PDB: 2avi) with the altered YRNA-streptavidin motif highlighted in red. Indicated protein structures were downloaded from PDB (protein database) and analyzed by PyMOL.

(H) IB analyses of Flag-IPs and WCL derived from HEK293T cells transfected with indicated DNA constructs. Cells were collected 48 h posttransfection. Data represent results from two independent experiments.

(I–K) Representative images from confocal imaging of EA.hy926 cells stably expressing either EV or streptavidin transfected with Cy3-ISD45 using a cGAS antibody at 1 h (I), 2 h (J), and 4 h (K) posttransfection. Scale bars represent 50 μm in I, J, and K.

(L) Quantification of colocalized cGAS/Cy3-DNA foci in I, J, and K. ∗p < 0.05 from Student's t tests. Quantifications were performed by counting at least 100 cells/each experiment and represent percentage of cells with puncta.

(M) A cycloheximide (CHX) chase experiment to measure endogenous STING protein half-life in EA.hy926 cells stably expressing either EV or streptavidin transfected with ISD90. Where indicated, 200 μg/mL CHX was added to cell culture and ISD90 was transfected into cells before taking time points. Data represent results from two independent experiments.

(N) Quantification of STING protein abundance in (M).

A “YRNA” Motif in Streptavidin Is Important for Streptavidin Binding to cGAS

Although distinct in primary protein sequence, avidin and streptavidin share common core structures for biotin binding. Interestingly, avidin is abundant in egg white with an additional glycoprotein portion compared with streptavidin; however, unlike streptavidin, avidin displayed minimal binding capacity with cGAS in vitro (Figures 3E and S3J). Coupled with the fact that a deglycosylated form of avidin, neutravidin, also did not bind cGAS in vitro (Figure 1A), these data suggest additional structural features in avidin and neutravidin prevent cGAS interaction and further suggest that cGAS may bind to a non-biotin-binding motif in streptavidin.

Intrigued by the fact that although both streptavidin and avidin share a similar structure to bind biotin, while unlike streptavidin, avidin does not bind cGAS (Figure 3E), we reasoned that unique structural features in streptavidin contribute to streptavidin binding to cGAS. Through a primary protein sequence alignment between streptavidin and avidin (Figure S3K), we identified a “YRNA” motif that is missing in avidin and displays an extended structure in streptavidin (Figures 3F and 3G). Excitingly, deleting this “YRNA” motif disrupted streptavidin binding to cGAS (Figure 3H) and subsequently reduced DNA-induced activation of the cGAS/STING signaling (Figure S3L) in cells. In addition, YRNA deletion did not significantly affect streptavidin binding to biotin (-tagged histone peptides) in vitro (Figure S3M), further supporting that cGAS and biotin independently recognize streptavidin. Together, these data support that the “YRNA” motif in streptavidin plays an important role in bridging streptavidin binding to cGAS.

Streptavidin Enhances DNA Binding to cGAS to Promote cGAS Phase Separation

We found that streptavidin partially enhanced cGAS phase separation in vitro, which is necessary for cGAS activation (Du and Chen, 2018) induced by ISD45 (Figures S3N and S3O) or ISD90 (Figure S3P). In addition, streptavidin expression in cells significantly increased the number of cGAS puncta (indicators of cGAS activation) formed in cells upon ISD45 (Figures S3Q and S3R) or ISD90 stimulation (Figures S3S and S3T), which is presumably due to increased cGAS phase separation induced by streptavidin in a time-dependent dynamic manner (Figures 3I–3L). Overall, our data suggest that streptavidin directly binds cGAS to facilitate cGAS binding to DNA for activation. Intriguingly, we observed that STING expression levels were increased in streptavidin expressing EA.hy926 cells and upon ISD90 stimulation, streptavidin expression partially protected STING from degradation (Figures 3M and 3N), which was largely through attenuating STING ubiquitination-mediated STING degradation (Figure S3U). Given that streptavidin does not directly bind STING (Figure 1F), this observation suggests that streptavidin may have additional binding partners in indirectly regulating STING protein stability (eg, through regulating STING ubiquitination or STING binding with lysosomal components). Nevertheless, streptavidin may promote cGAS/STING activation in multiple ways.

Streptomyces avidinii Triggers Innate Immune Responses in Mice

As a soil bacterium, to date no report shows S avidinii as a potential human or animal pathogen to trigger innate immunity. Given that streptavidin is largely produced by S avidinii, we were curious to test if S avidinii could infect mice. We injected increasing doses of S avidinii via an IP route into C57BL/6 mice, and 24 h later mouse spleens were harvested. We found that 10 × 105 pfu of S avidinii triggered the increased innate immune response in C57BL/6 mice (compared with 10 × 103 and 10 × 107 pfu), as evidenced by increased TBK1-pS172 signals (Figure 4A) and IFNβ mRNA levels (Figures S4A and S4B). However, S avidinii-induced increase in pTBK1 was not only mediated by the cGAS/STING signaling pathway, given that in an in vitro co-culture model using S avidinii and MDA-MB-231 cells, but also promoted significant TBK1 phosphorylation in cGAS MDA-MB231 cells, compared with wild-type (WT)-MDA-MB-231 cells (Figures 4B and S4C). It is plausible that in addition to DNA, bacterial proteins and RNA could also modulate innate immune responses through innate immune pathways such as endosomal Toll-like receptors and RIG-I/MAVS signaling, respectively (Ni et al., 2018; Tan et al., 2018). Notably, co-culture with S avidinii caused toxicity largely to WT but not cGAS MDA-MB-231 cells in vitro (Figures 4C and S4D). This suggests that cGAS-mediated IFNβ and other cytokine production might be important to trigger cellular apoptosis, as evidenced by cleaved-PARP and cleaved-caspase 3 signals (Figure 4B). In addition, we recognize that a streptavidin-negative strain will help further evaluate the specific effects of streptavidin on modulating innate immunity; however, due to technical difficulties we could not obtain such a strain and this will be a subject of future investigation. Based on these observations, we decided to examine if streptavidin nanoparticles, or streptavidin proteins, instead of S avidinii, promote DNA-induced innate immune signaling in both in vitro and in vivo models.

Figure 4

Fluorescence-Conjugated Streptavidin or Avidin Detects Endogenous cGAS

(A) PBS or 10 × 103, 10 × 105, and 10 × 107S avidinii bacteria were injected into BL6 mice through IP injection. Twenty-four hours postinjection mouse spleens were harvested and lysed for IB analyses.

(B) 10 × 107S avidinii bacteria were co-cultured with WT or cGAS MDA-MB-231 cells, and WCLs were obtained at indicated periods for IB analyses. Data represent results from two independent experiments.

(C) Representative images of cells from (B).

(D) 10 × 103.5 HSV-1 were used to infect WT or cGAS MDA-MB-231 cells in the presence or absence of 20 μL streptavidin nanoparticles and 100 μg streptavidin proteins in 24-well plates. WCLs were obtained at indicated periods for IB analyses. Quantifications of IRF3-pS386 signals were performed by ImageJ. Data represent results from two independent experiments.

(E) A low dose (10 × 104 pfu) or a high dose (10 × 107 pfu) of HSV-1 was injected into BL6 mice through tail vein injection. Fifty microliters of streptavidin nanoparticles or 50 μg streptavidin proteins were co-injected where indicated. Twenty-four hours postinjection, mouse spleens were harvested for IB analyses.

(F) HSV-1 mRNA analyses using brain tissues from either WT or cGAS BL6 mice with indicated treatments for 24 h ∗p < 0.05 from Student's t tests.

(G) HSV-1 genome copy number analyses using brain tissues from either WT or cGAS BL6 mice with indicated treatments for 24 h. ∗p < 0.05 from Student's t tests.

(H) Representative images from confocal imaging of either cGAS or cGAS MDA-MB-231 cells using a fluorescence-labeled streptavidin probe (488-strep). The scale bar represents 50 μm.

(I) Quantification of (H). At least 100 cells were counted for each group for quantification.

To this end, we used the DNA virus HSV-1 that has been well characterized to induce cGAS/STING signaling activation (Ni et al., 2018; Reinert et al., 2016). We found that infection of MDA-MB-231 cells with HSV-1 triggered cGAS/STING signaling activation and that addition of either streptavidin nanoparticles or streptavidin proteins facilitated HSV-1-induced IRF3 phosphorylation, which was largely dependent on cGAS (Figure 4D). Furthermore, we found that higher titers of HSV-1 led to increased cGAS activation (as evidenced by TBK1-pS172 signals) in spleens of C57BL/6 mice harvested 24 h postinfection (Figure 4E). Furthermore, co-administration of either streptavidin nanoparticles or streptavidin proteins through tail-vein injection further augmented TBK1-pS172 signals (Figure 4E). These data support that streptavidin facilitates cGAS activation upon HSV-1 infection in vitro and in mice. To further investigate if streptavidin also facilitates innate immune activation to clear HSV-1 infection, we injected a high dose of HSV-1 (1 × 10 × 108 pfu) into both WT and cGAS C57BL/6 mice, in the presence or absence of streptavidin nanoparticles. We observed that cGAS was necessary for HSV-1-infection-induced TBK1 and IRF3 phosphorylation in mice (Figure S4E) and HSV-1-infected mice were sicker with significantly less movement and dramatically reduced body temperature. However, when co-injected with streptavidin nanoparticles, mice displayed a higher body temperature and increased movement. Twenty-four hours postinfection, brains were harvested for analyses on HSV-1 infection. We found that injection with streptavidin nanoparticles reduced HSV-1 mRNA expression in brains from WT but not cGAS mice infected with HSV-1 (Figure 4F). In addition, HSV-1 genome copy number was also slightly reduced in streptavidin-nanoparticle-treated mice (Figure 4G). Together, these data suggest that streptavidin nanoparticles may help HSV-1 clearance, presumably due to promoting activation of the cGAS/STING signaling pathway. This is consistent with the immune stimulant function of streptavidin observed previously (Weir et al., 2014). However, we recognize that the protective effects from streptavidin proteins or streptavidin nanoparticles are minor, which might be due to low sequence homology between mouse and human cGAS. Although both hcGAS (Figure 1A) and mcGAS (Figure S1I) proteins bound to streptavidin in vitro, whether streptavidin recognizes and modulates hcGAS and mcGAS differently requires further in-depth investigations.

Fluorescent-Conjugated Streptavidin Probes Detect Endogenous cGAS Expression

Fluorescent-labeled streptavidin (Alex Fluor 488-strep) has been widely used in biotechnology to amplify the signal magnitude for in vivo labeling and detection of a given protein/nucleotide target by imaging (Howarth and Ting, 2008). Given our observation that streptavidin binds cGAS in vitro and in cells, we further tested if the streptavidin Fluor probes cross-react with endogenous cGAS. Using both WT and cGAS MDA-MB-231 cells, we found that although streptavidin Fluor probes detected signals in cGAS cells, which are presumably biotin or other streptavidin-binding proteins, the presence of endogenous cGAS significantly enhanced signals generated by these probes (Figures 4H and S4F). These data suggest that streptavidin Fluor conjugates detect background signals including endogenous cGAS. Interestingly, staining endogenous cGAS by either a cGAS antibody or the streptavidin-488 probe resulted in largely noncolocalization of these two signals (Figure 4I), with streptavidin Fluor conjugates recognizing largely cytoplasmic cGAS species while the cGAS antibody detecting nuclear cGAS. Querying the information for this cGAS antibody (Cell Signaling Technology #15102) indicated that this antibody was generated using a cGAS peptide centered on Ala19 residue, which suggests that either the cGAS antibody and the streptavidin-488 probe recognize a similar motif in the N-terminus of cGAS or binding of either of them can cause a cGAS protein conformational change that is not favored by the other. To further examine if streptavidin has a preference for binding cytosolic cGAS, we fractionated BPH1 or EA.hy926 cells into both cytoplasmic and nuclear fractions (where cGAS was observed in both fractions) and subsequently performed streptavidin-agarose-beads-mediated pull-down assays. We found streptavidin agarose beads preferred cytoplasmic cGAS to nuclear cGAS for binding (Figures S4G–S4J). This could be due to differential posttranslational modifications on cytoplasmic or nuclear cGAS or because additional binding partners contribute to the observed binding preferences.

Discussion

S avidinii is a soil bacterium, and it is not currently known whether it is a human or animal pathogen. We found that injecting S avidinii into C57BL/6 mice led to the induction of innate immunity (Figure 4A), suggesting that S avidinii may be a pathogenic bacterium. S avidinii is famous for its ability to produce and secrete streptavidin. Due to the tight and selective interactions with biotin even under stringent conditions, streptavidin and its homologs are widely used in protein detection, labeling, immune stimulation, and drug delivery (Dundas et al., 2013). Specifically, streptavidin is used to detect and purify strep-tagged or biotin-labeled proteins or other molecules such as DNA. Our immunostaining using a Fluor-labeled streptavidin in MDA-MB-231 cells (Figure 4H) indicates that streptavidin is capable of detecting endogenous cGAS proteins and therefore use of streptavidin in cellular immune assays may be problematic. In addition, streptavidin has been used as an immune stimulant to enhance efficacy of tumor vaccines with unknown mechanism (Weir et al., 2014), and various streptavidin-immune modulator fusions have shown enhanced immune regulatory effects (Arribillaga et al., 2013). Our data suggest that streptavidin activates innate immunity by facilitating cGAS activation induced by HSV-1 infection in vitro (Figure 4D) and in mice (Figure 4E). Moreover, streptavidin-coated nanoparticles are used in delivering biomaterials including antibodies, RNAs, DNAs, and compounds labeled with biotin (D'Agata et al., 2017), as well as pretargeted immunotherapy to deliver radiation only to cancerous cells due to the strong cell adsorptive ability of streptavidin (Altai et al., 2017). Due to the ability of streptavidin in activating innate immunity through cGAS, we speculate that streptavidin nanoparticles used in either drug delivery or pretargeted immunotherapy would trigger innate immune responses to enhance effects of these therapies. However, whether this effect is streptavidin dose dependent requires additional validations in vivo.

Notably, through a streptavidin protein sequence homology search, we found that streptavidin-like proteins are present in other soil bacterial species including Actinokineospora spp., Kitasatospora sp., Pseudonocardia hispaniensis sp., Saccharomonospora marina, Hymenobacter sp., and others. Streptavidin-like proteins are also present in sea creatures such as corals and sponges (Figure S4K). These proteins are uncharacterized but share significant sequence homology with streptavidin (Figure S4L). These observations suggest that these streptavidin-like bacterial proteins and coral/sponge proteins may exert similar function as streptavidin in not only binding biotin but also inducing innate immune responses through interacting with cGAS, which warrants further in-depth investigations.

Limitations of the Study

Although expression of streptavidin in mammalian cells facilitated DNA-induced cGAS activation, it remains critical to further validate if streptavidin directly enhances cGAS activity in vitro in the presence of DNA and if the cGAS catalytic domain is sufficient for this regulation. Moreover, it is important to further determine the stoichiometry of streptavidin proteins internalized by mammalian cells in a S avidinii infection model and if these internalized streptavidin proteins would enhance DNA-induced cGAS activation in vivo without significantly affecting cGAS cellular localization. Endogenous biotin is a co-factor for five human carboxylases and modulates gene expression through biotinylating histones (Zempleni et al., 2009). Given that streptavidin presumably depletes cellular biotin pools, effects from loss of biotin function should also be accounted for in the streptavidin-mediated cGAS pathway regulation. On the other hand, this suggests that using biotin-deficient streptavidin (eg. N23A/S27D/S45A)-coated nanoparticles may provide advantages for in vivo applications. Our study raises the possibility that the soil bacterium S avidinii is recognized by the innate immune system as a foreign pathogen. However, relevant animal infection models are necessary for further validation.

Resource Availability

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Pengda Liu (pengda_liu@med.unc.edu).

Materials Availability

Reagents generated in this study will be available upon request.

Data and Code Availability

The data that support the findings of this study are available from the corresponding authors (B.D. and P.L.) upon reasonable request.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.