Research Article

MAVS, cGAS, and endogenous retroviruses in T-independent B cell responses

See allHide authors and affiliations

Science  19 Dec 2014:
Vol. 346, Issue 6216, pp. 1486-1492
DOI: 10.1126/science.346.6216.1486

This article has been retracted. Please see:

Abstract

Multivalent molecules with repetitive structures including bacterial capsular polysaccharides and viral capsids elicit antibody responses through B cell receptor (BCR) crosslinking in the absence of T cell help. We report that immunization with these T cell–independent type 2 (TI-2) antigens causes up-regulation of endogenous retrovirus (ERV) RNAs in antigen-specific mouse B cells. These RNAs are detected via a mitochondrial antiviral signaling protein (MAVS)–dependent RNA sensing pathway or reverse-transcribed and detected via the cGAS-cGAMP-STING pathway, triggering a second, sustained wave of signaling that promotes specific immunoglobulin M production. Deficiency of both MAVS and cGAS, or treatment of MAVS-deficient mice with reverse transcriptase inhibitors, dramatically inhibits TI-2 antibody responses. These findings suggest that ERV and two innate sensing pathways that detect them are integral components of the TI-2 B cell signaling apparatus.

Endogenous retroviruses trigger B cells

Scattered across our genome are endogenous retroviruses (ERVs), ancient “footprints” of previous viral infections. Scientists do not fully understand their functions, but Zeng et al. now report a role for ERVs in mobilizing a particular type of B cell–driven immune response in mice (T cell–independent, TID), which is usually mounted in response to viral capids or bacterial polysaccharides (see the Perspective by Grasset and Cerutti). Immunizing mice with a model TID antigen elicited an increase in ERV RNA and DNA in the cytoplasm of B cells. Innate immune receptors that recognize cytoplasmic nucleotides then triggered signaling cascades that resulted in the production of immunoglobulin M.

Science, this issue p. 1486; see also p. 1454

Specific antibody production is a hallmark of the B cell response to antigens. T cell–dependent (TD) antibody responses typically elicited by protein antigens require follicular helper T cells for full B cell activation, proliferation, and antibody production. In contrast, T cell–independent (TI) antigens stimulate antibody production in the absence of major histocompatibility complex (MHC) class II–restricted T cell help. TI antigens include the TI type 1 (TI-1) antigens, which engage Toll-like receptors (TLRs) in addition to the B cell receptor (BCR), and TI type 2 (TI-2) antigens, which engage the BCR in a manner that induces extensive crosslinking, leading to BCR activation and immunoglobulin M (IgM) production. TI-2 antigens are large multivalent molecules with highly repetitive structures, such as bacterial capsular polysaccharides and viral capsids (1).

B cell–intrinsic cytosolic DNA and RNA sensing in the TI-2 antibody response

We tested the requirement for innate immune sensing pathways in the antibody response to the model TI-2 antigen 4-hydroxy-3-nitrophenylacetyl-Ficoll (NP-Ficoll) by monitoring anti-NP IgM in the serum of mice after immunization (2). C57BL/6J mice mounted a robust NP-specific IgM response by day 4.5 after immunization, which peaked around day 14.5 after immunization (Fig. 1A and fig. S1). Similarly, mice that could not signal via NLRP3, TLR3, TLR7, TLR9, TLR2, TLR4, CD36, MyD88, TICAM1, or IRAK4, all nucleic acid–sensing TLRs (Unc93b13d/3d), or all TLRs (Ticam1Lps2/Lps2;Irak4otiose/otiose) produced normal levels of NP-specific IgM on day 4.5 after immunization (Fig. 1A). In contrast, Tmem173gt/gt mice and Mb21d1−/− mice, deficient in the cytosolic DNA-sensing pathway components stimulator of interferon gene (STING) and cGMP-AMP synthase (cGAS), respectively, exhibited suboptimal IgM responses to NP-Ficoll on day 4.5 and for up to 30 days after immunization (Fig. 1A and fig. S1). Mice lacking mitochondrial antiviral signaling protein (MAVS), an adaptor for the cytoplasmic RNA-sensing RIG-I–like helicases, also produced diminished amounts of NP-specific IgM (Fig. 1A and fig. S1). Antibody responses to the TI-1 antigen NP-LPS (LPS, lipopolysaccharide) (Fig. 1B), and the TD antigen β-galactosidase (β-gal) encoded by a nonreplicating recombinant Semliki Forest virus (rSFV) vector (3) (Fig. 1C), were normal in STING-, cGAS-, and MAVS-deficient mice.

Fig. 1 Cytosolic DNA and RNA sensing pathways are essential for induction of the TI-2 antibody response.

(A) Serum NP-specific IgM on day 4.5 after immunization with NP-Ficoll. (B) Serum NP-specific IgM on day 4.5 after immunization with NP-LPS. (C) Serum β-gal–specific IgG on day 14.5 after immunization with rSFV-encoded β-gal. (D) Serum NP-specific IgM on day 4.5 after immunization of Rag2−/− mice adoptively transferred 1 day before immunization with splenic and peritoneal B cells from donor mice of the indicated genotypes. Data points represent individual mice. P values were determined by one-way analysis of variance (ANOVA) and post hoc Tukey test; in (B) and (C), no significant difference was found between any mutant genotype and C57BL/6J. Results are representative of two or three independent experiments.

We evaluated marginal zone (MZ) and B-1 B cell populations in STING-, cGAS-, and MAVS-deficient mice and found no deficiencies in frequencies or numbers (fig. S2 and supplementary text), except in the NP-specific populations after NP-Ficoll immunization (fig. S3). Also, NP-Ficoll capture by MZ B cells and MZ macrophages was normal in the mutant mice (fig. S4).

We performed adoptive transfer of C57BL/6J, STING-, cGAS-, or MAVS-deficient splenic and peritoneal B cells into Rag2−/− mice and immunized recipient mice with NP-Ficoll one day after transfer. Despite similar reconstitution of the B cell compartment by donor cells (fig. S5), mice that received STING-, cGAS-, or MAVS-deficient B cells produced diminished amounts of NP-specific IgM on day 4.5 after immunization, as compared to mice that received C57BL/6J B cells (Fig. 1D). These data demonstrate that B cell–intrinsic MAVS and cGAS-STING signaling are necessary for antibody responses to TI-2 immunization.

B cell activation by cGAMP

The DNA sensor cGAS binds to cytosolic DNA and catalyzes the synthesis of cGMP-AMP (cGAMP), a cyclic dinucleotide that binds and activates STING, leading to type I interferon production (4). We found that the presence of DNA in the cytoplasm was sufficient to activate C57BL/6J, but not STING-deficient splenic B cells in vitro (Fig. 2A, fig. S6, and supplementary text). After NP-Ficoll immunization of C57BL/6J mice, cGAMP levels were elevated for at least 10 days in NP-specific B cells relative to levels in non–NP-specific or naïve B cells (Fig. 2, B and C). In vitro cGAMP treatment activated B cells from C57BL/6J but not STING-deficient mice (Fig. 2, D and E), whereas in vivo cGAMP treatment partially rescued NP-specific IgM levels in the serum of cGAS-deficient mice immunized with NP-Ficoll together with cGAMP (Fig. 2F). Thus, cytoplasmic DNA and cGAMP are sufficient to activate B cells in a STING-dependent manner. Human B cells were also activated by cGAMP treatment in vitro (Fig. 2G).

Fig. 2 cGAMP is elevated within antigen-specific B cells after TI-2 immunization and is sufficient to drive B cell activation in vitro and in vivo.

(A) CD86, MHC class II, and CD25 expression, and BrdU incorporation by GFP+ or GFP(GFP, green fluorescent protein) splenic CD19+ B cells, 36 hours after transfection with a GFP expression plasmid. N = 3 C57BL/6J mice and 3 Stinggt/gt mice. (B) cGAMP level measured by liquid chromatography–tandem mass spectrometry (LC-MS/MS) in 2 × 105 NP-specific or non–NP-specific splenic CD19+ B cells from C57BL/6J mice on day 4.5 after immunization with NP-Ficoll (N = 3) or in naïve mice (N = 3). Upper panels, chromatograms of cGAMP. Lower panel, cGAMP abundance normalized to cell number for each sample. (C) Time course of cGAMP levels in NP-specific or non–NP-specific B cells from C57BL/6J mice (N = 3) immunized with NP-Ficoll. (D and E) CD86 expression (D) or the percentage of BrdU+ splenic B cells (E) after treatment with cGAMP or vehicle for 2 days in vitro. N = 3 C57BL/6J mice and 3 Stinggt/gt mice. (F) Serum NP-specific IgM on day 5 after immunization with NP-Ficoll plus cGAMP or NP-Ficoll plus vehicle. (G) HLA-DR or CD69 expression by human B cells isolated from healthy donor peripheral blood and treated with cGAMP or vehicle for 2 days in vitro. MFI, mean fluorescence intensity. Data points represent individual mice or humans [(B), (F), and (G)]. P values were determined by one-way [(A), (B), and (D) to (F)] or two-way ANOVA and post hoc Tukey test (C) or Student’s t test (G). Results are representative of two or three independent experiments.

These findings implicate cytosolic DNA sensed by the cGAS-STING pathway in B cell activation by TI-2 antigens. The impaired antibody response of MAVS-deficient mice immunized with NP-Ficoll (Fig. 1A) suggested that cytosolic RNA may also activate B cells.

ERV transcription and reverse transcription in TI-2 antigen–specific B cells

No exogenous nucleic acid species are known to be introduced into B cells upon BCR engagement by TI-2 antigens, suggesting that endogenous nucleic acids are responsible for activation of the cGAS-STING and MAVS pathways. We hypothesized that RNA and reverse-transcribed DNA derived from endogenous retroviruses (ERVs) or retrotransposons might gain access to the B cell cytoplasm upon BCR signaling. The retroviral life cycle entails the transient localization of RNA and, via reverse transcription, DNA within the cytoplasm (5). Both cGAS-STING (6) and RIG-I–MAVS pathways (7) have been implicated in the sensing of HIV-1 and other retroviruses. We therefore investigated whether ERVs could be induced by TI-2 immunization.

We measured transcript levels of individual ERVs in B cells after immunization of C57BL/6J mice with NP-Ficoll; DNA contamination of the RNA samples was excluded (fig. S7A). Compared to non–NP-specific B cells on day 4.5 after immunization, NP-specific B cells displayed increased expression of multiple ERV mRNAs (Fig. 3A). We found that total RNA isolated from NP-specific B cells was significantly more potent than total RNA from non–NP-specific B cells in activating cytokine and chemokine gene expression in bone marrow–derived macrophages (BMDMs) (fig. S7, B to D). Gene expression induced by NP-specific B cell total RNA was partially dependent on MAVS (fig. S7C); ribonuclease treatment abrogated the effect (fig. S7D). To test the idea that a MAVS-dependent RNA sensing pathway could be activated by ERV RNAs in NP-specific B cells, we used RIG-I immunoprecipitation followed by reverse transcription polymerase chain reaction (RT-PCR) analysis and found that all tested ERV mRNA sequences bound to RIG-I molecules in NP-specific B cells from immunized mice; greater amounts of ERV mRNA were immunoprecipitated with RIG-I from NP-specific B cells than from non–NP-specific B cells (Fig. 3B). We also verified the selective binding of RIG-I to ERV mRNA versus cellular mRNA (fig. S7E).

Fig. 3 TI-2 antigen immunization induces expression of ERV mRNA and cDNA that is detected by cytosolic sensors in antigen-specific B cells.

Splenic NP-specific or non–NP-specific CD19+ B cells were collected from C57BL/6J mice (A, C, and D) or IghB1-8+ transgenic mice (B) 4.5 days after immunization with NP-Ficoll (N = 3 mice per experiment). (A) Transcript levels of the indicated ERVs measured by reverse transcription quantitative PCR (RT-qPCR) of mRNA isolated from NP-specific or non–NP-specific B cells. Data were normalized to glyceraldehyde phosphate dehydrogenase (GAPDH) mRNA levels in the same cells. Because of copy number differences, the magnitude of up-regulation of different ERVs cannot be directly compared in this experiment. (B) IghB1-8+ transgenic mice express a recombined variable region derived from an NP-binding antibody in place of the endogenous 3′ Igh-D element (DQ52) and the Igh-J elements. RT-qPCR of the indicated ERV mRNAs immunoprecipitated with RIG-I from NP-specific or non–NP-specific B cells. Data were normalized to the level of GAPDH mRNA bound to RIG-I in the same samples, which represents a nonspecific interaction equivalent in NP-specific and non–NP-specific cells as shown in fig. S7E. (C) qPCR of ERV DNA in the cytoplasmic fraction of NP-specific or non–NP-specific B cells. Data were normalized to GAPDH intronic DNA levels in the same cells. Endogenous ecotropic murine leukemia virus (eMLV) and mouse mammary tumor virus (MMTV) were amplified with primers targeting spliced cDNAs; these species probably represent a minority of the cytoplasmic eMLV and MMTV cDNAs and thus may not precisely reflect total eMLV and MMTV cDNA levels. (D) RT activity in the indicated B cells from C57BL/6J mice. (E and F) Splenic CD19+ B cells from mice of the indicated genotypes were treated with NP-Ficoll plus RT inhibitors (AZT, NVP, and ddI) or NP-Ficoll plus vehicle for 2 days, and CD86 (E) or GL7 expression (F) was measured in NP-specific B cells. N = 3 mice for each genotype. (G) Serum NP-specific IgM on day 4.5 after NP-Ficoll immunization of mice pretreated for 3 days with RT inhibitors (AZT and NVP) or vehicle. RT inhibitor treatment continued after immunization until measurement of serum IgM. (H) Serum NP-specific IgM on day 4.5 after immunization with NP-Ficoll. Data points represent individual mice [(D), (G), and (H)]. P values were determined by Student’s t test [(A) to (C)] or one-way ANOVA and post hoc Tukey test [(D) to (H)]. n.d., not detected. Results are representative of two or three independent experiments.

Reverse transcription of retroviral RNA occurs in the cytoplasm and produces cDNA that can be detected by cGAS (6). Measurement of cytoplasmic ERV DNA (fig. S8A) revealed elevated levels in NP-specific B cells as compared to non–NP-specific B cells from immunized C57BL/6J mice (Fig. 3C). Consistent with this finding, the activity of reverse transcriptase (RT) was up-regulated in NP-specific relative to non–NP-specific or naïve B cells (Fig. 3D). We used inhibitors of RT to determine whether reverse transcription is necessary for B cell activation and antibody production in mice immunized with NP-Ficoll. B cells were cultured in vitro and treated with NP-Ficoll and the RT inhibitors azidothymidine (AZT), nevirapine (NVP), and didanosine (ddI). RT inhibition significantly reduced CD86 and GL7 expression in C57BL/6J and MAVS-deficient NP-specific B cells relative to NP-specific B cells without RT inhibition (Fig. 3, E and F).

In vivo, C57BL/6J mice pretreated with AZT and NVP followed by immunization with NP-Ficoll showed varying but significant reductions in ERV DNA levels in the cytoplasm of NP-specific B cells compared to immunized mice treated with vehicle (fig. S8B), which is consistent with the known differential susceptibility of RT enzymes to distinct RT inhibitors (8, 9). C57BL/6J and MAVS-deficient mice treated with RT inhibitors exhibited impaired NP-specific IgM production relative to immunized vehicle-treated mice (Fig. 3G). However, no additive impairment of activation marker expression on NP-specific B cells in vitro, or of NP-specific IgM production in vivo, resulted from RT inhibition in STING-deficient B cells or mice (Fig. 3, E to G). Consistent with these findings, double knockout of MAVS and cGAS reduced NP-specific IgM production to nearly undetectable levels in the serum of immunized mice (Fig. 3H). We also immunized double knockout mice deficient in MAVS and STING and noted that, although reduced as compared to mice lacking either protein alone, residual NP-specific IgM production still occurred, suggesting that cGAS may have functions independent of cGAMP production or that alternate cGAMP receptor(s) remain to be identified (fig. S9). These data suggest that TI-2 immunization activates transcription and reverse transcription to produce ERV mRNAs and cDNAs. These nucleic acids probably then engage the RIG-I–MAVS and cGAS-STING pathways, leading to signaling necessary for B cell activation and antigen-specific antibody production. We note that the potential involvement in TI-2 antibody responses of LINE-1 elements, which are also susceptible to RT inhibition, remains to be tested.

NF-κB is required for ERV induction and is activated by BCR and MAVS signaling

TI-2 antigens engage the BCR and trigger signaling events, including the activation of Src kinases, the Tec family kinase Btk, and subsequently NF-κB (10), which is known to activate the transcription of exogenous retroviruses in other cells (11, 12). Because mutations affecting the BCR reduce or eliminate B cells, we investigated the involvement of BCR signaling in the induction of ERVs by measuring ERV mRNA levels in response to BCR engagement by anti-IgM in vitro. Anti-IgM treatment increased activation marker expression (Fig. 4A), caused strong up-regulation of multiple ERV mRNAs (Fig. 4B), and elevated cGAMP (Fig. 4C) in C57BL/6J B cells. Human B cells also up-regulated ERV mRNAs upon treatment with anti-IgM in vitro (Fig. 4D). These data indicate that BCR ligation is sufficient to induce ERV transcription in mouse and human B cells, as well as activation of cGAS in mouse B cells. MAVS- and cGAS-deficient B cells failed to fully up-regulate activation markers in response to anti-IgM (Fig. 4A); these findings are consistent with MAVS and cGAS activation induced by BCR signaling. Downstream from the BCR, Btk activation is a defining feature of the signaling induced by TI-2 antigens (1). Btk-deficient xid mice failed to produce NP-specific IgM on day 4.5 after immunization with NP-Ficoll (Fig. 4E), which is consistent with previously reported data (13). NP-specific B cells from immunized xid mice expressed similar levels of ERV mRNAs as non–P-specific B cells from either xid mice or from control mice (Fig. 4F). In human B cells in vitro, the BTK inhibitor Ibrutinib blocked both the expression of activation markers (Fig. 4G) and the elevation of ERV transcription induced by anti-IgM (Fig. 4D). These findings demonstrate that Btk is necessary for the induction of ERV mRNAs in mouse B cells by TI-2 immunization, and in human B cells by BCR ligation.

Fig. 4 cGAMP elevation and ERV transcription are induced by BCR signaling.

(A to C) Mouse splenic B cells were cultured in vitro and stimulated with anti-IgM or vehicle for 22 hours. (A) CD86 and MHC class II expression. (B) Transcript levels of the indicated ERVs measured by RT-qPCR of isolated mRNA. Data were normalized to GAPDH mRNA levels in the same cells. N = 3 mice per genotype. (C) Chromatograms of cGAMP in C57BL/6J B cells measured by LC-MS/MS. N = 3 mice. (D) Human B cells from healthy donor peripheral blood (N = 3 individuals) were cultured in vitro and stimulated with anti-IgM, anti-IgM + Ibrutinib, or vehicle for 48 hours. Transcript levels of the indicated human ERVs measured by RT-qPCR of isolated mRNA. (E) Serum NP-specific IgM on day 4.5 after immunization with NP-Ficoll. (F) Transcript levels of the indicated ERVs measured by RT-qPCR of mRNA isolated from NP-specific or non–NP-specific B cells. Data were normalized to GAPDH mRNA levels in the same cells. N = 3 mice per genotype. (G) Activation marker expression by human B cells treated as in (D), or with 0.06 μM cGAMP for 48 hours. Data points represent individual mice or humans [(A), (E), and (G)]. P values were determined by Student’s t test[ (B) and (D) to (F)] or one-way ANOVA and post hoc Tukey test [(A) and (G)]. Results are representative of two independent experiments.

Data from several experiments supported a requirement for NF-κB in the antibody response to TI-2 immunization, and specifically in the induction of ERV mRNAs. First, NF-κB–dependent cytokines were up-regulated in NP-specific B cells from C57BL/6J mice immunized with NP-Ficoll, whereas non–NP-specific B cells showed cytokine production similar to that of naïve B cells (Fig. 5A). Second, mutation of NEMO (IkbkgpanR2), a component of the IKK complex required for NF-κB activation, abrogated the NP-specific IgM response of C57BL/6J mice to NP-Ficoll immunization (Fig. 1A). Third, NP-specific B cells from NEMO-deficient mice expressed ERV mRNA levels similar to those of non–NP-specific B cells (Fig. 5B).

Fig. 5 NF-κB is required for ERV induction and is activated by BCR and MAVS signaling.

(A and B) Splenic NP-specific or non–NP-specific CD19+ B cells were collected from mice 4.5 days after immunization with NP-Ficoll. (A) Cytokine expression in the indicated B cells from C57BL/6J mice. (B) Transcript levels of the indicated ERVs measured by RT-qPCR of mRNA isolated from NP-specific or non–NP-specific B cells. N = 3 mice. No significant difference was found between NP+ and NP cells for any ERV tested. (C and D) Splenic B cells were cultured in vitro and stimulated with anti-IgM or vehicle for 22 hours. (C) Levels of phospho-p65 (left) and phospho-p105 (right) are shown. N = 3 mice. (D) Levels of p65 in the nuclear fraction of cells. (E) Levels of phospho-p105 in splenic NP-specific and non–NP-specific CD19+ B cells or naïve B cells on day 6 after NP-Ficoll immunization. N = 3 mice per genotype. Data points represent individual mice [(A) and (D)]. P values were determined by Student’s t test (B) or one-way ANOVA and post hoc Tukey test [(A) and (C) to (E)]. Results are representative of two independent experiments.

Signaling from the BCR, RIG-I–MAVS, and cGAS-STING pathways all activate NF-κB (10, 14, 15), and we sought to identify which pathway(s) drive NF-κB activation needed for the TI-2 antibody response. We found that anti-IgM failed to induce ERV mRNAs in NEMO-deficient B cells in vitro (Fig. 4B), suggesting that BCR signaling induces ERV mRNAs via NF-κB. As we have demonstrated, ERV transcripts trigger a second, sustained wave of signaling via the RIG-I–MAVS and cGAS-cGAMP-STING pathways that may also require NF-κB to activate B cells and stimulate antibody production. We evaluated NF-κB activation by measuring levels of phosphorylated p65 and p105. In response to anti-IgM, C57BL/6J, MAVS-, and cGAS-deficient B cells displayed elevated levels of phospho-p65 and phospho-p105 as compared to vehicle-treated cells; however, MAVS-deficient cells showed a diminished level of NF-κB activation relative to C57BL/6J cells (Fig. 5C). Analysis of p65 nuclear translocation corroborated these findings (Fig. 5D). Similarly, NP-specific B cells from immunized C57BL/6J, MAVS-, and cGAS-deficient mice displayed elevated phospho-p105 relative to non–NP-specific B cells; MAVS deficiency prevented NF-κB activation to the extent observed in C57BL/6J NP-specific B cells (Fig. 5E). These data suggest that MAVS signaling but not cGAS-STING signaling activates NF-κB in response to ERV nucleic acids induced by BCR engagement. The finding that NP-specific B cell RNA from immunized mice transcriptionally activated NF-κB target genes, including those encoding TNFα, IL-6, MCP-1, and COX-2, in a MAVS-dependent manner (fig. S7, C and D) further supports this conclusion.

cGAS-STING and MAVS signaling in antibody responses to Streptococcus pneumoniae capsular polysaccharides

We immunized cGAS-, STING-, MAVS-, or cGAS/MAVS-deficient mice with Streptococcus pneumoniae capsular polysaccharides (PSs) 1 and 3 or with the commercial vaccine Pneumovax 23 (PPV-23), containing 23 S. pneumoniae polysaccharide antigens. By day 4.5 after immunization, C57BL/6J mice mounted a PS1- and PS3-specific IgM response, which was significantly diminished in cGAS- or MAVS-deficient mice (Fig. 6A). Mice lacking both cGAS and MAVS displayed a further reduction in PS1- and PS-3–specific IgM to nearly undetectable levels (Fig. 6A). Similarly, upon immunization with Pneumovax 23, cGAS- or STING-deficient mice produced diminished amounts of PPV-23–specific IgM and IgG as compared to C57BL/6J mice (Fig. 6, B and C). Thus, the requirement for cGAS-STING and RIG-I–MAVS pathways, likely due to their detection of ERV nucleic acids, extends to antibody responses against TI-2 antigens of a clinically important human pathogen.

Fig. 6 TI-2 antibody responses to S. pneumoniae PS1 and PS3 and to the commercial vaccine Pneumovax 23 require cytosolic DNA and RNA sensing pathways.

(A) Serum PS1- and PS3-specific IgM on day 4.5 after immunization with S. pneumoniae PS1 and PS3. (B and C) Serum PPV-23–specific IgM (B) and IgG (C) on day 5.5 after immunization with Pneumovax 23. Data points represent individual mice. P values were determined by one-way ANOVA and post hoc Tukey test (A) or Student’s t test [(B) and (C)]. Results are representative of two or three independent experiments.

Discussion

With respect to ERVs, we hypothesize that the following events occur in B cells after TI-2 immunization: (i) ERV proviruses are transcribed, and the resulting unspliced and spliced RNAs are transported to the cytoplasm; (ii) ERV mRNAs are translated to produce RT(s) (Fig. 3D), Env glycoprotein(s) (fig. S10A), and possibly other ERV proteins; (iii) ERV mRNAs are reverse-transcribed to produce cDNAs by one or more of the RT species produced in step (ii); and (iv) ERV mRNAs and cDNAs present in the cytoplasm activate RIG-I and cGAS, respectively (fig. S11). Although retroviruses including HIV-1 and MLV typically proteolytically process RT from the Gag-Pol polyprotein during or after virus budding, RT has also been reported to be active in the precursor form (16, 17); similar ERV-derived precursors may provide the RT activity we measured in NP-specific B cell lysates. Overall, our data support the hypothesis that ERV nucleic acids induced in the TI-2 antigen–activated B cell are sufficient to trigger cytosolic innate sensors, leading to subsequent antibody production by the same cell (fig. S11). However, despite multiple experiments supporting the involvement of ERV nucleic acids, it remains possible that their presence and activity are incidental to other component(s) essential for the activation of MAVS, cGAS, STING, and ultimately TI-2 antibody responses. Determining whether ERV nucleic acids represent a necessary component of the antibody response will require comprehensive inactivation of ERV sequences across the genome.

The appropriation of ERV genes for host immunity has often been observed in defense against exogenous retroviruses, as exemplified by Fv1 (18) and Fv4 (19). Whereas the host recruits a single ERV-derived gene to counteract one particular step in the life cycle of an exogenous retrovirus, however, it activates a wide range of ERVs in antigen-specific B cells that then serve to trigger intracellular nucleic acid sensors necessary for the TI-2 antibody response. The fate of these ERV nucleic acids remains unknown. We found in NP-specific B cells from immunized mice that the expression of endogenous murine leukemia virus–specific Env glycoprotein was enhanced (fig. S10A), as was RT activity detected in the supernatants (fig. S10B). Why ERV proteins are produced and how they might come to exist outside the cell during the TI-2 antibody response are open questions.

Recent reports suggest that aberrant cellular accumulation of endogenous retroelement cDNA and its recognition by cytosolic sensors lead to autoimmune disease in mice (20, 21). An association of human autoimmune diseases with ERVs has also been postulated (22). Such diseases may be hypothesized to result from dysregulation of adaptive rather than strictly innate immune activation, in that retroviruses, MAVS, cGAS, and STING serve as components of the normal TI-2 B cell signaling apparatus.

Supplementary Materials

www.sciencemag.org/content/346/6216/1486/suppl/DC1

Materials and Methods

Supplementary Text

Figs. S1 to S11

Table S1

References (2332)

References and Notes

  1. Materials and methods are available as supplementary materials on Science Online.
  2. Acknowledgments: We thank L. Evans and F. Malik for the generous gift of the 83A25 antibody; P. Jurek for expert assistance in preparing the figures; and the Children’s Medical Center Research Institute Flow Cytometry Facility at the University of Texas Southwestern Medical Center for assistance with cell sorting. The data presented in this paper are tabulated in the main paper and in the supplementary materials. This work was supported by generous donations from the Lyda Hill Foundation and the Kent and JoAnn Foster Family Foundation; and by NIH grants P01 AI070167 and U19 AI100627 (to B.B.), R01 AI093967 (to Z.J.C.), and R01 CA157996 (to R.D.).
View Abstract

Stay Connected to Science

Navigate This Article