Naturally occurring short splice variant of CYLD positively regulates dendritic cell function

CYLD is a tumor suppressor gene that is mutated in familial cylindromatosis, an autosomal dominant predisposition to tumors of skin appendages.¹  CYLD removes lysine-63 polyubiquitin chains from distinct members of the NF-κB pathway^2-4  and mutations in CYLD dysregulate NF-κB activity

Complete deletion of CYLD in mice (CYLD^ko) renders them susceptible to skin tumors,⁵  but the CYLD^ko mice do not display alterations of the immune system, as we could show for B-⁶  and T-cell development (A.W., N.H., S. Reissig, manuscript in preparation). Other knockout mouse models of CYLD point to a role for flCYLD in immunity, including hyperinduction of IFNα in virus-infected dendritic cells (DCs)⁷  as well as protection from infections in its absence.⁸ 

However, mice exclusively expressing the naturally occurring short splice variant of CYLD, sCYLD, are characterized by lymphomegaly, splenomegaly, and dramatic increases in B-cell numbers⁶  caused by aberrant NF-κB signaling. Because this pathway is important for the function of DCs,⁹  which orchestrate innate and adaptive immune responses, we investigated the effect of sCYLD overexpression on DC function. We found CYLD^ex7/8 DCs to be hyperresponsive to LPS, capable of inducing superior T-cell expansion on DC immunization in vivo, and able to suppress tolerance induced by α-DEC-205:OVA administration. As a molecular basis for this phenotype, we identified increased nuclear Bcl-3, p50, and p65 in resting and stimulated CYLD^ex7/8 bone marrow–derived DCs (BMDCs).

C57BL/6 wild-type (WT), CYLD^ko,⁵  and CYLD^ex7/86  mice 6 to 8 weeks old were used as recipients and for BMDC generation as previously described.¹⁰  St42 TCR Tg mice have been previously described.¹¹  OT-I mice¹²  were obtained from Christian Kurts (Institute for Molecular Medicine & Experimental Immunology, Bonn, Germany). Approval for these studies was obtained from the review board of the Federal State of Rhineland-Palatinate, Germany. All animal experiments were in accordance with the guidelines of the Central Animal Facility Institution of the University of Mainz.

FACSCanto cytometer and FlowJo software (Tree Star, Ashland, OR) using α-CD8 and α-CD90.1/CD45.1 (eBioscience, San Diego, CA) was used. α–IL-6, IL-10, and TNF-α antibodies from Cytometric Bead Array Flex system Becton Dickinson (Franklin Lakes, NJ) were used and analyzed with FCAP Array Software (BD Biosciences, Mountain View, CA).

Recipient mice were given 10⁶ adoptively transferred OT-I cells, immunized with 20 μg DEC-205:OVA, and boosted with 50 μg OVA protein (Sigma-Aldrich, St Louis, MO) as previously described.¹³ 

Western blot analysis was performed as previously described.⁸  Nuclear and cytoplasmic fractions were prepared according to standard procedures.^14,15  NF-κB Consensus Oligonucleotide was purchased from Promega (Madison, WI). Western blots and electrophoretic mobility shift assay (EMSA) were performed with antibodies α-p50 (Assay Designs, Ann Arbor, MI), α-Bcl-3 (Santa Cruz Biotechnology, Santa Cruz, CA), α–β-actin (Sigma-Aldrich), α-p65, and α-HDAC1 (Cell Signaling Technology, Danvers, MA).

Fibroblasts were transfected with sCYLD plasmid, using the Nucleofector Kit (Amaxa:VPD-1005; Amaxa Biosystems, Gaithersburg, MD). The NF-κB–Reporter plasmid was provided by Ralf Marienfeld¹⁶  (Institute for Molecular Pathology, Wurzburg, Germany) and cotransfected with pRL-TK (Promega) into BMDCs with the Nucleofector Kit (VPA-1011; Amaxa). Luciferase reporter activity was analyzed as previously described.¹⁷ 

mRNA detection was performed as previously described¹⁸  with Bcl-3 forward, TGCCCATTTACTCTACCCCGACGA; Bcl-3 reverse, CAGCGATGTGGAGAGGCGTGTC; sCYLD forward, CTATTGGCAACTGGGATGGAAGG; and sCYLD reverse, CACTTAAATAGCCCCCAAATGCTTC.

Values represent means plus or minus SEMs analyzed with the t test or 1-way ANOVA (analysis of variance).

To analyze the role of CYLD and its naturally occurring short splice variant in DC function, we generated BMDCs and observed that immature CD11c⁺ BMDCs from CYLD^ex7/8 mice express higher levels of CD86, MHC II, and OX40L compared with WT or CYLD^ko mice BMDCs (Figure 1A). sCYLD increase could be observed in WT DCs on stimulation (Figure S1A, available on the Blood website; see the Supplemental Materials link at the top of the online article), suggesting a role for sCYLD in DC function. However, neither WT mice nor heterozygous CYLD^ex7/8 (WT/CYLD^ex7/8) mice display a hyperactive BMDC phenotype (Figure S1B), indicating that a certain threshold level of sCYLD expression is required for the hyperactive phenotype.

On LPS stimulation, CYLD^ex7/8 BMDCs further amplify cell surface markers (Figure 1A) and secrete increased levels of the proinflammatory cytokines IL-6 and TNF-α but less of the anti-inflammatory cytokine IL-10 compared with WT BMDCs (Figure 1B).

To understand the in vivo consequences of CYLD^ex7/8 BMDC hyperactivity, BMDCs were stimulated with LPS, loaded with SGP peptide, and injected into WT mice, which received an adoptive transfer of SGP peptide-specific St42 TCR transgenic T cells. We observed a significantly higher expansion of the transferred T cells when the stimulation occurred via BMDCs from CYLD^ex7/8 mice, compared with BMDCs prepared from the other mice (Figure 1C). Because of the hyperactive phenotype of immature CYLD^ex7/8 BMDCs, we analyzed the capacity of these mice to induce T-cell tolerance in comparison to WT or CYLD^ko animals. We injected α-DEC-205:OVA (which delivers ovalbumin in a steady state to DEC-205–expressing DCs¹³ ) to recipient mice adoptively transferred with OT-I cells. On challenge with ovalbumin in CFA, the OT-I T-cell population in the spleens of WT and CYLD^ko recipient mice displayed, as expected, a strong reduction in OT-I T-cell numbers, whereas in CYLD^ex7/8 mice the OT-I cells expanded more than 10-fold (Figure 1D). This finding suggests that CYLD^ex7/8 mice lack the capacity for DC-mediated induction of tolerance toward exogenous antigens because of their predisposed high activation status.

To understand the mechanism behind our phenotypic observations, we focused on NF-κB members involved in CYLD-influenced NF-κB signaling. Bcl-3 has recently been shown to interact with both flCYLD and sCYLD and to modulate NF-κB activity.^5,8  Therefore, we investigated Bcl-3 expression and observed a significant increase on both the mRNA and nuclear protein level in CYLD^ex7/8 BMDCs compared with WT and CYLD^ko (Figures 2A,B, S1C). To clarify the effect of sCYLD on Bcl-3 expression, we transfected CYLD^ko fibroblasts with a sCYLD encoding plasmid and observed a 5-fold increase in Bcl-3 expression compared with mock-transfected CYLD^ko cells (Figure 2C).

The enhanced Bcl-3 expression corresponded to higher NF-κB reporter activity in CYLD^ex7/8 BMDCs, which could already be observed in the unstimulated state (Figure 2D). To determine the NF-κB subunits involved, we turned to the well-known binding partner of Bcl-3, p50,^19,20  to assess its role in the observed NF-κB activation. Unstimulated CYLD^ex7/8 BMDCs displayed detectable NF-κB activation driven by p50 as shown by the supershifted complex in the EMSA (Figure 2E), which was not observed in WT or CYLD^ko BMDCs. On stimulation, NF-κB activity increased in all cell types, but the degree of increase was most profound in CYLD^ex7/8 BMDCs. The aforementioned observation of p50-driven NF-κB activation turned our interest to its most common and best-characterized binding partner, p65, which together with p50 forms a heterodimer p50/p65 that is known to activate gene transcription.^21,22  The increased nuclear p65 and p50 protein levels concomitant with decreased cytoplasmic p105 expression in resting and stimulated CYLD^ex7/8 BMDCs compared with WT and CYLD^ko (Figures 2F, S1D) suggest that the increased NF-κB activation in these cells is a result of sCYLD overexpression.

Although resting CYLD^ko BMDCs express higher Bcl-3 mRNA and protein levels compared with WT BMDCs, they do not display a hyperactive phenotype (Figure 2B). Moreover, overexpression of Bcl-3 in WT BMDCs caused only a minimal up-regulation of the surface marker CD86 (Figure S1E), supporting a model in which, only when sCYLD is overexpressed and flCYLD is absent, does the up-regulation of Bcl-3 followed by its nuclear translocation, correspond to a hyperactive phenotype. The interplay between sCYLD and Bcl-3 could further be verified by showing that nuclear Bcl-3 localization occurs exclusively in the presence of sCYLD and the absence of its full-length transcript (Figure S1F).

Taken together we propose a model in which the exclusive overexpression of sCYLD with high nuclear levels of Bcl-3 in BMDCs is accompanied by an increased NF-κB activation, resulting in a hyperactive phenotype. These results propose a stimulatory role for sCYLD and indicate that the relationship between flCYLD and sCYLD needs further investigation to deepen our understanding of deubiquitinating enzymes in NF-κB signaling and their distinct roles in positive and negative regulation.

We thank Drs Hans-Christian Probst, Mathias Krummen, Matthias Klein, Tobias Bopp, as well as Sonja Reißig, for advice with technical applications and helpful scientific discussions, and Kristian Schütze for excellent technical assistance.

This work was supported by the Marie Curie Research Training Network (MC-RTN-CT-2004-512585; C.C.S.), by the Deutsche Forschungsgemeinschaft (grant SFB432; A.W. and H.S.) and (grant SFB/TR 52; A.W.), by the immunology cluster of excellence of the government of the Rhineland Palatinate (H.S. and A.W.), and by funds from the Boehringer Ingelheim Stiftung (A.W.).

Contribution: C.C.S., J.M., and A.K.K. performed experiments; C.C.S. and J.M. created the figures; C.C.S., J.M., C.T., A.W., and H.S. designed the research; C.C.S. and J.M. analyzed and interpreted the data; D.S. provided useful technical expertise; C.C.S., J.M., A.W., N.H., and H.S. wrote the paper; and R.M. and K.M. provided valuable reagents.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Ari Waisman, First Medical Department, Johannes Gutenberg University Mainz, Verfügungsgebäude, D-55131, Mainz, Germany; e-mail: waisman@uni-mainz.de; or Hansjörg Schild, Institute for Immunology, Johannes Gutenberg University, Obere Zahlbacher Str 67, D-55131 Mainz, Germany; e-mail: schild@uni-mainz.de.