A novel RAG1 mutation reveals a critical in vivo role for HMGB1/2 during V(D)J recombination

Abstract

The Recombination Activating Genes, RAG1 and RAG2, are essential for V(D)J recombination and adaptive immunity. Mutations in these genes often cause immunodeficiency, the severity of which reflects the importance of the altered residue or residues during recombination. Here, we describe a novel RAG1 mutation that causes immunodeficiency in an unexpected way: The mutated protein severely disrupts binding of the accessory protein, HMGB1. Although HMGB1 enhances RAG cutting in vitro, its role in vivo was controversial. We show here that reduced HMGB1 binding by the mutant protein dramatically reduces RAG cutting in vitro and almost completely eliminates recombination in vivo. The RAG1 mutation, R401W, places a bulky tryptophan opposite the binding site for HMG Box A at both 12- and 23-spacer recombination signal sequences, disrupting stable binding of HMGB1. Replacement of R401W with leucine and then lysine progressively restores HMGB1 binding, correlating with increased RAG cutting and recombination in vivo. We show further that knockdown of HMGB1 significantly reduces recombination by wild-type RAG1, whereas its re-addition restores recombination with wild-type, but not the mutant, RAG1 protein. Together, these data provide compelling evidence that HMGB1 plays a critical role during V(D)J recombination in vivo.

Graphical abstract

Figure 1.

RAG1 mutations affect V(D)J recombination in vivo. (A) Schematic of RAG1 showing the main core RAG1 subdomains. The positions of the mutations in the patient are shown with the equivalent mouse amino acids in brackets beneath. (B) Plasmid pJH299 used to quantify V(D)J recombination in vivo. Here, 12- and 23-RSSs are indicated as triangles. After inversional recombination, primers 1+2 and 3+4 are used in a qPCR nested assay, with a hydrolysis probe across the RSS junction. Total plasmid amounts are measured using primers 5 and 6. (C) Increasing amounts of WT RAG1 expression plasmid were transfected into NIH3T3 cells, and the levels of recombination were determined. A direct correlation with the amount of RAG1 activity is observed. n = 3; error bars show standard error of the mean (SEM). (D) Western blot showing expression levels of the various RAG1 proteins in transfected NIH3T3 cells; a β-galactosidase expression vector was co-transfected as a loading control. (E) Recombination levels using WT and mutated RAG1 proteins, relative to WT RAG1. Recombination was normalized according to the total amount of recombination plasmid recovered; similar levels of RAG1 protein expression were verified by western blotting (D). n = 3; error bars show SEM. CTD, carboxy-terminal domain; DDBD, dimerization and DNA binding domain; NBD, nonamer binding domain; preR, pre-RNaseH domain; ZBD, zinc binding domain.

Figure 2.

RAG1 mutations affect cutting and binding in vitro. (A) Cutting at a 12-RSS (left) or 23-RSS (right) is shown. An oligonucleotide carrying a consensus 12- or 23-RSS, as indicated, was incubated with equivalent amounts of core RAG2 (cRAG2) plus the core RAG1 (cRAG1) proteins shown (C) in the presence of HMGB1, with a 10-fold excess of unlabeled 23- or 12- partner RSS. For cutting reactions on the far right of each gel, equal amounts of R401W and R504Q were mixed to give the same total amount of RAG1 as the other lanes. The percentage cutting is indicated beneath the gels; an asterisk indicates the labeled oligonucleotide. (B) Binding to a labeled 12-RSS (left) and 23-RSS (right). Binding reactions were performed with cRAG2 and the cRAG1 proteins indicated. Complexes SC1 and SC2, as well as HSC1 and HSC2, are shown. (C) The levels of purified RAG proteins are equivalent. Purified, maltose binding protein-tagged cRAG1 and cRAG2 proteins were separated by electrophoresis and the gel stained with Coomassie blue. The various RAG1 proteins are indicated above each lane. (D) Comparison of R401W and WT RAG1 binding to a 12-RSS. RAG binding reactions were performed in the presence or absence of HMGB1, as indicated. Whereas R401W forms equivalent complexes to WT with the 12-RSS, in the absence of HMGB1, R401W does not form HSC1.

Figure 3.

HMGB1 binding to the RAG/DNA complex. HMGB1 binding was determined from the published cryo-electron microscopy and X-ray crystal structures of mouse RAG/RSS complexes.^, Blue and green cartoons represent individual RAG1 monomers. (Left) HMGB1 Box A from the 23-RSS structure was superimposed over the density that likely corresponds to HMGB1 Box A at the 12-RSS, denoted by the asterisk. This falls directly opposite R401 of mouse RAG1, indicated in red. R401 contacts DNA in the 12-RSS spacer, close to the RSS nonamer. (Right) HMGB1 Box B binding is observed in the middle of the 23-RSS spacer, whereas an additional area of HMGB1 binding (Box A) lies opposite R401 (in red) at the 23-RSS. Data from PDB 6CG0.

Figure 4.

Amino acid substitutions progressively restore HMGB1 binding and RAG1 activity. (A) HMGB1 binds less well to R401W/12-RSS complexes. WT or R401W mutant RAG1 complexes were formed with a labeled 12-RSS in the presence or absence of His-tagged HMGB1. A 23-RSS oligonucleotide partner was present in all cases. Supershifted complexes formed on addition of an anti-His tag antibody are indicated. The mobilities of SC1, SC2, HSC1, and HSC2 are shown. (B) Increasing amounts of HMGB1 do not restore cutting with R401W. RAG cutting reactions were performed with WT RAG1 or R401W in the presence of increasing amounts of HMGB1 (0, 25, 50, 100, 200, 400, 800 nmoles). A 23-RSS substrate was used, as HMGB1 has the greatest effect at this type of RSS. It is notable that in the absence of HMGB1, cutting by WT RAG1 is not detectable (lane 2). (C) R401W has catalytic activity. RAG cutting reactions were performed using a single labeled 12-RSS in the presence of manganese. (D) HSC1 and HSC2 complexes become increasingly like WT RAG1 complexes with R401L and R401K. RAG binding to a labeled 12- or 23-RSS in the presence of an unlabeled partner RSS and HMGB1, as indicated. The gel was run for longer than Figure 2B to better separate the complexes; a cropped image is shown to highlight the mobility differences. (E) RAG cutting with R401W, R401L, and R401K. RAG cutting was performed with a labeled 12-RSS or 23-RSS, indicated by the asterisk, in the presence of unlabeled partner and the RAG1 proteins indicated, with or without HMGB1. The percentage cutting is given beneath the gel. (F) R401K recombination activity is close to WT RAG1 levels in vivo. Recombination was monitored using pJH299 and equivalent levels of RAG1 proteins (Figure 1B,D). Recombination levels are shown relative to WT RAG1. n = 3; error bars show SEM.

Figure 5.

Knockdown of HMGB1 significantly reduces V(D)J recombination in vivo. (A) Schematic of the steps used in CRISPR/Cas9-mediated knockdown of HMGB1. Transduction with lentiviruses and antibiotic selection steps are shown. (B) Representative western blots showing the levels of HMGB1 and full-length (fl) and C-terminal deletion (ΔC) HMGB2 relative to a β-tubulin control. Antibodies that specifically recognize HMGB2ΔC are unavailable, as the epitope lies at the C terminus. However, the core regions of HMGB1 and HMGB2 are 85.5% identical, and the anti-HMGB1 antibody cross-reacts with HMGB2 (indicated by the asterisk). (C) V(D)J recombination levels with the RAG1 proteins shown in WT NIH3T3 cells (blue bars), when the KRAB repressor as introduced in the absence of guide RNAs (purple bars), when HMGB1 has been knocked down (red bars), or when HMGB1 (dark green bars) or HMGB2 full-length (light green bars) or HMGB2ΔC (turquoise bars) have been reintroduced. n = 3; error bars show SEM. The changes in recombination activity are statistically significant. Student t test; **P = .01; ***P = .001. The endogenous HMGB2 in NIH3T3 cells could account for residual recombination activity after HMGB1 knockdown. ns, nonspecific.

Figure 6.

Schematic showing altered HMG binding to WT RAG1 and R401W. The DNA strands are shown with the heptamer, spacer, and nonamer indicated. A and B represent the likely positions of HMG Box A and B, respectively, and R1 and R2 represent RAG1 and RAG2. The bend angles within each RSS are shown.