Humanized V(D)J-rearranging and TdT-expressing mouse vaccine models with physiological HIV-1 broadly neutralizing antibody precursors

Antibody heavy chain (HC) and light chain (LC) variable region exons are assembled by V(D)J recombination. V(D)J junctional regions encode complementarity-determining-region 3 (CDR3), an antigen-contact region immensely diversified through nontemplated nucleotide additions (“N-regions”) by terminal deoxynucleotidyl transferase (TdT). HIV-1 vaccine strategies seek to elicit human HIV-1 broadly neutralizing antibodies (bnAbs), such as the potent CD4-binding site VRC01-class bnAbs. Mice with primary B cells that express receptors (BCRs) representing bnAb precursors are used as vaccination models. VRC01-class bnAbs uniformly use human HC V_H1-2 and commonly use human LCs Vκ3-20 or Vκ1-33 associated with an exceptionally short 5-amino-acid (5-aa) CDR3. Prior VRC01-class models had nonphysiological precursor levels and/or limited precursor diversity. Here, we describe VRC01-class rearranging mice that generate more physiological primary VRC01-class BCR repertoires via rearrangement of V_H1-2, as well as Vκ1-33 and/or Vκ3-20 in association with diverse CDR3s. Human-like TdT expression in mouse precursor B cells increased LC CDR3 length and diversity and also promoted the generation of shorter LC CDR3s via N-region suppression of dominant microhomology-mediated Vκ-to-Jκ joins. Priming immunization with eOD-GT8 60mer, which strongly engages VRC01 precursors, induced robust VRC01-class germinal center B cell responses. Vκ3-20-based responses were enhanced by N-region addition, which generates Vκ3-20-to-Jκ junctional sequence combinations that encode VRC01-class 5-aa CDR3s with a critical E residue. VRC01-class-rearranging models should facilitate further evaluation of VRC01-class prime and boost immunogens. These new VRC01-class mouse models establish a prototype for the generation of vaccine-testing mouse models for other HIV-1 bnAb lineages that employ different HC or LC Vs.

Diverse antibody variable region exons are assembled in developing B cells from Immunoglobulin (Ig) heavy chain (HC) V, D, and J gene segments and from Igκ or Igλ light chain (LC) V and J segments (1). In humans, there are 55 germline HC Vs (V_Hs) and 70 Igκ and Igλ LC Vs. Vs encode most of the HC and LC variable region, including the antigen contact CDR1 and CDR2 sequences that vary among different HC and LC Vs. Ig HC V(D)J recombination occurs at the progenitor (Pro) B cell developmental stage in the fetal liver and in the postnatal bone marrow (2, 3). Ig LC V to J recombination takes place in the subsequent precursor (Pre) B cell developmental stage in these same sites (1). T cell receptor (TCR) variable region exon assembly also occurs in the fetal liver and thymus and then in the postnatal thymus (4, 5). Mice also have similar sets of Ig HC and LC and TCR variable region gene segments as those found in humans and, in general, assemble them in the context of similar developmental processes (6, 7).Primary B cell receptor (BCR) diversity is achieved, in part, by assorting HC and LC Vs along with each of their distinct sets of CDR1 and CDR2 sequences. However, several V(D)J junctional diversification mechanisms play an even greater role in V(D)J diversity generation (8). In this regard, terminal deoxynucleotidyl transferase (TdT), a DNA polymerase that adds nucleotides to 3''DNA ends without a template (9), plays a key role. V(D)J junctional diversity is immensely augmented by TdT-based nontemplated nucleotide additions, referred to as N regions (10), that are added to V(D)J junctions. While N-region addition generates CDR3 length and sequence diversity, it also suppresses recurrent CDR3s resulting from microhomology (MH)-mediated V(D)J joining (10–13). TdT expression is absent during fetal B and T cell development, resulting in less diverse repertoires dominated by variable region exons promoted by recurrent MH-mediated joins (14–21). In contrast, TDT expression diversifies antigen receptor variable region repertoires generated in mouse and human developing B and T cells that develop postnatally, with the notable exception of LC variable region repertoires in mice (10, 22, 23). Thus, while TdT is expressed during LC V(D)J recombination in postnatal human Pre-B cells (24), it is not expressed in postnatal mouse pre-B cells (25, 26), leading to decreased junctional diversity and much more abundant MH-mediated joins in primary mouse LC repertoires compared to those of humans (22, 23). Lack of TdT expression in fetal repertoires also is known to promote recurrent MH-mediated V(D)J junctions, that are not dominant in postnatal repertoires due to TdT expression. Some such recurrent MH-mediated V(D)J joins in fetal T or B cell repertoires generate TCRs or BCRs critical for certain physiological responses (13, 14, 27, 28). However, the potential role of TdT and N regions in promoting specific responses has remained largely unaddressed.VRC01-class bnAb HCs employ human V_H1-2, which encodes residues that contact the HIV-1 envelope protein (Env) CD4 binding site (29–37). VRC01-class LC variable regions are known to be encoded by several Vs; but all are associated with an exceptionally short 5 amino acid (5-aa) CDR3, which avoids steric clash with Env and contributes to Env interaction (29–37). As both requirements can be achieved by V(D)J recombination, they are predicted attributes of primary VRC01-class precursor BCRs. However, inferred primary VRC01-class BCRs lack detectable affinity for naive Envs (38–41). In this regard, following BCR antigen-activation, primary B cells are driven into germinal center (GC) reactions where they undergo rounds of variable region exon somatic hyper-mutation (SHM) followed by selection of SHMs that increase BCR antigen-binding affinity. This process ultimately leads to high-affinity antibody production. Correspondingly, a third VRC01-class bnAb attribute is abundant variable region SHMs with only a subset contributing to broad Env-binding and potent VRC01-class bnAb activity (37, 42), consistent with VRC01-class bnAb evolution occurring over long HIV-1 infection times and many SHM/selection cycles.To elicit VRC01-class broadly neutralizing antibodies (bnAbs), sequential vaccine immunization approaches propose a priming immunogen to drive precursors into GCs followed by boost immunogens designed to lead them through rounds of SHM/affinity maturation. Based on a structurally designed eOD-GT8 immunogen that binds to the inferred VRC01 unmutated common ancestor (UCA) BCR, potential human VRC01-like precursor B cell frequency was estimated to be one in 400,000 or fewer (43, 44). To test the priming and sequential immunogens that could elicit VRC01-class bnAbs in humans, mouse models are needed that reflect as closely as possible the biology of human B cell responses. Early models expressed knock-in V_H1-2 HCs and, in some, VRC01-class LC Vs, both with mature CDR3s (45–47). These models were nonphysiologic as their BCR repertoire was dominated by a single human HC/LC combination or a single human HC with diverse mouse LCs. Mice with fully human HC and LC gene segment loci assembled by V(D)J recombination were also tested; but precursor frequencies were 150- to 900-fold lower than that of humans (48), likely due to inability to express immense human-like CDR3 repertoires in mice with orders of magnitude fewer B cells. A V_H1-2-rearranging mouse model generated diverse V_H1-2 HC CDR3s, but it employed a germline-reverted VRC01 precursor LC with a 5-aa CDR3 from mature VRC01 bnAb (49). While useful for HC maturation studies during sequential immunization, this model was limited by over-abundance of VRC01 lineage LC precursors. More recently, B cells from transgenic VRC01-class UCA or eOD-GT8-binding precursor knock-in mice were adoptively transferred into congenic recipient mice at human-like frequencies (50–53). While this elegant approach has been very useful, it still has certain limitations as it focused only on eOD-GT8-priming and tested just a small subset of potential VRC01 lineage precursors (50–53).