Ig Knock-In Mice Producing Anti-Carbohydrate Antibodies:Breakthrough of B Cells Producing Low AffinityAnti-Self Antibodies1
Lorenzo Benatuil,* Joel Kaye,* Nathalie Cretin,† Jonathan G. Godwin,* Annaiah Cariappa,‡
Shiv Pillai,‡ and John Iacomini2*
Natural Abs specific for the carbohydrate Ag Gal�1–3Gal�1–4GlcNAc-R (�Gal) play an important role in providing protectivehost immunity to various pathogens; yet little is known about how production of these or other anti-carbohydrate natural Abs isregulated. In this study, we describe the generation of Ig knock-in mice carrying functionally rearranged H chain and L chainvariable region genes isolated from a B cell hybridoma producing �Gal-specific IgM Ab that make it possible to examine thedevelopment of B cells producing anti-carbohydrate natural Abs in the presence or absence of �Gal as a self-Ag. Knock-in miceon a �Gal-deficient background spontaneously developed �Gal-specific IgM Abs of a sufficiently high titer to mediate rejection of�Gal expressing cardiac transplants. In the spleen of these mice, B cells expressing �Gal-specific IgM are located in the marginalzone. In knock-in mice that express �Gal, B cells expressing the knocked in BCR undergo negative selection via receptor editing.Interestingly, production of low affinity �Gal-specific Ab was observed in mice that express �Gal that carry two copies of theknocked in H chain. We suggest that in these mice, receptor editing functioned to lower the affinity for self-Ag below a thresholdthat would result in overt pathology, while allowing development of low affinity anti-self Abs. The Journal of Immunology, 2008,180: 3839–3848.
N atural Abs, those produced without intentional immuni-zation, play a major role in providing protective hostimmunity (1–3). However, the study of natural Abs has
been difficult because in many cases their Ag specificity is un-known. The specificity of many natural Abs remains undefined;however, a significant portion are specific for carbohydrates suchas blood group Ags. In addition to anti-ABO blood group Abs,natural Abs specific for a carbohydrate Ag Gal�1–3Gal�1–4Glc-NAc-R (�Gal), represent a significant population of natural Abs(4–11). The �Gal Ag is synthesized by the glucosyltransferaseUDP galactose:�-D-galactosyl-1,4-N-acetyl-D-glucosaminide�(1–3) galactosyltransferase (Enzyme classification E.C.2.4.1.151), or �GT. All placental mammals except humans, apesand Old World monkeys express a functional �GT enzyme and�Gal epitopes on most tissues (12), and are consequently tolerantto �Gal because it is recognized as part of self. In contrast, hu-mans, apes and Old World primates carry a nonfunctional �GTgene whose function appears to have been lost �30 million years
ago (5). Because these species do not recognize �Gal structures asself, they consequently produce �Gal-specific Abs.
�Gal-specific natural Abs are estimated to comprise 1–8% ofcirculating Ig in humans, and �1% of EBV-transformed peripheralblood B cells make Abs that bind �Gal (6, 13). In humans, �Gal-specific Abs are encoded for by a restricted set of Ig Vh genes fromthe VH3 family (14). Production of Abs specific for �Gal is be-lieved to be elicited in response to normal bacterial flora that col-onize the human gastrointestinal tract (11, 15). The presence ofanti-�Gal Ab in serum and secretory fluids, such as colostrum andsaliva, suggests that these Abs have evolved to play a protectiverole in primate immunity. Viruses produced in �GT-expressingcells that display �Gal-modified glycoproteins within their en-velop, such as lymphocytic choriomeningitis virus, Newcastle dis-ease virus and vesicular stomatitis virus, as well as C-type retro-viruses, have all been shown to be susceptible to inactivation byserum anti-�Gal Abs (16, 17). �Gal-specific Abs are thereforebelieved to play an important role in preventing cross-species in-fection by pathogens. �Gal-specific Abs have also been shown toplay an important role in rejection of xenogeneic tissue whentransplanted into non-human primates (18–22). Despite the impor-tance of �Gal-reactive Abs to host immunity little is known abouthow development of B cells producing �Gal-reactive or other anti-carbohydrate Abs is regulated.
�Gal-deficient mutant mice lacking a functional �GT gene(GT0/0 mice) generated by gene targeting in embryonic stem cellslack expression of �Gal epitopes and consequently develop �Gal-specific natural Abs, the majority of which are IgM (23–25). Theserum titer of �Gal-specific Abs in GT0/0 mice increases in anage-dependent fashion. �Gal-specific Abs in GT0/0 mice sharemany features with human �Gal-specific Abs, including usage ofrelated V genes (26–31). These mice therefore represent a smallanimal model in which �Gal-reactive Abs can be studied. How-ever, the frequency of B cells that produce �Gal-specific Abs in
*Transplantation Research Center, Brigham and Women’s Hospital, Children’s Hos-pital Boston and Harvard Medical School, Boston, MA 02115; †Novartis Pharma,Infectious Diseases, Transplantation & Immunology, Basel, Switzerland; and ‡Mas-sachusetts General Hospital, Center for Cancer Research, Charlestown, MA 02129
Received for publication December 31, 2007. Accepted for publication January4, 2008.
The costs of publication of this article were defrayed in part by the payment of pagecharges. This article must therefore be hereby marked advertisement in accordancewith 18 U.S.C. Section 1734 solely to indicate this fact.1 This work was supported by Grants R01AI044268-09 and R01 AI050602-06 fromthe National Institutes of Health (to J.I.).2 Address correspondence and reprint requests to Dr. John Iacomini, TransplantationResearch Center, Brigham and Women’s Hospital and Children’s Hospital Boston,Harvard Medical School, 221 Longwood Avenue, Room LM303, Boston, MA 02115.E-mail address: [email protected]
these mice is low, making direct analysis of B cells producing�Gal-specific Abs difficult. We and others have generated Ig trans-genic mice to study regulation of B cells producing �Gal-specificAbs (32, 33). However, the use of these mice to address severalaspect of B cell development is limited because the Ig transgenesare randomly integrated. They are not transcriptionally regulatedin an identical fashion to endogenous Ig genes, and in the case ofIg L chain transgenes, unlike endogenous self-reactive rearranged� L chain genes, are not subject to deletion during receptor editing.Therefore, to develop a mouse model in which regulation of �Gal-specific Ab production could be studied, we used gene targeting inembryonic stem cells to construct Ig gene knock-in mice. To thisend, the rearranged VH and VL regions encoding specificity for�Gal were isolated from M86, a hybridoma (IgM, � L chain),derived from GT0/0 mice (34). M86VHVL knock-in mice weregenerated on either an Ag-deficient GT0/0 or Ag-sufficient GT�/�
or GT�/� background to address fundamental issues in regulationof anti-carbohydrate natural Abs. Using these mice, we examinedthe source of B cells producing �Gal-specific Abs and mechanismsleading to negative selection of B cells producing anti-carbohydrate Abs. Our data indicate that restricting the ability of Bcells producing self-reactive anti-carbohydrate Ags to undergo re-ceptor editing significantly affects B cell development and allowsfor the production of low affinity anti-self Abs.
Materials and MethodsMice
C57BL/6 GT0/0 mice and derivation of the colony used in these studieshave been described in (35). C57BL/6 and BALB/c mice were used ascontrols and were obtained from The Jackson Laboratory. All mice werehoused in viral Ab-free microisolator conditions. All animal experimentswere conducted in accordance with Institutional guidelines.
Generation of M86VHVL knock-in mice
The functionally rearranged M86 VDJH and VJ� regions were isolatedusing standard genomic cloning techniques. To construct Ig H chain and Lchain targeting vectors the rearranged M86VH and VL gene segments werethen cloned using standard techniques into the plVhL2Neo (36) andpVKRNeo (37) H and L chains targeting vectors kindly provided by Dr. K.Rajewsky (Immune Disease Institute, Harvard Medical School, Boston,MA). Vector integrity was confirmed by restriction mapping. Gene target-ing and generation of chimeric mice was performed essentially as de-scribed (38). Briefly, each targeting vector was linearized and transfectedseparately into ES-J1 embryonic stem cells. Transfected cells were selectedin presence of G418 (300 mg/ml) and gancyclovir (2 mM). DNA preparedfrom double drug-resistant colonies was then screened for the presence ofhomologous recombination by Southern blotting. ES clones containing ei-ther the targeted M86VH or VL regions were injected into C57BL/6 blas-tocysts and then transferred into (BALB/c � C57BL/6) F1 foster mothers.Blastocyst injection, and embryo transfer, was performed by the Massa-chusetts General Hospital Microinjection Core Facility (Boston, MA). Chi-meric mice were mated to C57BL/6 mice. DNA prepared from tail biopsysamples of resulting offspring was analyzed by Southern blotting to con-firm germline transmission. Resulting knock-in mice were then crossedwith cre mice (provided by the Massachusetts General Hospital Core Fa-cility) to delete the Neor gene from targeted H chain and L chain loci.Resulting knock-in mice were then crossed to GT0/0 mice on the C57BL/6background to generate M86VHGT�/0 and M86VLGT�/0 mice. These micewere then bred to generate �Gal expressing M86VHVLGT�/0 and �Gal-deficient M86VHVLGT0/0 mice.
Flow cytometry and Abs
Single cell suspensions were prepared from blood or lymphoid tissues andthen stained and analyzed by flow cytometry as described (35). �Galepitopes were detected using the Gal-specific IB4 lectin (Sigma-Aldrich)from Bandeiraea simplicifolia (BS-I isolectin B4) (39). The following Absused in this study were purchased from BD Pharmingen: RA3-6B2 (anti-CD45R/B220), 187.1 (anti-mouse Ig�), R26-46 (anti-mouse �1, �2, and�3), 11-26c.2a (anti-IgD FITC), R6-60.2 (anti-IgM), 7G6 (anti-CD21), S7(anti-CD43), and anti-Mac-1 (anti-CD11b). Goat anti-mouse IgM was pur-chased from Jackson ImmunoResearch Laboratories. RS3.1 (anti-Igh-6a
(40)) and MB86 (anti-Igh-6b (41)) were provided by Dr. H. Wortis (TuftsUniversity Sackler School of Biomedical Sciences, Boston MA). B cellscapable of binding �Gal were detected by staining with FITC- or biotin-conjugated Gal-BSA (V-Labs).
ELISA
ELISAs were conducted as previously described (35). Briefly, ELISAplates (Corning) were coated overnight at 4°C with either �Gal conjugatedto BSA (Gal-BSA) or lactosamine conjugated to BSA (Lac-BSA; V-Labs)in carbonate buffer (pH 9.5), and then washed with PBS containing 0.05%Tween 20 (PBS-Tween 20). Lac-BSA shares all determinants with Gal-BSA, except for the terminal galactose structure, and serves as a specificitycontrol. The wells were blocked with 1% BSA in PBS-Tween for 1 h atroom temperature and then washed. Serum samples were serially diluted inPBS-Tween 20, added to the plates, and incubated for 1 h at 37°C. Theplates were then washed extensively with PBS-Tween 20, and bound Abswere detected using HRP-conjugated goat anti-mouse IgM (1/4000; Jack-son ImmunoResearch Laboratories). To determine the relative contributionof transgene-encoded vs endogenously encoded anti-�Gal, bound Abswere detected with purified biotinylated RS3.1 or MB86, followed byHRP-conjugated streptavidin (1/800; Amersham Biosciences). The plateswere incubated for 1 h at 37°C and then washed five times with PBS-Tween 20. A total of 0.01 mg/ml o-phenylenediamine dihydrochloride(Sigma-Aldrich) in substrate buffer was then added for 20 min at roomtemperature to develop the assays. The reaction was terminated by addingsulfuric acid to each well, and absorbency was read at 492 nm. Backgroundvalues obtained from Lac-BSA-coated plates were subtracted from thoseobtained using Gal-BSA-coated plates. Assays were performed in dupli-cate. In some instances, serum from immunized mice was used. Mice wereimmunized i.p. with 107 irradiated (3000 rad) pig PBMC as described (42).
Anti-�Gal B cell ELISPOT assay
Multiscreen-HA plates (Millipore, Bedford, MA) were coated with 10�g/ml of either Gal-BSA or Lac-BSA in PBS at 4°C overnight. The plateswere then washed three times with PBS, allowing the plates to soak for 5min between each wash. The plates were blocked with IMDM containing0.4% BSA and penicillin and streptomycin for 2 h at 37°C. The blockingmedium was then removed and 10-fold serial dilutions (starting at 1 � 106
cells per well) of spleen cells prepared in blocking IMDM were added tothe wells. The cells were incubated at 37°C in 5% CO2 for 24, 48, or 72 hin the presence or absence of LPS (0.5 �g/well). After culture, the plateswere washed three times in PBS, followed by three additional washes inPBS-Tween 20. HRP-conjugated goat anti-mouse IgM was then added toeach well and incubated for 2 h at 4°C. The plates were washed three moretimes with PBS-Tween 20, followed by PBS, at which point the assayswere developed by adding filtered chromogen substrate (3-amino-9-ethyl-carbazole) in acetate buffer (pH 5.0). Plates were incubated in the presenceof chromogen substrate at room temperature for 5 min and the reactionterminated by washing the plate with water. Spots were enumerated usingan automated ELISPOT reader (ImmunoSpot; Cellular Technology). In allassays, the number of background spots obtained on Lac-BSA-coatedplates was subtracted from the number obtained on corresponding Gal-BSA-coated plates. All samples were plated in duplicate.
Cell culture
B cell precursors from M86VHVL-GT0/0 mice were grown in vitro as pre-viously described (43, 44). Briefly, bone marrow cells were depleted oferythrocytes and were cultured in BMB220 medium (IMDM supplementedwith 10% FCS, 2-ME, L-glutamine, penicillin/streptomycin, and 50–100U/ml recombinant murine IL-7; R&D Systems) at a concentration of 2 �106 cells/ml for 5 days. Washed cells were then cultured for 2–42 h in theabsence of IL-7 in wells with irradiated (2000 rad) confluent primary ad-herent bone marrow-derived stroma from Ag-negative GT0/0 or Ag-bearingwild-type C57BL/6 mice. After culturing the cells on the stroma, nonad-herant pre-B cells were washed and frozen at �80°C until RNA was ex-tracted. Stromal cultures were initiated by plating erythrocyte-depletedbone marrow cells to confluence in stromal medium (RPMI 1640, 5% FCS,2-ME, L-glutamine, penicillin/streptomycin, sodium pyruvate, and nones-sential amino acids) for 3 days, then washing away nonadherent cells andallowing the adherent stroma to grow for at least 2 wk before use. Stromalcell lines were tested for the expression of �Gal by staining with �Gal-specific IB4 lectin.
Detection of RAG2 expression
RNA was prepared from cells using an RNeasy mini kit (Qiagen). Com-plementary DNA was prepared from DNaseI-treated (Invitrogen Life
Technologies) RNA with oligo(dT) primers with the Superscript first-strand synthesis kit (Invitrogen Life Technologies). Primer sequencesused are as follows: RAG2 forward primer, 5�-CACATCCACAAGCAGGAAGTACAC-3�; RAG2 reverse primer, 5�-GGTTCAGGGACATCTCCTACTAAG-3�; �-actin forward primer, 5�-ACCCCAAGGCCAACCGCGAGAAGATGACC-3�; �-actin reverse primer, 5�-GGTGATGACCTGGCCGTCAGGCAGCTCGTA-3�. PCR were performed in a finalvolume of 50 �l using 3–5 �l of cDNA and 2 U Taq polymerase (Fi-scher Scientific) on a GeneAmp PCR 2400 thermal cycler(PerkinElmer). Except for the first cycle, which had a 2 min 94°C de-naturation step, each cycle consisted of 1 min at 94°C, 1 min at 60°C,and 1 min at 72°C, with a 7 min 72°C extension cycle to end the PCR.For PCR analysis, 35 cycles was used. The PCR product was thenfractionated by agarose gel electrophoresis, transferred to Hybond-N nylon membranes (Amersham Biosciences) and RAG2 ampliconsdetected by hybridization to 32P-labeled RAG2 probes as described(45).
RAG2 expression was also analyzed by quantitative real-time PCR. Onemicrogram of total RNA was reverse transcribed using the SuperScript IIIFirst-Strand Synthesis SuperMix for quantitative real-time PCR kit (In-vitrogen Life Technologies) and used to quantitate relative levels of RAG2expression using the TaqMan predeveloped assay Mm00501300_m1 forRAG2, and 4352932E for GADPH (Applied Biosystems). The assays wereperformed in triplicates following standard protocols. The values shownare presented as the difference in cycle threshold (Ct) values normalized toGADPH for each sample (RQ).
Data analysis and reproducibility
Data throughout the study are representative of experiments containing atleast three age- and sex-matched mice per group. All experiments havebeen shown to be reproducible by multiple individuals and have been con-ducted and confirmed on multiple occasions.
ResultsConstruction of M86VHVL Ig knock-in mice
The low frequency of B cells that produce �Gal-specific Abs inGT0/0 mice (25, 46, 47) limits the use of these mice to study theregulation of �Gal-specific Ab production. To develop a mousemodel in which B cells producing �Gal-specific Abs could bedirectly tracked during their development, we generated Igknock-in mice that express an �Gal-specific BCR. To this end, therearranged Ig H chain and L chain variable region segments werecloned from the hybridoma M86. M86 was derived from GT0 miceand produces an �Gal-specific IgM Ab that uses a � L chain (26,32, 34). The H chain (rearranged to Jh4) and L chain (rearrangedto J�1) variable region gene segments were then cloned into theplVhL2Neo (36) and pVKRNeo (37) H and L chains Ig targetingvectors. Each targeting vector was then electroporated separatelyinto 129/Sv embryonic stem cells (ES-J1) as described (38). ESclones containing either the targeted M86VH or M86VL regionwere then injected separately into C57BL/6 blastocysts that werethen transferred into (BALB/c � C57BL/6) F1 foster mothers togenerate chimeric mice as described (38). Offspring carrying aknocked in M86VH or M86VL allele were then crossed with Crerecombinase transgenic mice to delete the neomycin resistancegene from targeted H chain and L chain loci. Resulting knock-inmice were then crossed to GT0/0 mice on the C57BL/6 backgroundto generate M86VHGT�/0 and M86VLGT�/0 mice (Fig. 1A) thatwere then bred to generate �Gal-expressing M86VHVLGT�/0
mice and �Gal-deficient M86VHVLGT0/0 mice.
FIGURE 1. Characterization of �Gal-specific Ab production in M86VHVL knock-in mice. A, Amplification of tail DNA from M86VHVL knock-in miceand normal controls by PCR. The knocked in H chain was detected using primers that amplify the recombined VHDJ4 segment (left panel). To analyze thepresence of one or two copies, an additional primer was introduced to amplify the intron between JH3 and JH4 (JH3-JH4, top band) present only in wild-typealleles. A similar approach was used to detect the L chain allelic expression (right panel). B and C, Anti-�Gal Ab production in naive M86VHVL Ig knock-inmice. In all cases, binding to Lac-BSA was used as a specificity control. Data shown are representative of multiple experiments containing at least three miceper group. B, Anti-�Gal serum titers in naive (marked by x) and immunized M86VHVLGT0/0 (Œ) and M86VHVLGT� mice (�). Also shown are anti-�Gal serumtiters in C57BL/6 (F) and naive GT0/0 (�) mice. C, Production of knocked in IgMa (�) or endogenous IgMb (�) anti-�Gal Abs in naive M86VHVLGT0/0 mice.D, ELISPOT analysis of �Gal-specific IgM production in naive M86VHVLGT0/0, M86VHVLGT�, and immunized GT0/0 mice. The frequency of Ab-secreting cellsper 106 total cells in each tissue is shown. E, Survival of GT�/� (�) or GT0/0 (E) hearts transplanted into naive M86VHVLGT0/0 mice.
Characterization of �Gal-specific Ab production in M86VHVL
knock-in mice
Analysis of natural serum Abs revealed that M86VHVLGT0/0 micecontain high titers of �Gal-specific Ab in their serum (Fig. 1B).The level of �Gal-specific Abs in M86VHVLGT0/0 mice was sig-nificantly higher than that observed in GT0/0 controls at all agesexamined (Fig. 1B). Immunization of M86VHVLGT0/0 mice with�Gal-expressing pig cells led to an increase in the titer of �Gal-specific Abs (Fig. 1B). In contrast, we were unable to detect pro-duction of �Gal-specific Abs in the serum of M86VHVLGT�/0 orM86VHVLGT�/� mice even after immunization (Fig. 1B). InM86VHVL mice the endogenous H chain locus is IgH-6b
(IgMb,�b), whereas the knocked in H chain locus is IgH-6a
(IgMa,�a), which can be detected using the anti-allotypic mAbsMB86 (41) and RS3.1 (40), respectively. �Gal-specific Abs inM86VHVLGT0/0 mice were encoded for by the knocked in IgMa
allotype, rather than the endogenous IgMb allotype (Fig. 1C). Bcells spontaneously producing �Gal-specific Abs were detected inthe spleen, lymph node, and bone marrow but not the peritoneumof M86VHVLGT0/0 mice (Fig. 1D). Immunization with pig cellsdid not result in the detection of �Gal-producing B cells in theperitoneum of immunized M86VHVLGT0/0 mice (data not shown).The frequency of �Gal-producing B cells in the spleen, bone mar-row, and lymph nodes of M86VHVLGT0/0 mice was at least 10-fold higher than in GT0/0 controls (Fig. 1D). We were unable todetect B cells producing �Gal-specific Abs in lymphoid tissuesfrom M86VHVLGT�/0 or M86VHVLGT�/� mice (Fig. 1D).
We next examined whether the titer of �Gal-specific Abs inM86VHVLGT0/0 mice was sufficient to induce Ab-mediatedtransplant rejection. MHC-matched hearts from littermateM86VHVLGT� or C57BL6 mice were heterotopically trans-planted into the abdomen of M86VHVLGT0/0 mice. Hearts fromM86VHVLGT� and C57BL/6 mice were uniformly rejected within24–72 h. In contrast, hearts from M86VHVLGT0/0 mice were ac-cepted long-term (Fig. 1E). These data suggest that �Gal-specific
FIGURE 2. Characterization andearly B cell development in M86VHVL
knock-in mice. A, Staining of bonemarrow cells from C57BL/6, M86VH
VLGT�, and M86VHVLGT0/0 mice forexpression of B220 and CD43. Shownare the frequency of B220�, CD43� Bcells and B220�, CD43� B cells as de-termined by flow cytometry. B, Analy-sis of sIgMa and sIgMb expression byB220� cells in bone marrow by flowcytometry. C, Analysis of �Gal-bind-ing by sIgMa and sIgMb expressing Bcells (B220�) in the bone marrow ofM86VHVLGT0 (top row) and M86VH
VLGT� (bottom row) mice. D, Anal-ysis of �Gal-binding in B220�,CD43� pre-B cells and B220�,CD43� pro-B cells from M86VHVL
GT0/0 mice. Data shown are re-presentative of multiple experimentscontaining at least three mice pergroup.
FIGURE 3. Self-antigenic stimulation induces expression of RAG2 inM86VHVLGT0/0 BM B cells. Analysis of RAG2 up-regulation by bonemarrow pre-B cells upon Ag encounter. M86VHVLGT0/0 bone marrowpre-B cells were grown in IL-7 for 5 days and then cultured on pre-estab-lished primary GT� or GT0/0 bone marrow stromal cells for the timesindicated. A, Real-time PCR analysis by PCR and Southern blot showedup-regulation of RAG2 mRNA in bone marrow pre-B cells cultured onGT� stroma. �-actin mRNA levels served as an internal control and wereevaluated by ethidium bromide staining. Films were exposed for 2 or 24 h.B, Analysis of RAG2 expression by quantitative real-time PCR.M86VHVLGT0/0 bone marrow pre-B cells were grown in IL-7 for 5 daysand then cultured for 2, 4, and 16 h on irradiated (2000 rad) GT� (f) orGT0/0 (�) bone marrow stromal cells. RAG2 levels were then assessed byquantitative real-time PCR. Data presented are the difference in cyclethreshold values normalized to GAPDH for each sample (RQ) and are theresults of two different experiments. All assays were performed in tripli-cate. The level of RAG2 transcripts was statistically higher in pre-B cellscultured in GT� stromal cells (p � 0.05, at 2, 4 and 16 h, by Student’s ttest, unpaired 95% confidence).
Abs in M86VHVLGT0/0 mice are functional and of sufficient titerto induce Ab-mediated rejection.
Characterization of early lymphocyte development in M86VHVL
knock-in mice
M86VHVLGT0/0 and M86VHVLGT0/� mice were sacrificed andlymphoid tissues analyzed to examine lymphocyte development.�� T cell development in the thymus was essentially normal whencompared with GT0/0 or GT� controls (data not shown). Similarly,T cell development in the spleen and lymph nodes was normal(data not shown). Characterization of B cell development in thebone marrow of M86VHVLGT0/0 or M86VHVLGT�/� knock-inmice demonstrated that pro-B cells (B220�,CD43�) and immature(B220�,CD43�) B cells, as defined in (48, 49), were present insimilar proportions to those observed in normal mice, although thefrequency of each of these fractions was reduced when comparedwith normal controls (Fig. 2A), as observed in other Ig knock-inmice (32, 33). The frequency of pre-B cells was similar inM86VHVLGT0/0 or M86VHVLGT�/� knock-in mice (data notshown).
The majority of B220� B cells in the bone marrow ofM86VHVLGT� and M86VHVLGT0/0 mice expressed the knockedin H chain allele (B220�, sIgMa�) (40.4 � 9.3% and 46.6 �10.1%, respectively) rather than endogenously encoded (B220�,
sIgMb�) allele reflecting exclusion by the rearranged M86VH gene(Fig. 2B). Essentially all of the B220�, sIgMa� B cells in the bonemarrow of M86VHVLGT0/0 mice were able to bind �Gal as de-termined by staining with fluorescently labeled BSA conjugated to�Gal (�Gal-BSA) (Fig. 2C). B220�, sIgMb� B cells able to bind�Gal-BSA were not detected in M86VHVLGT� or M86VHVL
GT0/0 mice (Fig. 2C). Staining of M86VHVLGT0/0 bone marrowcells for expression of B220 and CD43 revealed that B cells ca-pable of binding �Gal-BSA were CD43�, pre-B or newly formedB cells (Fig. 2D). B cells able to bind �Gal were not detected in thebone marrow of M86VHVLGT� mice, suggesting that B cells ex-pressing the knocked in transgene are tolerized during their devel-opment (Fig. 2C).
Receptor editing prevents the development of B cells producingself-reactive Abs
Because the frequency of B cells expressing surface IgMa (sIgMa)3
in the bone marrow of M86VHVLGT� is relatively high (Fig. 2B)even though we were unable to detect B cells that bind �Gal in thebone marrow of M86VHVLGT� mice (Fig. 2C) we reasoned thattolerance to �Gal in M86VHVLGT� mice was not deletional. We
3 Abbreviations used in this paper: sIgM, surface IgM; MZ, marginal zone.
FIGURE 4. �Gal-binding B cells in the periphery of M86VHVLGT0/0 mice. A, Analysis of B220�, sIgMa, and sIgMb expressing B cells in the spleenof M86VHVLGT0/0 and M86VHVLGT� mice. B, Analysis of allelic exclusion in splenic B cells. Spleen cells from M86VHVLGT0/0 and M86VHVLGT�
were surface stained with Abs specific for IgMa (RS3.1) and IgMb (MB86) and then analyzed by flow cytometry. Shown is sIgMa and sIgMb expressionwithin the lymphoid gate. C, Analysis of �Gal-binding by sIgMa� and sIgMb� B cells in M86VHVLGT0 (top) and M86VHVLGT� (bottom) mice. D,Analysis of �Gal-binding in the peritoneum of M86VHVLGT0 mice. Shown is staining of peritoneal exudate cells (PeC) for expression of B220 and CD11b(top left). B220�,CD11b� cells were also analyzed for expression of CD5 (bottom left) and �Gal-binding by B220�, CD11b�, CD5� B1a B cells (top right)and B220�, CD11b�, CD5� B1b B cells (bottom right) examined. E, ELISPOT analysis of �Gal-specific IgM production in naive M86VHVLGT0/0
following stimulation with 0.5 �g of LPS mice. The frequency of anti-�Gal Ab-secreting cells per 106 total cells in each tissue is shown.
therefore examined whether tolerance was the result of receptorediting (45, 50–52). To this end, IL-7 driven pre-B cells from thebone marrow of M86VHVLGT0/0 mice were cultured on irradiatedbone marrow stromal cells from GT� and GT0/0 mice and thelevels of RAG2 expression assayed by real-time PCR. Expressionof RAG2 was up-regulated in M86VHVLGT0/0 pre-B cells culturedon stromal cells from GT� cells when compared with pre-B cellsfrom the same mice cultured on GT0/0 stromal cells (Fig. 3A).Analysis of RAG2 expression by quantitative real-time PCR con-firmed this finding (Fig. 3B). These data suggest that pre-B cellsproducing self-reactive anti-carbohydrate Abs undergo tolerancethrough a mechanism that involves receptor editing.
Peripheral B cell development in M86VHVL knock-in mice
We next characterized B cell development in the periphery ofM86VHVL mice. In the spleen of M86VHVLGT0/0 andM86VHVLGT�/0 mice the majority of B cells were B220�,sIgMa� (51.8 � 8.3% and 69.4 � 7.6%, respectively) with thefrequency of B220�, sIgMa� B cells being slightly higher inM86VHVLGT�/0 mice (Fig. 4A). B220�, sIgMb� B cells weredetected in both types of mice (Fig. 4A). When compared withbone marrow (Fig. 2B), the proportion of B220�, sIgMb� B cellsobserved was higher in the spleen of both M86VHVLGT0/0 andM86VHVLGT�/0 mice (Fig. 4A). B cells expressing both sIgMa
and sIgMb were not detected in either M86VHVLGT0/0 orM86VHVLGT�/0 mice, suggesting allelic exclusion by theknocked in H chain (Fig. 4B). As observed in the bone marrow, Bcells capable of binding �Gal were detected in the spleen ofM86VHVLGT0/0 but not M86VHVLGT�/0 mice (Fig. 4C). Al-though we were unable to detect B cells that spontaneously secrete�Gal-specific Abs in the peritoneum of M86VHVLGT0/0 mice(Fig. 1D), B220�, CD11b�, CD5� B cells in the peritoneum werecapable of binding �Gal in M86VHVLGT0/0 mice (Fig. 4D). Inboth the spleen and peritoneum, B cells able to bind �Gal were
B220�, sIgMa�. However, B cells in the peritoneum did not pro-duce �Gal-specific Abs even after stimulation with LPS (Fig. 4E).Consistent with data in Fig. 1E, B cells secreting �Gal-specificAbs were only observed in the spleen, bone marrow, and lymphnodes.
Immature B cells in the bone marrow first emigrate to the redpulp in the spleen where they can be detected as sIgMhigh, sIgDlow,CD21low, CD23� newly formed transitional or T1 cells. This em-igration to the spleen is an Ag-regulated process (reviewed in Ref.53). T1 cells then colonize the lymphoid follicles present in thespleen, and up-regulate IgD to become sIgMhigh, sIgDhigh, CD21int
follicular B cell precursors. These cells give rise to sIgMlow,sIgDhigh, CD21int, CD23� naive follicular B cells, and after addi-tional Ag stimulation, sIgMhigh, sIgDhigh mature follicular B cells.There is also a subpopulation of B cells that express high levels ofCD21 and CD1d, and are sIgMhigh, sIgDlow, CD21high, CD23�,CD1dhigh marginal zone (MZ) B cells, which exist outside thefollicle (54, 55). Analysis of the spleen of M86VHVLGT0/0 miceby cell surface staining with Abs specific for IgM, IgD, and CD21and with FITC-labeled �Gal-BSA revealed that within the spleenB cells capable of binding �Gal were sIgMhigh, IgDlow, CD21high
MZ B cells (60 � 22% MZ B cells are �Gal specific, Fig. 5A). Wewere unable to detect sIgMlow, IgDhigh, CD21int follicular B cellscapable of binding �Gal-BSA (Fig. 5A). Analysis of tissue sec-tions from the spleens of M86VHVLGT0/0 mice revealed that es-sentially all B cells capable of binding �Gal-BSA reside in the MZand were excluded from the follicle (Fig. 5B). These data suggestthat in M86VHVLGT0/0 mice �Gal-binding B cells are committedto a MZ fate.
Increasing M86VHVL copy number alters the fate of�Gal-specific B cells when self-Ag is encountered
While breeding M86VHVL knock-in mice, we also generated micecarrying two copies of the knocked in M86VH region and either
FIGURE 5. Analysis of B cell de-velopment in the spleens ofM86VHVLGT0/0 mice. A, Spleen cellswere stained with anti-IgM, -IgD,-CD21, and �Gal-BSA. Based on cellsurface expression levels of IgM andIgD, B cells within the spleen lym-phoid gate were divided (far leftpanel) into newly formed (NF) andMZ fractions (fraction I), follicularprecursors (FP), and MZ precursors(MZP) (fraction II) and follicular cells(FO) (fraction III). These fractionswere then analyzed with respect toCD21 expression levels and the abil-ity to bind �Gal. Shown is �Gal-bind-ing within the CD21 gate for each Bcell population. B, Staining of spleensections from M86VHVLGT� (left)and M86VHVLGT0/0 (right) micewith anti-metallophilic macrophage(MOMA1, blue) and �Gal-BSA(green). All �Gal-binding cells are lo-cated in the MZ, outside of the mar-ginal sinus. Data shown are represen-tative experiments.
one or two copies of knocked in M86VL region (M86VH2VL orM86VH2VL2 mice) on a GT0/0 or GT� background. Additionally,we generated mice carrying two copies of the knocked in M86VL
region (M86VHVL2 mice) on a GT0/0 or GT� background. Inblood, the frequency of B cells in M86VH2VL, M86VH2VL2, andM86VHVL2 mice on the GT0/0 background was similar to the fre-quency observed in M86VHVLGT0/0 mice (Fig. 6). However, thefrequency of B220� B cells in the periphery of M86VH2VLGT�
(18.7 � 2.3%) mice was significantly reduced when comparedwith the frequency in M86VHVLGT� mice (24.7 � 2.9%; p �0.05, Student’s t test). M86VHVL2GT� and M86VH2VL2GT�
mice exhibited an even greater reduction in B220� cells (10.1 �0.3% and 6.6 � 0.6%, p � 0.05) when compared with thefrequency in M86VHVLGT� mice (Fig. 6). These data suggest thatthe presence of multiple knocked in H chain or L chain allelesaffects B cell development. Interestingly, the reduction in B cellnumbers is observed only in mice that express �Gal as a self-Ag,suggesting that the effect observed may be related to alterations innegative selection.
Production of autoreactive �Gal-specific Abs inM86VH2VL2GT� knock-in mice
Our data indicate that B cells producing �Gal-specific Abs wereefficiently tolerized in M86VHVLGT�; however, we consistentlyobserved the presence of �Gal-specific serum IgM Abs inM86VH2VL2GT� mice (Fig. 7A). These Abs did not bind the con-trol Ag Lac-BSA which shares all determinants with �Gal-BSAexcept for the terminal galactose residue, indicating the bindingobserved was specific for �Gal. Essentially all �Gal-specific Absin these mice were of the M86VH-encoded IgMa allotype (data notshown). The titer of anti-�Gal Abs in these mice was consistentlylower than that observed in littermate M86VH2VLGT0/0 andM86VH2VL2GT0/0 mice, but was similar to the titers observed inGT0/0 control mice (Fig. 7A). Furthermore, the titers of these Abscould be boosted by pig cell immunization (Fig. 7A). To confirmthese results, spleen cells from pig cell immunizedM86VH2VL2GT�/0, C57BL/6 GT� controls and GT0/0 mice wereanalyzed for their ability to produce �Gal-specific Ab in IgELISPOT assays using �Gal-BSA- or Lac-BSA-coated plates. Wedetected B cells producing �Gal-specific IgM Abs in the spleens ofM86VH2VL2GT� mice, at a frequency similar to that observed inimmunized GT0/0 controls (Fig. 7B). Similar results were observedin M86VH2VLGT� mice (data not shown). These data suggest thatB cells producing self-reactive anti-�Gal Abs are able to developin M86VH2VL2GT� mice.
The development of B cells producing �Gal-specific Ab inM86VH2VL2GT� mice may reflect lowered affinity for Ag
The ability to detect anti-�Gal Abs in M86VH2VL2GT� mice wasunexpected. To begin to characterize these Abs, we analyzed the
FIGURE 6. Increasing copy number ofthe knocked in VH and VL chains alters thefate of �Gal-specific B cells. Analysis ofB220 expression and �Gal-binding by Bcells in peripheral blood of M86VHVL,M86VH2VL, M86VHVL2, and M86VH2VL2mice on an �Gal-deficient (GT0/0) or posi-tive (GT�) background. The frequency ofcells within the lymphoid gates is shown.
FIGURE 7. Production of autoreactive �Gal-specific Abs inM86VH2VL2 mice. A, Production of �Gal-specific IgM Abs in naive (f)and pig cell immunized (Œ) M86VH2VL2-GT�/� mice, as well as naiveM86VHVLGT0/0 (�), naive GT0/0 (�) and C57BL/6 (F) controls. B,ELISPOT analysis of �Gal-specific Ab production in pig cell-immunizedC57BL/6 mice, and GT0/0 and M86VH2VL2GT�/0 mice. The frequency ofcells secreting anti-�Gal Ab in the spleen is shown. Binding to Lac-BSAwas used as a specificity control in all assays. C, �Gal-specific Abs inM86VH2VL2GT� mice bind Ag at a lower affinity. Binding of serum�Gal-specific IgM from M86VH2VL2GT� (square symbols) orM86VHVLGT0/0 (circle symbols) mice to ELISA plates coated with 10�g/ml of a 15:1 (open symbols) or 10:1 (closed symbols) molar ratio of�Gal to BSA. Data shown are representative.
capacity of �Gal-specific Abs from these mice to bind BSA con-jugated to different numbers of �Gal epitopes. We were readilyable to detect binding of �Gal-specific IgM from these mice whenusing ELISA plates coated with �Gal-BSA conjugates in whichthe molar ratio of �Gal to BSA molecules was high (15:1) (Fig.7C). However, the ability to detect binding of �Gal in these micewas significantly reduced when ELISA plates were coated with�Gal-BSA conjugates in which the molar ratio of �Gal to BSAwas lower (10:1) (Fig. 7C). Importantly, serum IgM fromM86VHVLGT0/0 (Fig. 7C) and M86VH2VL2GT0/0 (data notshown) was able to bind to both �Gal-BSA preparations to a sim-ilar degree. These data suggested that �Gal-specific Abs pro-duced in M86VH2VL2GT� mice exhibit different binding char-acteristics than those of littermate mice on the GT0/0
background. We suggest that B cells making self-reactive �Gal-specific Abs in M86VH2VL2GT� and M86VH2VLGT� mice(data not shown) may develop because the Ab they produceexhibits a lowered affinity for �Gal.
DiscussionAnti-carbohydrate natural Abs are thought to play a key role inproviding protective host immunity and have been well describedas being important in mediating transplantation rejection acrossblood group disparities as well as discordant species. Yet little isknown about B cells that produce anti-carbohydrate Abs, in partdue to a lack of small animal models in which the development andregulation of B cells producing anti-carbohydrate Abs can be stud-ied in a physiologically relevant manner. The development ofGT0/0 mice has made it possible to begin to address some of theseissues. However, the frequency of B cells producing �Gal specificAbs in these mice is too low to allow for their direct analysis. Weand others have previously attempted to overcome this difficulty bygenerating Ig transgenic mice that express transgenes encoding�Gal-specific Abs (32, 33). However, these models are limitedbecause the expressed transgenes were not subject to regulation byelements within the endogenous Ig loci, making it difficult to as-sess physiological relevance. To overcome these issues we usedgene targeting in embryonic stem cells to construct Ig knock-inmice that carry rearranged H chain and L chain V regions thatencode specificity for �Gal that are under regulatory control of theendogenous Ig H chain and � L chain loci. Breeding M86VHVL
mice to GT0/0 mice offered us the opportunity to directly examinethe development of B cells producing �Gal-specific Abs in thepresence or absence of �Gal as a self-Ag.
M86VHVLGT0 mice show high serum levels of �Gal-specificAbs produced by the knocked in allele. The titer of these Abs wassufficient to mediate rejection of H-2-matched heart transplantsfrom GT� donors. The titer of �Gal-specific Abs in M86VHVL
GT0 mice could be increased by immunization. �Gal-specific Abswere not detected in the serum of M86VHVLGT� mice, indicatingthat expression of �Gal in these mice prevented development of�Gal-specific Abs. However, we were still able to detect sIgMa�
B cells in the spleens of M86VHVLGT� mice, suggesting that Bcells expressing the knocked in M86VH region were not deletedduring their development. Because the frequency of B cells ex-pressing surface IgMa in the bone marrow of M86VHVLGT� andM86VHVLGT0/0 knock-in mice was similar even though we wereunable to detect B cells that bind �Gal in the bone marrow ofM86VHVLGT� mice, we reasoned that tolerance to �Gal inM86VHVLGT� mice was most likely the result of receptor editingrather than deletion. Analysis of IL-7 driven pre-B cells from thebone marrow of M86VHVLGT0/0 mice revealed that exposure to�Gal as a self-Ag led to an up-regulation of RAG2 expression inM86VHVLGT0/0 pre-B cells. These data are consistent with the
idea that pre-B cells producing self-reactive anti-carbohydrate Absundergo tolerance induction via receptor editing in the bone mar-row. Interestingly, although receptor editing has been suggested tolead to an increase in the frequency of immature B cells expressing� light chains (37, 45, 56), we did not detect an increase in thefrequency of B cells expressing a � L chain in the bone marrow orperiphery of M86VHVLGT� mice (data not shown). Therefore, inthis model, receptor editing most likely occurs preferentially on the� L chain. At least three distinct mechanisms that shape the naiveB cell repertoire have been described, including clonal deletion(57–61); anergy (62, 63); and receptor editing (45, 50, 51, 64). Themechanism by which B cells producing Abs to self-carbohydrateAgs are tolerized has been studied in relatively little detail. Al-though receptor editing has been described as a major mecha-nism of B cell tolerance, to our knowledge this is the first de-scription suggesting that receptor editing is the mechanism oftolerance for B cells producing anti-carbohydrate Abs.
Our analysis of B cell development in M86 knock-in miceshows that anti-�Gal B cells undergo editing at the H chain locuseven in the absence of �-Gal, resulting in the development ofsIgMb� B cells. When comparing the frequency of sIgMb� B cellsin bone marrow and spleen, it is apparent that the relative fre-quency of such cells is increased in spleen (see Figs. 2 and 4),suggesting a selection of B cells expressing sIgMb� in the periph-ery. One possibility is that the knocked in H chain does not excludewell and that a selection process in GT� mice ensures that only Bcells expressing the endogenous H chain differentiate into periph-eral mature lymphocytes. However, based on the analysis shown inFig. 4B in which we were unable to detect B cells producing bothsIgMa and IgMb, we would suggest that apparently allelic exclu-sion in this system is not very leaky. Rather, there seems to be aselection process by which B cells expressing endogenous sIgMb
are selected for in the periphery. It seems reasonable to suggestthat this selection would be important to allow for an increase inthe diversity of the Ig repertoire.
There is an unresolved issue regarding which B cell subsetsproduce anti-carbohydrate Abs. Using Ig transgenic mice, it hasbeen suggested previously that B cells producing �Gal-specificAbs are skewed to a MZ B cell phenotype (33). However, the useof Ig transgenic mice has made it difficult to determine whetherthis observation is of physiological relevance. Analysis of sponta-neous �Gal Ab production in lymphoid tissue of M86VHVLGT0/0
mice revealed the presence �Gal-specific Ab production in thespleen, bone marrow, and lymph nodes, but not the peritoneum. Inthe spleen of M86VHVLGT0/0 mice, B cells capable of binding�Gal were sIgMhigh, IgDlow, CD21high MZ B cells. We were un-able to detect sIgMlow, IgDhigh, CD21int follicular B cells capableof binding �Gal-BSA. These data together with analysis of �Gal-binding B cells in tissue sections suggest that B cells producing�Gal anti-carbohydrate Abs are committed to a MZ B cell fate.
Analysis of mice carrying two copies of the knocked in M86VH
region and either one or two copies of the knocked in M86VL
region revealed that copy number can significantly impact B celldevelopment in mice expressing �Gal as a self-Ag. Although thefrequency of B cells in M86VH2VL2, M86VH2VL, and M86VH
VL2 mice on the GT0/0 background was similar to the frequency inM86VHVLGT0 mice, the frequency of B cells in M86VH2VLGT�
mice was significantly reduced. M86VHVL2GT� and M86VH
2VL2GT� mice exhibited an even greater reduction in B cellswhen compared with M86VHVL animals. Interestingly, the great-est affect on B cell numbers was observed in M86VHVL2GT� andM86VH2VL2GT� mice that carry two copies of the knocked in Lchain allele. Because the effect of copy number on B cell numberwas seen only in mice expressing �Gal as a self-Ag, we suggest
that the effect observed is related to alterations in negative selec-tion and that increasing the copy number of knocked in alleles mayalter mechanisms of negative selection by restricting receptor ed-iting, thereby skewing tolerance toward a deletional mechanism.This would support the idea that B cells primarily use editing toeliminate self-reactive B cells, and that negative selection via de-letion is secondary to receptor editing.
Although B cells producing �Gal-specific Abs were efficientlytolerized in M86VHVL mice on the GT� background, we consis-tently observed the presence of �Gal-specific serum IgM Abs inM86VH2VL2GT� mice. This unanticipated finding prompted us toexamine the binding characteristics of �Gal specific Abs in theseanimals. Interestingly, although we were readily able to detectbinding of �Gal-specific IgM from these mice when using ELISAplates coated with �Gal-BSA conjugates in which the molar ratioof �Gal to BSA molecules was high, the ability to detect bindingof �Gal in these mice was significantly reduced when ELISAplates were coated with �Gal-BSA conjugates in which the molarratio of �Gal to BSA was relatively low. These data suggest that�Gal-specific Abs in M86VH2VL2GT� mice have a lower affinityfor �Gal than that observed in GT0 mice. Such low affinity Absappear to require an increase in binding avidity to be detected,which can be achieved by using BSA molecules that are highlysubstituted with �Gal epitopes. Why then would such Abs developin mice that express �Gal? Although this issue is under study, it isimportant to point out that production of �Gal-specific Abs areproduced only in mice containing two copies of the knocked in Hchain allele and either one or two copies of the knocked in L chainallele. In M86 mice, the knocked in H chain allele is rearranged toJh4, which prevents receptor editing but not V gene replacement.We suggest that in the context of multiple copies of the knocked inH chain, V gene replacement via homologous recombination mayselect for VH genes of similar sequences to the M86VH region,which can encode H chain variable regions that bind �Gal with anaffinity that is sufficiently low to allow for their development inGT� mice. We are currently evaluating this hypothesis.
The development of M86 Ig knock-in mice provides us with aunique model to study the development of B cells producing anti-carbohydrate Abs. The initial description of these mice has pro-vided a novel insight into mechanisms of tolerance for B cell pro-ducing anti-carbohydrate Abs. Importantly, unanticipated findingsresulting from this work may aid our understanding of self-non-self recognition. Although M86VH2VL2GT� mice producing�Gal-specific Abs appear to be grossly healthy, the presence of�Gal-specific Abs in these mice may provide us with a model toexamine how autoimmunity can be precipitated and lead topathology.
AcknowledgmentsWe thank Uri Galili, (University of Massachusetts) for providing the M86hybridoma, members of the Dr. Iacomini laboratory for helpful discus-sions, and John Kearney (University of Alabama, Birmingham) for instruc-tion in the staining of spleen sections.
DisclosuresThe authors have no financial conflict of interest.
References1. Briles, D. E., M. Nahm, K. Schroer, J. Davie, P. Baker, J. Kearney, and
R. Barletta. 1981. Antiphosphocholine antibodies found in normal mouse serumare protective against intravenous infection with type 3 Streptococcus pneu-moniae. J. Exp. Med. 153: 694–705.
2. Casali, P., and E. W. Schettino. 1996. Structure and function of natural antibod-ies. Curr. Top. Microbiol. Immunol. 210: 167–179.
3. Haury, M., A. Sundblad, A. Grandien, C. Barreau, A. Coutinho, and A. Nobrega.1997. The repertoire of serum IgM in normal mice is largely independent ofexternal antigenic contact. Eur. J. Immunol. 27: 1557–1563.
4. Cooper, D. K., E. Koren, and R. Oriol. 1994. Oligosaccharides and discordantxenotransplantation. Immunol. Rev. 141: 31–58.
5. Galili, U. 1993. Evolution and pathophysiology of the human natural anti-�-galactosyl IgG (anti-Gal) antibody. Springer Semin. Immunopathol. 15: 155–171.
6. Galili, U., F. Anaraki, A. Thall, C. Hill-Black, and M. Radic. 1993. One percentof human circulating B lymphocytes are capable of producing the natural anti-Galantibody. Blood 82: 2485–2493.
7. Good, A. H., D. K. Cooper, A. J. Malcolm, R. M. Ippolito, E. Koren,F. A. Neethling, Y. Ye, N. Zuhdi, and L. R. Lamontagne. 1992. Identification ofcarbohydrate structures that bind human antiporcine antibodies: implications fordiscordant xenografting in humans. Transplant. Proc. 24: 559–562.
8. Oriol, R., Y. Ye, E. Koren, and D. K. Cooper. 1993. Carbohydrate antigens of pigtissues reacting with human natural antibodies as potential targets for hyperacutevascular rejection in pig-to-man organ xenotransplantation. Transplantation 56:1433–1442.
9. Sandrin, M. S., and I. F. McKenzie. 1994. Gal�(1,3)Gal, the major xenoanti-gen(s) recognised in pigs by human natural antibodies. Immunol. Rev. 141:169–190.
10. Sandrin, M. S., H. A. Vaughan, P. L. Dabkowski, and I. F. McKenzie. 1993.Anti-pig IgM antibodies in human serum react predominantly with Gal(�1–3)Galepitopes. Proc. Natl. Acad. Sci. USA 90: 11391–11395.
11. Galili, U., R. E. Mandrell, R. M. Hamadeh, S. B. Shohet, and J. M. Griffiss. 1988.Interaction between human natural anti-�-galactosyl immunoglobulin G and bac-teria of the human flora. Infect. Immun. 56: 1730–1737.
12. Galili, U., S. B. Shohet, E. Kobrin, C. L. Stults, and B. A. Macher. 1988. Man,apes, and Old World monkeys differ from other mammals in the expression of�-galactosyl epitopes on nucleated cells. J. Biol. Chem. 263: 17755–17762.
13. McMorrow, I. M., C. A. Comrack, P. P. Nazarey, D. H. Sachs, andH. DerSimonian. 1997. Relationship between ABO blood group and levels ofGal�,3galactose-reactive human immunoglobulin G. Transplantation 64:546–549.
14. Wang, L., M. Z. Radic, and U. Galili. 1995. Human anti-Gal heavy chain genes:preferential use of VH3 and the presence of somatic mutations. J. Immunol. 155:1276–1285.
15. Hamadeh, R. M., G. A. Jarvis, U. Galili, R. E. Mandrell, P. Zhou, andJ. M. Griffiss. 1992. Human natural anti-Gal IgG regulates alternative comple-ment pathway activation on bacterial surfaces. J. Clin. Invest. 89: 1223–1235.
16. Takeuchi, Y., C. D. Porter, K. M. Strahan, A. F. Preece, K. Gustafsson,F. L. Cosset, R. A. Weiss, and M. K. Collins. 1996. Sensitization of cells andretroviruses to human serum by (�1–3) galactosyltransferase. Nature 379: 85–88.
17. Welsh, R. M., C. L. O’Donnell, D. J. Reed, and R. P. Rother. 1998. Evaluationof the Gal�1–3Gal epitope as a host modification factor eliciting natural humoralimmunity to enveloped viruses. J. Virol. 72: 4650–4656.
18. Bach, F. H., S. C. Robson, H. Winkler, C. Ferran, K. M. Stuhlmeier,C. J. Wrighton, and W. W. Hancock. 1995. Barriers to xenotransplantation. Nat.Med. 1: 869–873.
19. Galili, U., D. C. LaTemple, A. W. Walgenbach, and K. R. Stone. 1997. Porcineand bovine cartilage transplants in cynomolgus monkey: II. Changes in anti-Galresponse during chronic rejection. Transplantation 63: 646–651.
20. Lin, S. S., D. L. Kooyman, L. J. Daniels, C. W. Daggett, W. Parker, J. H. Lawson,C. W. Hoopes, C. Gullotto, L. Li, P. Birch, et al. 1997. The role of naturalanti-Gal�1–3Gal antibodies in hyperacute rejection of pig-to-baboon cardiac xe-notransplants. Transpl. Immunol. 5: 212–218.
21. Platt, J. L. 1997. Approaching the clinical application of xenotransplantation.Am. J. Med. Sci. 313: 315–321.
22. Stone, K. R., A. W. Walgenbach, J. T. Abrams, J. Nelson, N. Gillett, andU. Galili. 1997. Porcine and bovine cartilage transplants in cynomolgus monkey:I. A model for chronic xenograft rejection. Transplantation 63: 640–645.
23. Tearle, R. G., M. J. Tange, Z. L. Zannettino, M. Katerelos, T. A. Shinkel,B. J. Van Denderen, A. J. Lonie, I. Lyons, M. B. Nottle, T. Cox, et al. 1996. The�-1,3-galactosyltransferase knockout mouse: implications for xenotransplanta-tion. Transplantation 61: 13–19.
24. Thall, A. D., P. Maly, and J. B. Lowe. 1995. Oocyte Gal�1,3Gal epitopes im-plicated in sperm adhesion to the zona pellucida glycoprotein ZP3 are not re-quired for fertilization in the mouse. J. Biol. Chem. 270: 21437–21440.
25. Thall, A. D., H. S. Murphy, and J. B. Lowe. 1996. �1,3-Galactosyltransferase-deficient mice produce naturally occurring cytotoxic anti-Gal antibodies. Trans-plant. Proc. 28: 556–557.
26. Chen, Z. C., and U. Galili. 2000. Genes coding for anti-Gal in knock-out mice forthe �1,3galactosyltransferase gene: analysis by hybridomas. Transplant. Proc.32: 846–847.
27. Chen, Z. C., M. Z. Radic, and U. Galili. 2000. Genes coding evolutionary novelanti-carbohydrate antibodies: studies on anti-Gal production in �1,3galactosyl-transferase knock out mice. Mol. Immunol. 37: 455–466.
28. McKenzie, I. F., Y. Q. Li, K. Patton, A. D. Thall, and M. S. Sandrin. 1998. Amurine model of antibody-mediated hyperacute rejection by galactose-�(1,3)ga-lactose antibodies in Gal0/0 mice. Transplantation 66: 754–763.
29. Nozawa, S., P. X. Xing, G. D. Wu, E. Gochi, M. Kearns-Jonker, J. Swensson,V. A. Starnes, M. S. Sandrin, I. F. McKenzie, and D. V. Cramer. 2001. Charac-teristics of immunoglobulin gene usage of the xenoantibody binding to gal-�(1,3)gal target antigens in the gal knockout mouse. Transplantation 72:147–155.
30. Pearse, M. J., E. Witort, P. Mottram, W. Han, L. Murray-Segal, M. Romanella,E. Salvaris, T. A. Shinkel, D. J. Goodman, and A. J. d’Apice. 1998. Anti-Galantibody-mediated allograft rejection in �1,3-galactosyltransferase gene knock-out mice: a model of delayed xenograft rejection. Transplantation 66: 748–754.
31. Xu, H., A. Sharma, L. Chen, C. Harrison, Y. Wei, A. S. Chong, J. S. Logan, andG. W. Byrne. 2001. The structure of anti-Gal immunoglobulin genes in naive andstimulated Gal knockout mice. Transplantation 72: 1817–1825.
32. Cretin, N., and J. Iacomini. 2002. Immunoglobulin heavy chain transgenic miceexpressing Gal�(1,3)Gal-reactive antibodies. Transplantation 73: 1558–1564.
33. Xu, H., A. Sharma, Y. Lei, J. Okabe, H. Wan, A. S. Chong, J. S. Logan, andG. W. Byrne. 2002. Development and characterization of anti-Gal B cell receptortransgenic Gal�/� mice. Transplantation 73: 1549–1557.
34. Galili, U., D. C. LaTemple, and M. Z. Radic. 1998. A sensitive assay for mea-suring �-Gal epitope expression on cells by a monoclonal anti-Gal antibody.Transplantation 65: 1129–1132.
35. Bracy, J. L., and J. Iacomini. 2000. Induction of B-cell tolerance by retroviralgene therapy. Blood 96: 3008–3015.
36. Sonoda, E., Y. Pewzner-Jung, S. Schwers, S. Taki, S. Jung, D. Eilat, andK. Rajewsky. 1997. B cell development under the condition of allelic inclusion.Immunity 6: 225–233.
37. Pelanda, R., S. Schaal, R. M. Torres, and K. Rajewsky. 1996. A prematurelyexpressed Ig� transgene, but not V�J� gene segment targeted into the Ig� locus,can rescue B cell development in �5-deficient mice. Immunity 5: 229–239.
38. Mombaerts, P., A. R. Clarke, M. A. Rudnicki, J. Iacomini, S. Itohara,J. J. Lafaille, L. Wang, Y. Ichikawa, R. Jaenisch, M. L. Hooper, et al. 1992.Mutations in T-cell antigen receptor genes � and � block thymocyte developmentat different stages. Nature 360: 225–231.
39. Wood, C., E. A. Kabat, L. A. Murphy, and I. J. Goldstein. 1979. Immunochemi-cal studies of the combining sites of the two isolectins, A4 and B4, isolated fromBandeiraea simplicifolia. Arch. Biochem. Biophys. 198: 1–11.
40. Schuppel, R., J. Wilke, and E. Weiler. 1987. Monoclonal anti-allotype antibodytowards BALB/c IgM. Analysis of specificity and site of a V-C crossover inrecombinant strain BALB-Igh-Va/Igh-Cb. Eur. J. Immunol. 17: 739–741.
41. Nishikawa, S., Y. Sasaki, T. Kina, T. Amagai, and Y. Katsura. 1986. A mono-clonal antibody against Igh6–4 determinant. Immunogenetics 23: 137–139.
42. Cretin, N., J. Bracy, K. Hanson, and J. Iacomini. 2002. The role of T cell help inthe production of antibodies specific for Gal�1–3Gal. J. Immunol. 168:1479–1483.
43. Melamed, D., R. J. Benschop, J. C. Cambier, and D. Nemazee. 1998. Develop-mental regulation of B lymphocyte immune tolerance compartmentalizes clonalselection from receptor selection. Cell 92: 173–182.
44. Melamed, D., and D. Nemazee. 1997. Self-antigen does not accelerate immatureB cell apoptosis, but stimulates receptor editing as a consequence of develop-mental arrest. Proc. Natl. Acad. Sci. USA 94: 9267–9272.
45. Tiegs, S. L., D. M. Russell, and D. Nemazee. 1993. Receptor editing in self-reactive bone marrow B cells. J. Exp. Med. 177: 1009–1020.
46. Ohdan, H., K. G. Swenson, H. S. Kruger Gray, Y. G. Yang, Y. Xu, A. D. Thall,and M. Sykes. 2000. Mac-1-negative B-1b phenotype of natural antibody-pro-ducing cells, including those responding to Gal�1,3Gal epitopes in �1,3-galac-tosyltransferase-deficient mice. J. Immunol. 165: 5518–5529.
47. Ohdan, H., Y. G. Yang, K. G. Swenson, A. D. Thall, and M. Sykes. 2000. In vivoT-cell depletion enhances production of anti-GAL�1,3GAL natural antibodies in�1,3-galactosyltransferase-deficient mice. Transplantation 69: 910–913.
48. Hardy, R. R., C. E. Carmack, S. A. Shinton, J. D. Kemp, and K. Hayakawa. 1991.Resolution and characterization of pro-B and pre-pro-B cell stages in normalmouse bone marrow. J. Exp. Med. 173: 1213–1225.
49. Pillai, S. 1999. The chosen few? Positive selection and the generation of naive Blymphocytes. Immunity 10: 493–502.
50. Chen, C., E. L. Prak, and M. Weigert. 1997. Editing disease-associated autoan-tibodies. Immunity 6: 97–105.
51. Gay, D., T. Saunders, S. Camper, and M. Weigert. 1993. Receptor editing: anapproach by autoreactive B cells to escape tolerance. J. Exp. Med. 177:999–1008.
52. Radic, M. Z., J. Erikson, S. Litwin, and M. Weigert. 1993. B lymphocytes mayescape tolerance by revising their antigen receptors. J. Exp. Med. 177:1165–1173.
53. Cariappa, A., and S. Pillai. 2002. Antigen-dependent B-cell development. Curr.Opin. Immunol. 14: 241–249.
54. Amano, M., N. Baumgarth, M. D. Dick, L. Brossay, M. Kronenberg,L. A. Herzenberg, and S. Strober. 1998. CD1 expression defines subsets of fol-licular and marginal zone B cells in the spleen: �2-microglobulin-dependent andindependent forms. J. Immunol. 161: 1710–1717.
55. Roark, J. H., S. H. Park, J. Jayawardena, U. Kavita, M. Shannon, andA. Bendelac. 1998. CD1.1 expression by mouse antigen-presenting cells andmarginal zone B cells. J. Immunol. 160: 3121–3127.
56. Nussenzweig, M. C. 1998. Immune receptor editing: revise and select. Cell 95:875–878.
57. Chen, C., Z. Nagy, E. L. Prak, and M. Weigert. 1995. Immunoglobulin heavychain gene replacement: a mechanism of receptor editing. Immunity 3: 747–755.
58. Hartley, S. B., J. Crosbie, R. Brink, A. B. Kantor, A. Basten, and C. C. Goodnow.1991. Elimination from peripheral lymphoid tissues of self-reactive B lympho-cytes recognizing membrane-bound antigens. Nature 353: 765–769.
59. Nemazee, D., and K. Buerki. 1989. Clonal deletion of autoreactive B lympho-cytes in bone marrow chimeras. Proc. Natl. Acad. Sci. USA 86: 8039–8043.
60. Nemazee, D. A., and K. Burki. 1989. Clonal deletion of B lymphocytes in atransgenic mouse bearing anti-MHC class I antibody genes. Nature 337:562–566.
61. Russell, D. M., Z. Dembic, G. Morahan, J. F. Miller, K. Burki, and D. Nemazee.1991. Peripheral deletion of self-reactive B cells. Nature 354: 308–311.
62. Fulcher, D. A., and A. Basten. 1994. Whither the anergic B-cell? Autoimmunity19: 135–140.
63. Goodnow, C. C., J. Crosbie, S. Adelstein, T. B. Lavoie, S. J. Smith-Gill,R. A. Brink, H. Pritchard-Briscoe, J. S. Wotherspoon, R. H. Loblay, K. Raphael,et al. 1988. Altered immunoglobulin expression and functional silencing of self-reactive B lymphocytes in transgenic mice. Nature 334: 676–682.
64. Lang, J., M. Jackson, L. Teyton, A. Brunmark, K. Kane, and D. Nemazee. 1996.B cells are exquisitely sensitive to central tolerance and receptor editing inducedby ultralow affinity, membrane-bound antigen. J. Exp. Med. 184: 1685–1697.
