IgH V-Region Sequence Does Not Predict the Survival Fate of Human Germinal Center B Cells
http://www.100md.com
免疫学杂志 2005年第5期
Abstract
Germinal center (GC) B cell survival fate is governed in part by the outcome of successful/failed BCR-mediated interactions with accessory cells. However, the extent to which the BCR primary sequence influences such interactions is not fully understood. Over 1000 IgVH4 family cDNAs were sequenced from living (annexin V–) and apoptotic (annexin V+ or from within tingible body macrophages) GC B cell fractions from seven tonsils. Results surprisingly demonstrate that living and dying GC B cells do not significantly differ in IgVH, D, or JH gene segment use; HCDR3 length or positive charge; or mutation frequency. Additionally, equivalent IgH cDNA sequences were identified in both fractions, suggesting that BCR sequence alone is an unreliable predictor of GC B cell survival.
Introduction
The development of an efficiently functioning humoral immune system is one of the most important factors in survival among vertebrates. It is quite paradoxical that such a well-coordinated system is established and maintained by processes that incorporate a high degree of randomness and chance. Among B cells, these processes include, but are not limited to, Ig gene rearrangement in the bone marrow (1, 2, 3) and somatic hypermutation-based affinity maturation within germinal centers (GC)3 (4, 5, 6, 7). Multiple rounds of receptor-mediated selection during various stages of cell development, the survivors of which differentiate into key effector B cell subsets, buffer the potential detriments of randomness.
To explore various issues of selection, investigations have focused on multiple aspects of B cell selection within the bone marrow, and secondary lymphoid tissues such as tonsils. Selection within the bone marrow is 2-fold, including positive selection of B cells that successfully rearrange and express components of the BCR, and negative selection against those expressing BCRs that bind to self Ag with inappropriately high avidity (autoreactivity) (8, 9, 10).
Selection in peripheral sites, including GC, is based upon a B cell’s ability to not only bind foreign Ag, but also to compete in a Darwinian fashion for limited opportunistic binding with accessory cells, including follicular dendritic cells (FDC) and T cells (6, 11, 12, 13, 14, 15, 16, 17, 18). These cells deliver critical life-promoting secondary signals that are essential for continued B cell survival (19, 20, 21, 22, 23, 24, 25). Both processes are BCR dependent. Consequently, the current paradigm for GC B cell survival is primarily focused on the BCR itself. B cells bearing receptors that attain greater affinity for Ag are selected and proceed to further stages of development. Conversely, those with lower affinity, frameshifts, nonsense codons, and other factors that negatively influence Ag binding are presumably destined to die via apoptosis. In this study, we test this hypothesis directly by comparing quantifiable differences in the primary sequences of BCR H chain V regions between living B cells and those that are marked for death.
A hallmark of the GC reaction is the high frequency of cell death, with as few as 5% of GC B cells surviving to the effector cell stage in some instances (26, 27, 28). The basis of such high death rates in the GC is not fully understood. Centroblasts and centrocytes are Bcl-2 negative and are prone to die, presumably by default (29, 30, 31, 32, 33). In addition, recent investigations demonstrate that both Ag recognition and cell:cell communications are needed for persistent survival within the GC (32, 34, 35, 36, 37). Therefore, the level of cell death is likely to be influenced by factors that deleteriously affect either requirement.
Apoptosis occurs in multiple stages (reviewed in Ref. 38). Loss of cell membrane asymmetry is an early event, resulting in atypical phosphatidylserine (PS) exposure on the exoplasmic leaflet. Externalized PS on GC B cell surfaces is physiologically relevant, serving as the targeted ligand during receptor-mediated phagocytosis by tingible body macrophages (TBMac). Similarly, annexin V (36 kDa) binds PS on surfaces of dying cells and is routinely used as a reliable marker of apoptosis for various cell types, including GC B cells (39, 40, 41, 42). The pattern of Annexin V (AnV) binding has been previously shown to correlate with the percentage of GC B cells undergoing apoptosis (43, 44).
AnV+ GC B cells exhibit characteristics consistent with cells undergoing apoptosis, including down-regulated expression of various proteins associated with B cell survival (CD21, CD11a, CD49d, and CD54) (43). During in vitro culture, AnV+ B cells form membrane blebs, readily degrade DNA, and test positive for active caspase 3, all of which are significantly blocked by the addition of caspase inhibitors (44).
To determine the relative roles that either receptor modification or competition play in influencing cell death within the GC, we performed molecular IgH repertoire analyses using GC B cell subsets, including centroblasts and centrocytes, that were separated into AnV+ and AnV– fractions. Necrotic cells (AnV+/propidium iodide (PI)+/7-aminoactinomycin D+ (7-AAD+)) were excluded. Despite a substantial effort to document consistent differences between BCRs of these subsets, our results show that AnV+ and AnV– fractions are surprisingly similar, showing no significant difference in IgVH, D, or JH gene segment use; HCDR3 length or composition; or mutation pattern and frequency. Surprisingly, identical IgH V region amino acid sequences were routinely identified in both fractions. In fact, we obtained the same results when IgH sequences were analyzed from dying B cells that were previously engulfed by TBMac in situ, which therefore represent a physiologically germane category of apoptotic B lymphocytes. These data suggest that BCR sequence (H chain) is unlikely to be the principally decisive determinant in predicting the survival fate of GC B cells. As such, these results challenge the prevailing paradigm and suggest that Darwinian fitness must be defined by factors other than receptor quality alone.
Materials and Methods
Isolation of GC B cells
Tonsillar B cells were isolated, as previously described (4, 45). All protocols are Institutional Review Board approved. Briefly, children’s tonsils were collected during routine tonsillectomies and processed immediately on ice. Tissue was separated into single cell suspensions using the Collector tissue sieve (ThermoEC) and resuspended in FCS-supplemented (10%) RPMI 1640 (Invitrogen Life Technologies) or MACS buffer. Total mononuclear cells were collected following 20-min centrifugation in Lymphoprep (Axis-Shield), and washed once in MACS buffer. TBMac were directly isolated from this mononuclear fraction. B cells were enriched by magnetic negative depletion of non-B cells using a MACS B cell isolation kit (Miltenyi Biotec), according to the manufacturer’s instructions. Cells were immediately counted and stained for subsequent cytometric analysis. Special care was taken to minimize the time from tonsillectomy to cell separation so as to limit the extent of cell death during processing.
Flow cytometry and cell sorting
Tonsillar B cells were stained for distinct GC B cell subsets using strategies described earlier (4). Incubations with the appropriate concentrations of conjugated Abs were performed for 15 min at 4°C. Incubations with FITC-labeled AnV (Caltag Laboratories), PI, or 7-AAD (BD Pharmingen) were at room temperature for 20 min. Briefly, cells (1–3 x 106) were stained using allophycocyanin-labeled CD38 (Caltag Laboratories). GC B cells were identified as CD38+. PE-labeled anti-CD77 (BD Pharmingen) staining was used to further subdivide GC B cells into centroblast (CD38+CD77+) and centrocyte (CD38+CD77–) populations. TBMac were isolated from total mononuclear cells following staining with anti-CD68 PE, in addition to anti-CD19 FITC and anti-CD38 allophycocyanin (all from Caltag Laboratories) to exclude potential B cell contamination. The TBMac fraction collected for further investigation was CD68+CD19–CD38–. Individual cell populations were collected following multicolor cell sorting using a MoFlo flow cytometer (
DakoCytomation), or cytometrically analyzed on a FACSCalibur instrument (BD Biosciences). Necrotic cells (AnV+PI+/7-AAD+) were excluded in all studies. In several control experiments, patterns of coincident staining using fluorometric AnV, caspase 3 (BD Pharmingen), DiOC6 (Sigma-Aldrich), and 7-AAD cell death markers were assessed using apoptosis-induced CLO1 or Jurkat cell lines treated with 10 Gy gamma-irradiation or A23187 (Alexis Biochemicals), respectively.
Cloning and sequencing of IgVH4 genes
RNA from B cells collected during cell sorting was isolated using RNAwiz (Ambion). VH4 family genes were amplified from two independent tonsils (T2, T6) via RT-PCR using family-specific and isotype-specific primers, as described previously (45, 46, 47). In a separate analysis, VH4–34 sequences were specifically amplified from multiple sources using a VH4–34-specific 5' primer (5'-AGCTACAGCAGTGGGGCG-3') and the same 3' IgG primer as before. Sources included centroblast and centrocyte subsets from three tonsils (T7, T9, and T11) and phagocytized B cells (within TBMac) from two tonsils (T40 and T41). IgM and IgG products were cloned using the Qiagen One-Step RT-PCR cloning system, according to manufacturer’s instructions. Purified plasmids were isolated (Qiaprep Spin system; Qiagen) from transformed Escherichia coli (DH5) bacteria and were subsequently sequenced with an ABI 3730 DNA sequencer.
DNA sequence and statistical analysis
Germline VH4 family gene segment use among sequences was identified using National Center for Biotechnology Information IgBlast (www.ncbi.nlm.nih.gov/igblast/) and IMGT/V-QUEST (http://imgt.cines.fr:8104) databases. Sequence comparison and alignments were performed with VectorNTI and a specialized software program based on a Microsoft Visual Basic v6.0 platform designed by P. Wilson of the Oklahoma Medical Research Foundation. Statistical analyses were also done using this program, in addition to SYSTAT and Microsoft Excel for Windows.
Results
Identification and isolation of AnV+ and AnV– GC B cells
Isolation of apoptotic GC B cells was performed using two complementary strategies. To isolate B cells that are unequivocally undergoing apoptosis, we directly collected CD68+ TBMac that simultaneously stained negative for extracellular B cell-specific markers CD19 and CD38. This approach excluded TBMac that may have recently initiated phagocytosis, i.e., those with partially engulfed B cells whose exposed surfaces would most likely contain B cell-specific markers generating a triple-positive CD68+CD19+CD38+ TBMac phenotype. Although nucleic acids begin to degrade following engulfment by TBMac, purified B cell RNA and DNA integrity remained sufficient, evidenced by the successful amplification of rearranged IgH cDNA sequences (despite the absence of CD19 and CD38 among the sorted TBMac population).
Macrophage recognition (phagocytosis) of apoptotic cells and AnV binding are both PS-dependent processes. In a second approach designed to permit the isolation of intact, distinct GC B cell subsets in the earliest stages of apoptosis (impractical from within the lytic environment of TBMac), fluorometric AnV was used to bind externalized PS. Tonsillar B cells were stained with Abs to CD38 and IgD to identify GC B cells (Fig. 1A). For some experiments, anti-CD77 was used to further subdivide GC B cells into centroblast (CD38+CD77+, Bm3) and centrocyte (CD38+CD77–, Bm4) populations (Fig. 1B). In both cases, distribution patterns based on cell surface phenotypes were consistent with earlier reports (4, 48).
FIGURE 1. Cytometric analysis of GC B cells and corresponding AnV binding. Panels represent one of five independent experiments. A, Cell surface phenotype of tonsillar B cells after staining with anti-CD38 and anti-IgD. B, Costaining of GC B cells (different patient from A) with anti-CD38 and anti-CD77 to identify centroblasts (CD38+CD77+) and centrocytes (CD38+CD77–). AnV staining was determined among total GC B cells (C) or centroblasts and centrocytes (D). Necrotic cells (AnV+7-AAD+) were excluded.
The reproducible identification of dying GC B cells using AnV as the primary cell death marker has been demonstrated previously in multiple investigations (43, 44), reporting values of <25% AnV positivity (AnV+) within the first few hours following surgery. Cytometric and FACS analyses were performed on cells that were stained with AnV immediately following passive B cell enrichment. This was done to maximize the accuracy of identifying AnV+ cells that are likely to have externalized PS in response to physiologically relevant factors within the in situ GC environment. Cytometric results from five tonsils were consistent with previous reports. For example, 5–24% of CD38+ GC B cells in our analyses were also AnV+, with an average of 11.7% (SD = 8.5%). The seemingly high SD is attributable to a single tonsil (24.3% AnV+, not used in subsequent analyses). The remaining tonsils were clustered around the average (example shown in Fig. 1C). To determine whether AnV staining was uniform among CD38+ GC B cells, we compared AnV+ levels in centroblasts vs centrocytes (Fig. 1D). In the example shown, there was nearly a 3-fold increase in AnV staining in centroblasts (13.3%) vs centrocytes (4.8%). This suggests that the two cell types are differentially susceptible to cell death, an observation consistent with previous reports (49). Together, these results document our ability to reproducibly isolate various subpopulations of human B cells that are marked for death.
IgH repertoire analysis
Persistent GC B cell survival and development to effector B cell stages is thought to be a BCR-dependent process. To determine the potential role that the BCR primary sequence plays in influencing survival fate, we performed repertoire analyses by sequencing over 1000 (1046) IgVH4 family members cloned from seven tonsils. Total tonsillar CD38+ GC B cells were sorted into AnV+ and AnV– pools and subjected to isotype-specific RT-PCR amplification of VH4 transcripts. IgM and IgG sequences from both pools were evaluated based on VH and JH gene segment use, HCDR3 length and positively charged amino acid composition, and mutation frequency. Clones that were identical at the cDNA level were counted once. Influences from potential clonal relatedness were negligible because exclusion of these sequences from statistical analyses did not significantly alter results. With very few exceptions, AnV+ and AnV– GC B cells are surprisingly indistinguishable based solely on the characteristics of their IgH chain primary sequences assessed in this study. In fact, identical H chain sequences were isolated from AnV+ and AnV– centroblast and centrocyte fractions with appreciable frequency (examples in Fig. 5 below). Similarly, in two independent experiments, identical IgVH4–34 sequences were observed in living and dying GC B cell fractions comparing AnV– CD38+ and phagocytized GC B cell counterparts, respectively. Together, these findings strongly suggest that BCR quality (at least H chain V region sequence) is insufficient to guarantee survival within the competitive GC environment.
FIGURE 5. Identical clones identified in AnV– and AnV+ GC B cell fractions. Identical clones were identified among AnV+ and AnV– fractions in centroblasts and centrocytes. Germline genes are indicated. Only codons with mutations are listed. Amino acid differences from germline are in bold type. Replacement and silent mutations are in capital or lower case, respectively. HCDR3 is in italic underline. The number of times a given sequence was isolated is in parentheses. Nomenclature: T9Bgn8 = tonsil 9, centroblast, IgG, AnV–, clone 8. T7Cmp29 = tonsil 7, centrocyte, IgM, AnV+, clone 29.
Equivalent VH gene segment use in AnV+ and AnV– fractions
IgVH4 gene expression frequencies in T2 and T6 GC IgMAnV+ and IgMAnV– fractions were quantitatively compared using the 2 test (heteroscedastic). For each gene segment analyzed, frequencies were statistically equivalent between IgMAnV+ and IgMAnV– fractions. Calculated p values exceeded 0.05 in each case. This includes VH4–59 in T2 (T2VH4–59), which despite a 2-fold difference had a p value of 0.28. VH4–34 was 58% of the population in T2 IgMAnV+ and IgMAnV– fractions (Fig. 2A). Similarly, VH4–34 is equivalently represented in T6 IgMAnV+ (32%) and IgMAnV– (30%) fractions, although the expression level was reduced compared with T2, demonstrating potential environmental influences on expression levels.
FIGURE 2. IgH repertoire analysis among AnV+ and AnV– fractions of total GC B cells. The distribution of VH4 family member gene segment use was determined for AnV+ (striped bars) and AnV– (solid bars) GC B cell fractions of IgM (A) and IgG (B) sequences. JH gene segment use was similarly determined for the same IgM (C) and IgG (D) sequences.
Similar to IgM, VH4 expression frequencies in GC B cell IgGAnV+ and IgGAnV– fractions were statistically indifferent between all gene segments compared (p > 0.05) (Fig. 2B).
The hierarchical pattern of gene expression differed between IgM and IgG pools, which may represent preferential selection (including counterselection) of certain gene segments or IgH rearrangements during the GC reaction. For example, VH4–34 expression patterns in both tonsils were significantly reduced among IgG sequences, despite being the most frequently isolated gene segment among IgM sequences. Conversely, T2VH4–59 and T6VH4–61 frequencies are increased in the IgG pool relative to their IgM counterparts (Fig. 2B). In addition, elevated T2VH4–59 and T6VH4–61 frequencies in the IgGAnV– fraction were coincident with elevated representation in the IgGAnV+ fraction. This suggests that frequency of expression of a given VH4 gene segment in the AnV– fraction directly influences the level of expression of the same gene segment in the corresponding AnV+ fraction. Therefore, the potential selection of B cells with BCRs possessing beneficial qualities (such as increased affinity) does not necessarily confer protection against becoming AnV+.
Similarities in expression frequencies between AnV+ and AnV– fractions are unlikely to be the result of primer bias. VH4 frequencies varied between tonsils, yet were consistent for independent amplifications from a given GC fraction. In addition, expression frequencies are significantly different between IgM and IgG pools, whose amplification strategies differed only in the use of downstream isotype primer.
Equivalent JH gene segment use in AnV+ and AnV– fractions
We analyzed the frequency of JH gene segment use via the same strategy as above. Similar to results from VH gene segment analyses, there is no statistical difference in JH gene segment use (p > 0.05) between GC AnV+ and AnV– fractions for either isotype. JH4 was the predominant gene segment in IgMAnV– and IgGAnV– fractions for both tonsils, accounting for nearly 50% of the rearrangements (Fig. 2, C and D). JH4 was equivalently represented in the corresponding IgMAnV+ and IgGAnV+ fractions, again suggesting that the frequency of gene segment use among AnV+ cells is strongly influenced by the expression frequency in the AnV– pool.
Elevated JH6 gene segment use is commonly observed among B cells with autoreactive or structurally aberrant BCRs, which undergo receptor revision presumably as a means of escaping death. Surprisingly, we did not observe a significant JH6 bias among AnV+ B cells in either isotype. Therefore, the propensity to externalize PS (AnV+) and subsequently become targeted for phagocytosis does not directly correspond with JH gene segment use.
Long, positively charged HCDR3s are not enriched among GC B cell AnV+ fractions
Negative selection against long, positively charged HCDR3s has been demonstrated in multiple murine and human autoimmune models (50, 51, 52, 53). The majority of nonautoreactive human HCDR3s typically average between 14 and 18 aa in length. We therefore compared HCDR3s in GC AnV– and AnV+ B cell pools to assess whether positive charge and nonstandard lengths are enriched in the latter fraction. For both tonsils, average HCDR3 lengths in all cases ranged between 14 and 16 aa. Values were statistically equivalent (p > 0.05) between AnV– and AnV+ counterparts in IgM and IgG pools (data not shown).
The distribution of HCDR3 lengths in AnV+ and AnV– fractions is presented in Fig. 3, A–F (grouped in three amino acid units). Consistent with earlier reports, >50% of HCDR3s in all subsets were present in the 14–18 (±1) range (51). In general, correlation between AnV+ and AnV– fractions is more coincident among IgG sequences. In all cases, p values exceed 0.05. Nonetheless, wider disparity was observed among IgM sequences, particularly in the T2 18- to 20-aa (T218–20aa) and T615–17aa ranges.
FIGURE 3. Population distribution of HCDR3s within a given HCDR3 length range. HCDR3 lengths were grouped into ranges of 3 aa. The percentage of clones within each range is presented for AnV– (filled line) and AnV+ (dashed line) fractions of IgM (A, C, and E) and IgG sequences (B, D, and F). For increased sample size, corresponding subgroups were combined from T2 and T6. Combined results are presented for IgM (E) and IgG (F) sequences. Note: differences between AnV– and AnV+ fractions for T2IgM (15 16 17 ), T2IgM (18 19 20 ), and T6IgM (15 16 17 ) approached significance, yet each p value exceeded 0.05. Differences for combined T2 + T6IgM (15 16 17 ) were significant (p < 0.05).
To determine whether AnV+ B cell fractions are enriched for potential signs of HCDR3-related autoreactivity, we quantified the number of in-frame, positively charged amino acids (R, K, and H) in this region (excluding the germline-encoded arginine at the 3' end of framework 3). Results are presented in Fig. 4A. Between 70 and 80% of all clones have one or no positive residue, independent of isotype and AnV staining pattern. In fact, the frequency of IgM or IgG clones with three or more positive HCDR3 residues does not exceed 10% in any AnV fraction, strongly suggesting that there is no bias toward positive charge among AnV+ B cells. Coupled with the observation that longer HCDR3s are not enriched among AnV+ B cells, HCDR3 primary structure alone appears to have minimal influence on predicting cell survival fate.
FIGURE 4. Percentage of clones with positively charged amino acids. The percentage of clones with a given number of positively charged R, K, and/or H residues within their HCDR3 is graphically represented for IgM (A) and IgG (B) sequences. AnV+ and AnV– clones are represented by dashed and solid lines, respectively.
Somatic hypermutation frequencies are identical in the living and dying pools of GC B cells
The frequency of somatic hypermutation was determined for all AnV– and AnV+ fractions from T2 and T6. Results are presented in Table I. In a broad-spectrum analysis based only on shared isotype, higher mutation frequencies were not consistently biased toward either AnV fraction. For example, mutation frequencies were higher among T2 IgMAnV+ cells (6.1% ± 4.6) relative to T2 IgMAnV– (3.8 ± 2.3). Calculated differences were statistically significant (p < 0.05). However, the pattern was reversed in T6 (4.5 ± 6.0 vs 6.7 ± 6.5, respectively), demonstrating that AnV+ cells do not inherently contain more mutations. In fact, differences between AnV+ and AnV– fractions in T6IgM, T2IgG, and T6IgG cells were not statistically significant (p > 0.05) (Table I). Consistent with previous reports, IgG sequences typically had significantly higher mutation frequencies (p < 0.01) when compared with IgM sequences from the same tonsil (Table I).
Table I. Mutation frequency among AnV– and AnV+ isotypic fractionsa
SDs associated with calculated mutation frequencies exceeded 70% in most cases, indicating that groups contained sequences with a widely variable number of mutations. It is plausible that sequences with more mutations could belong to a particular subgroup of sequences that could potentially be further categorized by shared characteristics. We therefore determined mutation frequencies among isotype-specific AnV– and AnV+ sequences that were additionally subdivided by shared use of VH gene segment, JH gene segment, or HCDR3 length. Surprisingly, 85% (29/34) of AnV+ vs AnV– comparisons made in this more focused analysis showed no significant difference in mutation frequency (data not shown). Of the five that did, three were in the T2IgM pool (VH4–34, JH6, and HCDR3 length 9–11 aa) in which significant differences were previously observed in the isotype-only calculation. However, a consistent pattern was not evident because none of the five was duplicated in both tonsils. Therefore, in the overwhelming majority of cases, AnV– and AnV+ fractions do not differ in mutation frequency, although this deserves further investigation (i.e., in systems in which mounted immune responses are elicited by a single known Ag).
IgVH regions with identical amino acid sequences exist in living and dying fractions of both centroblasts and centrocytes
As noted above, we observed no significant difference between AnV– and AnV+ fractions on the level of VH or JH gene segment use, HCDR3 length or charge, or somatic mutation load. This suggests that B cells bearing surface BCRs with equivalent characteristics are similarly prone to externalize PS (and therefore be engulfed by GC-resident TBMac). To test this hypothesis, we surveyed both AnV fractions for the presence of BCRs with identical amino acid sequences. To increase the chance of isolating clonally related sequences from a given pool, VH4 family members were amplified from CD38+CD77+ sorted centroblasts (clonally proliferating) or centrocytes (CD38+CD77–) using an upstream VH4–34-specific primer (a subset of sequences was also amplified with a VH4 universal primer).
Three separate tonsils were used in independent analyses (T7, T9, and T11). Clonality was determined based on shared VH, D, and JH gene segment use and identical VD and DJ joins. DNA sequences with accompanying amino acid changes are displayed in Fig. 5. In all three tonsils, clones with identically mutated amino acid sequences were successfully identified in both AnV fractions from centroblast (group 1) and centrocyte (groups 2–6) subsets. For example, four identical (amino acid level) centroblast sequences using VH4–31, DH4–17, and JH4 were isolated from the T7IgG pool. All four shared the same six replacement mutations. T7Bgn8 and T7Bgn14 are AnV–. T7Bgp5 and T7Bgp7 are AnV+. Note that T7Bgp5 and T7Bgp7 differ by a single silent mutation. The same observation was made using phagocytized B cells as the source of dying cells (described below). It is unlikely that reproducible sequence isolation was the by-product of potential RT-PCR bias, as most clonally related groups contained sequences that differed by one or more silent mutations, thereby indicating that their source was from distinct B cells. In fact, identical sequences (cDNA level) T7Cmn5 and T7C34p24 (group 2) were isolated following amplification with completely different 5' and 3' primer pairs (VH4:IgM and P434:IgG, respectively).
Evidence of clonal progression from centroblast to centrocyte was observed in group 6. Six identical amino acid sequences sharing three replacement mutations (S, T, and R) were isolated from the T9 centroblast AnV– fraction (i.e., T9B34n11). The same sequence was detected once in T9 centrocyte AnV– fraction and eight times in T9 centrocyte AnV+ fraction. The presence of silent mutations indicates that sequences were amplified from independent B cells. In this example, cells that successfully survive during the centroblast stage are able to further differentiate in centrocytes. However, continued centrocyte survival is not guaranteed because the same sequence was also identified in the centrocyte AnV+ fraction. Therefore, cell death within the GC occurs in both the centroblast and centrocyte stages of development, and persistent survival appears to require influences from additional factors at various time points.
IgG VH sequences from phagocytized GC B cells (tingible bodies) cannot be distinguished from the living B cell pool
An inherent disadvantage to performing apoptosis studies ex vivo is the difficulty in distinguishing between physiologically relevant cell death that was initiated in vivo, and that which unavoidably occurs following removal of any tissue from its nurturing environment. Despite having expediently processed each tonsil, there remains the possibility that a small fraction of the sorted GC B cells could represent cells that began to externalize PS following surgery. In a more stringent assessment of the potential correlation between the BCR primary sequence and survival fate in vivo, CD68+ TBMac were independently sorted from two tonsils (T40 and T41). Simultaneous staining with anti-CD19 and anti-CD38 (CD68+CD19–CD38–) permitted the cytometric exclusion of B cells. This includes those that would otherwise colocalize with TBMac due to having partially exposed surfaces stemming from their incomplete engulfment. Total RNA was immediately isolated to circumvent nucleic acid degradation, reverse transcribed, and amplified using VH4–34- and IgG-specific primers, as before. IgH cDNA sequences from engulfed apoptotic B cells were compared with those amplified from viable AnV–CD38+ GC B cells collected from their respective tonsils.
Results using this more direct assay for apoptotic cells (i.e., absolute dying vs prephagocytosis) were consistent with our previous observations based on AnV staining alone. As shown in Fig. 6, phagocytized and AnV– B cells did not significantly differ in JH gene use (JH4 majority in both fractions) or HCDR3 length (average length between 11 and 13 aa for each fraction). Comparable somatic hypermutation frequencies were observed between dying and living fractions from T41 (respectively, 5.5 ± 5.0 and 4.8 ± 5.4; p = 0.67; Table I). Calculations for T40 suggest that dying B cells mutate at a slightly higher frequency (respectively, 5.5 ± 3.5 and 3.6 ± 3.5; p = 0.05), although a difference of two substitutions is unlikely to have physiological significance, considering the wide range of mutations permissible (high SD). Both T40 fractions contained multiple sequences with 0 (germline) to 10 or more mutations. In fact, the IgH sequence with most mutations (17 substitutions) was present in the living (AnV–) fraction. Therefore, mutation load, JH gene segment use, and HCDR3 length each continue to be an unreliable gauge for predicting GC B cell survival fate, even among cells that cannot escape death due to engulfment and ultimate degradation within TBMac.
FIGURE 6. JH gene segment use and CDR3 length comparison between phagocytized and AnV– GC B cell fractions. VH4–34 sequences from T40 and T41 dying (MAC) and living (AnV–) fractions were compared based on JH gene segment use (A) and HCDR3 length (B), as before. Phagocytized (MAC) and AnV– fractions are indicated by and , respectively. The percentage of clones within each range is presented for AnV– (filled line) and phagocytized (dashed line) fractions.
Discussion
Factors influencing the differential survival fate of clonally related GC B cell progeny are not fully understood. However, the BCR has long been thought to play a central role in the initiation and maintenance of the GC reaction in response to T-dependent Ags (6, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25). For example, pre-GC B cells are activated by Ag-specific interactions between the BCR and Th cells (6). Consequently, all founder B cells that initially colonize GC have already successfully demonstrated BCR functional, structural, and Ag recognition competency, and are positively selected for survival. However, our findings suggest, as have others before, that benefits from this initial survival signal are most likely transient, periodically requiring subsequent survival signals during developmental progression in the GC. We observed identical IgH amino acid sequences in AnV– and AnV+ fractions of both centroblasts and centrocytes. It is therefore plausible that AnV+ (and phagocytized) GC B
cells represent those that received life-promoting signals at one time point, but failed to receive additional signals within the appropriate time window, leading to PS externalization and clearance by TBMac. Therefore, sequence alone is insufficient to predict survival fate at either centroblast or centrocyte stages, and does not guarantee the continued survival of clones bearing BCRs that otherwise appear functional at the H chain level. Potential influences from the L chain are currently under investigation.
PS exposure on dying GC B cells is physiologically relevant. In reports by Krahling et al. (42) and Callahan et al. (54), preincubation with AnV sufficiently inhibited PS recognition by macrophages, resulting in radically reduced levels of phagocytosis. Under these circumstances, apoptotic GC B cells were atypically associated with FDC and served as potential sources of self Ag and the onset of autoimmunity, demonstrating that continued survival of GC PS+ cells can be detrimental to the host. Therefore, PS exposure within the GC environment, and that observed by Dillon et al. (55, 56) on viable cells in the bone marrow, may not represent the same phenomenon. It is unlikely that viable GC B cells would externalize PS (even if transiently) because this would lead to their rapid clearance by proximally positioned TBMac (barring the theoretical possibility of viable cells coexpressing inhibitors of phagocytosis on their surfaces). In fact, continued survival of PS+ bone marrow B cells (nonphagocytized) may be a factor contributing to the relatively high degree of polyreactivity (self-recognition) among early B cells (57).
Nonetheless, annexin V has proven to be the most readily detectable marker of early apoptosis for GC B cells (42, 43). In our experiments, the percentage of tonsillar AnV+ GC B cells typically ranged between 5 and 15%. This is comparable to levels reported by Brieva and colleagues (43) and Groot and colleagues (21). Earlier investigations show that DNA degradation and caspase 3 activation generally do not occur until after B cells are engulfed by macrophages (58, 59), therefore rendering TUNEL assays and use of fluorometric caspase substrates less reliable for isolating early apoptotic, nonphagocytized GC B cells. Consistent with these reports, our efforts to sort intact GC B cells based on simultaneous staining of AnV and caspase 3 (active form) proved to be problematic (data not presented). However, gamma-irradiated GC-like CL01 cells (generously provided by P. Wilson) and A23187-treated JKT cells (E6-1) both successfully bound AnV in control experiments (data not shown). AnV+ control cells intercalated 7-AAD, successfully bound anticaspase 3 (active form), and showed signs of mitochondrial membrane breakdown (DiOC6 negative), indicative of their entry into the apoptosis pathway. In contrast, AnV– CL01 cells excluded 7-AAD, and were caspase 3 negative and DiOC6 positive (data not shown).
To determine the extent to which IgH primary sequence could predict survival fate, we examined 1046 IgH sequences from seven tonsils for the presence of consistent immunogenetic patterns that were exclusive to either AnV– or AnV+ fraction. A sample size this large should represent a true distribution of IgH sequences. Underrepresentation of rare IgH sequences is less likely to occur in a sample size of this magnitude. Sequences were compared based on VH and JH gene segment use, HCDR3 length and amino acid charge, and frequency of somatic hypermutation. For a given tonsil, VH and JH gene segment use did not significantly differ between AnV– and AnV+ fractions of the same isotype. Instead, high representation in the AnV– fraction directly coincided with high representation in the AnV+ fraction (see T2IgM VH4–34 and T2IgM JH4).
Sequence characteristics commonly associated with potential autoreactivity were not enriched in the AnV+ GC B cell pool, including long, positively charged HCDR3s and elevated JH6 use. Average HCDR3 lengths were statistically similar in AnV– and AnV+ fractions, ranging between 14 and 16 aa in all cases. Long HCDR3s (>18 aa) typically represented <15% of any given AnV fraction (regardless of isotype) and were not generated by the biased use of JH6 gene segments (5' JH repeats of tyrosine residues). Less than 10% of sequences in any AnV fraction contained three or more positively charged R, K, or H residues. Therefore, while B cells bearing BCRs with autoreactive properties do not typically survive to the end stages of the GC reaction in normal individuals, our data suggest that preferential deletion of such cells is unlikely to account for the majority of B cell death within the GC environment (based on criteria assessed in this study).
No consistent difference in somatic hypermutation frequencies was observed between AnV– and AnV+ sequences that were grouped according to isotype (IgM or IgG). Germline and highly mutated (>20 substitutions) sequences were isolated from each pool, suggesting that mutation quantity does not inherently determine survival fate. Earlier reports demonstrate that increased affinity is better associated with the quality of mutation and not necessarily the quantity, in that a single amino acid change in the IgH chain alone can increase affinity 10-fold during anti-4-hydroxy-3-nitrophenylacetyl Ab responses in mice (60). To account for a potential overgeneralization in our mutation analysis, we further subcategorized IgM and IgG sequences into more specific groups based on shared VH or JH gene segment use or HCDR3 length. Results did not change in this more focused approach. No significant difference was observed in 29 of 34 comparisons, reinforcing our finding that the level of somatic mutation cannot predict the decision between life and death.
Although there are many aspects of the GC reaction that remain uncertain, including the normal life expectancy of Ag-stimulated vs unstimulated B cells, it is generally accepted that the apoptotic B cell pool may be influenced by at least two processes, which are not necessarily mutually exclusive. One possibility is that BCRs from dying B cells are rendered unresponsive due to a loss of Ag specificity or the acquisition of structural aberrancies resulting from frameshifts and stop codons generated during somatic hypermutation and/or receptor modification/editing. Although stop codons and frameshifts were observed among AnV+ B cells, they surprisingly disrupted fewer than 5% of the sequences. Instead, the overwhelming majority were in-frame and encoded otherwise functional IgVH sequences, suggesting that loss of function is not the major cause of GC B cell death. Unfortunately, we were unable to determine the contribution made by the loss or reduction of Ag specificity for cognate Ag because immunogens are unknown in routine tonsillectomies.
A second possibility affecting GC B cell survival is based on stochastic dynamics in which fate is determined by successful competition for limited binding sites with FDC and T cells (36, 37). Our data are more consistent with this explanation. Under this model, competition not only exists between clones with potentially different affinities for cognate Ag (as determined by different IgH V region amino acid sequences), but also between clones with equal affinity (identical amino acid sequences). Therefore, GC B cells in AnV– and AnV+ fractions could represent clones that successfully received FDC/T cell-dependent survival signals, and those that failed, respectively.
Shown in Fig. 5, we isolated identical IgH V region amino acid sequences from AnV– and AnV+ centroblast and centrocyte fractions, strongly suggesting that survival fate is determined by factors other than IgH primary sequence alone. In the example shown in Fig. 5, group 6, a sequence identified among AnV– centroblasts successfully progressed to the centrocyte stage, presumably after receiving CD40:CD40L-mediated survival signals from Th cells. However, centrocytes bearing this BCR were subsequently isolated from AnV– and AnV+ fractions, indicating that continual survival during the centrocyte stage requires the receipt of additional survival signals from FDC/Th cells. Nearly half of the sequences in group 6 contained one or more silent mutations, thus preserving the significance of our findings even after the conservative consideration of potential PCR amplification biases.
Our inability to identify consistent differences in AnV– and AnV+ B cell fractions, nor in AnV– and TBMac, based on the BCR primary sequence, is in agreement with recent reports that suggest that BCR signal strength is one of the major governing factors influencing B cell survival (61, 62, 63). It is impossible to prove a negative, although we put forth considerable effort in analyzing the sequence data under multiple permutations. We are currently performing several experiments to address potential pitfalls within our data and our experimental approach. Potential influences on survival contributed by differential Ig H and L chain pairing were not discernable in our experimental model. To address this issue, the existence of common IgH:IgL combinations in AnV– and AnV+ fractions is being evaluated through single cell PCR amplification of IgH and IgL transcripts.
We conclude that GC B cell death (as determined by AnV binding) can occur during centroblast and centrocyte stages, regardless of singular factors such as V region gene segment use, HCDR3 length and charge, or quantity of incorporated mutations. B cells bearing BCRs that confer selective advantage to entering the centrocyte pool merely carry the means to survival, but must subsequently have it verified by successful interactions with resident FDC/T cells to ensure continual survival.
Disclosures
The authors have no financial conflict of interest.
Acknowledgments
We appreciate the technical expertise and helpful discussions provided by Dr. P. C. Wilson of Oklahoma Medical Research Foundation and Dr. P. W. Williamson of Amherst College (Amherst, MA).
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 This work was supported by grants AI 12127 and COBRE P20RR15577 from the National Institutes of Health.
2 Address correspondence and reprint requests to Dr. J. Donald Capra, Molecular Immunogenetics, Oklahoma Medical Research Foundation, 825 Northeast 13th Street, Oklahoma City, OK 73104. E-mail address: The_Capras{at}msn.com
3 Abbreviations used in this paper: GC, germinal center; 7-AAD, 7-aminoactinomycin D; FDC, follicular dendritic cell; PI, propidium iodide; PS, phosphatidylserine; TBMac, tingible body macrophage; AnV, annexin V.
Received for publication August 20, 2004. Accepted for publication November 5, 2004.
References
Tonegawa, S.. 1983. Somatic generation of antibody diversity. Nature 302:575.
Alt, F. W., E. M. Oltz, F. Young, J. Gorman, G. Taccioli, J. Chen. 1992. VDJ recombination. Immunol. Today 13:306.
Kolar, G. R., J. D. Capra. 2003. Immunoglobulin structure and function. W. E. Paul, ed. Fundamental Immunology 47. Lippincott Williams and Wilkins, Philadelphia.
Pascual, V., Y. J. Liu, A. Magalski, O. de Bouteiller, J. Banchereau, J. D. Capra. 1994. Analysis of somatic mutation in five B cell subsets of human tonsil. J. Exp. Med. 180:329.
Neuberger, M. S., C. Milstein. 1995. Somatic hypermutation. Curr. Opin. Immunol. 7:248.
MacLennan, I. C.. 1994. Germinal centers. Annu. Rev. Immunol. 12:117.
Jacob, J., G. Kelsoe, K. Rajewsky, U. Weiss. 1991. Intraclonal generation of antibody mutants in germinal centres. Nature 354:389
Shih, T. A., E. Meffre, M. Roederer, M. C. Nussenzweig. 2002. Role of BCR affinity in T cell dependent antibody responses in vivo. Nat. Immunol. 3:570.
Kline, G. H., L. Hartwell, G. B. Beck-Engeser, U. Keyna, S. Zaharevitz, N. R. Klinman, H. M. Jack. 1998. Pre-B cell receptor-mediated selection of pre-B cells synthesizing functional μ heavy chains. J. Immunol. 161:1608.
Brandtzaeg, P.. 1996. The B-cell development in tonsillar lymphoid follicles. Acta Otolaryngol. Suppl. 523:55.
Tsunoda, R., E. Heinen, N. Sugai. 2000. Follicular dendritic cells in vitro modulate the expression of Fas and Bcl-2 on germinal center B cells. Cell Tissue Res. 299:395
Koopman, G., R. M. Keehnen, E. Lindhout, D. F. Zhou, C. de Groot, S. T. Pals. 1997. Germinal center B cells rescued from apoptosis by CD40 ligation or attachment to follicular dendritic cells, but not by engagement of surface immunoglobulin or adhesion receptors, become resistant to CD95-induced apoptosis. Eur. J. Immunol. 27:1.
Grouard, G., O. de Bouteiller, J. Banchereau, Y. J. Liu. 1995. Human follicular dendritic cells enhance cytokine-dependent growth and differentiation of CD40-activated B cells. J. Immunol. 155:3345.
Van Nierop, K., C. de Groot. 2002. Human follicular dendritic cells: function, origin and development. Semin. Immunol. 14:251
Liu, Y. J., G. Grouard, O. de Bouteiller, J. Banchereau. 1996. Follicular dendritic cells and germinal centers. Int. Rev. Cytol. 166:139.
Petrasch, S. G., M. H. Kosco, C. J. Perez-Alvarez, J. Schmitz, G. Brittinger. 1991. Proliferation of germinal center B lymphocytes in vitro by direct membrane contact with follicular dendritic cells. Immunobiology 183:451.
Liu, Y. J., C. Arpin, O. de Bouteiller, C. Guret, J. Banchereau, H. Martinez-Valdez, S. Lebecque. 1996. Sequential triggering of apoptosis, somatic mutation and isotype switch during germinal center development. Semin. Immunol. 8:169.
Choi, Y. S.. 1997. Differentiation and apoptosis of human germinal center B-lymphocytes. Immunol. Res. 16:161.
Reynolds, J. D.. 1986. Evidence of extensive lymphocyte death in sheep Peyer’s patches. I. A comparison of lymphocyte production and export. J. Immunol. 136:2005
Liu, Y. J., C. Arpin. 1997. Germinal center development. Immunol. Rev. 156:111.
Tuscano, J. M., K. M. Druey, A. Riva, J. Pena, C. B. Thompson, J. H. Kehrl. 1996. Bcl-x rather than Bcl-2 mediates CD40-dependent centrocyte survival in the germinal center. Blood 88:1359.
Klein, U., Y. Tu, G. A. Stolovitzky, J. L. Keller, J. Haddad, Jr, V. Miljkovic, G. Cattoretti, A. Califano, R. Dalla-Favera. 2003. Transcriptional analysis of the B cell germinal center reaction. Proc. Natl. Acad. Sci. USA 100:2639.
De Boer, R. J., A. A. Freitas, A. S. Perelson. 2001. Resource competition determines selection of B cell repertoires. J. Theor. Biol. 212:333.
Kesmir, C., R. J. De Boer. 2003. A spatial model of germinal center reactions: cellular adhesion based sorting of B cells results in efficient affinity maturation. J. Theor. Biol. 222:9.
Danial, N. N., S. J. Korsmeyer. 2004. Cell death: critical control points. Cell 116:205.
Koopman, G., C. P. Reutelingsperger, G. A. Kuijten, R. M. Keehnen, S. T. Pals, M. H. van Oers. 1994. Annexin V for flow cytometric detection of phosphatidylserine expression on B cells undergoing apoptosis. Blood 84:1415.
Vermes, I., C. Haanen, H. Steffens-Nakken, C. Reutelingsperger. 1995. A novel assay for apoptosis: flow cytometric detection of phosphatidylserine expression on early apoptotic cells using fluorescein labelled Annexin V. J. Immunol. Methods 184:39.
Verhoven, B., S. Krahling, R. A. Schlegel, P. Williamson. 1999. Regulation of phosphatidylserine exposure and phagocytosis of apoptotic T lymphocytes. Cell Death Differ. 6:262.
Krahling, S., M. K. Callahan, P. Williamson, R. A. Schlegel. 1999. Exposure of phosphatidylserine is a general feature in the phagocytosis of apoptotic lymphocytes by macrophages. Cell Death Differ. 6:183.
Segundo, C., F. Medina, C. Rodriguez, R. Martinez-Palencia, F. Leyva-Cobian, J. A. Brieva. 1999. Surface molecule loss and bleb formation by human germinal center B cells undergoing apoptosis: role of apoptotic blebs in monocyte chemotaxis. Blood 94:1012.
Van Eijk, M., C. de Groot. 1999. Germinal center B cell apoptosis requires both caspase and cathepsin activity. J. Immunol. 163:2478
Zheng, N. Y., K. Wilson, X. Wang, A. Boston, G. R. Kolar, S. M. Jackson, Y. J. Liu, V. Pascual, J. D. Capra, P. C. Wilson. 2004. Human immunoglobulin selection associated with class switch and possible tolerogenic origins for C- class-switched B cells. J. Clin. Invest. 113:1188.
Pascual, V., P. Wilson, Y. J. Liu, J. Banchereau, J. D. Capra. 1997. Biased VH4 gene segment repertoire in the human tonsil. Chem. Immunol. 67:45.
Zhang, Z., Y. H. Wang, M. Zemlin, H. W. Findley, S. L. Bridges, P. D. Burrows, M. D. Cooper. 2003. Molecular mechanism of serial VH gene replacement. Ann. NY Acad. Sci. 987:270
Zemlin, M., K. Bauer, M. Hummel, S. Pfeiffer, S. Devers, C. Zemlin, H. Stein, H. T. Versmold. 2001. The diversity of rearranged immunoglobulin heavy chain variable region genes in peripheral blood B cells of preterm infants is restricted by short third complementarity-determining regions but not by limited gene segment usage. Blood 97:1511.
Radic, M. Z., J. Mackle, J. Erikson, C. Mol, W. F. Anderson, M. Weigert. 1993. Residues that mediate DNA binding of autoimmune antibodies. J. Immunol. 150:4966.
Aguilera, I., J. Melero, A. Nunez-Roldan, B. Sanchez. 2001. Molecular structure of eight human autoreactive monoclonal antibodies. Immunology 102:273
Callahan, M. K., M. S. Halleck, S. Krahling, A. J. Henderson, P. Williamson, R. A. Schlegel. 2003. Phosphatidylserine expression and phagocytosis of apoptotic thymocytes during differentiation of monocytic cells. J. Leukocyte Biol. 74:846.
Dillon, S. R., A. Constantinescu, M. S. Schlissel. 2001. Annexin V binds to positively selected B cells. J. Immunol. 166:58
Dillon, S. R., M. Mancini, A. Rosen, M. S. Schlissel. 2000. Annexin V binds to viable B cells and colocalizes with a marker of lipid rafts upon B cell receptor activation. J. Immunol. 164:1322.
Wardemann, H., S. Yurasov, A. Schaefer, J. W. Young, E. Meffre, M. C. Nussenzweig. 2003. Predominant autoantibody production by early human B cell precursors. Science 301:1374
Ninomiya, E., Y. Ito, M. A. Shibata, K. Kawashima, T. Sakamoto, E. Maruyama, H. Doi, K. Tokitsu, Y. Otsuki. 2003. The activation of caspase-3 and DNA fragmentation in B cells phagocytosed by macrophages. Med. Electron Microsc. 36:87.
Nakamura, M., H. Yagi, S. Kayaba, T. Ishii, T. Gotoh, S. Ohtsu, T. Itoh. 1996. Death of germinal center B cells without DNA fragmentation. Eur. J. Immunol. 26:1211.
Allen, D., T. Simon, F. Sablitzky, K. Rajewsky, A. Cumano. 1988. Antibody engineering for the analysis of affinity maturation of an anti-hapten response. EMBO J. 7:1995.
Casola, S., K. L. Otipoby, M. Alimzhanov, S. Humme, N. Uyttersprot, J. L. Kutok, M. C. Carroll, K. Rajewsky. 2004. B cell receptor signal strength determines B cell fate. Nat. Immunol. 5:317
Niiro, H., E. A. Clark. 2002. Regulation of B-cell fate by antigen-receptor signals. Nat. Rev. Immunol. 2:945.
Pittner, B. T., E. C. Snow. 1998. Strength of signal through BCR determines the fate of cycling B cells by regulating the expression of the Bcl-2 family of survival proteins. Cell. Immunol. 186:55.(Stephen M. Jackson and J.)
Germinal center (GC) B cell survival fate is governed in part by the outcome of successful/failed BCR-mediated interactions with accessory cells. However, the extent to which the BCR primary sequence influences such interactions is not fully understood. Over 1000 IgVH4 family cDNAs were sequenced from living (annexin V–) and apoptotic (annexin V+ or from within tingible body macrophages) GC B cell fractions from seven tonsils. Results surprisingly demonstrate that living and dying GC B cells do not significantly differ in IgVH, D, or JH gene segment use; HCDR3 length or positive charge; or mutation frequency. Additionally, equivalent IgH cDNA sequences were identified in both fractions, suggesting that BCR sequence alone is an unreliable predictor of GC B cell survival.
Introduction
The development of an efficiently functioning humoral immune system is one of the most important factors in survival among vertebrates. It is quite paradoxical that such a well-coordinated system is established and maintained by processes that incorporate a high degree of randomness and chance. Among B cells, these processes include, but are not limited to, Ig gene rearrangement in the bone marrow (1, 2, 3) and somatic hypermutation-based affinity maturation within germinal centers (GC)3 (4, 5, 6, 7). Multiple rounds of receptor-mediated selection during various stages of cell development, the survivors of which differentiate into key effector B cell subsets, buffer the potential detriments of randomness.
To explore various issues of selection, investigations have focused on multiple aspects of B cell selection within the bone marrow, and secondary lymphoid tissues such as tonsils. Selection within the bone marrow is 2-fold, including positive selection of B cells that successfully rearrange and express components of the BCR, and negative selection against those expressing BCRs that bind to self Ag with inappropriately high avidity (autoreactivity) (8, 9, 10).
Selection in peripheral sites, including GC, is based upon a B cell’s ability to not only bind foreign Ag, but also to compete in a Darwinian fashion for limited opportunistic binding with accessory cells, including follicular dendritic cells (FDC) and T cells (6, 11, 12, 13, 14, 15, 16, 17, 18). These cells deliver critical life-promoting secondary signals that are essential for continued B cell survival (19, 20, 21, 22, 23, 24, 25). Both processes are BCR dependent. Consequently, the current paradigm for GC B cell survival is primarily focused on the BCR itself. B cells bearing receptors that attain greater affinity for Ag are selected and proceed to further stages of development. Conversely, those with lower affinity, frameshifts, nonsense codons, and other factors that negatively influence Ag binding are presumably destined to die via apoptosis. In this study, we test this hypothesis directly by comparing quantifiable differences in the primary sequences of BCR H chain V regions between living B cells and those that are marked for death.
A hallmark of the GC reaction is the high frequency of cell death, with as few as 5% of GC B cells surviving to the effector cell stage in some instances (26, 27, 28). The basis of such high death rates in the GC is not fully understood. Centroblasts and centrocytes are Bcl-2 negative and are prone to die, presumably by default (29, 30, 31, 32, 33). In addition, recent investigations demonstrate that both Ag recognition and cell:cell communications are needed for persistent survival within the GC (32, 34, 35, 36, 37). Therefore, the level of cell death is likely to be influenced by factors that deleteriously affect either requirement.
Apoptosis occurs in multiple stages (reviewed in Ref. 38). Loss of cell membrane asymmetry is an early event, resulting in atypical phosphatidylserine (PS) exposure on the exoplasmic leaflet. Externalized PS on GC B cell surfaces is physiologically relevant, serving as the targeted ligand during receptor-mediated phagocytosis by tingible body macrophages (TBMac). Similarly, annexin V (36 kDa) binds PS on surfaces of dying cells and is routinely used as a reliable marker of apoptosis for various cell types, including GC B cells (39, 40, 41, 42). The pattern of Annexin V (AnV) binding has been previously shown to correlate with the percentage of GC B cells undergoing apoptosis (43, 44).
AnV+ GC B cells exhibit characteristics consistent with cells undergoing apoptosis, including down-regulated expression of various proteins associated with B cell survival (CD21, CD11a, CD49d, and CD54) (43). During in vitro culture, AnV+ B cells form membrane blebs, readily degrade DNA, and test positive for active caspase 3, all of which are significantly blocked by the addition of caspase inhibitors (44).
To determine the relative roles that either receptor modification or competition play in influencing cell death within the GC, we performed molecular IgH repertoire analyses using GC B cell subsets, including centroblasts and centrocytes, that were separated into AnV+ and AnV– fractions. Necrotic cells (AnV+/propidium iodide (PI)+/7-aminoactinomycin D+ (7-AAD+)) were excluded. Despite a substantial effort to document consistent differences between BCRs of these subsets, our results show that AnV+ and AnV– fractions are surprisingly similar, showing no significant difference in IgVH, D, or JH gene segment use; HCDR3 length or composition; or mutation pattern and frequency. Surprisingly, identical IgH V region amino acid sequences were routinely identified in both fractions. In fact, we obtained the same results when IgH sequences were analyzed from dying B cells that were previously engulfed by TBMac in situ, which therefore represent a physiologically germane category of apoptotic B lymphocytes. These data suggest that BCR sequence (H chain) is unlikely to be the principally decisive determinant in predicting the survival fate of GC B cells. As such, these results challenge the prevailing paradigm and suggest that Darwinian fitness must be defined by factors other than receptor quality alone.
Materials and Methods
Isolation of GC B cells
Tonsillar B cells were isolated, as previously described (4, 45). All protocols are Institutional Review Board approved. Briefly, children’s tonsils were collected during routine tonsillectomies and processed immediately on ice. Tissue was separated into single cell suspensions using the Collector tissue sieve (ThermoEC) and resuspended in FCS-supplemented (10%) RPMI 1640 (Invitrogen Life Technologies) or MACS buffer. Total mononuclear cells were collected following 20-min centrifugation in Lymphoprep (Axis-Shield), and washed once in MACS buffer. TBMac were directly isolated from this mononuclear fraction. B cells were enriched by magnetic negative depletion of non-B cells using a MACS B cell isolation kit (Miltenyi Biotec), according to the manufacturer’s instructions. Cells were immediately counted and stained for subsequent cytometric analysis. Special care was taken to minimize the time from tonsillectomy to cell separation so as to limit the extent of cell death during processing.
Flow cytometry and cell sorting
Tonsillar B cells were stained for distinct GC B cell subsets using strategies described earlier (4). Incubations with the appropriate concentrations of conjugated Abs were performed for 15 min at 4°C. Incubations with FITC-labeled AnV (Caltag Laboratories), PI, or 7-AAD (BD Pharmingen) were at room temperature for 20 min. Briefly, cells (1–3 x 106) were stained using allophycocyanin-labeled CD38 (Caltag Laboratories). GC B cells were identified as CD38+. PE-labeled anti-CD77 (BD Pharmingen) staining was used to further subdivide GC B cells into centroblast (CD38+CD77+) and centrocyte (CD38+CD77–) populations. TBMac were isolated from total mononuclear cells following staining with anti-CD68 PE, in addition to anti-CD19 FITC and anti-CD38 allophycocyanin (all from Caltag Laboratories) to exclude potential B cell contamination. The TBMac fraction collected for further investigation was CD68+CD19–CD38–. Individual cell populations were collected following multicolor cell sorting using a MoFlo flow cytometer (
DakoCytomation), or cytometrically analyzed on a FACSCalibur instrument (BD Biosciences). Necrotic cells (AnV+PI+/7-AAD+) were excluded in all studies. In several control experiments, patterns of coincident staining using fluorometric AnV, caspase 3 (BD Pharmingen), DiOC6 (Sigma-Aldrich), and 7-AAD cell death markers were assessed using apoptosis-induced CLO1 or Jurkat cell lines treated with 10 Gy gamma-irradiation or A23187 (Alexis Biochemicals), respectively.
Cloning and sequencing of IgVH4 genes
RNA from B cells collected during cell sorting was isolated using RNAwiz (Ambion). VH4 family genes were amplified from two independent tonsils (T2, T6) via RT-PCR using family-specific and isotype-specific primers, as described previously (45, 46, 47). In a separate analysis, VH4–34 sequences were specifically amplified from multiple sources using a VH4–34-specific 5' primer (5'-AGCTACAGCAGTGGGGCG-3') and the same 3' IgG primer as before. Sources included centroblast and centrocyte subsets from three tonsils (T7, T9, and T11) and phagocytized B cells (within TBMac) from two tonsils (T40 and T41). IgM and IgG products were cloned using the Qiagen One-Step RT-PCR cloning system, according to manufacturer’s instructions. Purified plasmids were isolated (Qiaprep Spin system; Qiagen) from transformed Escherichia coli (DH5) bacteria and were subsequently sequenced with an ABI 3730 DNA sequencer.
DNA sequence and statistical analysis
Germline VH4 family gene segment use among sequences was identified using National Center for Biotechnology Information IgBlast (www.ncbi.nlm.nih.gov/igblast/) and IMGT/V-QUEST (http://imgt.cines.fr:8104) databases. Sequence comparison and alignments were performed with VectorNTI and a specialized software program based on a Microsoft Visual Basic v6.0 platform designed by P. Wilson of the Oklahoma Medical Research Foundation. Statistical analyses were also done using this program, in addition to SYSTAT and Microsoft Excel for Windows.
Results
Identification and isolation of AnV+ and AnV– GC B cells
Isolation of apoptotic GC B cells was performed using two complementary strategies. To isolate B cells that are unequivocally undergoing apoptosis, we directly collected CD68+ TBMac that simultaneously stained negative for extracellular B cell-specific markers CD19 and CD38. This approach excluded TBMac that may have recently initiated phagocytosis, i.e., those with partially engulfed B cells whose exposed surfaces would most likely contain B cell-specific markers generating a triple-positive CD68+CD19+CD38+ TBMac phenotype. Although nucleic acids begin to degrade following engulfment by TBMac, purified B cell RNA and DNA integrity remained sufficient, evidenced by the successful amplification of rearranged IgH cDNA sequences (despite the absence of CD19 and CD38 among the sorted TBMac population).
Macrophage recognition (phagocytosis) of apoptotic cells and AnV binding are both PS-dependent processes. In a second approach designed to permit the isolation of intact, distinct GC B cell subsets in the earliest stages of apoptosis (impractical from within the lytic environment of TBMac), fluorometric AnV was used to bind externalized PS. Tonsillar B cells were stained with Abs to CD38 and IgD to identify GC B cells (Fig. 1A). For some experiments, anti-CD77 was used to further subdivide GC B cells into centroblast (CD38+CD77+, Bm3) and centrocyte (CD38+CD77–, Bm4) populations (Fig. 1B). In both cases, distribution patterns based on cell surface phenotypes were consistent with earlier reports (4, 48).
FIGURE 1. Cytometric analysis of GC B cells and corresponding AnV binding. Panels represent one of five independent experiments. A, Cell surface phenotype of tonsillar B cells after staining with anti-CD38 and anti-IgD. B, Costaining of GC B cells (different patient from A) with anti-CD38 and anti-CD77 to identify centroblasts (CD38+CD77+) and centrocytes (CD38+CD77–). AnV staining was determined among total GC B cells (C) or centroblasts and centrocytes (D). Necrotic cells (AnV+7-AAD+) were excluded.
The reproducible identification of dying GC B cells using AnV as the primary cell death marker has been demonstrated previously in multiple investigations (43, 44), reporting values of <25% AnV positivity (AnV+) within the first few hours following surgery. Cytometric and FACS analyses were performed on cells that were stained with AnV immediately following passive B cell enrichment. This was done to maximize the accuracy of identifying AnV+ cells that are likely to have externalized PS in response to physiologically relevant factors within the in situ GC environment. Cytometric results from five tonsils were consistent with previous reports. For example, 5–24% of CD38+ GC B cells in our analyses were also AnV+, with an average of 11.7% (SD = 8.5%). The seemingly high SD is attributable to a single tonsil (24.3% AnV+, not used in subsequent analyses). The remaining tonsils were clustered around the average (example shown in Fig. 1C). To determine whether AnV staining was uniform among CD38+ GC B cells, we compared AnV+ levels in centroblasts vs centrocytes (Fig. 1D). In the example shown, there was nearly a 3-fold increase in AnV staining in centroblasts (13.3%) vs centrocytes (4.8%). This suggests that the two cell types are differentially susceptible to cell death, an observation consistent with previous reports (49). Together, these results document our ability to reproducibly isolate various subpopulations of human B cells that are marked for death.
IgH repertoire analysis
Persistent GC B cell survival and development to effector B cell stages is thought to be a BCR-dependent process. To determine the potential role that the BCR primary sequence plays in influencing survival fate, we performed repertoire analyses by sequencing over 1000 (1046) IgVH4 family members cloned from seven tonsils. Total tonsillar CD38+ GC B cells were sorted into AnV+ and AnV– pools and subjected to isotype-specific RT-PCR amplification of VH4 transcripts. IgM and IgG sequences from both pools were evaluated based on VH and JH gene segment use, HCDR3 length and positively charged amino acid composition, and mutation frequency. Clones that were identical at the cDNA level were counted once. Influences from potential clonal relatedness were negligible because exclusion of these sequences from statistical analyses did not significantly alter results. With very few exceptions, AnV+ and AnV– GC B cells are surprisingly indistinguishable based solely on the characteristics of their IgH chain primary sequences assessed in this study. In fact, identical H chain sequences were isolated from AnV+ and AnV– centroblast and centrocyte fractions with appreciable frequency (examples in Fig. 5 below). Similarly, in two independent experiments, identical IgVH4–34 sequences were observed in living and dying GC B cell fractions comparing AnV– CD38+ and phagocytized GC B cell counterparts, respectively. Together, these findings strongly suggest that BCR quality (at least H chain V region sequence) is insufficient to guarantee survival within the competitive GC environment.
FIGURE 5. Identical clones identified in AnV– and AnV+ GC B cell fractions. Identical clones were identified among AnV+ and AnV– fractions in centroblasts and centrocytes. Germline genes are indicated. Only codons with mutations are listed. Amino acid differences from germline are in bold type. Replacement and silent mutations are in capital or lower case, respectively. HCDR3 is in italic underline. The number of times a given sequence was isolated is in parentheses. Nomenclature: T9Bgn8 = tonsil 9, centroblast, IgG, AnV–, clone 8. T7Cmp29 = tonsil 7, centrocyte, IgM, AnV+, clone 29.
Equivalent VH gene segment use in AnV+ and AnV– fractions
IgVH4 gene expression frequencies in T2 and T6 GC IgMAnV+ and IgMAnV– fractions were quantitatively compared using the 2 test (heteroscedastic). For each gene segment analyzed, frequencies were statistically equivalent between IgMAnV+ and IgMAnV– fractions. Calculated p values exceeded 0.05 in each case. This includes VH4–59 in T2 (T2VH4–59), which despite a 2-fold difference had a p value of 0.28. VH4–34 was 58% of the population in T2 IgMAnV+ and IgMAnV– fractions (Fig. 2A). Similarly, VH4–34 is equivalently represented in T6 IgMAnV+ (32%) and IgMAnV– (30%) fractions, although the expression level was reduced compared with T2, demonstrating potential environmental influences on expression levels.
FIGURE 2. IgH repertoire analysis among AnV+ and AnV– fractions of total GC B cells. The distribution of VH4 family member gene segment use was determined for AnV+ (striped bars) and AnV– (solid bars) GC B cell fractions of IgM (A) and IgG (B) sequences. JH gene segment use was similarly determined for the same IgM (C) and IgG (D) sequences.
Similar to IgM, VH4 expression frequencies in GC B cell IgGAnV+ and IgGAnV– fractions were statistically indifferent between all gene segments compared (p > 0.05) (Fig. 2B).
The hierarchical pattern of gene expression differed between IgM and IgG pools, which may represent preferential selection (including counterselection) of certain gene segments or IgH rearrangements during the GC reaction. For example, VH4–34 expression patterns in both tonsils were significantly reduced among IgG sequences, despite being the most frequently isolated gene segment among IgM sequences. Conversely, T2VH4–59 and T6VH4–61 frequencies are increased in the IgG pool relative to their IgM counterparts (Fig. 2B). In addition, elevated T2VH4–59 and T6VH4–61 frequencies in the IgGAnV– fraction were coincident with elevated representation in the IgGAnV+ fraction. This suggests that frequency of expression of a given VH4 gene segment in the AnV– fraction directly influences the level of expression of the same gene segment in the corresponding AnV+ fraction. Therefore, the potential selection of B cells with BCRs possessing beneficial qualities (such as increased affinity) does not necessarily confer protection against becoming AnV+.
Similarities in expression frequencies between AnV+ and AnV– fractions are unlikely to be the result of primer bias. VH4 frequencies varied between tonsils, yet were consistent for independent amplifications from a given GC fraction. In addition, expression frequencies are significantly different between IgM and IgG pools, whose amplification strategies differed only in the use of downstream isotype primer.
Equivalent JH gene segment use in AnV+ and AnV– fractions
We analyzed the frequency of JH gene segment use via the same strategy as above. Similar to results from VH gene segment analyses, there is no statistical difference in JH gene segment use (p > 0.05) between GC AnV+ and AnV– fractions for either isotype. JH4 was the predominant gene segment in IgMAnV– and IgGAnV– fractions for both tonsils, accounting for nearly 50% of the rearrangements (Fig. 2, C and D). JH4 was equivalently represented in the corresponding IgMAnV+ and IgGAnV+ fractions, again suggesting that the frequency of gene segment use among AnV+ cells is strongly influenced by the expression frequency in the AnV– pool.
Elevated JH6 gene segment use is commonly observed among B cells with autoreactive or structurally aberrant BCRs, which undergo receptor revision presumably as a means of escaping death. Surprisingly, we did not observe a significant JH6 bias among AnV+ B cells in either isotype. Therefore, the propensity to externalize PS (AnV+) and subsequently become targeted for phagocytosis does not directly correspond with JH gene segment use.
Long, positively charged HCDR3s are not enriched among GC B cell AnV+ fractions
Negative selection against long, positively charged HCDR3s has been demonstrated in multiple murine and human autoimmune models (50, 51, 52, 53). The majority of nonautoreactive human HCDR3s typically average between 14 and 18 aa in length. We therefore compared HCDR3s in GC AnV– and AnV+ B cell pools to assess whether positive charge and nonstandard lengths are enriched in the latter fraction. For both tonsils, average HCDR3 lengths in all cases ranged between 14 and 16 aa. Values were statistically equivalent (p > 0.05) between AnV– and AnV+ counterparts in IgM and IgG pools (data not shown).
The distribution of HCDR3 lengths in AnV+ and AnV– fractions is presented in Fig. 3, A–F (grouped in three amino acid units). Consistent with earlier reports, >50% of HCDR3s in all subsets were present in the 14–18 (±1) range (51). In general, correlation between AnV+ and AnV– fractions is more coincident among IgG sequences. In all cases, p values exceed 0.05. Nonetheless, wider disparity was observed among IgM sequences, particularly in the T2 18- to 20-aa (T218–20aa) and T615–17aa ranges.
FIGURE 3. Population distribution of HCDR3s within a given HCDR3 length range. HCDR3 lengths were grouped into ranges of 3 aa. The percentage of clones within each range is presented for AnV– (filled line) and AnV+ (dashed line) fractions of IgM (A, C, and E) and IgG sequences (B, D, and F). For increased sample size, corresponding subgroups were combined from T2 and T6. Combined results are presented for IgM (E) and IgG (F) sequences. Note: differences between AnV– and AnV+ fractions for T2IgM (15 16 17 ), T2IgM (18 19 20 ), and T6IgM (15 16 17 ) approached significance, yet each p value exceeded 0.05. Differences for combined T2 + T6IgM (15 16 17 ) were significant (p < 0.05).
To determine whether AnV+ B cell fractions are enriched for potential signs of HCDR3-related autoreactivity, we quantified the number of in-frame, positively charged amino acids (R, K, and H) in this region (excluding the germline-encoded arginine at the 3' end of framework 3). Results are presented in Fig. 4A. Between 70 and 80% of all clones have one or no positive residue, independent of isotype and AnV staining pattern. In fact, the frequency of IgM or IgG clones with three or more positive HCDR3 residues does not exceed 10% in any AnV fraction, strongly suggesting that there is no bias toward positive charge among AnV+ B cells. Coupled with the observation that longer HCDR3s are not enriched among AnV+ B cells, HCDR3 primary structure alone appears to have minimal influence on predicting cell survival fate.
FIGURE 4. Percentage of clones with positively charged amino acids. The percentage of clones with a given number of positively charged R, K, and/or H residues within their HCDR3 is graphically represented for IgM (A) and IgG (B) sequences. AnV+ and AnV– clones are represented by dashed and solid lines, respectively.
Somatic hypermutation frequencies are identical in the living and dying pools of GC B cells
The frequency of somatic hypermutation was determined for all AnV– and AnV+ fractions from T2 and T6. Results are presented in Table I. In a broad-spectrum analysis based only on shared isotype, higher mutation frequencies were not consistently biased toward either AnV fraction. For example, mutation frequencies were higher among T2 IgMAnV+ cells (6.1% ± 4.6) relative to T2 IgMAnV– (3.8 ± 2.3). Calculated differences were statistically significant (p < 0.05). However, the pattern was reversed in T6 (4.5 ± 6.0 vs 6.7 ± 6.5, respectively), demonstrating that AnV+ cells do not inherently contain more mutations. In fact, differences between AnV+ and AnV– fractions in T6IgM, T2IgG, and T6IgG cells were not statistically significant (p > 0.05) (Table I). Consistent with previous reports, IgG sequences typically had significantly higher mutation frequencies (p < 0.01) when compared with IgM sequences from the same tonsil (Table I).
Table I. Mutation frequency among AnV– and AnV+ isotypic fractionsa
SDs associated with calculated mutation frequencies exceeded 70% in most cases, indicating that groups contained sequences with a widely variable number of mutations. It is plausible that sequences with more mutations could belong to a particular subgroup of sequences that could potentially be further categorized by shared characteristics. We therefore determined mutation frequencies among isotype-specific AnV– and AnV+ sequences that were additionally subdivided by shared use of VH gene segment, JH gene segment, or HCDR3 length. Surprisingly, 85% (29/34) of AnV+ vs AnV– comparisons made in this more focused analysis showed no significant difference in mutation frequency (data not shown). Of the five that did, three were in the T2IgM pool (VH4–34, JH6, and HCDR3 length 9–11 aa) in which significant differences were previously observed in the isotype-only calculation. However, a consistent pattern was not evident because none of the five was duplicated in both tonsils. Therefore, in the overwhelming majority of cases, AnV– and AnV+ fractions do not differ in mutation frequency, although this deserves further investigation (i.e., in systems in which mounted immune responses are elicited by a single known Ag).
IgVH regions with identical amino acid sequences exist in living and dying fractions of both centroblasts and centrocytes
As noted above, we observed no significant difference between AnV– and AnV+ fractions on the level of VH or JH gene segment use, HCDR3 length or charge, or somatic mutation load. This suggests that B cells bearing surface BCRs with equivalent characteristics are similarly prone to externalize PS (and therefore be engulfed by GC-resident TBMac). To test this hypothesis, we surveyed both AnV fractions for the presence of BCRs with identical amino acid sequences. To increase the chance of isolating clonally related sequences from a given pool, VH4 family members were amplified from CD38+CD77+ sorted centroblasts (clonally proliferating) or centrocytes (CD38+CD77–) using an upstream VH4–34-specific primer (a subset of sequences was also amplified with a VH4 universal primer).
Three separate tonsils were used in independent analyses (T7, T9, and T11). Clonality was determined based on shared VH, D, and JH gene segment use and identical VD and DJ joins. DNA sequences with accompanying amino acid changes are displayed in Fig. 5. In all three tonsils, clones with identically mutated amino acid sequences were successfully identified in both AnV fractions from centroblast (group 1) and centrocyte (groups 2–6) subsets. For example, four identical (amino acid level) centroblast sequences using VH4–31, DH4–17, and JH4 were isolated from the T7IgG pool. All four shared the same six replacement mutations. T7Bgn8 and T7Bgn14 are AnV–. T7Bgp5 and T7Bgp7 are AnV+. Note that T7Bgp5 and T7Bgp7 differ by a single silent mutation. The same observation was made using phagocytized B cells as the source of dying cells (described below). It is unlikely that reproducible sequence isolation was the by-product of potential RT-PCR bias, as most clonally related groups contained sequences that differed by one or more silent mutations, thereby indicating that their source was from distinct B cells. In fact, identical sequences (cDNA level) T7Cmn5 and T7C34p24 (group 2) were isolated following amplification with completely different 5' and 3' primer pairs (VH4:IgM and P434:IgG, respectively).
Evidence of clonal progression from centroblast to centrocyte was observed in group 6. Six identical amino acid sequences sharing three replacement mutations (S, T, and R) were isolated from the T9 centroblast AnV– fraction (i.e., T9B34n11). The same sequence was detected once in T9 centrocyte AnV– fraction and eight times in T9 centrocyte AnV+ fraction. The presence of silent mutations indicates that sequences were amplified from independent B cells. In this example, cells that successfully survive during the centroblast stage are able to further differentiate in centrocytes. However, continued centrocyte survival is not guaranteed because the same sequence was also identified in the centrocyte AnV+ fraction. Therefore, cell death within the GC occurs in both the centroblast and centrocyte stages of development, and persistent survival appears to require influences from additional factors at various time points.
IgG VH sequences from phagocytized GC B cells (tingible bodies) cannot be distinguished from the living B cell pool
An inherent disadvantage to performing apoptosis studies ex vivo is the difficulty in distinguishing between physiologically relevant cell death that was initiated in vivo, and that which unavoidably occurs following removal of any tissue from its nurturing environment. Despite having expediently processed each tonsil, there remains the possibility that a small fraction of the sorted GC B cells could represent cells that began to externalize PS following surgery. In a more stringent assessment of the potential correlation between the BCR primary sequence and survival fate in vivo, CD68+ TBMac were independently sorted from two tonsils (T40 and T41). Simultaneous staining with anti-CD19 and anti-CD38 (CD68+CD19–CD38–) permitted the cytometric exclusion of B cells. This includes those that would otherwise colocalize with TBMac due to having partially exposed surfaces stemming from their incomplete engulfment. Total RNA was immediately isolated to circumvent nucleic acid degradation, reverse transcribed, and amplified using VH4–34- and IgG-specific primers, as before. IgH cDNA sequences from engulfed apoptotic B cells were compared with those amplified from viable AnV–CD38+ GC B cells collected from their respective tonsils.
Results using this more direct assay for apoptotic cells (i.e., absolute dying vs prephagocytosis) were consistent with our previous observations based on AnV staining alone. As shown in Fig. 6, phagocytized and AnV– B cells did not significantly differ in JH gene use (JH4 majority in both fractions) or HCDR3 length (average length between 11 and 13 aa for each fraction). Comparable somatic hypermutation frequencies were observed between dying and living fractions from T41 (respectively, 5.5 ± 5.0 and 4.8 ± 5.4; p = 0.67; Table I). Calculations for T40 suggest that dying B cells mutate at a slightly higher frequency (respectively, 5.5 ± 3.5 and 3.6 ± 3.5; p = 0.05), although a difference of two substitutions is unlikely to have physiological significance, considering the wide range of mutations permissible (high SD). Both T40 fractions contained multiple sequences with 0 (germline) to 10 or more mutations. In fact, the IgH sequence with most mutations (17 substitutions) was present in the living (AnV–) fraction. Therefore, mutation load, JH gene segment use, and HCDR3 length each continue to be an unreliable gauge for predicting GC B cell survival fate, even among cells that cannot escape death due to engulfment and ultimate degradation within TBMac.
FIGURE 6. JH gene segment use and CDR3 length comparison between phagocytized and AnV– GC B cell fractions. VH4–34 sequences from T40 and T41 dying (MAC) and living (AnV–) fractions were compared based on JH gene segment use (A) and HCDR3 length (B), as before. Phagocytized (MAC) and AnV– fractions are indicated by and , respectively. The percentage of clones within each range is presented for AnV– (filled line) and phagocytized (dashed line) fractions.
Discussion
Factors influencing the differential survival fate of clonally related GC B cell progeny are not fully understood. However, the BCR has long been thought to play a central role in the initiation and maintenance of the GC reaction in response to T-dependent Ags (6, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25). For example, pre-GC B cells are activated by Ag-specific interactions between the BCR and Th cells (6). Consequently, all founder B cells that initially colonize GC have already successfully demonstrated BCR functional, structural, and Ag recognition competency, and are positively selected for survival. However, our findings suggest, as have others before, that benefits from this initial survival signal are most likely transient, periodically requiring subsequent survival signals during developmental progression in the GC. We observed identical IgH amino acid sequences in AnV– and AnV+ fractions of both centroblasts and centrocytes. It is therefore plausible that AnV+ (and phagocytized) GC B
cells represent those that received life-promoting signals at one time point, but failed to receive additional signals within the appropriate time window, leading to PS externalization and clearance by TBMac. Therefore, sequence alone is insufficient to predict survival fate at either centroblast or centrocyte stages, and does not guarantee the continued survival of clones bearing BCRs that otherwise appear functional at the H chain level. Potential influences from the L chain are currently under investigation.
PS exposure on dying GC B cells is physiologically relevant. In reports by Krahling et al. (42) and Callahan et al. (54), preincubation with AnV sufficiently inhibited PS recognition by macrophages, resulting in radically reduced levels of phagocytosis. Under these circumstances, apoptotic GC B cells were atypically associated with FDC and served as potential sources of self Ag and the onset of autoimmunity, demonstrating that continued survival of GC PS+ cells can be detrimental to the host. Therefore, PS exposure within the GC environment, and that observed by Dillon et al. (55, 56) on viable cells in the bone marrow, may not represent the same phenomenon. It is unlikely that viable GC B cells would externalize PS (even if transiently) because this would lead to their rapid clearance by proximally positioned TBMac (barring the theoretical possibility of viable cells coexpressing inhibitors of phagocytosis on their surfaces). In fact, continued survival of PS+ bone marrow B cells (nonphagocytized) may be a factor contributing to the relatively high degree of polyreactivity (self-recognition) among early B cells (57).
Nonetheless, annexin V has proven to be the most readily detectable marker of early apoptosis for GC B cells (42, 43). In our experiments, the percentage of tonsillar AnV+ GC B cells typically ranged between 5 and 15%. This is comparable to levels reported by Brieva and colleagues (43) and Groot and colleagues (21). Earlier investigations show that DNA degradation and caspase 3 activation generally do not occur until after B cells are engulfed by macrophages (58, 59), therefore rendering TUNEL assays and use of fluorometric caspase substrates less reliable for isolating early apoptotic, nonphagocytized GC B cells. Consistent with these reports, our efforts to sort intact GC B cells based on simultaneous staining of AnV and caspase 3 (active form) proved to be problematic (data not presented). However, gamma-irradiated GC-like CL01 cells (generously provided by P. Wilson) and A23187-treated JKT cells (E6-1) both successfully bound AnV in control experiments (data not shown). AnV+ control cells intercalated 7-AAD, successfully bound anticaspase 3 (active form), and showed signs of mitochondrial membrane breakdown (DiOC6 negative), indicative of their entry into the apoptosis pathway. In contrast, AnV– CL01 cells excluded 7-AAD, and were caspase 3 negative and DiOC6 positive (data not shown).
To determine the extent to which IgH primary sequence could predict survival fate, we examined 1046 IgH sequences from seven tonsils for the presence of consistent immunogenetic patterns that were exclusive to either AnV– or AnV+ fraction. A sample size this large should represent a true distribution of IgH sequences. Underrepresentation of rare IgH sequences is less likely to occur in a sample size of this magnitude. Sequences were compared based on VH and JH gene segment use, HCDR3 length and amino acid charge, and frequency of somatic hypermutation. For a given tonsil, VH and JH gene segment use did not significantly differ between AnV– and AnV+ fractions of the same isotype. Instead, high representation in the AnV– fraction directly coincided with high representation in the AnV+ fraction (see T2IgM VH4–34 and T2IgM JH4).
Sequence characteristics commonly associated with potential autoreactivity were not enriched in the AnV+ GC B cell pool, including long, positively charged HCDR3s and elevated JH6 use. Average HCDR3 lengths were statistically similar in AnV– and AnV+ fractions, ranging between 14 and 16 aa in all cases. Long HCDR3s (>18 aa) typically represented <15% of any given AnV fraction (regardless of isotype) and were not generated by the biased use of JH6 gene segments (5' JH repeats of tyrosine residues). Less than 10% of sequences in any AnV fraction contained three or more positively charged R, K, or H residues. Therefore, while B cells bearing BCRs with autoreactive properties do not typically survive to the end stages of the GC reaction in normal individuals, our data suggest that preferential deletion of such cells is unlikely to account for the majority of B cell death within the GC environment (based on criteria assessed in this study).
No consistent difference in somatic hypermutation frequencies was observed between AnV– and AnV+ sequences that were grouped according to isotype (IgM or IgG). Germline and highly mutated (>20 substitutions) sequences were isolated from each pool, suggesting that mutation quantity does not inherently determine survival fate. Earlier reports demonstrate that increased affinity is better associated with the quality of mutation and not necessarily the quantity, in that a single amino acid change in the IgH chain alone can increase affinity 10-fold during anti-4-hydroxy-3-nitrophenylacetyl Ab responses in mice (60). To account for a potential overgeneralization in our mutation analysis, we further subcategorized IgM and IgG sequences into more specific groups based on shared VH or JH gene segment use or HCDR3 length. Results did not change in this more focused approach. No significant difference was observed in 29 of 34 comparisons, reinforcing our finding that the level of somatic mutation cannot predict the decision between life and death.
Although there are many aspects of the GC reaction that remain uncertain, including the normal life expectancy of Ag-stimulated vs unstimulated B cells, it is generally accepted that the apoptotic B cell pool may be influenced by at least two processes, which are not necessarily mutually exclusive. One possibility is that BCRs from dying B cells are rendered unresponsive due to a loss of Ag specificity or the acquisition of structural aberrancies resulting from frameshifts and stop codons generated during somatic hypermutation and/or receptor modification/editing. Although stop codons and frameshifts were observed among AnV+ B cells, they surprisingly disrupted fewer than 5% of the sequences. Instead, the overwhelming majority were in-frame and encoded otherwise functional IgVH sequences, suggesting that loss of function is not the major cause of GC B cell death. Unfortunately, we were unable to determine the contribution made by the loss or reduction of Ag specificity for cognate Ag because immunogens are unknown in routine tonsillectomies.
A second possibility affecting GC B cell survival is based on stochastic dynamics in which fate is determined by successful competition for limited binding sites with FDC and T cells (36, 37). Our data are more consistent with this explanation. Under this model, competition not only exists between clones with potentially different affinities for cognate Ag (as determined by different IgH V region amino acid sequences), but also between clones with equal affinity (identical amino acid sequences). Therefore, GC B cells in AnV– and AnV+ fractions could represent clones that successfully received FDC/T cell-dependent survival signals, and those that failed, respectively.
Shown in Fig. 5, we isolated identical IgH V region amino acid sequences from AnV– and AnV+ centroblast and centrocyte fractions, strongly suggesting that survival fate is determined by factors other than IgH primary sequence alone. In the example shown in Fig. 5, group 6, a sequence identified among AnV– centroblasts successfully progressed to the centrocyte stage, presumably after receiving CD40:CD40L-mediated survival signals from Th cells. However, centrocytes bearing this BCR were subsequently isolated from AnV– and AnV+ fractions, indicating that continual survival during the centrocyte stage requires the receipt of additional survival signals from FDC/Th cells. Nearly half of the sequences in group 6 contained one or more silent mutations, thus preserving the significance of our findings even after the conservative consideration of potential PCR amplification biases.
Our inability to identify consistent differences in AnV– and AnV+ B cell fractions, nor in AnV– and TBMac, based on the BCR primary sequence, is in agreement with recent reports that suggest that BCR signal strength is one of the major governing factors influencing B cell survival (61, 62, 63). It is impossible to prove a negative, although we put forth considerable effort in analyzing the sequence data under multiple permutations. We are currently performing several experiments to address potential pitfalls within our data and our experimental approach. Potential influences on survival contributed by differential Ig H and L chain pairing were not discernable in our experimental model. To address this issue, the existence of common IgH:IgL combinations in AnV– and AnV+ fractions is being evaluated through single cell PCR amplification of IgH and IgL transcripts.
We conclude that GC B cell death (as determined by AnV binding) can occur during centroblast and centrocyte stages, regardless of singular factors such as V region gene segment use, HCDR3 length and charge, or quantity of incorporated mutations. B cells bearing BCRs that confer selective advantage to entering the centrocyte pool merely carry the means to survival, but must subsequently have it verified by successful interactions with resident FDC/T cells to ensure continual survival.
Disclosures
The authors have no financial conflict of interest.
Acknowledgments
We appreciate the technical expertise and helpful discussions provided by Dr. P. C. Wilson of Oklahoma Medical Research Foundation and Dr. P. W. Williamson of Amherst College (Amherst, MA).
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 This work was supported by grants AI 12127 and COBRE P20RR15577 from the National Institutes of Health.
2 Address correspondence and reprint requests to Dr. J. Donald Capra, Molecular Immunogenetics, Oklahoma Medical Research Foundation, 825 Northeast 13th Street, Oklahoma City, OK 73104. E-mail address: The_Capras{at}msn.com
3 Abbreviations used in this paper: GC, germinal center; 7-AAD, 7-aminoactinomycin D; FDC, follicular dendritic cell; PI, propidium iodide; PS, phosphatidylserine; TBMac, tingible body macrophage; AnV, annexin V.
Received for publication August 20, 2004. Accepted for publication November 5, 2004.
References
Tonegawa, S.. 1983. Somatic generation of antibody diversity. Nature 302:575.
Alt, F. W., E. M. Oltz, F. Young, J. Gorman, G. Taccioli, J. Chen. 1992. VDJ recombination. Immunol. Today 13:306.
Kolar, G. R., J. D. Capra. 2003. Immunoglobulin structure and function. W. E. Paul, ed. Fundamental Immunology 47. Lippincott Williams and Wilkins, Philadelphia.
Pascual, V., Y. J. Liu, A. Magalski, O. de Bouteiller, J. Banchereau, J. D. Capra. 1994. Analysis of somatic mutation in five B cell subsets of human tonsil. J. Exp. Med. 180:329.
Neuberger, M. S., C. Milstein. 1995. Somatic hypermutation. Curr. Opin. Immunol. 7:248.
MacLennan, I. C.. 1994. Germinal centers. Annu. Rev. Immunol. 12:117.
Jacob, J., G. Kelsoe, K. Rajewsky, U. Weiss. 1991. Intraclonal generation of antibody mutants in germinal centres. Nature 354:389
Shih, T. A., E. Meffre, M. Roederer, M. C. Nussenzweig. 2002. Role of BCR affinity in T cell dependent antibody responses in vivo. Nat. Immunol. 3:570.
Kline, G. H., L. Hartwell, G. B. Beck-Engeser, U. Keyna, S. Zaharevitz, N. R. Klinman, H. M. Jack. 1998. Pre-B cell receptor-mediated selection of pre-B cells synthesizing functional μ heavy chains. J. Immunol. 161:1608.
Brandtzaeg, P.. 1996. The B-cell development in tonsillar lymphoid follicles. Acta Otolaryngol. Suppl. 523:55.
Tsunoda, R., E. Heinen, N. Sugai. 2000. Follicular dendritic cells in vitro modulate the expression of Fas and Bcl-2 on germinal center B cells. Cell Tissue Res. 299:395
Koopman, G., R. M. Keehnen, E. Lindhout, D. F. Zhou, C. de Groot, S. T. Pals. 1997. Germinal center B cells rescued from apoptosis by CD40 ligation or attachment to follicular dendritic cells, but not by engagement of surface immunoglobulin or adhesion receptors, become resistant to CD95-induced apoptosis. Eur. J. Immunol. 27:1.
Grouard, G., O. de Bouteiller, J. Banchereau, Y. J. Liu. 1995. Human follicular dendritic cells enhance cytokine-dependent growth and differentiation of CD40-activated B cells. J. Immunol. 155:3345.
Van Nierop, K., C. de Groot. 2002. Human follicular dendritic cells: function, origin and development. Semin. Immunol. 14:251
Liu, Y. J., G. Grouard, O. de Bouteiller, J. Banchereau. 1996. Follicular dendritic cells and germinal centers. Int. Rev. Cytol. 166:139.
Petrasch, S. G., M. H. Kosco, C. J. Perez-Alvarez, J. Schmitz, G. Brittinger. 1991. Proliferation of germinal center B lymphocytes in vitro by direct membrane contact with follicular dendritic cells. Immunobiology 183:451.
Liu, Y. J., C. Arpin, O. de Bouteiller, C. Guret, J. Banchereau, H. Martinez-Valdez, S. Lebecque. 1996. Sequential triggering of apoptosis, somatic mutation and isotype switch during germinal center development. Semin. Immunol. 8:169.
Choi, Y. S.. 1997. Differentiation and apoptosis of human germinal center B-lymphocytes. Immunol. Res. 16:161.
Reynolds, J. D.. 1986. Evidence of extensive lymphocyte death in sheep Peyer’s patches. I. A comparison of lymphocyte production and export. J. Immunol. 136:2005
Liu, Y. J., C. Arpin. 1997. Germinal center development. Immunol. Rev. 156:111.
Tuscano, J. M., K. M. Druey, A. Riva, J. Pena, C. B. Thompson, J. H. Kehrl. 1996. Bcl-x rather than Bcl-2 mediates CD40-dependent centrocyte survival in the germinal center. Blood 88:1359.
Klein, U., Y. Tu, G. A. Stolovitzky, J. L. Keller, J. Haddad, Jr, V. Miljkovic, G. Cattoretti, A. Califano, R. Dalla-Favera. 2003. Transcriptional analysis of the B cell germinal center reaction. Proc. Natl. Acad. Sci. USA 100:2639.
De Boer, R. J., A. A. Freitas, A. S. Perelson. 2001. Resource competition determines selection of B cell repertoires. J. Theor. Biol. 212:333.
Kesmir, C., R. J. De Boer. 2003. A spatial model of germinal center reactions: cellular adhesion based sorting of B cells results in efficient affinity maturation. J. Theor. Biol. 222:9.
Danial, N. N., S. J. Korsmeyer. 2004. Cell death: critical control points. Cell 116:205.
Koopman, G., C. P. Reutelingsperger, G. A. Kuijten, R. M. Keehnen, S. T. Pals, M. H. van Oers. 1994. Annexin V for flow cytometric detection of phosphatidylserine expression on B cells undergoing apoptosis. Blood 84:1415.
Vermes, I., C. Haanen, H. Steffens-Nakken, C. Reutelingsperger. 1995. A novel assay for apoptosis: flow cytometric detection of phosphatidylserine expression on early apoptotic cells using fluorescein labelled Annexin V. J. Immunol. Methods 184:39.
Verhoven, B., S. Krahling, R. A. Schlegel, P. Williamson. 1999. Regulation of phosphatidylserine exposure and phagocytosis of apoptotic T lymphocytes. Cell Death Differ. 6:262.
Krahling, S., M. K. Callahan, P. Williamson, R. A. Schlegel. 1999. Exposure of phosphatidylserine is a general feature in the phagocytosis of apoptotic lymphocytes by macrophages. Cell Death Differ. 6:183.
Segundo, C., F. Medina, C. Rodriguez, R. Martinez-Palencia, F. Leyva-Cobian, J. A. Brieva. 1999. Surface molecule loss and bleb formation by human germinal center B cells undergoing apoptosis: role of apoptotic blebs in monocyte chemotaxis. Blood 94:1012.
Van Eijk, M., C. de Groot. 1999. Germinal center B cell apoptosis requires both caspase and cathepsin activity. J. Immunol. 163:2478
Zheng, N. Y., K. Wilson, X. Wang, A. Boston, G. R. Kolar, S. M. Jackson, Y. J. Liu, V. Pascual, J. D. Capra, P. C. Wilson. 2004. Human immunoglobulin selection associated with class switch and possible tolerogenic origins for C- class-switched B cells. J. Clin. Invest. 113:1188.
Pascual, V., P. Wilson, Y. J. Liu, J. Banchereau, J. D. Capra. 1997. Biased VH4 gene segment repertoire in the human tonsil. Chem. Immunol. 67:45.
Zhang, Z., Y. H. Wang, M. Zemlin, H. W. Findley, S. L. Bridges, P. D. Burrows, M. D. Cooper. 2003. Molecular mechanism of serial VH gene replacement. Ann. NY Acad. Sci. 987:270
Zemlin, M., K. Bauer, M. Hummel, S. Pfeiffer, S. Devers, C. Zemlin, H. Stein, H. T. Versmold. 2001. The diversity of rearranged immunoglobulin heavy chain variable region genes in peripheral blood B cells of preterm infants is restricted by short third complementarity-determining regions but not by limited gene segment usage. Blood 97:1511.
Radic, M. Z., J. Mackle, J. Erikson, C. Mol, W. F. Anderson, M. Weigert. 1993. Residues that mediate DNA binding of autoimmune antibodies. J. Immunol. 150:4966.
Aguilera, I., J. Melero, A. Nunez-Roldan, B. Sanchez. 2001. Molecular structure of eight human autoreactive monoclonal antibodies. Immunology 102:273
Callahan, M. K., M. S. Halleck, S. Krahling, A. J. Henderson, P. Williamson, R. A. Schlegel. 2003. Phosphatidylserine expression and phagocytosis of apoptotic thymocytes during differentiation of monocytic cells. J. Leukocyte Biol. 74:846.
Dillon, S. R., A. Constantinescu, M. S. Schlissel. 2001. Annexin V binds to positively selected B cells. J. Immunol. 166:58
Dillon, S. R., M. Mancini, A. Rosen, M. S. Schlissel. 2000. Annexin V binds to viable B cells and colocalizes with a marker of lipid rafts upon B cell receptor activation. J. Immunol. 164:1322.
Wardemann, H., S. Yurasov, A. Schaefer, J. W. Young, E. Meffre, M. C. Nussenzweig. 2003. Predominant autoantibody production by early human B cell precursors. Science 301:1374
Ninomiya, E., Y. Ito, M. A. Shibata, K. Kawashima, T. Sakamoto, E. Maruyama, H. Doi, K. Tokitsu, Y. Otsuki. 2003. The activation of caspase-3 and DNA fragmentation in B cells phagocytosed by macrophages. Med. Electron Microsc. 36:87.
Nakamura, M., H. Yagi, S. Kayaba, T. Ishii, T. Gotoh, S. Ohtsu, T. Itoh. 1996. Death of germinal center B cells without DNA fragmentation. Eur. J. Immunol. 26:1211.
Allen, D., T. Simon, F. Sablitzky, K. Rajewsky, A. Cumano. 1988. Antibody engineering for the analysis of affinity maturation of an anti-hapten response. EMBO J. 7:1995.
Casola, S., K. L. Otipoby, M. Alimzhanov, S. Humme, N. Uyttersprot, J. L. Kutok, M. C. Carroll, K. Rajewsky. 2004. B cell receptor signal strength determines B cell fate. Nat. Immunol. 5:317
Niiro, H., E. A. Clark. 2002. Regulation of B-cell fate by antigen-receptor signals. Nat. Rev. Immunol. 2:945.
Pittner, B. T., E. C. Snow. 1998. Strength of signal through BCR determines the fate of cycling B cells by regulating the expression of the Bcl-2 family of survival proteins. Cell. Immunol. 186:55.(Stephen M. Jackson and J.)