In case of reinfection with the same pathogen, memory T cells will mount a prompt response by immediately producing effector cytokines and by rapidly proliferating into a large number of secondary effectors [ 1 — 4 ]. This substantial increase in antigen-responsiveness of both effector and memory T cells upon reencounter with the pathogen is a fundamental property of adaptive immunity.
Lymphocytes recognize antigens through specialized antigen receptors. During the cause of an immune response, a high number of point mutations take place in the BCR genes of the dividing B cells. This result in a panel of B cells expressing BCR with varying affinities against the antigen, and the B cells carrying BCR with the highest affinity are selectively expanded. As a consequence, high-efficiency B cells are selected during the immune response in a process known as affinity maturation [ 5 ].
Unlike B cells, T cells lack the capacity to mutate their TCR genes after T-cell activation, and thus classical affinity maturation does not take place in T cells.
- Biochemistry | Book Download Site. | Page 3.
- Doch, ich kenne meine Größe! - Junger Mann verzweifelt am Shoppen (German Edition).
- The biochemistry of somatic hypermutation.
- Associated Data.
They found that only primed T cells produced IL-2 and proliferated in vitro in response to TCR triggering induced by anti-CD3 antibodies and monocytes [ 14 ]. Similar observations were later reported by others [ 7 , 9 — 13 , 15 ]. This finding was supported in a subsequent study, where Croft et al. In an equivalent study also examining T-cell responses to infection, Pihlgren et al. An overview of studies indicating the existence of functional avidity maturation is given in Table 1. Today, it is widely accepted that T-cell activation should not be considered as a single signal process, but as a sum of interdependent signals.
TCR signaling takes place at the interface between the T-cell and the antigen presenting cell. At this contact zone, often referred to as the immunological synapse IS , TCR-signaling components including the TCR itself as well as intracellular-signaling molecules are continuously accumulated during antigen contact [ 26 ]. Although somewhat controversial [ 26 , 27 ], formation of an IS correlates with generation of a robust immune response, and is considered a prerequisite for T-cell activation [ 28 , 29 ].
Even so, new insight into the biology of immunological synapses has revealed that TCR signaling is already initiated in TCR microclusters prior to IS formation. At the IS, CD28 signaling both induces structural stabilization and enlargement of the area itself [ 32 , 33 ]. Eventhough the exact implication of CD28 signaling in T-cell activation is still elusive, it is generally agreed that CD28 amplifies intracellular signaling induced by antigen-triggering of the TCR through modulation of morphological features and TCR signals [ 32 , 33 ].
CD45 is a transmembrane tyrosine phosphatase that maintains Lck activity by promoting dephosphorylation of an inhibitory carboxy-terminal tyrosine residue of Lck. Lck activity is a necessity for initiation of TCR signal transduction [ 39 ]. This finding parallels the study of Kersh et al. The vast majority of studies contributing to the current model for TCR signaling were performed using immortal T-cell lines or primed T cells propagated in vitro. The discrepancy between the two human studies might be due to two different primed T-cell populations studied effector and memory cells, resp.
Adachi and Davis used high concentrations of soluble anti-CD3 and anti-CD28 antibodies cross-linked by secondary antibodies to stimulate the T cells. By using cross-linked antibodies for stimulation, a very strong receptor signaling is achieved. As illustrated in a series of mouse virus studies, the strength of TCR signaling determines the requirement for additional activation signals like CD28 signaling and also results in somewhat different responses [ 19 ].
Both scenarios could be relevant for human immunity where a wide range of pathogens with different origins is encountered. Unfortunately, mouse and man seem to differ when it comes to some of the signaling molecules involved in TCR signaling. Early studies describing a need for a third-signal cytokine came from a series of in vitro and in vivo experiments performed by Mesher and co-workers.
Eventhough IL has a role in skewing the CD4 T-cell response, it has no effect on CD4 T-cell proliferation and differentiation in response to antigen. Analysis of B cell clonal expansion, diversification, and selection processes thus critically depends on an accurate background model for SHM micro-sequence targeting i.
1. INTRODUCTION
Here, we combine high-throughput Ig sequencing with new computational analysis methods to produce improved models of SHM targeting and substitution that are based only on synonymous mutations, and are thus independent of selection. The estimated profiles can explain almost half of the variance in observed mutation patterns, and clearly show that both mutation targeting and substitution are significantly influenced by neighboring bases. While mutability and substitution profiles were highly conserved across individuals, the variability across motifs was found to be much larger than previously estimated.
The model and method source code are made available at http: During the course of an immune response, B cells that initially bind antigen with low affinity through their immunoglobulin Ig receptor are modified through cycles of proliferation, somatic hypermutation SHM , and affinity-dependent selection to produce high-affinity memory and plasma cells. G mismatches in the Ig V D J sequence.
AID targeting is dependent on RNA polymerase II pausing
These pathways operate in an error-prone manner to introduce the full spectrum of mutations at the initial lesion, as well as spreading mutations to the surrounding bases. While the process of SHM appears to be stochastic, there are clear intrinsic biases, both in the bases that are targeted 5 , 6 as well as the substitutions that are introduced 7 , 8.
Accurate background models for SHM micro-sequence targeting i. In addition, targeting and substitution models could provide important insights into the relative contributions of the various error-prone DNA repair pathways that mediate SHM. Computational models and analyses of SHM have separated the process into two independent components 7 — In experimentally derived Ig sequences, observed mutation patterns are influenced by selection.
The affinity maturation process selects for affinity-increasing mutations, while many mutations at structurally important positions in the framework regions are selected against To avoid the confounding influences of selection, most existing models are built using mutation data from intronic regions flanking the V gene 13 and non-productively rearranged Ig genes 6 — 10 , A single cold-spot motif has also been recognized: Despite the wide recognition of these specific hot-spot and cold-spot motifs, it is clear that a hierarchy of mutabilities exists that is highly dependent on the surrounding bases 7 , More recently, it has been recognized that the profile of nucleotide substitutions may also be dependent on the surrounding bases 8 , Modeling SHM targeting and substitution is important for the analysis of mutation patterns since these intrinsic biases can give the appearance of selection due to the particular codon usage and base composition in Ig sequences 17 , Moreover, having such a model could shed light on the molecular mechanisms underlying SHM, which are not fully understood.
Previous work has attempted to model the dependencies on surrounding bases, but has been limited to at most the targeted base and three surrounding bases 19 , mainly due to the relatively small data sets available. The use of intronic regions has also limited the number of motifs that can be modeled because of the limited diversity of these regions , and non-productively rearranged Ig genes may still be influenced by selection e. In this study, we take advantage of the wealth of data available from high-throughput Ig sequencing technologies to build improved targeting and substitution models for SHM.
To avoid the biasing effects of selection, we have developed a new methodology for constructing models from synonymous mutations only, thus avoiding the need to limit analysis to non-productive Ig sequences. The increased data set size allows modeling of dependencies on the surrounding four bases two bases upstream and downstream of the mutation.
We also find that the nucleotide substitution profiles at all bases are dependent on the surrounding nucleotides.
The S5F targeting and substitution models can be employed as background distributions for mutation analysis, such as the detection and quantification of affinity-dependent selection in Ig sequences 11 , These models improve dramatically the ability to analyze mutation patterns in Ig sequences, and provide insights into the SHM process. These data were derived from 7 human blood and lymph node samples, and Ig sequencing was carried out using both Roche and Illumina MiSeq next-generation sequencing technologies.
These high-fidelity sequences were clustered to identify clones sequences related by a common ancestor and one effective sequence was constructed per clone so that each observed mutation corresponded to an independent event. Overall, this process produced a set of , synonymous mutations that were used to model somatic hypermutation targeting and substitution. The different filters applied to arrive at the number of synonymous mutations used for the targeting and substitution models are described in the text. All three studies relate to manuscripts in preparation. A nucleotide substitution model specifies the probability of each base A, T, G, or C mutating to each of the other three possibilities.
These probabilities may depend on the surrounding bases i. To derive a nucleotide substitution model, the set of mutations was filtered to include only those that occurred in positions where none of the possible base substitutions lead to amino acid exchanges. Focusing on positions where only synonymous mutations were possible removes the confounding influence of selection. For each of the possible 5-mers M , a substitution model was derived by calculating S B M , the probability that the central base in the 5-mer motif M mutates to base B.
In this case, the number of observed mutations that led to each of the other three possible nucleotides C, G, or T was recorded: The maximum likelihood value for the probability that A is substituted by base B is then calculated as:. Comparison of the substitution profiles for different 5-mer motifs with the same central base clearly showed the significant influence of surrounding bases. If one ignores neighboring nucleotides, the substitution profiles were qualitatively similar to previous estimates 7 , although significant quantitative differences were apparent presumably due to the much larger size of the dataset compiled here.
Thus, nucleotide substitution profiles at every base are significantly affected by adjacent nucleotides, including at least two bases on either side of the mutating base. The substitution profile is significantly influenced by surrounding bases. Substitution profiles for various micro-sequence contexts are shown for substitutions at A guanine and B adenosine, cytidine, and thymidine. Horizontal lines in A indicate the substitution values for the S1F model following the color scheme shown in the legend.
Motifs that fall into one of the standard hot or cold-spots categories are indicated by the motif above the column. It is not possible to estimate substitution profiles for all 5-mer motifs using the above methodology because: Among the 11 datasets used here, these issues prevent estimation of the substitution profiles for of the 5-mers. To infer values for the missing motifs, four methods were evaluated.
In the second and third methods, missing values were replaced by averaging over motifs sharing the two bases upstream and downstream of the mutated base, respectively. To choose between these four methods, we compared their performance on 5-mers that could be directly estimated from the data. Pearson and Spearman coefficients were both used in order to be robust to the linear dependency assumption, and they yielded comparable results. In contrast to previous studies 8 , there was no significant correlation between substitution values of 5-mers and their reverse complements Pearson correlation of 0.
The mutability of a motif is defined here as the non-normalized probability of the central base in the motif being targeted for SHM relative to all other motifs. Similar to the substitution model, the targeting model was based on 5-mer motifs, including the two nucleotides immediately upstream and downstream of the mutated base. The use of a 5-mer model is motivated by the well-known WR C Y hot-spot where the underlined C is targeted for mutation , and its reverse-complement R G YW which, when taken together, create dependencies with the two bases on either side of the mutating base.
To see why this is the case, consider the extreme example of a sequence composed of all C nucleotides. When calculating mutabilities it is also important to avoid statistical artifacts due to heterogeneity e.
Thus, Ig sequences were first analyzed individually since each has a different background 5-mer distribution. These individual-sequence targeting models were then combined into a single aggregated targeting model for each data set. Estimating the relative mutabilities of 5-mer motifs for an individual Ig sequence involves two steps: To avoid the confounding influence of selection, only mutations that were synonymous i.
Note that these criteria are slightly different from those used in the substitution model. For each of the possible 5-mers motifs M in each Ig sequence, the background frequency B M was calculated as follows:. A similar array was also calculated for the mutated positions:. The resulting vector was renormalized so that the mean mutability was one. It was not possible to estimate mutabilities for of the possible 5-mer motifs because not all 5-mers appeared within the set of Ig sequences.
The same four methods tested for inferring missing values in the substitution model were also tested to infer these mutabilities see 2. The S5F targeting model is consistent across individuals. Estimated values for all 5-mer motifs derived using lymph node samples from three individuals LN, LN, and LN are shown along the diagonal. Symbols indicate the mutated nucleotide in the center of the 5-mer.
The biochemistry of somatic hypermutation.
Correlations between the mutabilities for all 5-mer motifs across individuals are shown in the upper log-log scale and lower linear scale triangles. A 5-mer mutabilities estimated directly from the Ig sequencing data. B The complete S5F targeting model after inferring values for missing 5-mer motifs.
Bars radiating from each circle depict the mutability as a function of surrounding bases. Each plot corresponds to a different mutated nucleotide. There is a The mutabilities estimated by the S5F approach paint a qualitatively different picture of SHM when compared with those estimated by the existing tri-nucleotide model of Shapiro et al.
Using the tri-nucleotide model, hot-spots were only 1. The mutabilities estimated by the S5F model better predicted the positional-distribution of in vivo mutations. The Pearson correlation between the expected mutability and observed mutation frequency calculated over IMGT-numbered positions in 12, sequences derived from a variety of germline segments was 0.
The observation of position-specific signals suggests that there is something generic about the Ig structure at these positions, and may help refine traditional definitions of the complementarity determining regions CDR and framework regions FWR see also Overall, the S5F targeting model provides a new view of SHM with hot-spots being significantly more targeted and significantly more variable than previously thought.
Boxes borders correspond to the first and fourth quartiles while the horizontal bar inside the box corresponds to the median of the distribution. Comparison between expected and observed somatic hypermutation targeting. The correlation across positions points is shown in the inset of B. Two positions with evidence of negative selection red circles and one position with positive selection blue circles are indicated.
The threshold for calling a position with significant selection was set to 3 SD away from the linear regression line shown as a solid line in the inset, with thresholds plotted as dashed lines. Lines connect reverse-complement motifs for cases where both could be directly estimated from the data.
A total of 11 human Ig repertoires were sequenced from blood and lymph node samples from 7 different individuals. These samples were originally collected and sequenced as part of three ongoing studies manuscripts in preparation. Human lymph node specimens were collected under an exempt protocol approved by the Human Research Protection Program at Yale School of Medicine. Tissues were processed and RNA isolated as previously described Blood samples were collected under the approval of the Personal Genome Project To carry out sequencing, mRNA was reverse transcribed into cDNA using gene-specific primers mapping to the constant region of the Ig heavy chain.
The amplified library was tagged with barcodes for sample multiplexing, PCR enriched, and annealed to the required Illumina clustering adapters.
High-throughput base-pair paired-end sequencing was performed using the Illumina MiSeq platform. Raw reads were exported without the sample barcodes and Illumina clustering adapters. Ig heavy chain mRNA were reverse-transcribed using a pool of 6 primers specific to the Ig constant regions and cDNA was amplified using 16 cycles of PCR with a pool of 46 V-region-specific primers and 6 nested constant region primers.
Following ligation of compatible sequencing adapters, the expected heavy chain V gene fragments were purified using PAGE. Each sample was uniquely barcoded during the ligation process, allowing subsequent mixing of all the samples into one common reaction sample performed independently for each replicate run. Raw sequencing reads were filtered in several steps to identify and remove low-quality sequences. Conservative thresholds were applied in all cases to increase the reliability of the resulting mutation calls, at the potential expense of excluding some real mutations. We have constructed new SHM targeting and substitution models using a collection of more than , synonymous mutations from next-generation Ig sequencing studies.
The exclusive use of synonymous mutations allowed us to include mutations from functional Ig sequences without the biasing influence of selection. The large size of the resulting mutation data set allowed us to model targeting and substitution dependencies on the mutating base as well as on two bases upstream and downstream of the mutation. This high variance demonstrates the importance of including higher order dependencies, as we have done. It has been suggested that nucleotide substitution profiles are also dependent on the micro-sequence context of the mutating base 8 , We confirm that the substitution profiles at all nucleotides are highly dependent on neighboring bases and these dependencies are conserved across individuals.
Interestingly, the fact that substitution rates depend on surrounding bases may resemble the situation in meiotic mutations as was suggested in the past 9.