The JI
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     
 

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Related articles in The JI
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow Request Permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Arudchandran, A.
Right arrow Articles by Max, E. E.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Arudchandran, A.
Right arrow Articles by Max, E. E.
Right arrowPubmed/NCBI databases
*Compound via MeSH
*Substance via MeSH
The Journal of Immunology, 2004, 173: 3223-3229.
Copyright © 2004 by The American Association of Immunologists

Single-Stranded DNA Breaks Adjacent to Cytosines Occur during Ig Gene Class Switch Recombination

Arulvathani Arudchandran, Ralph M. Bernstein and Edward E. Max1

* Division of Therapeutic Proteins, Office of Biotechnology Products, Center for Drug Evaluation and Research, Food and Drug Administration, Bethesda, MD 20892


   Abstract
Top
 Abstract
Introduction
 Materials and Methods
 Results
 Discussion
 References
 
Class switch recombination (CSR) at the DNA level underlies ability of B lymphocytes to switch from expressing IgM to expressing IgG, IgA, or IgE. The mechanism of CSR is largely unknown, but it is clear that CSR is stimulated by T cell signals and is mediated in part by activation-induced deaminase (AID), an enzyme that is also required for somatic hypermutation of Ig genes. In one current model, AID is proposed to initiate CSR by deaminating cytosines in the unpaired nontemplate strand of DNA displaced from its complementary strand by the "sterile" RNA transcript across the switch region. We have used LM-PCR to analyze single-strand breaks in CH12F3-2, a murine cell line that switches in vitro to IgA expression. In contrast to the above model, we have detected CSR-associated ssDNA breaks in the template strand of the H chain {alpha} switch region, the strand thought to be complexed with RNA. Most breaks are adjacent to cytosines, consistent with mediation by AID, and occur within the novel consensus sequence C*AG, which occurs much more frequently on the template strand than on the putatively displaced nontemplate strand. These results suggest that AID may target the DNA strand bound to RNA, perhaps resembling APOBEC-3G, a cytosine deaminase related to AID that inhibits HIV replication by mutating viral DNA. Furthermore, the absence of detectable breaks in the nontemplate strand within the DNA segment under study suggests that the two DNA strands are handled differently in the generation or processing of strand breaks.


   Introduction
Top
 Abstract
 Introduction
Materials and Methods
 Results
 Discussion
 References
 
Resting mature lymphocytes express surface IgM. On activation by exposure to Ag, B lymphocytes migrate to germinal centers, where they encounter an environment that triggers somatic hypermutation (SHM)2 and isotype switching. SHM generates B cells that produce mutated Ig proteins, which undergo selection for Ag binding, leading to affinity maturation. As a result of isotype switching, B lymphocytes cease IgM production and initiate synthesis of new isotypes, IgG, IgA, or IgE, that retain the same Ag specificity as the original IgM but acquire new effector functions. This switch results from a DNA recombination event, class switch recombination (CSR), that deletes the Cµ region from its position downstream of the rearranged V(D)J region, and replaces it with a downstream H chain C region that is then expressed as the C-terminal portion of the H chain.

Although SHM and CSR were not initially considered mechanistically related, several observations have made it apparent that these processes share several key features. Most striking has been the discovery that both SHM and CSR depend on the integrity of the enzyme activation-induced deaminase (AID), which was originally discovered by subtractive cloning as a protein induced under conditions promoting CSR (1). This protein bears sequence similarity to APOBEC-1, a cytosine deaminase discovered as an mRNA editing enzyme, and AID itself was shown to have deaminase activity in vitro. Mice (2) and humans (3) with defective AID genes show dramatic impairment in CSR and SHM. Although AID might hypothetically promote CSR and SHM by editing mRNA transcripts required for these processes, recent evidence has indicated that AID can deaminate cytosines to uracil in ssDNA (4, 5, 6, 7). V region DNA would be temporarily single stranded during transcription of these regions, allowing deamination that ultimately results in SHM; this would be consistent with the known requirement of transcription for SHM.

CSR also requires transcription across the repetitive G-rich "switch" (S) regions located upstream of the Ig H chain C regions Cµ, C{gamma}, C{epsilon}, and C{alpha}. The T cell stimuli that promote switching to a particular isotype invariably also promote transcription initiating upstream of the corresponding S region and continuing downstream through the associated CH region. In vitro transcripts across S regions have been found to form unusual stable RNA-DNA hybrids with the template strand (8, 9). The RNA transcript of the S region displaces the nontemplate strand as an unpaired "R-loop"; and evidence has been reported that similar structures may form in vivo (10). Thus, there is some evidence to support the idea that the nontemplate strand of S regions could be a target for AID.

After deamination by AID causes a C to U mutation in DNA, the nonstandard DNA base U can be excised by uracil DNA glycosylase (UDG). Defects in the murine (11) or human (12) genes encoding UDG are associated with impaired CSR and abnormalities in the V region mutations resulting from SHM, implying a role for this enzyme in both processes. The abasic site in the affected DNA strand might then be a substrate for cleavage by apurinic endonuclease (APE), leading to a single-strand break (or nick). The subsequent events leading to recombination are poorly understood, but switching defects in various genes have suggested roles for several components of the DNA break repair machinery. Null mutants of the nonhomologous end-joining factors, Ku70 (13) and Ku80 (14), are dramatically impaired in CSR, and partial impairments have been reported for DNA protein kinase (15, 16), for the histone variant H2AX (17), for components of the mismatch repair machinery, such as Msh2 (18, 19), and for exonuclease-1 (20).

Several previous reports have described blunt end double-strand breaks (dsbs) in the DNA of Ig S regions from cells undergoing CSR. In the initial report (21), ligation-mediated PCR (LM-PCR) (see Fig. 1A) was used to amplify blunt dsbs in the S{gamma}3 region of splenic B cells switching to IgG3 expression. Two predominant B cell-specific cleavages in S{gamma}3 were observed, correlated with the onset of in vitro CSR. A similar strategy has been used to detect breaks in the human Sµ region (22, 23), yielding ladders of multiple amplified products. Other investigators have proposed CSR models that would predict not blunt, but staggered, cleavages targeted near stem-loop structures formed in ssDNA (24); certainly staggered ends would be produced if all cleavages occur adjacent to cytosines deaminated by AID.



View larger version (15K):
[in this window]
[in a new window]
  		FIGURE 1. Ligation-mediated PCR strategies. A, Blunt-end detection. A blunt double-stranded cleavage in genomic DNA (top line) is detected by ligation of a double-stranded linker; then, DNA is amplified between a gene-specific primer and a primer representing one strand of the linker, the linker primer (LP). The resulting amplified product can be detected by Southern blotting or by primer extension. B, Detecting single-strand cuts. Genomic DNA containing a nick on the lower strand is denatured, and a gene-specific primer (P1) is hybridized and extended to the position of the nick (first strand extension), leaving a blunt end. Then, (as in A above) a double-stranded linker is ligated to the newly created blunt end, and amplification is performed between a second gene-specific primer (P2) and the linker primer. The amplified product is detected by primer extension using a third, radiolabeled, gene-specific primer (P3). C, Detecting single-strand cleavages in the 5' region of murine S{alpha}. The three primers, 5'P1, 5'P2, and 5'P3, were used to detect lower (template) strand cleavages; the downstream primers, 3'P1, 3'P2, and 3'P3, were used to detect upper strand cleavages. The graphic shows the positions of recognition sites for several restriction endonucleases used in this study.

 
To investigate CSR-associated DNA breaks without restricting our analysis to blunt ends, we used an LM-PCR strategy designed to detect single-strand breaks, as described in the original report of LM-PCR. As shown in the strategy diagram of Fig. 1B, genomic DNA containing single-strand nicks is denatured, and a sequence-specific primer (P1) is used to extend the DNA, forming a blunt end at the position of the nick. This is followed by blunt-end ligation to a linker, then PCR amplification, and detection by electrophoresis and autoradiography. We have used this technique to study nicks in the H chain {alpha} switch region (S{alpha} region) of the murine cell line CH12F3-2, which undergoes efficient switching to IgA expression with appropriate cytokine stimulation (25). We show that this technique can detect DNA breaks resulting from both blunt and staggered cleavages. However, all the S{alpha} breaks we detect during CSR derive from single-strand nicks, mostly adjacent to cytosines in the template strand.

As this study was being completed, another laboratory described AID-dependent dsbs with staggered ends occurring in Sµ in splenic B cells (26).


   Materials and Methods
Top
 Abstract
 Introduction
 Materials and Methods
Results
 Discussion
 References
 
Induction of class switching in CH12 F3-2 cells

CH12F3-2A cells were obtained from Dr. T. Honjo, and maintained as described (25). At a density of 2 x 104/ml, cells were stimulated with 2 ng/ml IL-4 (PeproTech, Rocky Hill, NJ), 1 ng/ml TGF{beta}1 (R&D Systems, Minneapolis, MN), and with 0.5 µg/ml anti-mouse CD40 mAb (BD Pharmingen, San Diego, CA) for 3 days. After culture for 24, 48, or 72 h, cells were double-stained with FITC-conjugated rat anti-mouse IgA mAb (BD Pharmingen) and with R-PE–conjugated rat anti-mouse IgM mAb, and were analyzed by FACSCalibur (BD Biosciences, San Diego, CA).

Detection of breaks in the S{alpha} region

CH12F3-2 cells were treated with cytokines as described above (except where noted) for varying time periods. Genomic DNA was prepared from both cytokine-treated and -untreated control cells by standard procedures. Primers were designed based on the published S{alpha} sequence (GenBank D11468) according to published criteria (27), and were synthesized and HPLC-purified by Sigma-Genosys (The Woodlands, TX). LM-PCR was conducted as described (27). Vent DNA polymerase (New England Biolabs, Beverly, MA) was used in all extension steps. Two micrograms of genomic DNA were used in each reaction. Briefly, for detection of lower strand breaks using primers 5' of S{alpha}, first strand primer extension was conducted with the 5'P1 gene-specific primer (TGACCCATCCACAGGCAATCAC) for one cycle (94°C, 5 min; 60°C, 30 min; 76°C, 10 min), followed by ligation of the linker (annealed linker oligos 5'HO-GCGGTGACCCGGGAGATCTGAATTC (linker primer) and 5'HO-GAATTCAGATC) (27) at 16°C overnight. PCR amplification was conducted for 23 cycles with the linker primer and the second gene-specific primer (5'P2, CAGCAACAGGAGACTCCCAGGC) at the annealing temperature of 51.5°C for 2 min (the success of the lower strand LM-PCR was exquisitely sensitive to the annealing temperature). Extensions were at 76°C for 3 min, with 5 s increase in every other step. The third gene-specific primer(5'P3, GACTCCCAGGCTAGACAGAGGCAAGG) was end-labeled with [33P]ATP (Amersham Biosciences, Piscataway, NJ) using T4 polynucleotide kinase (New England Biolabs) according to the published protocol (27). This primer was used in the third extension step: two cycles of denaturation at 94°C for 1 min, annealing 64°C for 2 min, and extension at 76°C for 10 min. For detection of upper strand breaks, all reaction conditions were identical except that the 3'P1 primer was CAAACCACCCTAGTCTAGCCCAACTC, 3'P2 primer (GCCCATGCTAGCTTAGCCTAGCTC) was annealed at 60°C for 2 min in the PCR amplification steps, and 3'P3 was CTAGCTCAGCCCAGTTTAGCCCAGTCCAC. Sequence analysis of LM-PCR clones indicated that the 3'P3 primer had two mismatches with the corresponding CH12F3-2 genomic template sequence, but the mismatches are in the 5' half of the primer, and apparently allowed it to function adequately (as shown in Fig. 4.). Amplification products were electrophoresed on denaturing polyacrylamide gels and visualized on Fuji FLA 3000 scanner (Stamford, CT).



View larger version (41K):
[in this window]
[in a new window]
  		FIGURE 4. Assay for upper strand breaks. Genomic DNA from CH12F3-2 cells incubated for 24 h with TGF{beta}, IL-4, and anti-CD40 was subjected to the LM-PCR protocol designed to detect breaks in the upper (nontemplate) DNA strand (see Fig. 1C, 3' primer set), in the same region of S{alpha} where lower strand breaks were observed. As shown in the first lane, no bands were detected. As positive controls, CH12F3-2 DNA samples (from untreated cells) were digested with several restriction enzymes and then subjected to the upper strand LM-PCR protocol in parallel with the undigested DNA from cytokine-treated cells.

 
Annexin-V treatment

CH12F3-2 cells treated with cytokines for 24 h were harvested and stained with FITC-conjugated annexin-V (BD Pharmingen) according to the manufacturer’s protocol. Stained cells were sorted into annexin-V-positive and -negative populations in FACSVantage SE (BD Biosciences) and washed twice with 1x PBS.

Characterization of the DNA breaks

To detect whether the breaks observed in the S{alpha} region were either staggered- or blunt-end cuts, we conducted LM-PCR on genomic DNA obtained from the 24-h cytokine-treated and -untreated cells with and without carrying out the initial denaturation and primer extension. Ten microgram samples of undenatured genomic DNA were treated with 30 U of T4 DNA polymerase (New England Biolabs) in 500 µl reaction volume at 12°C for 20 min as described by the manufacturer to convert any staggered ends to blunt ends that could be ligated directly to the linker. The subsequent amplification and the labeling steps of LM-PCR were same as previously described. Restriction enzyme-digested genomic DNA samples were also used in this experiment as controls.

Cloning and sequencing of S{alpha} break fragments

LM-PCR as described above was conducted with the genomic DNA from the induced CH12F3-2 cells (24 h after treatment with cytokines) with the following modification. After the amplification step with P2 primer, DNA was extracted with phenol:chloroform, precipitated with 2 vol of ethanol, resuspended in 25 µl of water and subjected to a third amplification step: 30 cycles of denaturing at 94°C for 1 min, annealing at 55°C for 30 s, and extension at 72°C for 1.5 min, with the P3 primer and the large linker primer. The amplified product in 2 µl of the PCR was cloned into the TOPO TA vector (Invitrogen Life Technologies, Carlsbad, CA) using the manufacturer’s protocols, and the white colonies were subjected to colony PCR using the P3 and linker primers; clones yielding a single intense amplified product were sequenced. Sequencing reactions were conducted with the Bigdye reaction mix (Applied Biosystems, Foster City, CA) and sequences were obtained using an ABI prism sequencer (Applied Biosystems). The sequence of the 5' region of the S{alpha} region of CH12F3-2 (submitted to GenBank with accession number AY640116) was determined by analyses of LM-PCR clones and of a fragment amplified between PCR primers 5'P3 and 3'P3.


   Results
Top
 Abstract
 Introduction
 Materials and Methods
 Results
Discussion
 References
 
In designing a strategy to detect DNA breaks in S{alpha}, we initially concentrated on the 5' portion of S{alpha}, where recombination junctions have been reported to occur preferentially (28), and where primers could be designed upstream of S{alpha}, avoiding the highly repetitive region. A more recent and larger compilation of recombination breakpoints has described a more even distribution within S{alpha} (29). To detect breaks or nicks in the lower (template) strand we used the primer set 5'P1, 5'P2, and 5'P3, as shown in Fig. 1C. Samples of CH12F3-2 DNA digested with several restriction enzymes known to cleave the 5' target region of S{alpha} were used as positive controls and to generate size markers (Fig. 2A). In several experiments with this 5' primer set applied to DNA from cells cultured with IL-4, TGF{beta}, and an anti-CD40 mAb (here designated anti-CD40), we detected a ladder of bands apparently reflecting preferred cleavage sites in the lower strand, in a pattern that was similar from experiment to experiment. The intensity of the bands was maximal at 24 h after the initiation of cytokine stimuli (Fig. 2A). A decline in detectable breaks after 24 h would be expected based on the time course of the percentage of cells with surface IgA expression of CH12F3-2 cells, which is substantial by 48 h and levels off at ~50% by 72 h (data not shown); cells that are IgA positive would be expected to have completed CSR and to have sealed any DNA breaks. We chose the 24-h time point for subsequent experiments.



View larger version (49K):
[in this window]
[in a new window]
  		FIGURE 2. Single-strand cleavages in the lower strand of S{alpha}. A, Electrophoresis bands from LM-PCR products derived from CH12F3-2 cells treated with TGF{beta}, IL-4, and anti-CD40 for varying times. Arrows in the left panel highlight LM-PCR products from untreated (0 h) CH12F3-2 genomic DNA digested with SpeI or BsmI; these bands serve as size markers and positive controls. B, CH12F3-2 cells were treated for 24 h with TGF{beta}, IL-4, and anti-CD40, and then sorted for annexin-V before DNA isolation and LM-PCR. C, LM-PCR pattern from CH12F3-2 cells treated for 24 h with varying combinations of cytokines. D, DNAs from untreated (0 h) or 24-h cytokine-treated CH12F3-2 cells were assayed for blunt- or staggered-end cleavages by direct linker ligation (with or without prior T4 polymerase treatment) and PCR.

 
In the previously reported analysis of breaks in the S{gamma}3 switch region, several cleavage hotspots were found to be due to apoptosing cells in the studied population (21). To explore whether apoptosis-related DNA breaks contributed to the ladder we observed in CH12 cells, we sorted cytokine-stimulated cells for surface staining with annexin-V, a characteristic of apoptosing cells. Annexin-V positive cells represented ~5% of our cytokine-treated CH12 cells. Equal quantities of DNA from the annexin-V-positive and -negative cell fractions were subjected to LM-PCR. The cytokine-treated cells sorted for absence of annexin-V staining showed essentially the same bands as unsorted cells, while annexin-V-staining cells yielded a diffuse smear with a pattern of faint bands different from that pattern shared by DNA from unsorted and annexin-V-negative cells (Fig. 2B). To study the cytokine dependence of the cleavages, we incubated cells with various combinations of IL-4, TGF{beta}, and anti-CD40. As shown in Fig. 2C, all three of these agents individually produced weak bands at many of the same positions seen with the combination, while the combination of all three stimuli produced the familiar ladder of intense bands. The greater intensity of bands in the IL-4 plus anti-CD40 lane was not a consistent finding. The ability of single stimuli to induce a low level of DNA cleavage is consistent with the reported ability of these single agents to induce low levels of switching (25) and modest stimulation of mRNA for AID (1). Thus, the reproducible LM-PCR pattern of lower strand breaks in S{alpha} requires stimuli known to be associated with isotype switching, occurs with a time course expected for an intermediate in CSR, and is not generated by apoptosing cells. Taken together, these observations suggest that the reproducible breaks in S{alpha} are associated with switch recombination.

Previous reports of S region breaks detected by LM-PCR (21, 22, 23) had indicated that the breaks were blunt ended, in that they could be detected with the strategy of Fig. 1A involving direct ligation of the blunt linker to double-stranded genomic DNA. Our routine strategy would be expected to detect single-strand cleavages deriving from single-strand nicks, as well as from dsbs that were either blunt or staggered. To determine whether the bands visualized in our LM-PCR experiments derived from blunt dsbs, we investigated whether we could detect breaks in S{alpha} using the direct ligation strategy of Fig. 1A, which detects only blunt cuts. As a positive control, we showed that blunt ends generated by cleaving the double-stranded genomic DNA with StuI could easily be detected by a direct ligation strategy (Fig. 2D, lane 2). However, as shown in Fig. 2D, lane 6, no bands were detected with ligation of the linkers directly to genomic DNA from cytokine-treated CH12F3-2 cells; in parallel control lanes, our usual ladder was obtained by our routine strategy of ligation after primer extension using the 5'P1 primer, which generates a blunt end (Fig. 2D, lane 5). These results suggested that the breaks we detect in the lower strand of S{alpha} derived from either single-strand nicks or staggered-ended DNA breaks. To distinguish between the latter possibilities, we attempted to ligate the blunt linker to double-stranded CH12F3-2 DNA (from 24-h cytokine-treated cells) that had been "polished" by various treatments designed to convert staggered ends to blunt. As positive controls, we used CH12F3-2 DNA that had been digested with restriction enzymes that cleave DNA leaving a 5' overhang (SpeI) or 3' overhang (MwoI). As shown in Fig. 2D, lane 2, ends produced by SpeI or MwoI did not generate an LM-PCR band if directly ligated, but if the ends were first blunted by T4 DNA polymerase treatment, they generated the expected LM-PCR bands (lane 3). In contrast, double-stranded genomic DNA from cytokine-treated CH12F3-2 cells generated no bands even after treatment with T4 polymerase (lane 7). Similar negative results were observed using the Klenow fragment of Escherichia coli DNA polymerase (data not shown). These results suggest that the reproducible bands detected in our routine protocol derive from nicks in the lower (template) strand, rather than from staggered dsbs.

To characterize more precisely the position of the cuts in the lower strand of S{alpha}, we modified our procedure so that the LM-PCR products could be cloned into a TOPO vector and the ends sequenced. Of the clones with inserts (i.e., obtained from white colonies), more than two thirds yielded smears, no bands or multiple faint bands after colony PCR, and apparently arose from PCR or cloning artifacts. Of 22 clones yielding intense single bands, sequence analysis indicated that 21 were derived from the S{alpha} region, as shown in Fig. 3A. This figure presents the sequence of the upper strand of CH21F3-2 S{alpha}, in the usual 5' to 3' direction corresponding to the transcriptional orientation of the IgH locus. The 3' breakpoints of the 21 clones are marked. Most of the clones represent amplified fragments that are smaller than the most intense bands in our reproducible electrophoresis ladder; small clones may have been favored by bias resulting from the additional PCR cycles required for cloning and from more efficient cloning of small inserts. Nevertheless, for 11 of the 21 clones, the insert length as estimated from the positions of lower strand cuts by restriction enzymes (Fig. 3B) corresponded to bands detected on our electrophoresis ladder. Presumably the electrophoresis bands represent only the few preferred cleavage sites, but a substantial number of cleavages occur at other positions as represented by our clones.



View larger version (54K):
[in this window]
[in a new window]
  		FIGURE 3. Sequences of cloned LM-PCR products. A, The CH12F3-2 genomic sequence of the 5' end of S{alpha} is shown, with marks identifying the 3' ends of the 21 clones sequenced in this investigation. The CH12F3-2 sequence contains several differences, including two deletions totaling ~70 bp, compared with previously published S{alpha} sequences; in particular, two nucleotides where the 3'P3 primer we designed (based on GenBank sequence D11468) differs from the CH12F3-2 sequence are indicated (between position 430 and 440). In the 4835 bp sequenced in these clones, nine discrepancies from the CH12F3-2 sequence were observed, possibly representing PCR errors. B, Relationship between clones and LM-PCR electrophoresis bands. The LM-PCR bands from the indicated restriction enzyme digests of CH12F3-2 DNA were used to estimate the sizes of bands observed in DNA from 24-h cytokine-treated CH12F3-2 cells. Clone sequences that match the length of electrophoresis bands are indicated by arrows. C, Consensus sequence at DNA breaks in the template strand. Sequences of the lower strand are presented 5'-3' (reverse complement of the orientation in A) aligned at the position of the break, which is represented by the thick vertical line. The sequences to the right of this line represent the lower strand sequence telomeric to the position of the lower strand break, which lies in our clones adjacent to the linker primer as shown in the schematic below. The sequences to the left of the vertical line represent the sequences in CH12F3-2 genomic DNA that lie directly contiguous with the clone sequences.

 
When we examined the genomic S{alpha} sequences of the lower strand aligned according to the position of the junctions between S{alpha} and the linker primer sequence in our clones (Fig. 3C), we found a striking consensus. In 19 of the 21 clones, the genomic sequence replaced by the primer terminates with a C within the consensus sequence CAG. This suggests the possibility that the breaks in the lower S{alpha} strand are targeted primarily at C nucleotides within the consensus sequence C*AG (where * represents the cleavage position). An attractive model to explain this finding would be that the breaks result from AID-catalyzed deamination of these C nucleotides, where the resulting U is removed by UDG, leaving an apurinic site that is cleaved by APE. The two clones with breakpoints that are not adjacent to a C, i.e., clones va17(228) and va15(88), may reflect some exonuclease trimming after the initial deamination-initiated cleavage at a nearby C.

In contrast, an alternative to our model that we cannot at present rule out is that the genomic DNA ends that ligated to the linker in our assay did not reflect the position of initial DNA cleavage, but resulted from subsequent exonuclease trimming after the initial cleavage. The consensus sequence we observe might then reflect a consensus pause site of such an exonuclease.

As discussed above, several reports have indicated that transcription across Ig S regions leads to a unique complex between the transcribed RNA and the complementary template strand of DNA, leaving the nontemplate (or "coding" strand) in a single-stranded state (8, 9, 10). Because AID targets ssDNA in vitro and in E. Coli model systems (4, 5, 6, 7), several investigators have proposed that CSR may be initiated by AID-mediated deamination of C residues on this nontemplate strand. To investigate this model, we have looked for single-strand breaks in the "upper" strand of S{alpha} using primers 3' of the positions where we detected lower strand breaks (see Fig. 1C). As a positive control to test whether our 3' primers could detect upper strand breaks, we used genomic DNA cleaved with restriction endonucleases. As shown in Fig. 4, we were able to detect bands in genomic DNA cleaved with several restriction endonucleases. The lighter bands above and below each main band probably represent products of hybridization of the 3'P3 oligonucleotide to partial sequence matches in the repeats upstream or downstream of the 3'P3 oligonucleotide target sequence. However, we could not detect any bands from the upper strand of undigested DNA from 24-h cytokine-treated CH12F3-2 cells, although the same DNA samples yielded many bands in the complementary region of the lower strand. When this DNA was amplified with the 3' primers using the PCR procedure modified to allow cloning of any amplification products from the upper strand, only a single clone from the {alpha} locus was obtained, corresponding to an upper strand break outside and upstream of S{alpha}. An additional set of primers was synthesized corresponding to template strand sequence downstream of the repetitive region of S{alpha} and designed to detect upper strand breaks at the 3' end of S{alpha}; this primer set easily detected upper strand breaks caused by restriction endonuclease digestion, but failed to detect upper strand breaks in genomic DNA of 24-h cytokine-stimulated cells (data not shown).


   Discussion
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
References
 
To elucidate the mechanism of CSR, it would clearly be desirable to detect DNA intermediates in the process. Broken DNA ends must be formed during recombination, and might be detectable transiently before they are repaired or ligated in the process of recombination. Like previous investigators, we have used LM-PCR to amplify broken ends in an Ig S region, but we designed our strategy to detect ssDNA ends, which would be present as a result of nicks, staggered dsbs, or blunt dsbs.

In the murine cell line CH12F3-2, which switches to IgA expression with high efficiency when cultured with IL-4, TGF{beta}, and anti-CD40, we detected a reproducible pattern of LM-PCR bands from the 5' region of S{alpha}. This pattern was dependent on the stimuli that promote isotype switching, and its time course was consistent with expectations for a CSR intermediate. The bands appear to reflect nicks in the lower strand in that their appearance required the initial primer extension steps that create blunt ends suitable for ligation to the LM-PCR linker. In particular, we did not detect the bands after subjecting the genomic DNA to treatments that should have detected either blunt dsbs (direct ligation to the linker) or staggered dsbs (ligation after T4 polymerase treatment). The inference that the LM-PCR bands reflect single-strand nicks is consistent with our failure to observe any DNA breaks in the corresponding region of the upper nontemplate strand.

The absence of detectable S{alpha} upper strand breaks in our experiments does not imply that such breaks do not occur in S{alpha}. Indeed, the switch recombination event itself implies that both DNA strands must be broken; and cleavage of the upper strand, though undetected by our LM-PCR assay, may well occur exactly as has been proposed: following deamination of the ssDNA displaced by an S{alpha} transcript. One explanation for absence of LM-PCR products from upper strand breaks might be that these breaks are scattered without predominant cleavage "hot spots" to generate visible electrophoresis bands; however, this hypothesis does not explain why no cloned products corresponding to cuts in the upper strand of S{alpha} were recovered. Another possibility is that upper strand breaks occur before or after the 24-h time point we used for most of our experiments. To explore this possibility, we have examined samples from 6, 12, 24, and 48-h time points using the 3' primer set, but no bands were detected in any of these samples (data not shown). A possibility that we currently favor is that each break in the upper strand of S{alpha} is rapidly processed into some structure that precludes ligation to the linker oligonucleotides in the LM-PCR protocol. For example, the upper strand ends from S{alpha} might be immediately ligated to ends from Sµ. In contrast, cleavages in the lower strand may be catalyzed differently and processed more slowly compared with those in the upper strand, generating a pool of unsealed single-strand breaks that is detectable by LM-PCR. The previous reports of CSR-associated blunt dsbs in S{gamma}3 and in Sµ may reflect recombinase mechanisms at these S regions somewhat different from S{alpha} CSR; or alternatively, initially staggered cuts at S{gamma}3 or Sµ were perhaps trimmed into blunt ends.

Strikingly, 19 of the 21 ends of the LM-PCR lower strand cuts that we cloned were consistent with cleavage at a C within the consensus sequence C*AG, consistent with involvement of AID, which deaminates C residues to U. As suggested above, the U residues might be removed by UDG (which is known to participate in CSR) and the resulting abasic site could be cleaved by APE. A similar model of cleavage mediated by AID, UDG, and APE has been proposed to explain somatic hypermutation (6, 7), although the C within a different consensus sequence, WRCY (where W = A or T; R = purine; Y = pyrimidine), has been suggested as a deamination target. Conceivably AID is targeted to the CAG consensus in the context of the lower strand of S{alpha} by a CSR-associated sequence-specific DNA recognition protein. Such a protein might have a precedent in the targeting mechanism of APOBEC-1, the RNA-editing deaminase that shows sequence similarity to AID and is encoded adjacent to the AID gene in mouse and human (30). APOBEC-1 is targeted to a specific sequence in apolipoprotein B mRNA by another protein designated the "APOBEC-1 complementation factor" protein (31). A CSR-specific S region targeting protein recognizing the CAG consensus might be the hypothetical cofactor that has been proposed to bind to the C-terminal residues of AID (32), based on the observation that AID constructs with engineered C-terminal deletions (33) or mutations (32) are competent to support somatic hypermutation, but not CSR.

With regard to a differential mechanism of lower vs upper strand cleavage, it is notable that the consensus sequence we observed for the lower strand cuts occurs much more frequently in the lower than in the upper strand. In the 446 bp between the 5'P3 oligo and the 3'P3 oligo used in our experiments, the minimal sequence CAG occurs 34 times in the lower strand and only 6 in the upper. In most of the cleavage sites identified in our clones (Fig. 3C), the C*AG falls within a 5-bp repeat (consensus C*AGCY) flanked by similar repeats, and it is possible that the functional target sequence for cleavage includes features of the repeats surrounding the cleavage or deamination site. It may be significant that the AGCY part of the repeat falls into the previously reported WRCY consensus sequence. Several candidate consensus sites longer than CAG occur frequently in the lower strand but not at all in the upper strand.

To summarize, the cleavages we have detected in S{alpha} are confined to the template DNA strand, which is thought to be hybridized to the sterile RNA transcript initiating at a promoter upstream of the S{alpha} sequence, though we have no evidence that deamination or cleavage occurred in DNA nucleotides that were hybridized to RNA at the time of the cleavage. If cleavages occurring in this template strand are initiated by AID as we have suggested, this process would bear an interesting similarity to the deamination of HIV DNA by another member of the APOBEC/AID family, APOBEC-3G (also known as CEM15) as recently reported by several laboratories (34, 35). Evidence in these reports suggests that APOBEC-3G induces lethal mutations in several viral genomes by deaminating nascent cDNA reverse-transcribed from the viral RNA, possibly while the DNA is still partially complexed with its RNA template. Thus, it is possible that both APOBEC-3G and its paralog AID specifically recognize and deaminate C residues of DNA in the strand involved in an RNA-DNA complex.

As this manuscript was being completed, Rush et al. (26) reported detecting AID-dependent DNA dsbs in Sµ from splenic B cells switching in vitro under cytokine stimulation for 2 or 4 days. They detected dsbs in the 5' and 3' regions of Sµ at approximately equal frequency, using a linker ligation strategy that would not detect single-strand breaks. They found that T4 polymerase polishing of double-stranded genomic DNA from switching cells increased the number of LM-PCR bands detected, implying that the predominant dsbs had staggered ends. The sequences at the ends of their cloned DNA breaks did not show an obvious consensus. The experiments of Rush et al. (26) differed from ours in that they used different cells, later time points, different culture conditions, a different DNA isolation technique, and a different LM-PCR method applied to different S regions. Future experiments will be necessary to determine which differences explain why their experiments detected dsbs when ours did not; but it may be significant that despite efficient detection of dsbs in Sµ, Rush et al. (26) were able to detect dsbs consistently only in Sµ, and not in S{gamma}1 or S{gamma}3, in cells switching to these isotypes, suggesting that events occurring during CSR at Sµ may differ from what occurs at downstream target S regions. This notion is consistent with the fact that mutations occur at Sµ during CSR at a much higher frequency than in S{gamma} CSR targets (36).


   Acknowledgments
 
We are grateful to S. Bauer, A. Nussenzweig, and M. Nussenzweig for critically reading this manuscript.


   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 Address correspondence and reprint requests to Dr. Edward E. Max, Division of Therapeutic Proteins, Office of Biotechnology Products, Center for Drug Evaluation and Research, Food and Drug Administration, HFM-541, 8800 Rockville Pike, Bethesda, MD 20892. E-mail address: max{at}cber.fda.gov Back

2 Abbreviations used in this paper: SHM, somatic hypermutation; AID, activation-induced deaminase; APE, apurinic endonuclease; CSR, class switch recombination; dsb, double-strand break; LM, ligation mediated; S region, switch region; S{alpha} region, H chain {alpha} switch region; UDG, uracil DNA glycosylase. Back

Received for publication April 7, 2004. Accepted for publication June 17, 2004.


   References
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 References
 

  1. Muramatsu, M., V. S. Sankaranand, S. Anant, M. Sugai, K. Kinoshita, N. O. Davidson, T. Honjo. 1999. Specific expression of activation-induced cytidine deaminase (AID), a novel member of the RNA-editing deaminase family in germinal center B cells. J. Biol. Chem. 274:18470.[Abstract/Free Full Text]
  2. Muramatsu, M., K. Kinoshita, S. Fagarasan, S. Yamada, Y. Shinkai, T. Honjo. 2000. Class switch recombination and hypermutation require activation-induced cytidine deaminase (AID), a potential RNA editing enzyme. Cell 102:553.[Medline]
  3. Revy, P., T. Muto, Y. Levy, F. Geissmann, A. Plebani, O. Sanal, N. Catalan, M. Forveille, R. Dufourcq-Labelouse, A. Gennery, et al 2000. Activation-induced cytidine deaminase (AID) deficiency causes the autosomal recessive form of the Hyper-IgM syndrome (HIGM2). Cell 102:565.[Medline]
  4. Chaudhuri, J., M. Tian, C. Khuong, K. Chua, E. Pinaud, F. W. Alt. 2003. Transcription-targeted DNA deamination by the AID antibody diversification enzyme. Nature 422:726.[Medline]
  5. Bransteitter, R., P. Pham, M. D. Scharff, M. F. Goodman. 2003. Activation-induced cytidine deaminase deaminates deoxycytidine on single-stranded DNA but requires the action of RNase. Proc. Natl. Acad. Sci. USA 100:4102.[Abstract/Free Full Text]
  6. Pham, P., R. Bransteitter, J. Petruska, M. F. Goodman. 2003. Processive AID-catalysed cytosine deamination on single-stranded DNA simulates somatic hypermutation. Nature 424:103.[Medline]
  7. Ramiro, A. R., P. Stavropoulos, M. Jankovic, M. C. Nussenzweig. 2003. Transcription enhances AID-mediated cytidine deamination by exposing single-stranded DNA on the nontemplate strand. Nat. Immunol. 4:452.[Medline]
  8. Daniels, G. A., M. R. Lieber. 1995. RNA:DNA complex formation upon transcription of immunoglobulin switch regions: implications for the mechanism and regulation of class switch recombination. Nucleic Acids Res. 23:5006.[Abstract/Free Full Text]
  9. Tian, M., F. W. Alt. 2000. Transcription-induced cleavage of immunoglobulin switch regions by nucleotide excision repair nucleases in vitro. J. Biol. Chem. 275:24163.[Abstract/Free Full Text]
  10. Yu, K., F. Chedin, C. L. Hsieh, T. E. Wilson, M. R. Lieber. 2003. R-loops at immunoglobulin class switch regions in the chromosomes of stimulated B cells. Nat. Immunol. 4:442.[Medline]
  11. Rada, C., G. T. Williams, H. Nilsen, D. E. Barnes, T. Lindahl, M. S. Neuberger. 2002. Immunoglobulin isotype switching is inhibited and somatic hypermutation perturbed in UNG-deficient mice. Curr. Biol. 12:1748.[Medline]
  12. Imai, K., G. Slupphaug, W. I. Lee, P. Revy, S. Nonoyama, N. Catalan, L. Yel, M. Forveille, B. Kavli, H. E. Krokan, et al 2003. Human uracil-DNA glycosylase deficiency associated with profoundly impaired immunoglobulin class-switch recombination. Nat. Immunol. 4:1023.[Medline]
  13. Manis, J. P., Y. Gu, R. Lansford, E. Sonoda, R. Ferrini, L. Davidson, K. Rajewsky, F. W. Alt. 1998. Ku70 is required for late B cell development and immunoglobulin heavy chain class switching. J. Exp. Med. 187:2081.[Abstract/Free Full Text]
  14. Casellas, R., A. Nussenzweig, R. Wuerffel, R. Pelanda, A. Reichlin, H. Suh, X. F. Qin, E. Besmer, A. Kenter, K. Rajewsky, M. C. Nussenzweig. 1998. Ku80 is required for immunoglobulin isotype switching. EMBO J. 17:2404.[Medline]
  15. Cook, A. J., L. Oganesian, P. Harumal, A. Basten, R. Brink, C. J. Jolly. 2003. Reduced switching in SCID B cells is associated with altered somatic mutation of recombined S regions. J. Immunol. 171:6556.[Abstract/Free Full Text]
  16. Manis, J. P., D. Dudley, L. Kaylor, F. W. Alt. 2002. IgH class switch recombination to IgG1 in DNA-PKcs-deficient B cells. Immunity 16:607.[Medline]
  17. Reina-San-Martin, B., S. Difilippantonio, L. Hanitsch, R. F. Masilamani, A. Nussenzweig, M. C. Nussenzweig. 2003. H2AX is required for recombination between immunoglobulin switch regions but not for intra-switch region recombination or somatic hypermutation. J. Exp. Med. 197:1767.[Abstract/Free Full Text]
  18. Schrader, C. E., J. Vardo, J. Stavnezer. 2002. Role for mismatch repair proteins Msh2, Mlh1, and Pms2 in immunoglobulin class switching shown by sequence analysis of recombination junctions. J. Exp. Med. 195:367.[Abstract/Free Full Text]
  19. Ehrenstein, M. R., M. S. Neuberger. 1999. Deficiency in Msh2 affects the efficiency and local sequence specificity of immunoglobulin class-switch recombination: parallels with somatic hypermutation. EMBO J. 18:3484.[Medline]
  20. Bardwell, P. D., C. J. Woo, K. Wei, Z. Li, A. Martin, S. Z. Sack, T. Parris, W. Edelmann, M. D. Scharff. 2004. Altered somatic hypermutation and reduced class-switch recombination in exonuclease 1-mutant mice. Nat. Immunol. 5:224.[Medline]
  21. Wuerffel, R. A., J. Du, R. J. Thompson, A. L. Kenter. 1997. Ig S{gamma}3 DNA-specific double strand breaks are induced in mitogen-activated B cells and are implicated in switch recombination. J. Immunol. 159:4139.[Abstract]
  22. Catalan, N., F. Selz, K. Imai, P. Revy, A. Fischer, A. Durandy. 2003. The block in immunoglobulin class switch recombination caused by activation-induced cytidine deaminase deficiency occurs prior to the generation of DNA double strand breaks in switch µ region. J. Immunol. 171:2504.[Abstract/Free Full Text]
  23. Imai, K., N. Catalan, A. Plebani, L. Marodi, O. Sanal, S. Kumaki, V. Nagendran, P. Wood, C. Glastre, F. Sarrot-Reynauld, et al 2003. Hyper-IgM syndrome type 4 with a B lymphocyte-intrinsic selective deficiency in Ig class-switch recombination. J. Clin. Invest. 112:136.[Medline]
  24. Chen, X., K. Kinoshita, T. Honjo. 2001. Variable deletion and duplication at recombination junction ends: implication for staggered double-strand cleavage in class-switch recombination. Proc. Natl. Acad. Sci. USA 98:13860.[Abstract/Free Full Text]
  25. Nakamura, M., S. Kondo, M. Sugai, M. Nazarea, S. Imamura, T. Honjo. 1996. High frequency class switching of an IgM+ B lymphoma clone CH12F3 to IgA+ cells. Int. Immunol. 8:193.[Abstract/Free Full Text]
  26. Rush, J. S., S. D. Fugmann, D. G. Schatz. 2004. Staggered AID-dependent DNA double strand breaks are the predominant DNA lesions targeted to S micro in Ig class switch recombination. Int. Immunol. 16:549.[Abstract/Free Full Text]
  27. Mueller, P. R., B. Wold, P. A. Garrity. 2003. Ligation-mediated PCR for genomic sequencing and footprinting. F. M. Ausubel, and R. Brent, and R. E. Kent, and D. D. Moore, and J. G. Seidman, and J. A. Smith, and K. Struhl, eds. Current Protocols in Molecular Biology John Wiley & Sons, New York. Unit 153.
  28. Iwasato, T., H. Arakawa, A. Shimizu, T. Honjo, H. Yamagishi. 1992. Biased distribution of recombination sites within S regions upon immunoglobulin class switch recombination induced by transforming growth factor {beta} and lipopolysaccharide. J. Exp. Med. 175:1539.[Abstract/Free Full Text]
  29. Arakawa, H., T. Iwasato, H. Hayashida, A. Shimizu, T. Honjo, H. Yamagishi. 1993. The complete murine immunoglobulin class switch region of the {alpha} heavy chain gene-hierarchic repetitive structure and recombination breakpoints. J. Biol. Chem. 268:4651.[Abstract/Free Full Text]
  30. Muto, T., M. Muramatsu, M. Taniwaki, K. Kinoshita, T. Honjo. 2000. Isolation, tissue distribution, and chromosomal localization of the human activation-induced cytidine deaminase (AID) gene. Genomics 68:85.[Medline]
  31. Mehta, A., M. T. Kinter, N. E. Sherman, D. M. Driscoll. 2000. Molecular cloning of apobec-1 complementation factor, a novel RNA-binding protein involved in the editing of apolipoprotein B mRNA. Mol. Cell. Biol. 20:1846.[Abstract/Free Full Text]
  32. Ta, V. T., H. Nagaoka, N. Catalan, A. Durandy, A. Fischer, K. Imai, S. Nonoyama, J. Tashiro, M. Ikegawa, S. Ito, et al 2003. AID mutant analyses indicate requirement for class-switch-specific cofactors. Nat. Immunol. 4:843.[Medline]
  33. Barreto, V., B. Reina-San-Martin, A. R. Ramiro, K. M. McBride, M. C. Nussenzweig. 2003. C-terminal deletion of AID uncouples class switch recombination from somatic hypermutation and gene conversion. Mol. Cell 12:501.[Medline]
  34. Mangeat, B., P. Turelli, G. Caron, M. Friedli, L. Perrin, D. Trono. 2003. Broad antiretroviral defence by human APOBEC3G through lethal editing of nascent reverse transcripts. Nature 424:99.[Medline]
  35. Zhang, H., B. Yang, R. J. Pomerantz, C. Zhang, S. C. Arunachalam, L. Gao. 2003. The cytidine deaminase CEM15 induces hypermutation in newly synthesized HIV-1 DNA. Nature 424:94.[Medline]
  36. Schrader, C. E., S. P. Bradley, J. Vardo, S. N. Mochegova, E. Flanagan, J. Stavnezer. 2003. Mutations occur in the Ig Sµ region but rarely in S{gamma} regions prior to class switch recombination. EMBO J. 22:5893.[Medline]

Related articles in The JI:

IN THIS ISSUE

The JI 2004 173: 2893-2894. [Full Text]  



This article has been cited by other articles:


Home page
 Int ImmunolHome page
A. Arudchandran, R. M. Bernstein, and E. E. Max
Single-strand DNA breaks in Ig class switch recombination that depend on UNG but not AID
Int. Immunol., November 1, 2008; 20(11): 1381 - 1393.
[Abstract] [Full Text] [PDF]

Home page
 JEMHome page
N. Feldhahn, N. Henke, K. Melchior, C. Duy, B. N. Soh, F. Klein, G. von Levetzow, B. Giebel, A. Li, W.-K. Hofmann, et al.
Activation-induced cytidine deaminase acts as a mutator in BCR-ABL1-transformed acute lymphoblastic leukemia cells
J. Exp. Med., May 14, 2007; 204(5): 1157 - 1166.
[Abstract] [Full Text] [PDF]

Home page
 J. Immunol.Home page
J. E. Butler and N. Wertz
Antibody Repertoire Development in Fetal and Neonatal Piglets. XVII. IgG Subclass Transcription Revisited with Emphasis on New IgG3
J. Immunol., October 15, 2006; 177(8): 5480 - 5489.
[Abstract] [Full Text] [PDF]

Home page
 Proc. Natl. Acad. Sci. USAHome page
J. S. Rush, M. Liu, V. H. Odegard, S. Unniraman, and D. G. Schatz
Expression of activation-induced cytidine deaminase is regulated by cell division, providing a mechanistic basis for division-linked class switch recombination
PNAS, September 13, 2005; 102(37): 13242 - 13247.
[Abstract] [Full Text] [PDF]

Home page
 JEMHome page
C. E. Schrader, E. K. Linehan, S. N. Mochegova, R. T. Woodland, and J. Stavnezer
Inducible DNA breaks in Ig S regions are dependent on AID and UNG
J. Exp. Med., August 15, 2005; 202(4): 561 - 568.
[Abstract] [Full Text] [PDF]

Home page
 JEMHome page
B. Kavli, S. Andersen, M. Otterlei, N. B. Liabakk, K. Imai, A. Fischer, A. Durandy, H. E. Krokan, and G. Slupphaug
B cells from hyper-IgM patients carrying UNG mutations lack ability to remove uracil from ssDNA and have elevated genomic uracil
J. Exp. Med., June 20, 2005; 201(12): 2011 - 2021.
[Abstract] [Full Text] [PDF]

Home page
 J. Immunol.Home page
S. A. Martomo, D. Fu, W. W. Yang, N. S. Joshi, and P. J. Gearhart
Deoxyuridine Is Generated Preferentially in the Nontranscribed Strand of DNA from Cells Expressing Activation-Induced Cytidine Deaminase
J. Immunol., June 15, 2005; 174(12): 7787 - 7791.
[Abstract] [Full Text] [PDF]

Home page
 JEMHome page
J. M. Lumsden, T. McCarty, L. K. Petiniot, R. Shen, C. Barlow, T. A. Wynn, H. C. Morse III, P. J. Gearhart, A. Wynshaw-Boris, E. E. Max, et al.
Immunoglobulin Class Switch Recombination Is Impaired in Atm-deficient Mice
J. Exp. Med., November 1, 2004; 200(9): 1111 - 1121.
[Abstract] [Full Text] [PDF]

Home page
 Proc. Natl. Acad. Sci. USAHome page
Z. Li, Z. Luo, and M. D. Scharff
Differential regulation of histone acetylation and generation of mutations in switch regions is associated with Ig class switching
PNAS, October 26, 2004; 101(43): 15428 - 15433.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Related articles in The JI
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow Request Permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Arudchandran, A.
Right arrow Articles by Max, E. E.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Arudchandran, A.
Right arrow Articles by Max, E. E.
Right arrowPubmed/NCBI databases
*Compound via MeSH
*Substance via MeSH

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS