Author's response to reviews Title: Functional analysis of the C-reactive protein (CRP) gene -717A>G polymorphism associated with coronary heart disease Authors: Laiyuan Wang ([email protected]) Xiangfeng Lu ([email protected]) Yun Li ([email protected]) Hongfan Li ([email protected]) Shufeng Chen ([email protected]) Dongfeng Gu ([email protected]) Version: 2 Date: 26 March 2009 Author's response to reviews: see over

Dear Editor and Reviewer: Thank you very much for your careful review of our work. We believe your comments have improved our manuscript greatly. As you have suggested, we have discussed all of the points you have raised and made a revision to our manuscript as below.

Reviewer: David Samols 1. Comment: P3, par 2, line 8. The authors state that CRP “has been shown to play important roles in the process of atherosclerosis”. However, the title of the reference they give merely refers to CRP as a marker, rather than playing an active role. The evidence that CRP plays a pathogenetic role in atherosclerosis is slim indeed [see recent paper by Casas et al]. Response: According to the paper by Casas et al, the experimental evidence from animals indicates that human CRP can definitely be pathogenic after ischaemic infarction, but any relevance to the slow process of atherogenesis remains unproved. However, CRP has been shown to play proatherogenic roles in the process of atherosclerosis via effects on monocytes and endothelial cells: CRP directly induced the expression of adhesion molecules and monocyte chemoattractant protein-1 (MCP-1) by endothelial cells [Ref 9, 10], increased plasminogen activator inhibitor-1 expression and activity in endothelial cells [Ref 11], decreased eNOS mRNA, protein abundance and enzyme activity in endothelial cells, and preincubation of cells with CRP also significantly increased the adhesion of monocytes to endothelial cells [Ref 12]. Moreover, CRP deposited in the arterial wall in early atherosclerotic lesions [Ref 13], and might contribute to the formation of foam cells in atherosclerotic lesions by causing the aggregation of LDL molecules that are then taken up by macrophages through a CD32-independent pathway [Ref 14]. Revision: We have provided evidences that CRP plays a pathogenetic role in atherosclerosis and made some revisions as flowed: (Background, Page 3, Line 16-24; Page 4, Line 1-3) “C-reactive protein (CRP), a sensitive, but nonspecific marker of inflammation (an acute-phase reactant) [8] has been shown to play proatherogenic roles in the process of atherosclerosis via effects on monocytes and endothelial cells: CRP directly induced the expression of adhesion molecules and monocyte chemoattractant protein-1 (MCP-1) by endothelial cells [9, 10], increased plasminogen activator inhibitor-1 expression and activity in endothelial cells [11], decreased eNOS mRNA, protein abundance and enzyme activity in endothelial cells, and preincubation of cells with CRP also significantly 1

increased the adhesion of monocytes to endothelial cells [12]. CRP deposited in the arterial wall in early atherosclerotic lesions [13], and might contribute to the formation of foam cells in atherosclerotic lesions by causing the aggregation of LDL molecules that are then taken up by macrophages through a CD32-independent pathway [14].”

2. Comment: It is of interest that the authors point out that previous studies have not shown a difference in blood levels of CRP in the A and G carriers [Ref 4, 15, 17]. This suggests that the association cited in my first sentence may not be mediated through circulating CRP at all. Response: Our previous study [Ref 4] focused on the association between -717A>G and CHD status, and no association between -717A>G and baseline CRP was found in secondary analysis. This might be due to several factors. First, prior work from population studies indicates that obesity, age, smoking, diet, chronic infection, and statins are reported to be associated with serum CRP concentration. Second, the number of individuals screened for associations between serum CRP and genetic variation is small. A well designed study should be conducted to define the association between -717A>G and baseline CRP. E.Ben-Assayag et al. reported that -717A>G was associated with triggered CPR levels during acute stroke/TIA [Ref 21]. This study indicated that -717A>G located at the promoter region of the CRP gene was far more than just a marker associated with CHD, but might have native biological function during some stimulating conditions. Our present study showed an increased transcriptional activity of the promoter of the CRP gene driven by A allele, therefore we deduced that the association of -717A>G with altered plasma levels of CRP [Ref 21] might be due to the exchange of G to A allele at the site of rs2794521, thus resulting in the increase of the CRP levels during some stimulating conditions. Additional studies aimed to confirm this finding are warranted.

3. Comment: Several aspects of the EMSAs in Figs 2 and 3 are unclear. What is the identity of Band 2? Band 2 was not marked in Fig 3 but it appears as if only Band 1 was supershifted by the antibody against GR. The hypothesis that Band 2 is a monomer of GR is thus not well supported. In general, monomers of GR do not bind DNA but require a hetero or homodimeric partners. In Fig 2 lane 5, what is the strong band under Band 2? The authors state that the G allele bound more GR than did the A allele (Band 1 in Fig 2). However, more oligo was used in lanes 4-6 than lanes 1-3 so 2

the difference in intensities in the Band 1 between lanes 2 and 5 may reflect the amount of oligo shifted and not differences in GR binding affinity. Response: In present study, the sequence containing G allele at the site of rs2794521 is a not well-matched GR binding motif, and the G to A exchange at the site of rs2794521 resulted in a well-matched GR binding motif. The GR binds to a glucocorticoid response element (GRE) as a monomer and a dimer as previously reported, and the incubation of nuclear extract of HepG2 cells with probe A showed two kinds of DNA-protein complex with the GR motif oligonucleotide probes, which probably represent monomer (Band 2) and dimer (Band 1) of GR (Lane 2). The band 1 was supershifted by the antibody against GR, but the band 2 formed on probe A or G was not supershifted by anti-GR antibody probably due to the heavy amount of the band 2, we still could not draw a conclusion that the band 2 is not GR-containing band. In Fig 2 lane 5, there is a strong band below band 2 formed on probe G, and this band might be a novel protein binding specifically to nonsusceptible G allele, which is under study. It appeared that more oligo was used in lanes 4-6 than lanes 1-3, but in fact, the quantity of biotin-labeled probe used in EMSA is the same in 6 lanes in Fig. 2, and the EMSA experiments were repeated three times with similar results. The G to A exchange at the site of rs2794521 resulted in a well-matched GR binding motif, and the EMSA experiment showed that the probe A with a well-matched GR binding motif showed a much higher affinity for monomer of GR (band 2) than that of the probe G. Therefore, the affinity of A or G allele for GR differed drastically.

4. Comment: Since corticosteroids have been shown to enhance induction of CRP by cytokines, that finding that the A allele bound GR less than the G allele would suggest that CRP expression should be less in the former group. This does not hang together with the other major finding in this study. Response: The G to A exchange at the site of rs2794521 resulted in a well-matched GR binding motif, and the EMSA experiment indicated that the affinity of A or G allele for GR differed drastically: the probe A with a well-matched GR binding motif showed a much higher affinity for monomer of GR (band 2) than that of the probe G. Therefore, we deduced that the monomer of GR might bind to the A allele of rs2794521with a higher affinity than G allele, and affect the transcription activity of the promoter of the CRP gene through the formation of heterodimers with other nucleus receptors, which needs to be further studied.

3

5. Comment: Replace dimer with dimer throughout the text. Response: Thank you! We have replaced dimer with dimer throughout the text.

Reviewer: Alok Agrawal 1.Comment: Figs. 2B and 3: The results of EMSA are shown in these figures. The band 1 formed on probe A is not clear in Fig. 3 (it was clear in Fig. 2B). Response: The supershifted EMSA experiment shown in Fig.3 was repeated three times with similar results, and Fig.3 is less clear than Fig.2B, but we can confirmly found that the DNA-protein complex was successfully supershifted with the anti-GR antibody (Fig. 3, lane 3), but not the rabbit IgG (Fig. 3, lane 4).

2. Comment: Fig. 2B: Three bands are formed on probe G. What are those bands? Response: In fact, the sequence containing G allele at the site of rs2794521 is a not well-matched GR binding motif, and the G to A exchange at the site of rs2794521 resulted in a well-matched GR binding motif. Therefore, our previous conclusion that the G to A exchange at the site of rs2794521 generated a potential binding site of glucocorticoid receptor (GR) protein is improper. We performed electrophoretic mobility shift assays to determine the nuclear factors that might bind to oligonucleotides corresponding to genomic sequence of the A or G allele of rs2794521, and determine whether the binding of the transcription factor(s) differed for the A and G alleles. Although the sequence containing G allele at the site of rs2794521 is a not well-matched GR binding motif, the EMSA experiment showed that the GR still bound to G allele, but the binding of GR differed for the A and G alleles. The incubation of nuclear extract of HepG2 cells with probe G showed a large increase in the dimer (Lane 5, band 1), and a dramatic reduction in monomer (Lane 5, band 2). What’ more, another band below the band 2 appeared to bind specifically to the G allele, and this band might be a novel binding protein, which is under study. Revision: We have made some revisions in the paper as followed: (Results, Page 7, Line 18-20) “the sequence containing G allele at the site of rs2794521 is a not well-matched GR binding motif, and the G to A exchange at the site of rs2794521 resulted in a well-matched GR binding motif.” (Results, Page 8, Line 9, 10) “Because the G to A exchange at the site of rs2794521 resulted in a well-matched GR binding motif and the GR bound to both the A and G alleles” 4

(Discussion, Page 9, Line 16-18) “There is a well-matched GR binding motif at the site of rs2794521 after the exchange of G to A allele.”

3. Comment: Figs. 2B and 3: Two bands are formed on probe A. The authors conclude that the 2 bands are the monomers and dimers of GR. Then, why only the band 1 was supershifted with anti-GR antibodies? Why was the band 2 not supershifted? Response: The G to A exchange at the site of rs2794521 resulted in a well-matched GR binding motif, and the EMSA experiment showed that the probe A with a well-matched GR binding motif showed a much higher affinity for monomer of GR (band 2), therefore we deduced that the amount of the band 2 binding to the probe A in our experiment was so heavy that the anti-GR antibodies was unable to supershift the band 2.

4. Comment: Figs. 2B and 3: Supershift assays must also be performed with probe G. Where is the data to show that the bands formed on probe G contain GR? What is the possibility that the band 2 on probe A and all 3 bands on probe G are not GR-containing bands? The authors mention in the “discussion” section that the probe A (with -717A), and not the G, has the potential GR-site. Response: Because the G to A exchange at the site of rs2794521 resulted in a well-matched GR binding motif and the GR bound to both the A and G alleles, we performed the supershift EMSA experiment with probe A. In similar study, Shimokawa et al reported that there is a -6415 CTCTAA/GC polymorphic site in IL-12B gene, and a transcription factor Sp 1 bound to the region including the GC allele with a higher affinity than that of the CTCTAA allele in EMSA experiment. The supershift EMSA was then performed with only one probe, -6415GC probe because of the higher similarity of -6415/GC sequence to Sp1 binding motif[Shimokawa N, et al. Functional analysis of a polymorphism in the promoter region of the IL-12/23p40 gene.Clin Exp Allergy. 2009;39(2):228-235.]. Given the fact mentioned above, we do not think that the supershift EMSA is necessarily to be performed with probe G in our study. As we pointed out, the sequence containing G allele at the site of rs2794521 is a not well-matched GR binding motif, the GR binds to a GRE as a monomer and a dimer, and the EMSA experiment showed that the GR still bound to G allele. Compared with probe A, probe G showed higher affinity for the dimer (Lane 5, band 1), and lower affinity for monomer (Lane 5, band 2). Although the supershift EMSA was not performed with probe G, we could not drawn a conclusion that the band 5

1and band 2 formed on probe G did not contain GR. Another band below the band 2 appeared to bind specifically to the G allele, and this band might be novel regulatory protein(s), which is under study. The band 2 formed on probe A was not supershifted by anti-GR antibody probably due to the heavy amount of the band 2, we still could not draw a conclusion that the band 2 is not GR-containing band. In fact, the sequence containing G allele at the site of rs2794521 is a not well-matched GR binding motif, and the G to A exchange at the site of rs2794521 resulted in a well-matched GR binding motif. Therefore, our previous conclusion that the probe A (with -717A), and not the G, has the potential GR-site is improper.

5. Comment: Fig. 1A should be deleted. The authors have already mentioned the nucleotide sequence of the promoter in the “materials and methods” section. If the authors do want to present this information in the form of a figure also, then they should correct the position of nucleotide -717 in the figure. As shown, the position is incorrect. Fig. 2A should also be deleted. The authors have already shown the nucleotide sequence of the probes in the “materials and methods” section. Response: Thank you! We have made some revisions in the position of nucleotide -717 in the Fig. 1A.

Fig. 2A and its figure legend have been deleted.

Revision: We have made some revisions as followed: (Figure Legends，Page 16, Line 3-5): “Figure 2.

Alleles of rs2794521 differentially bind

transcription factor(s). EMSA was performed by using biotin-labeled probe A or probe G with HepG2 cell nuclear extract, with or without competition from unlabeled probe A or probe G.”

6. Comment: In the ‘abstract’ and ‘background’ sections, the authors state that “CRP has been shown to play important roles in the process of atherosclerosis”. The authors should define those important roles: are they proatherogenic or atheroprotective? In the abstract, define the abbreviation “GR”. Response: We have provided evidences that C-reactive protein (CRP) plays proatherogenic roles in the pathogenesis of developing atherosclerosis in our paper. We have defined the abbreviation of “GR” in the abstract. Revision: We have made some revisions as followed: 6

(Abstract, Page 2, Line 4-6) “C-reactive protein (CRP), a sensitive, but nonspecific marker of inflammation has been shown to play proatherogenic roles in the process of atherosclerosis.” (Abstract, Page 2, Line 14-17) “The G to A exchange at the site of rs2794521 resulted in an increased transcriptional activity of the promoter of CRP gene, and glucocorticoid receptor (GR) protein factor bound drastically differently to the A and G alleles at the site of rs2794521.” (Background, Page 3, Line 16-24; Page 4, Line 1-3) “C-reactive protein (CRP), a sensitive, but nonspecific marker of inflammation (an acute-phase reactant) [8] has been shown to play proatherogenic roles in the process of atherosclerosis via effects on monocytes and endothelial cells: CRP directly induced the expression of adhesion molecules and monocyte chemoattractant protein-1 (MCP-1) by endothelial cells [9, 10], increased plasminogen activator inhibitor-1 expression and activity in endothelial cells [11], decreased eNOS mRNA, protein abundance and enzyme activity in endothelial cells, and preincubation of cells with CRP also significantly increased the adhesion of monocytes to endothelial cells [12]. CRP deposited in the arterial wall in early atherosclerotic lesions [13], and might contribute to the formation of foam cells in atherosclerotic lesions by causing the aggregation of LDL molecules that are then taken up by macrophages through a CD32-independent pathway [14].”

In addition to the questions by reviewers, we have made some revisions as followed: (Abstract, Page 2, Line 8, 9)“In the present study, we tried to investigate the biological significance of this genetic variation in vitro.” (Background, Page 4, Line 15-18) “Is there any difference for nuclear proteins binding to A or G allele at the site of rs2794521? In the present study, the biological relevance of the SNP rs2794521 significantly associated with CHD in Chinese subjects[4] was studied by various functional assays in vitro.” (Discussion, Page 8, Line 19-23; page 9, Line 1-6) “Our previous report[4] found that the A allele of rs2794521 located in the promoter region of the CRP gene was independently associated with CHD in Chinese subjects. Individuals carrying the A allele had an approx. 6.8-fold higher risk of developing CHD compared with those not carrying this allele. This finding has been confirmed, as the same SNP was found to correlate with occurrence of myocardial infarction or thromboembolic stroke in the Physician’s Health Study[19]. In the present study, we found that transcriptional activity of the promoter of CRP gene containing the A allele was higher than that of 7

the promoter of CRP gene containing the G allele. Among several case-control studies involving the SNP rs2794521 with CHD [4, 19, 21, 23], our present study is the first to confirm that the SNP rs2794521 is to be functional.” (Discussion, Page 10, Line 10-13) “According to the reference [22, 24] and our results, we deduced that the monomer of GR might bind to the A allele of rs2794521, and affect the transcription activity of the promoter of the CRP gene through the formation of heterodimers with other nucleus receptors, which needs to be further studied.” (Conclusion, Page 10, Line 15-17) “In summary, we found that the G to A exchange at the site of rs2794521 resulted in an increased transcriptional activity of the promoter of CRP gene, and the GR protein factor bound drastically differently to the A and G alleles.” (Acknowledgements, Page 11, Line 12-17) “We thank Pro. Lanying Chen for her nice suggestion for the manuscript. This work was supported by National Basic Research Program of China (Grant No. 2006CB503805), National High Technology Research and Development Program of China (Grant No. 2006AA02Z170), Beijing municipal Commission of Education (Young Teachers Program) and National Natural Science Foundation of China (Grant No. 30600357).”

Thank you very much for your time and consideration. Sincerely, Dongfeng Gu, MD, PhD Department of Evidence Based Medicine and Division of Population Genetics, Cardiovascular Institute and Fu Wai Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, 167 Beilishi Rd, Beijing, 100037, People’s Republic of China Tel: +86 (10) 68331752 Fax: +86 (10) 88363812 E-mail: [email protected]

8