The chemokine CXCL13 in lung cancers associated with environmental polycyclic aromatic hydrocarbons pollution

Thank you for submitting your work entitled "The chemokine CXCL13 in lung cancers associated with environmental polycyclic aromatic hydrocarbons pollution" for peer review at eLife. Your submission has been favorably evaluated by Sean Morrison (Senior editor) and three reviewers, one of whom served as Guest Reviewing editor.

The reviewers have discussed the reviews with one another and the Reviewing editor has drafted this decision to help you prepare a revised submission.

Summary:

All three reviewers felt the work was interesting and of importance, but had concerns about several aspects of the paper that need to be addressed before considering publication. In particular, all three reviewers asked for more detailed analysis of Cxcl13 and Cxcr5 mouse knockout studies, an essential component of the paper. In addition, analyses of CXCL13 levels in orthogonal datasets were requested for Figure 1, and additional controls were recommended for ChIP data in Figure 3. Another issue raised was the source of CXCL13, whether it was derived from epithelial cells or associated immune cells.

Essential revisions:

1) Figure 1 – is CXCL13 enriched in orthogonal datasets or is this specific to the Chinese cohorts? We would suggest analysis of TCGA data of tumor versus normal tissue and smokers versus non-smokers. It would also be of interest to perform a similar comparison in microarray data deposited from genetically engineered mouse models of lung cancer to see whether in these models, which are not carcinogen induced, if CXCL13 is not enriched. For example, mouse GEO datasets GSE6135, GSE21581, GSE54353 are from lung tumors derived following LKB1 inactivation. It would be interesting to know whether or not CXCL13 is involved in this context.

2) What are the expression levels of CXCR5 in human and mouse samples with lung cancers?

3) It is unclear which cell type is staining positive for CXCL13 in Figure 2D. It would be useful to add IHC panels to help determine whether it is being produced from tumor cells or tumor associated macrophages. The authors could use TTF1 stain to mark lung cancer cells, and CD68 stain to mark macrophages. Alternatively they could perform flow cytometry and sort by EPCAM or CD68.

4) Figure 3B needs additional controls. The authors should use qPCR instead of PCR to present the ChIP data. In addition, although the specific induction of binding by BAP is nice, the authors also need to test another gene without the XRE element, perhaps CXCL12, which was used in another experiment. For the EMSA, it seems that the lane with the shift is overloaded. This blot is not convincing and needs to be repeated. Alternatively, if the results are not robust, they should be removed from the paper, since it erodes confidence in the ChIP result.

5) Figure 4 which is a key component of this paper should include quantification of tumor burden by microCT (as presented in Figure 2—figure supplement 1B). In the CT images presented it appears there are tumors in Cxcl13^-/- mice. Also at what time point were CTs performed? There should also be more rigorous quantification of tumor nodules in histologic sections between mouse groups (e.g. see Tan et al. PLOS One 2013; PMID 24260500). This data needs to be presented for the Cxcr5 mice as well. Also for Figure 3—figure supplement 1 that is used to document Cxcl13 inactivation; which band is it? If it is the lower band why is it higher in the hetereozygous mice? Same for Cxcr5 mice.

6) As in point 2, additional IHC controls for Figure 6G would help to confirm that macrophages are indeed the cell type positive for SPP1 in the BaP induced tumors. As above, the authors could use TTF1 stain to mark lung cancer cells and CD68 stain to mark macrophages, or perform flow cytometry and sort by EPCAM or CD68.

Summary:

All three reviewers felt the work was interesting and of importance, but had concerns about several aspects of the paper that need to be addressed before considering publication. In particular, all three reviewers asked for more detailed analysis of cxcl13 and cxcr5 mouse knockout studies, an essential component of the paper. In addition, analyses of CXCL13 levels in orthogonal datasets were requested for Figure 1, and additional controls were recommended for ChIP data in Figure 3. Another issue raised was the source of CXCL13, whether it was derived from epithelial cells or associated immune cells.

We performed additional and more detailed analysis in Cxcl13 and Cxcr5 knockout mice, analyzed CXCL13 levels in orthogonal datasets, investigated the source of CXCL13, and used additional controls for ChIP data. In addition to the above revisions, 14 new references were added to the paper, which now has 3 tables, 6 figures, 77 references, 2 supplementary tables and 5 supplementary figures.

Essential revisions:

1) Figure 1 – is CXCL13 enriched in orthogonal datasets or is this specific to the Chinese cohorts? We would suggest analysis of TCGA data of tumor versus normal tissue and smokers versus non-smokers. It would also be of interest to perform a similar comparison in microarray data deposited from genetically engineered mouse models of lung cancer to see whether in these models, which are not carcinogen induced, if CXCL13 is not enriched. For example, mouse GEO datasets GSE6135, GSE21581, GSE54353 are from lung tumors derived following LKB1 inactivation. It would be interesting to know whether or not CXCL13 is involved in this context.

We thank the reviewers for the important suggestions, and evaluated the expression of CXCL13 in tumor versus normal tissue and smokers versus non-smokers in TCGA data and a cancer microarray database Oncomine (www.oncomine.org). In two works of TCGA (Nature 2014; Nature 2012), CXCL13 expression was assessed by microarray or RNA-seq, and the results showed that CXCL13 was not up-regulated in tumor samples compared to normal tissues (data not shown). Nevertheless, CXCL13 in tumor samples of smokers was not increased compared to that in nonsmokers (data not shown). However, in several works documented in the Oncomine database, CXCL13 was elevated in tumor compared to normal tissues (Figure 1H), and was increased in smokers compared to non-smokers (Figure 1I). These results suggested that overexpression of CXCL13 was not specific to the Chinese cohorts.

In mouse GEO dataset GSE54353, Cxcl13 expression in wild type (normal), Kras^G12D, and Lkb1(Stk11)^-/-Pten^-/- mice was not significantly different (Figure 1J). In GSE6135, Cxcl13 in Kras^G12DStk11^{L/+, -/-, L/-} mice was slightly but not statistically significantly higher than in Kras^G12DStk11^wt mice (p=0.08; Figure 1J). In GSE21581 dataset, Cxcl13 was slightly elevated in primary tumors and significantly increased in metastatic tumors of the Kras^G12DStk11^-/- mice, as compared to that of the Kras^G12D mice (Figure 1J). These results suggested that Cxcl13 might play a role in Stk11-related lung carcinogenesis which warrants further investigation.

2) What are the expression levels of CXCR5 in human and mouse samples with lung cancers?

We tested the expression of CXCR5 in 24 NSCLCs by qPCR, and found that in tumor samples it was slightly higher than in paired normal lung tissues (Figure 4F). In tumor samples of mice treated with BaP at 50 or 100 mg/kg, Cxcr5 was also slightly up-regulated (Figure 4G).

3) It is unclear which cell type is staining positive for CXCL13 in Figure 2D. It would be useful to add IHC panels to help determine whether it is being produced from tumor cells or tumor associated macrophages. The authors could use TTF1 stain to mark lung cancer cells, and CD68 stain to mark macrophages. Alternatively they could perform flow cytometry and sort by EPCAM or CD68.

We thank the reviewers for the comments and performed IHC and immunofluorescence assays to determine the source of Cxcl13. By IHC assay, we found that both the Cd68 positive macrophages and Ttf1 positive lung cancer cells were stained positive for Cxcl13, but Ttf1 positive cells constituted more than 95% of the cellular component of the lung tumor tissues of the mice (Figure 2G). This observation was confirmed by immunofluorescence assay (Figure 2H) using antibodies against Cxcl13 (green), Cd68 (red), and Ttf1 (white). DAPI was used to counter stain the nucleus. These results indicate that Ttf1 positive lung cancer cells were the main source of Cxcl13 in mice exposed to BaP.

4) Figure 3B needs additional controls. The authors should use qPCR instead of PCR to present the ChIP data. In addition, although the specific induction of binding by BAP is nice, the authors also need to test another gene without the XRE element, perhaps CXCL12, which was used in another experiment. For the EMSA, it seems that the lane with the shift is overloaded. This blot is not convincing and needs to be repeated. Alternatively, if the results are not robust, they should be removed from the paper, since it erodes confidence in the ChIP result.

We thank the reviewers for the comments, and performed qPCR to test the expression of CXCL13, CXCL12, CXCL14, CXCL11 and CXCL2 in ChIP experiment of cells co-incubated with or without BaP. The results showed that among these chemokines, only CXCL13 was enriched and significantly up-regulated by BaP treatment (Figure 5 here and Figure 3B of the paper). We used this data to replace the PCR results of the original Figure 3B. The EMSA result was removed from the paper.

5) Figure 4 which is a key component of this paper should include quantification of tumor burden by microCT (as presented in Figure 2—figure supplement 1B). In the CT images presented it appears there are tumors in cxcl13^-/- mice. Also at what time point were CTs performed? There should also be more rigorous quantification of tumor nodules in histologic sections between mouse groups (e.g. see Tan et al. PLOS One 2013; PMID 24260500). This data needs to be presented for the cxcr5 mice as well. Also for Figure 3–figure supplement 1 that is used to document cxcl13 inactivation; which band is it? If it is the lower band why is it higher in the hetereozygous mice? Same for cxcr5 mice.

We thank the reviewers for the important comments and suggestions, and evaluated the tumor burden of the Cxcl13- and Cxcr5-deficiency mice. Firstly, tumor nodules in histologic sections of mice upon BaP treatment were analyzed as describe (Tan et al. PLOS One 2013), and the result showed that at treatment time points of 120, 180, and 240 days, Cxcl13^-/- mice had much less lesions than Cxcl13^+/- and Cxcl13^+/+ mice (Figure 4A). For microCT analysis of the mice treated with BaP for 240 days, the tumor volume of Cxcl13^-/- mice was significantly lower than Cxcl13^+/- and Cxcl13^+/+ mice (Figure 4B, C). Tumor burden of Cxcr5^-/- mice, reflected by tumor nodules and microCT assays, was also significantly lower than Cxcr5^+/- and Cxcr5^+/+ mice (Figure 4H, I, J).

The Cxcl13^-/- mice were obtained from the Jackson Laboratory (https://www.jax.org/strain/005626). The mice were also called B6.129X1-Cxcl13^tm1Cys/J. A targeting vector was constructed in which base pairs 18-116 of exon 2 from the endogenous gene were replaced with an in-frame stop codon, a Mengo virus internal ribosome entry site, an enhanced green fluorescent protein gene (EGFP), and a loxP-flanked neomycin resistance gene. The construct was electroporated into 129X1/SvJ derived JM-1 embryonic stem (ES) cells. Correctly targeted ES cells were injected into C57BL/6J blastocysts and the resulting chimeric males were backcrossed for germ-line transmission to C57BL/6J females. Offspring were mated with Cre-expressing C57BL/6J to remove the neomycin resistance gene. The neo-excised heterozygotes were backcrossed to C57BL/6J for ten generations before being made homozygous. Mice that are homozygous for the targeted mutation are viable, fertile and normal in size. No endogenous gene product (mRNA or protein) is detected, EGFP is not expressed. The primers to detect Cxcl13 were designed by the Jackson Laboratory, and the expected results were: mutant=~240 bp, heterozygote =~240 bp and ~200 bp, wild type=~200 bp (https://www2.jax.org/protocolsdb/f?p=116:2:::NO:2:P2_MASTER_PROTOCOL_ID,P2_JRS_CODE:945,005626). Our results were in consistent with the references provided by the Jackson Laboratory, e.g., the higher one of Figure 4—figure supplement 1A was the inactivation band.

For Cxcr5^-/- mice (https://www.jax.org/strain/006659), a targeting vector was designed to replace the 350 bp coding region of exon 2 of the targeted gene with a neomycin resistance gene. The construct was electroporated into 129S2/SvPas-derived D3 embryonic stem (ES) cells. Correctly targeted ES cells were aggregated with morulae from outbred CD-1 mice and then transferred into pseudopregnant CD-1 females. Chimeric mice were bred with CD-1. The donating investigator reported that mutant mice were backcrossed to C57BL/6 mice for 8 generations prior to arrival at The Jackson Laboratory. Homozygous (CXCR5-deficient) mice are viable and fertile. No endogenous gene product (mRNA or protein) is detected. In Figure 4—figure supplement 1C, our results were in consistent with the references provided by the Jackson Laboratory, e.g., the lower one was the inactivation band.

6) As in point 2, additional IHC controls for Figure 6G would help to confirm that macrophages are indeed the cell type positive for SPP1 in the BaP induced tumors. As above, the authors could use TTF1 stain to mark lung cancer cells and CD68 stain to mark macrophages, or perform flow cytometry and sort by EPCAM or CD68.

We thank the reviewers for the comments and performed IHC and immunofluorescence assays to determine the source of Spp1. By IHC assay, we found that the Cd68 positive macrophages were strongly stained by Spp1, whereas Ttf positive lung cancer cells were weakly stained by Spp1 (Figure 6H). This observation was confirmed by immunofluorescence assay (Figure 7B) using antibodies against Spp1 (green), Cd68 (red), and Ttf (white). DAPI was used to counter stain the nucleus. These results indicate that Cd68 positive macrophages were the main source of Spp1 in mice exposed to BaP.