EMBO Molecular Medicine
Peer Review Process File - EMM-2010-00562
Manuscript EMM-2010-00562
Lysyl oxidase-like 2 (LOXL2), a new regulator of cell polarity required for metastatic dissemination of basal-like breast carcinomas Gema Moreno-Bueno, Fernando Salvador, Alberto Martin, Alfredo Floristan, Eva P. Cuevas, Vanesa Santos, Amalia Montes, Saleta Morales, Maria Angeles Castilla, Alejandro Rojo-Sebastían, Alejandra Martínez, David Hardisson, Katalin Csiszar, Francisco Portillo, Héctor Peinado, José Palacios and Amparo Cano Corresponding author: Amparo Cano, Instituto de Investigaciones Biomédicas, CSIC-UAM
Review timeline:
Submission date: Editorial Decision: Rebuttal: Appeal received: Editorial Decision: Revision received: Editorial Decision: Revision received: Editorial Decision Accepted:
03 December 2010 25 January 2011 31 January 2011 01 February 2011 09 February 2011 02 May 2011 25 May 2011 06 June 2011 07 June 2011 07 June 2011
Transaction Report: (Note: With the exception of the correction of typographical or spelling errors that could be a source of ambiguity, letters and reports are not edited. The original formatting of letters and referee reports may not be reflected in this compilation.)
1st Editorial Decision
25 January 2011
Thank you for the submission of your manuscript "Lysyl oxidase-like 2 (LOXL2) is required for metastatic dissemination of basal-like breast carcinomas" to EMBO Molecular Medicine. We have now heard back from the three reviewers whom we asked to evaluate your manuscript. As you will see, the referees acknowledge that the overall topic of the manuscript is potentially interesting. However, they raise significant concerns, which, I am afraid, preclude publication of the manuscript in EMBO Molecular Medicine. The major concerns raised by the reviewers refer to the limited conceptual novelty and conclusiveness of the presented findings. Thus, reviewer #1 notes that the study remains incomplete while reviewer #2 and #3 are not convinced that the reported results represent a sufficient advance and feel that the potentially most novel findings are not sufficiently well supported by the data. As such, two out of three referees would not support publication of the manuscript in EMBO Molecular Medicine. Since EMBO Molecular Medicine can only invite revision of papers that receive enthusiastic support from a majority of referees, I am afraid that we cannot offer to publish your manuscript.
© European Molecular Biology Organization
1
EMBO Molecular Medicine
Peer Review Process File - EMM-2010-00562
I am sorry to have to disappoint you on this occasion. I hope that this negative decision does not prevent you from considering EMBO Medicine for the publication of future studies and that the referee comments are helpful in your continued work in this area. Yours sincerely, Editor EMBO Molecular Medicine ***** Reviewer's comments ***** Referee #1 (Novelty/Model system): The manuscript presents a mixture of experimental results which, as it stands, do not sufficiently support the conclusions. Besides some additional experimental data, as specified below, the manuscript needs to be focused on one or two major aspects only. Referee #1 (Other comments): Moreno-Bueno and co-workers report the analysis of the functional contribution of LOXL2 to breast cancer invasion and metastasis and the correlation of LOXL2 expression and subcellular localization with metastatic basal-like breast cancer. Previously, the authors' and other laboratories have demonstrated that LOXL2 functionally contributes to epithelial-mesenchymal transition and tumor malignancy in a several cancer types. Here, a detailed bioinformatical study reveals that LOXL2 is specifically expressed in triple-negative, basal-like breast cancers and is part of a specific gene signature for this breast cancer subtype. Loss of function studies in established basal-like cell lines demonstrates that LOXL2 is required for the maintenance of a mesenchymal phenotype and the invasion and metastatic capabilities of these cells. Moreover, LOXL2 expression significantly correlates with poor prognosis and metastasis in patients with basal-like breast cancer. While the manuscript is concisely presented in its different parts, overall it represents a mixture of results which by themselves are not complete and in parts do not convincingly support the conclusions drawn by the authors. For example, while the bioinformatical data, which by itself is novel and very exciting, goes into some details of classification of various breast cancer subtypes, it stops short in using these data to test a specific gene signature on patient populations. Moreover, while the functional study of LOXL2 contribution clearly shows the requirement of LOXL2 for EMT, tumor cell invasion and metastasis, it does not provide insights into the mechanisms by which LOXL2 exerts this function. Notably, circumstantial evidence suggests that LOXL2 may act in a Snail- and E-cadherin-independent function, yet this observation is not corroborated on, and no experimentation has been performed to directly test this possibility. The actual targets of LOXL2 remain unknown. Also, it is claimed that LOXL2 may act via modulating tight junction rather than adherens junction components, yet no direct evidence is provided for this conclusion. Testing the effects of LOXL2 activity on tight junctions and adherens junctions may need the use of short-term reversible EMT systems and not cell lines that are cultured for years as mesenchymal cells and that may have gathered a variety of genetic and epigenetic changes. Thus, as it stands, the manuscript appears preliminary and incomplete in its variety of results and conclusions. Yet, most of the experimental data are novel, exciting and certainly important for a wide audience and, thus, the manuscript needs to be revised to focus on the interpretation of the actual experimental data and to draw the appropriate conclusions exclusively from substantiated experimental results. Specific comments: Page 8, Fig. 2C: a,b, or c are not in the figure or the figure legend. Page 9, Fig. 3F: It is stated that proliferation is not affected by the knock-down of LOXL2, yet there is a moderate if not significant effect?
© European Molecular Biology Organization
2
EMBO Molecular Medicine
Peer Review Process File - EMM-2010-00562
Page 9, 2. Paragraph: Fig. S3A not Fig. 3A. The significance of the correlation between the 231-shLoxl2 and BBC gene signatures is not clear. The fact that there are some similarities between gene ontology and only a few common genes is not very meaningful. There is also no evidence whatsoever that some of these genes are "likely candidates to be regulated by LOXL2 in basal tumors". Fig. 5: The results show an effect of LOXL2 on primary tumor growth and tumor recurrence, not on metastasis. This is disappointing, since these cell line is supposed to be metastatic. Is it possible that LOXL2 acts also on tumor cell proliferation and less so on EMT and tumor invasion? Is there a difference in tumor histology in the presence and absence of LOXL2? Lung colonization is shown in Fig. 5 C-D and not as mentioned in Fig. 5 B-D. There is no labeling of Figure 6B "a"?
Referee #2: The authors investigated the role of lysyl oxidase-like 2 (LOXL2) for invasion and metastastic dissemination of basal-type of breast cancers. LOXL2 is well known to induce an EMT in tumor cells and was already shown to be associated with poor prognosis of other cancers, such as squamous cell carcinomas. Basal-type breast cancers are particularly aggressive and show an EMT-phenotype. Here the authors detected a strong association of LOXL2 with basal-type breast cancers, in particular the aggressive claudin-low subtype. A perinuclear staining pattern was associated with increased distant metastasis of breast cancers. They further showed that a knockdown decreased the invasiveness and metastastic potential of basal-type of breast cancer cell lines (e.g. MDA-MB231). The functional assays largely confirm previous data showing the effect of LOXL2 on migration and invasion of breast cancer cell lines (see Kirschmann et al, Cancer res, 2002) and extend these finding by showing an effect of in vivo tumor and metastasis formation. The really new and relevant finding is the association of LOXL2 expression with metastasis of human breast cancers. This indicates the potential usage of LOXL2 as a prognostic marker. To further strengthen and foster this findings, the immunohistochemical stainings should be repeated with other available antibodies/antisera against LOXL2. This is of particular importance because different staining patterns are described for LOXL2 (e.g. Barry-Hamilton et al., Nature Med, 2010). Minor points: - the supplementary tables should be better explained (e.g. abbreviations in the headlines) - page 9: should read ...'HBL100 basal cell lines (Fig. S3A)...' Referee #3 (Novelty/Model system): Most if not all of the observations presented in this article have been published previously either by the current authors or by others. This includes demonstration that LOXL2 can regulate breast tumor progression, migration, invasion and tumorigenesis/metastasis in vivo, demonstration that LOXL2 can regulate EMT and appears to do so through a novel yet currently unresolved mechanism, that LOXL2 is highly expressed in human breast tissues. No new mechanistic insight is provided linking LOXL2 to EMT.
© European Molecular Biology Organization
3
EMBO Molecular Medicine
Peer Review Process File - EMM-2010-00562
Referee #3 (Other comments): Moreno-Bueno and colleagues explore the role of LOXL2 in mammary tumor aggression and metastasis. They argue that LOXL2 is a marker of basal-like breast tumors and suggest that LOXL2 mediates breast tumor metastasis through a cell intrinsic mechanism that depends upon EMT induction through an E cadherin/snail independent mechanism. Prior studies by these authors and others have already shown that LOXL2 regulates tumor progression and metastasis and that LOXL2 can induce an EMT phenotype through an E cadherin/snail independent mechanism. Included in the article is a collection of cell line studies that essentially reproduce what has previously been published by others attesting to the effect of LOXL2 on cell migration and invasion in culture and tumorigenicity and metastasis in vivo. The authors virtually ignore data showing that loss of LOXL2 has no effect on cell growth in culture but a profound effect in vivo - data which strongly underscore the possibility that LOXL2 elicits its effects on tumor behavior in part through effects on the tissue microenvironment; this observation seems to have escaped the authors. As such the only novelty of this report is the authors contention that they have discovered a novel marker and regulator of basallike breast tumors. This finding while potentially interesting and important seems better suited for a clinically specialized journal such as clinical cancer research. Moreover, the data presented on this topic seem premature. Thus, while the authors make a reasonable case that there is an association between LOXL2 and human breast cancer through RT-PCR and IHC analysis of human biopsies they group together all basal cancers - and fail to address the incredible complexity of this subtype. They make no attempt to link LOXL2 expression with any EMT marker through costaining studies or other histophenotype markers of breast tumor aggression. They do discuss their observation that LOXL2 immunostaining appears membrane bound and perinuclear in the more aggressive breast tumor specimens - but fail to address the mechanistic potential of this observation. Indeed, the article lacks mechanism for the purported LOXL2 dependent EMT phenotype in that the authors do not explore whether membrane or pernuclear LOXL2 are important and if so why. In this instance prior studies suggest membrane associated LOX can modify cell phenotype through effects on integrin adhesions while others have shown that nuclear associated LOXL2 can influence a variety of cellular behaviors through independent means.
Rebuttal
31 January 2011
We are profoundly disappointed with your decision of not considering our ms "Lysyl oxidase-like 2 (LOXL2) is required for metastatic dissemination of basal-like breast carcinomas" suitable for revision and eventual publication in EMBO Molecular Medicine based on the reviewer's comments. This has been particularly disappointing after the long time that our ms has been under revision. We would like to make several considerations taking into account the specific reviewer's comments: 1. After careful reading of the comments, we only see that one (reviewer # 3) out of three reviewers does not clearly support the ms. Indeed, reviewer #1 after raising his/her concerns on the content of the ms and the incomplete parts, it ends its report by stating"Yet, most of the experimental data are novel, exciting and certainly important for a wide audience and, thus, the manuscript needs to be revised to focus on the interpretation of the actual experimental data and to draw the appropriate conclusions exclusively from substantiated experimental results". This statement strongly suggests that this reviewer is supporting a revised version of the ms. 2. Indeed, the queries raised by reviewer #1 can be answered in a revised version, in particular those more general related to analysis of the basal-like signature on patient populations and the use of short term EMT systems, as well as the more specific
© European Molecular Biology Organization
4
EMBO Molecular Medicine
Peer Review Process File - EMM-2010-00562
questions indicated by this reviewer in his/her report. Similarly, the only one query raised by reviewer #2 can be answered by staining the samples with a different anti-LOXL2 antibody, as he/she indicated. 3. Regarding reviewer #3, we would like to respectfully indicate that some of his/her assertions are not completely correct. The overall comment of this reviewer, extended further in the remarks to authors, indicate "Most if not all of the observations presented in this article have been published previously either by the current authors or by others. This includes demonstration that LOXL2 can regulate breast tumor progression, migration, invasion and tumorigenesis/metastasis in vivo, demonstration that LOXL2 can regulate EMT and appears to do so through a novel yet currently unresolved mechanism, that LOXL2 is highly expressed in human breast tissues. No new mechanistic insight is provided linking LOXL2 to EMT". We honestly think that some of those comments are at least partly incorrect, particularly at the time of submission of our ms, and we would like to explain them further: 1. The reviewer is right when affirming that the influence of LOXL2 in the invasive behaviour of some breast tumor cells was previously reported by other authors. However, those previous studies have not distinguished between breast tumor cells representing distinct tumor phenotypes. We have extended the study to show that the expression of LOXL2 is restricted to basal-like carcinoma cell lines and analysed its effect of invasiveness of basal like carcinoma cells. The effect on tumorigenic and metastatic behaviour was reported in a recent paper by Barry-Hamilton et al. (2010) (published when our ms was under preparation) showing the effect of LOXL2 inhibition in the tumorigenic behaviour of MDA-MB-435 cells (a cell line that is still under debate if it represents breast or melanoma derived cells; see Rae et al., Breast Cancer Res Treat., 104: 13-9, 2007; Chambers et al., Cancer Res, 69: 5293-3, 2009) and in bone metastasis of MDA-MB-231 cells (a situation not occurring in human breast basal like tumors; reviewed in Foulkes et al., New Engl J Med 363:1938-48, 2010); those previous results were discussed in our ms. Surprisingly to us, two weeks ago and while our ms was under review, an additional article came out (Barker et al., Cancer Res, Published OnlineFirst January 13, 2011, DOI:10.1158/0008-5472.CAN-10-2868) showing that LOXL2 knockdown or inhibition decreased spontaneous metastasis of mouse breast 4T1 cells and human breast MDA-MB-231 cells. No such information was therefore publically available when our ms. was submitted and under revision in EMM. 2. The statement related to the previous demonstration that "LOXL2 is highly expressed in human breast tissues" only stand for meta-analyses studies of public dataset of microarrays, including our own previous observations in N0 breast carcinomas showing a correlation of LOXL2 transcripts with poor survival (Peinado et al, Cancer Res, 2008) and the study of selected ER- tumors (that will include both ER- and ErbB2+ tumors) from the NKI dataset in which high LOXL2 mRNA is correlated with metastasis that was included in the recent paper by Barker et al. (2011), referred above. A small and phenotypically undefined breast adenocarcinoma series was analysed for LOXL2 expression by IHQ in the
© European Molecular Biology Organization
5
EMBO Molecular Medicine
Peer Review Process File - EMM-2010-00562
Barry-Hamilton et al.'s paper (as discussed in our ms, see below point 4.e). In none of the previous studies the protein expression of LOXL2 by IHQ has been analysed in an extensive series of grade 3 breast carcinoma such as the one included in our ms. More importantly, we have been able to show for the first time that LOXL2 cytoplasmic and perinuclear staining pattern significantly correlates with metastasis of basal-like carcinomas. Indeed, as this and the other two reviewers point out, our results for LOXL2 expression and localization in the tumor series and its correlation with metastatic basal-like breast tumors are completely novel and highly relevant. No such information is provided in none of the previous publications. In fact, when reading the ms by Barker et al (2011) the immediate question is: how is the protein expression of LOXL2 in breast tumors? This is the precisely the information that we already presented in our ms for the first time, being these results completely novel. 3. In the remarks to authors, reviewer #3 state "Prior studies by these authors and others have already shown that LOXL2 regulates tumor progression and metastasis and that LOXL2 can induce an EMT phenotype through an E cadherin/snail independent mechanism". Our previous works indeed indicated that LOXL2 can mediate EMT by Snail1 dependent and independent mechanisms (Peinado et al, EMBO J 2005; Peinado et al., Cancer Res, 2008), but we did not previously observed neither reported on LOXL2 effects independent of E-cadherin. To our knowledge, no information on this specific aspect has been reported by other authors. 4. Also in the remarks to authors, the following statement is included: "Included in the article is a collection of cell line studies that essentially reproduce what has previously been published by others attesting to the effect of LOXL2 on cell migration and invasion in culture and tumorigenicity and metastasis in vivo". As commented above, the previous information existed on some breast cancer cell lines. However, what the reviewer appears to have ignored is the fact that the collection of cell lines we have analysed were selected depending of their phenotype (luminal, ErbB2/neu+ and basal-like) showing that the expression of LOXL2 is specific of basal-like carcinoma cell lines, a fact that is completely novel and not reported in any previous study. 5. Another comment in the remarks to author: "The authors virtually ignore data showing that loss of LOXL2 has no effect on cell growth in culture but a profound effect in vivo - data which strongly underscore the possibility that LOXL2 elicits its effects on tumor behavior in part through effects on the tissue microenvironment; this observation seems to have escaped the authors". The potential effect of LOXL2 on the microenvironment was indeed considered in a long paragraph of the discussion section of our ms (page 17, 2^nd paragraph ending in page 18, 3 first lines), where it was specifically stated: "/The specific role of LOXL2 in lung metastasis of basal-like breast carcinomas remains to be established. In this regard, previous studies have indicated an important role for LOX in hypoxia-induced metastasis (Erler et al, 2006) and the specific
© European Molecular Biology Organization
6
EMBO Molecular Medicine
Peer Review Process File - EMM-2010-00562
participation of LOX in the establishment of the pre-metastatic niche through extracellular matrix remodeling (Erler et al, 2009). A role for extracellular LOXL2 in the development of pathological microenvironment and the effectiveness of an anti-LOXL2 antibody in inhibition of tumor burden of carcinoma MDA-MB-435 cells as well as in fibrosis has been recently reported (Barry-Hamilton et al, 2010). Other recent studies have also reported the participation of secreted LOXL2 in gastric carcinoma metastasis (Peng et al, 2009). Our results point to an intracellular function for LOXL2 in basal-like breast carcinoma metastasis in agreement with our previous studies in squamous carcinoma tumors (Peinado et al, 2008) and in mammary gland tissue (Hollosi et al, 2009). It is worthy to emphasize that previous studies on LOXL2 on breast tumors have been restricted to a few samples of breast adenocarcinomas of non-specific histological subtypes and lacking information on other clinico-pathological features (Barry-Hamilton et al, 2010; Hollosi et al, 2009). Our results provide data not only for LOXL2 expression in a specific breast tumor subtype, the basal-like carcinomas, but also for its clinical association with in vivo distant metastasis. The distinct staining pattern of LOXL2 observed in different studies might be related to the use of different antibodies recognizing intracellular and secreted LOXL2 proteins with distinct domains (Hollosi et al, 2009), use of different immunostaining protocols or even might suggest a dual role and/or substrates for LOXL2 in regulating intracellular and extracellular components that may prevail in distinct pathological contexts. Although the participation of extracellular LOXL2 functions in basal-like breast carcinomas cannot be formally excluded, our results provide a new dimension to the role of LOXL2 in tumor progression by influencing intracellular functions. . 6. Reviewer #3 also remarks: "They do discuss their observation that LOXL2 immunostaining appears membrane bound and perinuclear in the more aggressive breast tumor specimens but fail to address the mechanistic potential of this observation. Indeed, the article lacks mechanism for the purported LOXL2 dependent EMT phenotype in that the authors do not explore whether membrane or pernuclear LOXL2 are important and if so why. In this instance prior studies suggest membrane associated LOX can modify cell phenotype through effects on integrin adhesions while others have shown that nuclear associated LOXL2 can influence a variety of cellular behaviors through independent means". Regarding this comment, we would like to firstly stress that the specific staining we observed associated to basal-like tumors is cytoplasmic and perinuclear, but never membrane-bound or extracellular. Secondly, to our knowledge no nuclear LOXL2 stain has been previously reported in the literature, while nuclear LOX (the prototypical isoform) has been observed in distinct situations and indeed involved in different cell behaviours, as well as LOX-membrane associated or secreted forms (this information is discussed in previous reviews referenced in our ms., see Lucero and Kagan, 2006; Mäki, 2009; Payne et al., 2007).
© European Molecular Biology Organization
7
EMBO Molecular Medicine
Peer Review Process File - EMM-2010-00562
7. The rest of the comments of this reviewer, such as discussion of different basal-like tumor subtypes (indeed a matter of present debate) and costaining with EMT markers are issues that can be easily addressed in a revised version. 4. Because of the coincidences of part of the results reported in our ms (effect of LOXL2 on metastasis of breast tumor cells) with those reported in the Barker's article (published while our ms was under revision in EMM), and the long delay that one of the reviewers took to submit his/her report, we would appreciate very much if you can make sure that reviewer #3 had no conflict of interest with the content of our ms. 5. Finally, we would like to mention that last week A. Cano presented the results reported in our ms in an invited talk at the Keystone meeting on "Epithelial plasticity and EMT" hold in Vancouver (21-25 January 2011), where they were recognised as highly novel and relevant by many specialist in the field. The attendance to this meeting is also the reason for the delay in answering to your decision message. On the bases of all the above arguments, we would like to respectfully ask you to reconsider your decision and provide us the opportunity to submit a revised version in which the main and objectives queries of the reviewers can be addressed. Also, because of the long delay we have already experienced with this ms., we would also very much appreciate if you can provide an answer to this letter as soon as possible, in order to avoid further delays in submission of our ms to other journal in case you are not able to modify your decision. We do appreciate your time and consideration and apologise for this long reply, that we consider absolutely necessary to clarify our opinion.
2nd Editorial Decision
09 February 2011
Thank you for your message asking us to reconsider our decision and please accept my apologies for not replying earlier. I now had the opportunity to carefully re-read your manuscript and considered your letter as well as the related literature and reviewer reports again. I have also discussed them with my colleagues and our Chief Editor as well as consulted with an Editorial Board member. You will see from the advisor's report below that he/she feels we could consider a major revision of your manuscript. The advisor highlights that it is crucial to provide conclusive mechanistic insight, especially considering the new paper, which reports conflicting evidence regarding primary tumor growth. Finally, the advisor highlights the need to strengthen the IHC data as suggested by Reviewer 3. His/her report is pasted below. Revised manuscripts should be submitted within three month of a request for revision. They will otherwise be treated as new submissions, unless arranged differently with the editor. I look forward to reading a revised version of your manuscript as soon as possible. Yours sincerely, Editor EMBO Molecular Medicine
© European Molecular Biology Organization
8
EMBO Molecular Medicine
Peer Review Process File - EMM-2010-00562
Advisor report: For a revision they would have to consider the new paper that came out - in particular Barker in Can Res 2011 since this group shows that LOX KD only impacts on metastasis - & even makes the 4T1 primary tumors grow better. This group also used the MB231 & did not see an effect on primary tumor growth - which is different from what Moreno et al see. Reviewers 1 & 2 do bring up some important points - in particular the fact that there's no real mechanistic insight in their work - (this contrasts with Barker et al). IHC - all the controls need to be shown & if they have another good tested antibody then it needs to be shown too. The literature is full of bad IHC on tumor sections!
1st Revision - authors' response
02 May 2011
Detailed point-by-point response to the editor, external advisor and the reviewers: Editor and external advisor: Editor: “The advisor highlights that it is crucial to provide conclusive mechanistic insight, especially considering the new paper, which reports conflicting evidence regarding primary tumor growth. Finally, the advisor highlights the need to strengthen the IHC data as suggested by Reviewer 3” External advisor: a)“For a revision they would have to consider the new paper that came out - in particular Barker in Can Res 2011 since this group shows that LOX KD only impacts on metastasis - & even makes the 4T1 primary tumors grow better. This group also used the MB231 & did not see an effect on primary tumor growth - which is different from what Moreno et al see. As suggested by the advisor, we have thoroughly considered the recent Barker et al.’s article (Cancer Res, 2011) in relation to our own data. Firstly, we reassessed the expression of LOXL2 and other LOX members in the mouse breast carcinoma 4T1 cells that were not included in our original ms. We have obtained 4T1 cells from the ATCC and analysed the expression of the five LOX family members by RT-PCR and qPCR. As shown in the following figure 1 (a), we could not detect expression of LOXL2 transcript in 4T1 cells, instead these cells appear to express LOXL3, LOXL4 and to a lower extent LOXL1 isoforms. To confirm the specificity of this analysis, we have transfected 4T1 cells with a mouse LOXL2 cDNA in transient assays in several independent experiments, observing the expression of LOXL2 transcript (figure 1, b).
© European Molecular Biology Organization
9
EMBO Molecular Medicine
Peer Review Process File - EMM-2010-00562
These results preclude any further analyses of the effect of LOXL2 knockdown in 4T1 cells and strongly suggest that the previous reported data by Barker et al. (Cancer Res, 2011) either rely on a derivative 4T1 cell line that has acquired LOXL2 expression with culture passages or that their knockdown analyses were not specific to LOXL2. Indeed, a careful inspection of the Barker et al. article indicates that no control for the specificity of the LOXL2 KD is provided. A band of about 150 kDa is also detected in the extracellular medium of 4T1 cells in the Barker et al. article (see Fig. 4A, Suppl Fig S2B in that workusing a commercial anti-LOXL2 antisera. To confirm the specificity of our home-made antibody (see previous antibody characterization in Fong et al., 2007; Hollosi et al., 2009) HEK293T cells were transiently transfected with LOXL2 cDNA and analysed by WB. As can be observed in the enclosed figure 2, the expected 87-95 kDa band for human LOXL2 was detected in in whole cells extracts of HEK293T-LOXL2-Flag cells, but absent in control HEK293T cells, , together with a nonspecific band of about 150-170 kDa (figure 2b). The antibody also detected the 87-95 kDa band in the conditioned medium of HEK293TLOXL2 cells, together with two non-specific bands of about 115 and 180 kDa, also detected in conditioned medium from control HEK293T cells (figure 2a). A band of about 150-170 kDa was also detected in the conditioned medium of MCF7 and MDA-MB-231 cells regardless of LOXL2 expression (see enclosed figure 3). These data suggest, although do not formally prove, that the bands detected in the extracellular medium of 4T1 cells in the Barker et al’s report likely correspond to non-specific bands instead of the secreted LOXL2.
© European Molecular Biology Organization
10
EMBO Molecular Medicine
Peer Review Process File - EMM-2010-00562
It is also worth noting that the inhibition studies with D-penicillamine reported in the Barker et al. article are not indicative of LOXL2 targeting, since this is a general copper chelator that can target other copper-dependent enzymes and indeed it has been reported to inhibit LOX activity (Misiorowski & Werner, BBRC, 85: 809-14, 1978; Roncheti et al., Am J Pathol, 124: 436-47, 1986), apart from LOXL2 (Valdaz et al., J Pathol., 43: 499-507, 2005). Regarding the tumorigenic behaviour of MDA-MB-231 cells, we observed that in the Barker’s article 1x107 cells were injected into the fat mammary pad, in contrast to 1x106 cells injected in our previous study (and as usual dose in most studies). This fact could potentially explain the differences in tumorigenicity observed in both studies. Nevertheless, we have repeated the tumorigenesis assays by injecting both 1x107 and 1x106 cells of parental MDA-MB-231, control shEGFP and shLOXL2 cells and repeately observed the differences in tumorigenicity (see new Fig. 6A). In relation to this point, we would like to comment that significant differences in tumour growth after inhibition or silencing of LOXL2 in MDA-MB-435 or gastric carcinoma cells have been also reported in other recent publications (Barry-Hamilton et al., Nat. Medicine, 2010; Peng et al., Carcinogenesis, 2009), further supporting an effect of LOXL2 in tumour growth potential of different cell lines. These data have been also considered in the discussion of the revised version (see page 22, lines 12-15). Finally, regarding the metastasis data of MDA-MB-231 cells also reported in Barker et al. article, we would also like to mention for the editor/advisor’s
consideration
that
parental
MDA-MB-231 cells are considered by most specialists non-metastatic when injected into the mammary fat pad. This has been reported in the original articles characterizing this cell line (Price et al., Cancer Res, 50:717-21, 1990) as well as in more recent studies, including those from Massagué’s group (Minn et al, Nature, 436:51824, 2005) and our own previous study on Snail1 silencing that extended the analysis up to 6 month after injection (Olmeda et al., Cancer Res, 67:11721-31, 2007). Indeed, many groups that study the spontaneous metastatic behaviour of MDAMB-231 cells use highly metastatic derivative cell lines, like 231/LM2-4 cells (Munoz et al., Cancer Res, 66:3386-91, 2006), or organ-specific metastatic derivatives as those generated by Massagué’s group (Minn et al, 2005; Kang et al., Cancer Cell, 3:537-49, 2003). Therefore, it is rather surprising that spontaneous lung metastasis of MDA-MB-231 cells are observed in the recent Barker’s article, cited above. It is nevertheless likely that the high cell number injected in that previous work could promote the metastatic spread.
© European Molecular Biology Organization
11
EMBO Molecular Medicine
Peer Review Process File - EMM-2010-00562
b) Reviewers 1 & 2 do bring up some important points - in particular the fact that there's no real mechanistic insight in their work - (this contrasts with Barker et al). As described in detail in response to reviewers 1 & 2, our new results included in this revised version provide new mechanistic insights into the intracellular action of LOXL2, that is basically derived from the regulation of cell polarity and focal adhesion complexes and a novel involvement of transcriptional regulation of cell polarity and tight junction genes independent of Snail1 and LOXL2 catalytic activity. We firmly believe that these new data, together with added IHQ analyses in an extended tumour series provide strong basis to support an intracellular action of LOXL2 in basal-like carcinomas. These data, however, do not preclude an additional extracellular function of LOXL2 on the microenvironment as we discussed in the original ms. and in the present revised version (see page 23, last 7 lines, and page 24, lines 1-8). c) IHC - all the controls need to be shown & if they have another good tested antibody then it needs to be shown too. The literature is full of bad IHC on tumor sections!” We have included the proper controls for the polyclonal antibody used (new Suppl Figs S9 & S10) and also stained our tumour series with an additional non-commercial tested anti-LOXL2 antibody (reported in Barry-Hamilton, Nat Medicine, 2010). The results are included in the new Suppl. Fig. S9 and discussed in the text. Intracellular LOXL2 stain has been detected in a subset of basal tumours with the later antiserum, further supporting the intracellular LOXL2 detection with an independent antibody. We decided not to test additional commercial antibodies because of their lack of reliability. Indeed, we would like here to further comment on the specificity of our home-made polyclonal antibody. We have generated these antibodies based on two decades of accumulated knowledge in the K.Csiszar’s lab on the similarities between the LOXL proteins and biochemical data on the processing of LOXL2 and its processing in tumour cells. We have thoroughly tested these LOXL2 antibodies in normal and tumour cells and also in immunohistochemistry using a comprehensive tissue panel and also series of tumour tissues (see for instance, Fong et al., Genes Chromosomes & Cancer, 46:644-55, 2007; Hollosi et al., Int J Cancer, 125;318-27, 2009). Regarding the commercial anti-LOXL2 antibodies we have reviewed all available commercial antihuman LOXL2 antibodies and summarized these in the attached table 1. As can be observed from the epitopes there are only a few genuine antibodies, most of the others are remakes. There are diverse bands detected by the different antibodies. There are also massive inconsistencies with antibodies generated against overlapping epitopes that detect different size, 87 and 37 kD, bands. The overlapping antibodies detected different size proteins in cell extracts and when recombinant protein was used for test. None of these antibodies were tested in tissues neither normal or cancerous. Cell-based tests were limited for only 3 antibodies, and in these tests various protein bands were detected with inconsistencies.
© European Molecular Biology Organization
12
EMBO Molecular Medicine
Peer Review Process File - EMM-2010-00562
Table 1. Commercial human anti-LOXL2 antibodies Epitope in Human LOXL2
Species
LOXL2 (H-65): sc66950
161-225
Rabbit†
Santa Cruz Biotechnology, Inc#
LOXl2 Mab 262418
26-744
Mouse‡
R&D Systems
SAB1100132
31-45
Rabbit
98,43,35
HEK-293T
Ab96233
454-712
Rabbit†
87
293T, A431, H1299
ABCAM
S2057
454-712
Rabbit†
87
293T, A431, H1299
EPITOMICS
GTX105085
454-712
Rabbit†
87
NBP1-32954
454-712
Rabbit†
87
SAB1100133
606-620
Rabbit
96,72
PA1-41263
61-73
Rabbit†
PAB14599
61-73
Rabbit†
Thermo Scientific Pierce Antibodies ABNOVA
IMIX-5126
61-73
Rabbit†
MGENEX
ab55470
675-774
Mouse‡
37
ab60753
675-774w taq
Mouse‡
37
AT2736a
675-774w taq
Mouse‡
ABGENT
AT2735a
675-774w taq
Mouse‡
ABGENT
H00004017-M05
675-774w taq
Mouse‡
37
H00004017-M02
675-774w taq
Mouse‡
37
H00004017-M01
675-774w taq
Mouse‡
37
H00004017-A01
675-774w taq
Mouse‡
37
LOXL2 Ab (3C5)
675-774w taq
Mouse‡
WH0004017M1
675-774w taq
Mouse‡
Name
LOXL2 (F-13): sc48723 LOXL2 (N-15): sc48724
Internal región of LOXL2* N-terminus of LOXL2**
Goat†
Band (kDa)
Cell Line
293T, A431, H1299 293T, A431, H1299 293T
Recombinant Protein Recombinant Protein
Recombinant Protein Recombinant Protein Recombinant Protein Recombinant Protein
Company
SIGMA-ALDRICH
GENETEX NOVUS BIOLOGICALS SIGMA-ALDRICH
ABCAM ABCAM
ABNOVA ABNOVA ABNOVA ABNOVA NOVUS BIOLOGICALS
37
Recombinant Protein
SIGMA-ALDRICH Santa Cruz Biothechnology, Inc# Santa Cruz Biothechnology, Inc#
*Lelievre E et al (2008) VE-stain/egfl7 regulates vascular elastogenesis by interacting with lysyl oxidases. EMBO J 27(12): 1658-1670. ** Kamarasamy A et al. (2009). Lysyl oxidase is dysregulated during impaired alveolization of mouse and human lungs. Am. J. Respir. Crit. Care Med. 180: 1239-1252. †Polyclonal.; ‡Monoclonal. #One of those three used in Barker et al. LOXL2-mediated matrix remodelling in metastasis and mammary gland involution. Cancer Res, 71: 1561-1572 (2011).
In the corresponding technical sheets of the various antibodies there are no explanations for the different size bands nor any consideration of the underlying mechanisms that may produce these. No correlation with any sort of potential processing of LOXL2, nor hint/test whether the detected form(s) may be catalytically active or not. Honestly, looking at this table, we thought that it would be a distraction and a complete side avenue to repeat the IHC with any of these unreliable
© European Molecular Biology Organization
13
EMBO Molecular Medicine
Peer Review Process File - EMM-2010-00562
commercial antibodies that can be the origin of bad data on tumours currently present in the literature, as the advisor mentioned. We hope that our IHQ data using the largely increased tumour series with two tested and validated good LOXL2 antibodies, together with the above arguments on the commercial antibodies will satisfy the advisor concerns on this important issue. As we explained in the answer to reviewer 3, IHQ analyses on additional EMT markers have also been performed and included in the present revised version (new Suppl. Fig. S10). Reviewer 1 We thank the reviewer for the constructive comments that have been addressed as follows: General comments: 1.
While the manuscript is concisely presented in its different parts, overall it represents a mixture of results which by themselves are not complete and in parts do not convincingly support the conclusions drawn by the authors. For example, while the bioinformatical data, which by itself is novel and very exciting, goes into some details of classification of various breast cancer subtypes, it stops short in using these data to test a specific gene signature on patient populations.
The ms. has been extensively rewritten and reorganized in order to more clearly present the data and the major conclusions based on the previous and additional evidence. We hope the reviewer will like this reorganization. We have addressed his/her specific comment on the bioinformatical data by analysing our “basal-like gene signature” in several public databases for breast tumours. The signature has been confirmed to be present and indeed define the basal like phenotype in the Van’t Veer et al. (2002) and Wang et al. (2005) tumour series and also in the Charafe-Jauffret (2006) breast carcinoma cell lines series (new Fig. S1B). Furthermore, upregulation of LOXL2 was also confirmed in basal-like tumours and/or associated to metastasis in the Wang et al. (2005), Hess et al. (2006), Sortiriu et al. (2006) and Van de Vijver et al. (2002) series (new Fig. S2A, B), in agreement with our previous observation that LOXL2 associated to decreased overall survival in the Wang et al. series (see Peinado et al, Cancer Res, 2008). In summary, to emphasise the novel observations of this analysis: a) the identified “basal-like gene expression signature” defines the basal phenotype in several tumour series, apart from our own; b) LOXL2 overexpression defines the basal-like phenotype; and c) LOXL2 overexpression associates with metastatic basal-like tumours. 2.
Moreover, while the functional study of LOXL2 contribution clearly shows the requirement of LOXL2 for EMT, tumor cell invasion and metastasis, it does not provide insights into the mechanisms by which LOXL2 exerts this function.
We have explored in further detail the mechanism involved in LOXL2 action to modulate EMT and provide evidence for its participation in the regulation of cell polarity and tight junctions in a negative fashion. WB and confocal immunofluorescence analyses indicate the upregulation and/or increased organization of tight junction components (claudin1, ZO1) after LOXL2 silencing in basal-like carcinoma cells, and their downregulation/disorganization following LOXL2 expression in MCF7 cells (new Fig. 4). Furthermore, the expression and organization of cell polarity complexes
© European Molecular Biology Organization
14
EMBO Molecular Medicine
Peer Review Process File - EMM-2010-00562
is also modulated by LOXL2. Indeed, while LOXL2-expressing cells show a complete disorganization of Par3 and Lgl2 components, the Lgl2 and Par3 containing complexes are clearly redistributed to the upper lateral membranes of LOXL2-silenced cells (new Fig. 4A). The reverse situation on cell polarity components was detected in MCF7-LOXL2 cells (new Fig. 4B). In contrast, the organization of focal adhesion complexes and F-actin are modified in the reverse way depending on LOXL2 expression: focal adhesions are strongly decreased and reorganization of F-actin from stress fibers/filopodia to cortical F-actin occurs following LOXL2 silencing in MDAMB-231 cells, while the opposite behaviour was observed in MCF7 cells overexpressing LOXL2 (new Fig. 3A and Suppl Fig. S5). Furthermore, the FAK and Src signalling pathways are also modulated in a parallel fashion to focal contact organization depending on LOXL2 expression, decreased phospho-FAK and phosphor-Src in MDA-MB-321-shLOXL2 cells (new Fig. 3C, D; Suppl Fig. S5). These results provide new mechanistic insights into the action of LOXL2 and indicate that it acts by downmodulating apico-basal cell polarity, a finding that to our knowledge has not been previously reported and that can explain the maintenance of the mesenchymal phenotype by LOXL2. On the other hand, LOXL2 has a strong positive influence in the organization of focal adhesion complexes, a fact that can explain the increased invasive and migratory behaviour of basallike breast carcinoma cells. The noted modulation of FAK pathway by LOXL2 is also in agreement with recent data reported on gastric carcinoma cells (Peng et al. Carcinogenesis, 2009), as further discussed in the text. 3.
Notably, circumstantial evidence suggests that LOXL2 may act in a Snail- and E-cadherinindependent function, yet this observation is not corroborated on, and no experimentation has been performed to directly test this possibility. The actual targets of LOXL2 remain unknown.
The main evidence presented in the original ms. to support the independent action of LOXL2 from Snail1 in basal-like carcinoma cells was derived from the observation that despite the induction of MET and reduced migration and invasion in MDA-MB-231-shLOXL2 cells (new Fig. 2H-I)) no changes in Snail1 expression levels were detected compared to parental and control shEGFP cells (new Fig. 2E). We agree with the reviewer that this can be considered only circumstantial evidence. However, the same situation was observed after LOXL2 knockdown in BT549 cells that do not express endogenous Snail1, at any detectable level (Suppl. Figs. S3 and S4A, C). Indeed, we choose this additional basal-like carcinoma cell line for further studies because of its strong expression of LOXL2 and its complete lack of endogenous Snail1 expression to further corroborate the findings in MDA-MB-231 cells. We firmly believe that these data provide strong support for a LOXL2 action in the maintenance of the mesenchymal-like phenotype of basal-like carcinoma cells independent of Snail1. The new functional experiments on claudin1 and Lgl2 promoters presented in this revised version also indicate a novel repressor action of LOXL2 on both genes that appear to be independent of Snail1 (see below, answer to point 4). Although additional experimental work can be designed to further address this issue, for instance by manipulating Snail1 and/or LOXL2 expression in other cellular models, we honestly think they could not provide stronger evidence than the data we have already obtained in a cell system fully devoid of endogenous Snail1 expression.
© European Molecular Biology Organization
15
EMBO Molecular Medicine
Peer Review Process File - EMM-2010-00562
Regarding E-cadherin, our data in the MDA-MB-231 and BT549 cells also support the LOXL2 action as independent of E-cadherin modulation or organization. Although some weak re-expression of E-cadherin was detected in MDA-MB-231-shLOXL2 cells in WB (new Fig. 2E), the confocal analyses clearly indicated that E-cadherin was not properly organized at cell-cell contacts in most cells or apparently involved in the reorganization of the cell polarity complexes observed in shLOXL2 cells (new Fig. 4A). Furthermore, no reexpression of E-cadherin could be detected in the BT549-shLOXL2 cells or in basal-like HBL100 cells (Suppl Fig. S3A). In addition to those data presented in the original version of the ms, the independence of LOXL2 action on E-cadherin in breast cells has also been corroborated in the MCF7-LOXL2 cells, since those cells maintain detectable levels of E-cadherin (new Suppl Fig. S6C) properly organized at cell-cell contacts (new Fig. 4Ba,b) while acquiring a mesenchymal-like phenotype (new Fig 3B) and increased migratory and invasive properties, as reported by us and others (Hollosi et al., Int J Cancer, 125;318-27, 2009; Akiri et al., Cancer Res, 63:1657-66, 2003). Interestingly, a similar situation has been also reported for Zeb1 mediated downregulation of cell polarity in MDA-MB-231 cells that appears to be also independent of E-cadherin (Aigner et al, Oncogene, 2007). Furthermore, staining of the tumours for LOXL2 and E-cadherin shows the co-expression of perinuclear LOXL2 and membranous Ecadherin in a high proportion of basal-like carcinomas (new Suppl. Fig. S10; Table S3), also reinforcing the LOXL2 independence of E-cadherin in this specific breast tumour subtype. As discussed in the present revised version, these data provide support to the induction of the epithelial phenotype by LOXL2 knockdown in basal-like cells independent of E-cadherin modulation. This finding can be considered at odds with the established opinion in the field that adherens junctions mediated by E-cadherin are required for the proper organization of tight junctions and cell polarity complexes (see for instance, Perez-Moreno et al, Cell, 112: 535-48, 2003). However, our finding also raises the interesting notion that the action of E-cadherin can be overcome in specific cell contexts, like the basal-like carcinoma cells when deprived of LOXL2 as described in the present study. Indeed, some situations in which other classical cadherins, like N-cadherin, can replace Ecadherin as in embryonic intestinal epithelium organization have been recently reported (Libusova et al., Development, 137: 2297-305, 2010). This important issue has been further discussed in the present revised version (see page 25, lines 2-4). The answer to the reviewer comment on the actual LOXL2 targets is presented in the next point. 4.
Also, it is claimed that LOXL2 may act via modulating tight junction rather than adherens junction components, yet no direct evidence is provided for this conclusion. Testing the effects of LOXL2 activity on tight junctions and adherens junctions may need the use of short-term reversible EMT systems.
As indicated in the answer to point 2, we have further extended our previous study on the tight junction components and included new analyses on cell polarity components. The more extensive WB and IF confocal studies presented in this revised version showed increased expression and organization of tight junctions and cell polarity complexes after LOXL2 silencing in MDA-MB-231 cells, while the reverse situation was observed in MCF7 cells overexpressing LOXL2 (new Fig 4 and Suppl Fig. S6). In addition, the new RT-PCR data showing increased expression of claudin1
© European Molecular Biology Organization
16
EMBO Molecular Medicine
Peer Review Process File - EMM-2010-00562
and Lgl2 transcripts (new Fig. 5A), led us to analyse in further detail the regulation of those genes by LOXL2. This has been performed analysing the influence of LOXL2 on the activity of the human claudin1 and Lgl2 promoters characterized previously by us and other groups (Martinez-Estrada et al, Biochem J, 2006; Brabletz et al, Cancer Res, 2008). The results indicate that LOXL2 represses the activity of the human Lgl2 (two independent constructs) and claudin1 promoters in transient transfection assays to about 60 and 40%, respectively (new Fig. 5C,D). The LOXL2 mediated repression of claudin1 and Lgl2 promoters is independent of Snail1 and, importantly, of the catalytic domain of LOXL2 (new Fig. 5D-F), thus indicating a novel LOXL2 action on claudin1 and Lgl2 expression at transcriptional level and reinforcing the proposed Snail1-independent action of LOXL2. Furthermore, the ability of LOXL2 to downregulate claudin1 and Lgl2 is also supported by the upregulation of the promoters observed in MDA-MB-231shLOXL2 cells compared to parental and control cells (new Fig. 5B). Taken together, these results provide strong support for the regulation of LOXL2 on tight junction and cell polarity genes at transcriptional level, thus providing novel insights into the mechanisms involved in LOXL2 maintenance of the mesenchymal phenotype of basal-like carcinoma cells. We firmly believe that these data also provide for the first time information on novel intracellular targets of LOXL2 and support catalytic domain or activityindependent actions of LOXL2. Finally, to answer the criticism of the reviewer on using established cell lines and the suggestion for using short-term reversible EMT systems, we have tried to knockdown LOXL2 in MDA-MB-231 cells in a transient and reversible fashion. Unfortunately, and after repeated attempts, we have been unable to obtain viable cell populations after lentiviral infections precluding the development of the suggested system within the short revision period available to us. Despite this caveat, we firmly believe that the transient transfection assays commented above clearly support the action of LOXL2 after a short expression period (24 h) and thus overcome the criticism for the use of only established cell lines. Specific comments: a)
Page 8, Fig. 2C: a,b, or c are not in the figure or the figure legend.
This has been corrected in the text. b) Page 9, Fig. 3F: It is stated that proliferation is not affected by the knock-down of LOXL2, yet there is a moderate if not significant effect? Statistical analyses of proliferation data (new Fig. 2J) indicated that the differences observed between control and shLOXL2 cells (20%) are non-significant. This has been indicated in the figure (n.s.) and legend. c) Page 9, 2. Paragraph: Fig. S3A not Fig. 3A. This has been corrected in the text. d) The significance of the correlation between the 231-shLoxl2 and BBC gene signatures is not clear. The fact that there are some similarities between gene ontology and only a few common genes
© European Molecular Biology Organization
17
EMBO Molecular Medicine
Peer Review Process File - EMM-2010-00562
is not very meaningful. There is also no evidence whatsoever that some of these genes are "likely candidates to be regulated by LOXL2 in basal tumors". We agree with this comment and, accordingly, we have eliminated the microarray analysis of 231shLOXL2 cells and its comparison with BBC signature, as well as Fig. 4 and Supplemental Fig. S3 of the original ms. e) Fig. 5: The results show an effect of LOXL2 on primary tumor growth and tumor recurrence, not on metastasis. This is disappointing, since these cell line is supposed to be metastatic. Is it possible that LOXL2 acts also on tumor cell proliferation and less so on EMT and tumor invasion? Is there a difference in tumor histology in the presence and absence of LOXL2? The results obtained in the mouse tail vein injection experiments indicate that LOXL2 silencing has a strong effect on experimental metastasis. Unfortunately, the effect of LOXL2 on spontaneous metastasis could not be ascertained, but this is likely because of the low potential of spontaneous metastasis of MDA-MB-231 cells. We would like to respectfully comment to the reviewer that, in contrast to his comment, parental MDA-MB-231 cells are considered as non-metastatic when injected into the mammary fat pad. This has been reported in the original article characterizing this cell line (Price et al., Cancer Res, 50:717-21, 1990) as well as in more recent studies, including those from Massagué’s group (Minn et al, Nature, 2003, 436: 518-24, 2005) and our own previous study on Snail1 silencing that extended the analysis up to 6 month after injection (Olmeda et al., Cancer Res, 67: 11721-31, 2007). Indeed, many groups that study the spontaneous metastatic behaviour of MDA-MB-231 cells, use highly metastatic derivative cell lines, like 231/LM2-4 cells (Munoz et al., Cancer Res, 66:3386-91, 2006) or organ-specific metastatic derivatives as those generated by Massagué’s group (Minn et al, 2005; Kang et al., Cancer Cell, 3: 537-49, 2003). A recent report by Barker et al (Cancer Res, 2011; appeared on-line on January 13, 2011, while our ms. was under revision) surprisingly show spontaneous lung metastasis of MDA-MB-231 cells after mammary fat pad injection. In this regard, it is worth mentioning that the high cell number injected (i.e., 1x107 cells, 10-times higher than in our study and in most regular studies) in that previous work could favour the metastatic spread. We have repeated the tumorigenic analyses using 1x107 and 1x106 cells per injection site. Significant differences were still observed in the tumour growth between control and shLOXL2 cells in both conditions, although we prefer to present the data for 1x106 cells as in the original version, but showing the data of a new independent experiment. Our data are also in agreement with other reports showing differences in tumour growth after inhibition or silencing of LOXL2 in MDA-MB-435 or gastric carcinoma cells (Barry-Hamilton et al., Nat. Medicine, 2010; Peng et al., Carcinogenesis, 2009). These data have been discussed in the present revised ms (see pages 22, lines 12-15). No differences in the tumour histology have been observed between control and shLOXL2 cells, and this has been also mentioned in the text. A similar situation was observed in our previous studies on MDA-MB-231 cells after Snail1 silencing (Olmeda et al, Cancer Res, 2007).
© European Molecular Biology Organization
18
EMBO Molecular Medicine
Peer Review Process File - EMM-2010-00562
f) Lung colonization is shown in Fig. 5 C-D and not as mentioned in Fig. 5 B-D. This has been corrected in the revised version (referred as present Fig. 6 C, D) g) There is no labeling of Figure 6B "a"? This has been corrected in the revised version (present Fig. 7) We thank the reviewer for his/her comments and hope he/she will find our responses satisfactorily and that we have addressed the main concerns. Reviewer 2 We thank the reviewer for the constructive and encouraging comments on the ms. The specific concerns have been addressed as follows: 1.
To further strengthen and foster this findings, the immunohistochemical stainings should be repeated with other available antibodies/antisera against LOXL2. This is of particular importance because different staining patterns are described for LOXL2 (e.g. BarryHamilton et al., Nature Med, 2010).
We have extended the IHQ analyses to an additional series of 104 grade 3 breast tumours, confirming our previous finding on the significant association of perinuclear LOXL2 expression and metastatic basal-like tumours (new Fig 7B, C and Tables S3). As suggested by the reviewer, the IHQ staining have been repeated used an independent anti-LOXL2 antibody. The samples have been stained with the LOXL2 antiserum previously described in the Barry-Hamilton et al. (Nature Med, 2010), with the help of Dr. V. Smith and Dr. V. Barry-Hamilton (Gilead Science Inc). The results are included in the new Fig. S9 and discussed in the text. Intracellular stain has been detected in a subset of basal tumours with the later antiserum, further supporting the intracellular LOXL2 detection with an independent antibody. We would like to further comment on the specificity of our home-made polyclonal antibody. These antibodies were generated based on two decades of accumulated knowledge on the K. Csiszar’s lab on the similarities between the LOXL proteins and biochemical data on the processing of LOXL2 and its processing in tumour cells. We have thoroughly tested these LOXL2 abs in normal and tumour cells and also in immunohistochemistry using a comprehensive tissue panel and also series of tumour tissues (Fong et al., Genes Chromosomes & Cancer, 46:644-55, 2007; Hollosi et al., Int J Cancer, 125;318-27, 2009). We have not used commercial anti-LOXL2 antibodies because of their lack of reliability. We have reviewed all available commercial anti-human LOXL2 antibodies and summarized these in the attached table for reviewer consideration. As can be observed from the epitopes there are a few genuine antibodies, most of the others are remakes. There are diverse bands detected by the different antibodies. There are also massive inconsistencies with antibodies generated against overlapping epitopes that detect different size, 87 and 37 kD, bands. The overlapping antibodies detected different size proteins in cell extracts and when recombinant protein was used for test. None of these antibodies were tested in tissues neither normal or cancerous. Cell-
© European Molecular Biology Organization
19
EMBO Molecular Medicine
Peer Review Process File - EMM-2010-00562
based tests were limited for only 3 abs, and in these tests various protein bands were detected with inconsistencies. Table 1. Commercial human anti-LOXL2 antibodies Epitope in Human LOXL2
Species
161-225
Rabbit†
Santa Cruz Biotechnology, Inc
26-744
Mouse‡
R&D Systems
31-45
Rabbit
98,43,35
HEK-293T
Ab96233
454-712
Rabbit†
87
293T, A431, H1299
ABCAM
S2057
454-712
Rabbit†
87
293T, A431, H1299
EPITOMICS
GTX105085
454-712
Rabbit†
87
NBP1-32954
454-712
Rabbit†
87
SAB1100133
606-620
Rabbit
96,72
PA1-41263
61-73
Rabbit†
PAB14599
61-73
Rabbit†
IMIX-5126
61-73
Rabbit†
Name LOXL2 (H-65): sc66950 LOXl2 Mab 262418 SAB1100132
Band (kDa)
Cell Line
293T, A431, H1299 293T, A431, H1299 293T
Company
SIGMA-ALDRICH
GENETEX NOVUS BIOLOGICALS SIGMA-ALDRICH Thermo Scientific Pierce Antibodies ABNOVA MGENEX
Recombinant Protein Recombinant Protein
ab55470
675-774
Mouse‡
37
ab60753
675-774w taq
Mouse‡
37
AT2736a
675-774w taq
Mouse‡
ABGENT
AT2735a
675-774w taq
Mouse‡
ABGENT
H00004017-M05
675-774w taq
Mouse‡
37
H00004017-M02
675-774w taq
Mouse‡
37
H00004017-M01
675-774w taq
Mouse‡
37
H00004017-A01
675-774w taq
Mouse‡
37
LOXL2 Ab (3C5)
675-774w taq
Mouse‡
WH0004017M1
675-774w taq
Mouse‡
Recombinant Protein Recombinant Protein Recombinant Protein Recombinant Protein
ABCAM ABCAM
ABNOVA ABNOVA ABNOVA ABNOVA NOVUS BIOLOGICALS
37
Recombinant Protein
SIGMA-ALDRICH
LOXL2 (F-13): scInternal región of Santa Cruz Goat† 48723 LOXL2* Biothechnology, Inc LOXL2 (N-15): scN-terminus of Santa Cruz 48724 LOXL2** Biothechnology, Inc *Lelievre E et al (2008) VE-stain/egfl7 regulates vascular elastogenesis by interacting with lysyl oxidases. EMBO J 27(12): 1658-1670. ** Kamarasamy A et al. (2009). Lysyl oxidase is dysregulated during impaired alveolization of mouse and human lungs. Am. J. Respir. Crit. Care Med. 180: 1239-1252. †Polyclonal. ‡Monoclonal. #One of those three used in Barker et al (2011). LOXL2-mediated matrix remodelling in metastasis and mammary gland involution. Cancer Res, 71: 1561-1572 (2011).
© European Molecular Biology Organization
20
EMBO Molecular Medicine
Peer Review Process File - EMM-2010-00562
We hope that our IHQ data using the largely increased tumour series with two tested and validated good LOXL2 antibodies, together with the above arguments on the commercial antibodies will satisfy the reviewer concerns on this important issue. 2. Minor points: - the supplementary tables should be better explained (e.g. abbreviations in the headlines) We have revised the supplementary tables and improved their explanation as suggested by the reviewer. - page 9: should read ...'HBL100 basal cell lines (Fig. S3A)...' This has been corrected in the revised ms., now corresponding to Fig. S3B We hope that the reviewer find our responses fully satisfactory. Reviewer 3 We appreciate the reviewer´s comments and we have tried to address the specific issues raised to the best of our knowledge. Nevertheless, we also respectfully disagree with other general comments of this reviewer, and would like to explain our arguments here: Overall comment on novelty/model systems: Most if not all of the observations presented in this article have been published previously either by the current authors or by others. This includes demonstration that LOXL2 can regulate breast tumor progression, migration, invasion and tumorigenesis/metastasis in vivo, demonstration that LOXL2 can regulate EMT and appears to do so through a novel yet currently unresolved mechanism, that LOXL2 is highly expressed in human breast tissues. No new mechanistic insight is provided linking LOXL2 to EMT. We honestly think that some of those comments are partly incorrect, particularly at the time of submission of our ms., and we would like to explain them further: a. The reviewer is right when mentioning that the influence of LOXL2 in the invasive behaviour of some breast tumour cells was previously reported by other authors. However, those previous studies have not distinguished between breast tumour cells representing distinct tumour phenotypes. We have extended the study to show that the expression of LOXL2 is restricted to basal-like carcinoma cell lines and analysed its effect on migration and/or invasiveness of three independent basal like breast cells. The effect on tumorigenic and metastatic behaviour was reported in a recent paper by BarryHamilton et al. (Nat Medicine, 16: 1009-17, 2010) (published when our ms. was under preparation) showing the effect of LOXL2 inhibition in the tumorigenic behaviour of MDA-MB-435 cells (a cell line that is still under debate if it represents breast or melanoma derived cells; see Rae et al., Breast Cancer Res Treat., 104: 13-9, 2007; Chambers et al., Cancer Res, 69: 5293-3, 2009) and in bone metastasis of MDA-MB-231; those previous results were discussed in our ms. Surprisingly, while
© European Molecular Biology Organization
21
EMBO Molecular Medicine
Peer Review Process File - EMM-2010-00562
our ms. was under review, an additional article came out (Barker et al., Cancer Res, Published OnlineFirst January 13, 2011, DOI:10.1158/0008-5472.CAN-10-2868) showing that LOXL2 knockdown or inhibition decreased spontaneous metastasis of mouse breast 4T1 cells and human breast MDA-MB-231 cells. No such information was therefore publicly available when our ms. was submitted and was under revision in EMM. b. The statement related to the previous demonstration that “LOXL2 is highly expressed in human breast tissues” only stand, to our knowledge, for meta-analyses studies of public databases of microarrays, including our own previous observations in Wang et al (2005)’s series of N0 breast carcinomas the showed a correlation of LOXL2 transcripts with poor survival (Peinado et al, Cancer Res, 2008) and the study of selected ER- tumours (that will include both ER- and ErbB2+ tumours) from the NKI database in which high LOXL2 mRNA is correlated with metastasis that was included in the recent paper by Barker et al. (2011), referred above. A small and phenotypically undefined breast adenocarcinoma series was analysed for LOXL2 expression by IHQ in Barry-Hamilton et al. (2010) and Hollosi et al (2009) (as discussed in our original ms., see below point 3 in the Remarks to authors’s response). In none of the previous studies the protein expression of LOXL2 by IHQ was analysed in an extensive series of grade 3 breast carcinoma such as the one included in our original ms. More importantly, we have been able to show for the first time that LOXL2 cytoplasmic and perinuclear staining pattern significantly correlates with metastasis of basal-like carcinomas. This finding has been further confirmed in an extended series of additional 104 grade 3 breast tumours showing the strong and unequivocal association of intracellular LOXL2 with metastatic basal-like tumours (presented in the new Fig. 7, and Table S3). Indeed, as this and the other two reviewers point out, our results for LOXL2 expression and localization in the tumour series and its correlation with metastatic basal-like breast tumors are completely novel and highly relevant. No such information was provided in any of the previous publications, and this is precisely the information that we presented in our original ms. for the first time and further extended in the present revised version. We firmly believe that these results are completely novel. The answer to the concerns on the lack of mechanistic insights is provided below (points 1 & 6). Remarks to authors: 1.
“Prior studies by these authors and others have already shown that LOXL2 regulates tumor progression and metastasis and that LOXL2 can induce an EMT phenotype through an E cadherin/snail independent mechanism”.
Our previous works indeed indicated that LOXL2 can mediate EMT by Snail1 dependent and independent mechanisms (Peinado et al, EMBO J 2005; Peinado et al., Cancer Res, 2008), but we did not previously observed neither reported on LOXL2 effects independent of E-cadherin. To our knowledge, no information on this specific aspect has been reported by other authors. Our original and present data reinforce the LOXL2 independent action of E-cadherin and Snail1. As commented in the answer to reviewer 1, the main evidence presented in the original ms. to support the independent action
© European Molecular Biology Organization
22
EMBO Molecular Medicine
Peer Review Process File - EMM-2010-00562
of LOXL2 from Snail1 in basal-like carcinoma cells was derived from the observation that no changes in Snail1 expression levels were detected following LOXL2 silencing in MDA-MB-231 cells (new Fig. 2E), a situation in which cells acquire a noticeable epithelial phenotype (new Fig. 2G). Moreover, the same situation was observed after LOXL2 knockdown in BT549 cells (reverting to a striking epithelial phenotype) that do not express endogenous Snail1, at any detectable level (Suppl Figs. S3C and S4A). Indeed, we chose this additional basal-like carcinoma cell line for further studies because of its strong expression of LOXL2 and its complete lack of endogenous Snail1 expression to further corroborate the findings in MDA-MB-231 cells. We firmly believe that these data provide strong support for a LOXL2 action in the maintenance of the mesenchymal-like phenotype of basal-like carcinoma cells independent of Snail1. Our data in the MDA-MB-231 and BT549 cell systems also support the LOXL2 action independent of E-cadherin modulation or organization. Although some weak re-expression of Ecadherin was detected by WB in MDA-MB-231-shLOXL2 cells (Fig. 2E), the confocal analyses clearly indicated that E-cadherin was not properly organized at cell-cell contacts in most cells or apparently involved in the reorganization of the cell polarity complexes observed in shLOXL2 cells (new Fig. 4). Furthermore, no reexpression of E-cadherin could be detected in the BT549-shLOXL2 cells or in basallike HBL100 cells (Suppl Fig. S3B). In addition to those data, presented in the original version of the ms, the independence of LOXL2 action on E-cadherin in breast cells has also been corroborated in the MCF7-LOXL2 cells, since those cells maintain detectable levels of E-cadherin (new Suppl Fig. S6C) properly organized at cell-cell contacts (new Fig. 4Ba,b) while acquiring a mesenchymal-like phenotype (Fig. 3B) and increased migratory and invasive properties, as reported by us and others (Hollosi et al., Int J Cancer, 125;318-27, 2009; Akiri et al., Cancer Res, 63:1657-66, 2003). Furthermore, staining of the tumours for LOXL2 and E-cadherin shows the co-expression of perinuclear LOXL2 and membranous E-cadherin in a high proportion of basal-like carcinomas (new Suppl. Fig. S10; Table S3), also reinforcing the LOXL2 independence of E-cadherin in this specific breast tumour subtype. As discussed in the present revised version, these data provide support to the induction of the epithelial phenotype by LOXL2 knockdown in basal-like cells independent of E-cadherin modulation. This finding can be considered at odds with the established opinion in the field that adherens junctions mediated by E-cadherin are required for the proper organization of tight junctions and cell polarity complexes (see for instance, Perez-Moreno et al, Cell, 112: 535-48, 2003). However, this finding also raises the interesting notion that the action of E-cadherin can be overcome in specific cell contexts, like the basal-like carcinoma cells when deprived of LOXL2 as described in the present study. Indeed, some situations in which other classical cadherins, like N-cadherin, can replace E-cadherin as in embryonic intestinal epithelium organization have been recently reported (Libusova et al., Development, 137: 2297-305, 2010). This important issue has been further discussed in the present revised version (see page 25, lines 2-4). 2.
”Included in the article is a collection of cell line studies that essentially reproduce what has previously been published by others attesting to the effect of LOXL2 on cell migration and invasion in culture and tumorigenicity and metastasis in vivo”.
© European Molecular Biology Organization
23
EMBO Molecular Medicine
Peer Review Process File - EMM-2010-00562
As commented above, previous information on migration and invasion, and to much lower extent on tumorigenesis and metastasis, existed on some breast cancer cell lines. However, we would like to stress for the reviewer’s consideration that the collection of cell lines we have analysed were selected because of their phenotype (luminal, ErbB2/neu+ and basal-like) and this analysis showed that LOXL2 expression is restricted to the basal-like carcinoma cell lines, a fact that is completely novel and, to our knowledge, not reported in any previous study. Consequently, we extended the analysis of LOXL2 silencing to three independent basal breast cell lines. 3.
“The authors virtually ignore data showing that loss of LOXL2 has no effect on cell growth in culture but a profound effect in vivo - data which strongly underscore the possibility that LOXL2 elicits its effects on tumor behavior in part through effects on the tissue microenvironment; this observation seems to have escaped the authors”.
The potential effect of LOXL2 on the microenvironment was considered in a long paragraph of the discussion section of our original ms (page 17, 2nd paragraph ending in page 18, 3 first lines), where it was specifically stated: “The specific role of LOXL2 in lung metastasis of basal-like breast carcinomas remains to be established. In this regard, previous studies have indicated an important role for LOX in hypoxiainduced metastasis (Erler et al, 2006) and the specific participation of LOX in the establishment of the pre-metastatic niche through extracellular matrix remodeling (Erler et al, 2009). A role for extracellular LOXL2 in the development of pathological microenvironment and the effectiveness of an anti-LOXL2 antibody in inhibition of tumor burden of carcinoma MDA-MB-435 cells as well as in fibrosis has been recently reported (Barry-Hamilton et al, 2010). Other recent studies have also reported the participation of secreted LOXL2 in gastric carcinoma metastasis (Peng et al, 2009). Our results point to an intracellular function for LOXL2 in basal-like breast carcinoma metastasis in agreement with our previous studies in squamous carcinoma tumors (Peinado et al, 2008) and in mammary gland tissue (Hollosi et al, 2009). It is worthy to emphasize that previous studies on LOXL2 on breast tumors have been restricted to a few samples of breast adenocarcinomas of nonspecific histological subtypes and lacking information on other clinico-pathological features (Barry-Hamilton et al, 2010; Hollosi et al, 2009). Our results provide data not only for LOXL2 expression in a specific breast tumor subtype, the basal-like carcinomas, but also for its clinical association with in vivo distant metastasis. The distinct staining pattern of LOXL2 observed in different studies might be related to the use of different antibodies recognizing intracellular and secreted LOXL2 proteins with distinct domains (Hollosi et al, 2009), use of different immunostaining protocols or even might suggest a dual role and/or substrates for LOXL2 in regulating intracellular and extracellular components that may prevail in distinct pathological contexts. Although the participation of extracellular LOXL2 functions in basal-like breast carcinomas cannot be formally excluded, our results provide a new dimension to the role of LOXL2 in tumor progression by influencing intracellular functions. We have further discussed this important issue in the present revised version, together with the new results obtained in the IHQ of tumours with an additional LOXL2 antibody tested recently in human samples (Barry-Hamilton et al, 2010) (new Fig. S9) (see page 24, last 9 lines, and page 24, lines 1-8). 4.
“As such the only novelty of this report is the authors contention that they have discovered a novel marker and regulator of basal-like breast tumors. This finding while potentially interesting and important seems better suited for a clinically specialized journal such as clinical cancer research. Moreover, the data presented on this topic seem premature. Thus, while the authors make a reasonable case that there is an association between LOXL2 and human breast cancer through RT-PCR and IHC analysis of human biopsies they group together all basal cancers - and fail to address the incredible complexity of this subtype”.
© European Molecular Biology Organization
24
EMBO Molecular Medicine
Peer Review Process File - EMM-2010-00562
We agree with the reviewer on the complexity of basal-like cancers and the recent identification of several molecular subtypes among this breast cancer group, including claudin-low and metaplastic tumors. We also agree with other specialists in the breast cancer field that recently reviewed this issue either in the context of triple negative breast tumors or in relation with stemness and EMT properties (see for instance, Foulkes et al., New Engl J Med, 2010; Blick et al., J Mammary Gland Biol Neoplasia, 2010) and which propose that there is still insufficient information as to clearly distinguish whether the proposed different basal subtypes have clinical significance. Nevertheless, we have further addressed this point by broadening the discussion on this important issue in the revised ms., in particular in relation with triple negative and claudin-low tumours (see page 21). 5.
They make no attempt to link LOXL2 expression with any EMT marker through costaining studies or other histophenotype markers of breast tumor aggression.
We appreciate this comment and we have performed additional IHQ staining analyses of the histophenotype and aggressiveness of our breast tumors. Apart from analyses of ER, PR, ErbB2, proliferation markers (Ki67) and p53 performed in the original study, IHQ analyses of established basal/triple negative markers (CK5, EGFR) as well some EMT markers (E-cadherin, cadherin 11, SPARC), that we previously characterized as expressed in basal like breast tumours (Sarrio et al, Cancer Res, 2008), have been performed on serial sections in the extended tumour sample. Results indicate that LOXL2 is associated to CK5+ and EGFR+ tumours in about 83% and 64% cases, respectively (new Suppl Fig S10 and Table S3), indicating that LOXL2 may define a subgroup of CK5+/EGFR+ basallike tumours. These new data have been commented in the main text (see page 18, end of second paragraph) 6.
“They do discuss their observation that LOXL2 immunostaining appears membrane bound and perinuclear in the more aggressive breast tumor specimens - but fail to address the mechanistic potential of this observation. Indeed, the article lacks mechanism for the purported LOXL2 dependent EMT phenotype in that the authors do not explore whether membrane or pernuclear LOXL2 are important and if so why. In this instance prior studies suggest membrane associated LOX can modify cell phenotype through effects on integrin adhesions while others have shown that nuclear associated LOXL2 can influence a variety of cellular behaviors through independent means”.
Regarding this comment, we would like to firstly stress that the specific staining detected with our home-made antisera associated to basal-like tumours is mainly cytoplasmic and perinuclear, but no membrane-bound or extracellular. Secondly, while nuclear LOX (the prototypical isoform) has been observed in distinct situations and indeed involved in different cell behaviours, as well as LOXmembrane associated or secreted forms no clear information on nuclear LOXL2 localization is available. This information is discussed in previous reviews referenced in our original ms. and present revised version (see Lucero and Kagan, 2006; Mäki, 2009; Payne et al., 2007). We have also discussed in the present revised version the recent work by Zhao et al. (JBC, 2009) showing that secreted LOXpropeptide negatively modulates fibronectin binding and focal adhesion (FAK) signalling as well as the previous work by Peng et al (2009) reporting that secreted LOXL2 activates phospho-Src and phosphoFAK pathways in gastric cancer cells. Apart from confirming the LOXL2 dependent regulation of the
© European Molecular Biology Organization
25
EMBO Molecular Medicine
Peer Review Process File - EMM-2010-00562
FAK and Src pathways, we provide here additional evidence for the remarkable change in focal contacts organization associated to LOXL2 expression in basal-like breast cancer cells (new Figs. 3 and S5). Moreover, we provide additional evidence in the present revised ms. for a new mechanistic potential of intracellular LOXL2 in the negative regulation of tight junction and cell polarity components. Indeed, we show that LOXL2 represses the activity of claudin-1 and Lgl2 promoters in transient transfection assays, strongly supporting an intracellular action of LOXL2 in transcriptional regulation. Furthermore, the LOXL2 mediated repression of claudin-1 and Lgl2 promoters is fully independent of Snail1 and LOXL2 catalytic domain (new Fig. 5). The repressor action of LOXL2 on claudin1 and Lgl2 is further supported by data from the promoter assays performed on shLOXL2 cells, where upregulation of both promoters is also detected. These data provide novel mechanistic insights for the intracellular action of LOXL2 in basal-like carcinoma cells and identify novel intracellular LOXL2 targets, tight junction and cell polarity components, regulated at transcriptional level and independent of LOXL2 catalytic activity. We firmly believe these novel finding will contribute to a better understanding of the role and molecular action of LOXL2 in breast tumours, in particular in the highly aggressive basal-like subtype. We hope that the reviewer will find appropriate and satisfactory our responses and that they fully answer his/her concerns.
3rd Editorial Decision
25 May 2011
Thank you for the submission of your revised manuscript "Lysyl oxidase-like 2 (LOXL2) a new regulator of cell polarity required for metastatic dissemination of basal-like breast carcinomas" to EMBO Molecular Medicine. We have now received the enclosed reports from the referees and the advisor (Referee #4) that were asked to re-assess it. As you will see the reviewers are now globally supportive and I am pleased to inform you that we will be able to accept your manuscript pending the following final amendments: We would strongly encourage you to shorten the text where possible. As also highlighted by reviewers #1 and #4, this would improve readability for the non-specialist. In addition, please provide the GEO accession number for the microarray analysis (Supporting information p 4). We also note that the Response to Reviewers contains figures, which would be published in the Review Process File (for more information on our transparent editorial process, please see below). Please indicate whether you agree with the publication of a Review Process File, would like to exclude the figures or would like to opt out. Please submit your revised manuscript within two weeks. I look forward to seeing a revised form of your manuscript as soon as possible. Yours sincerely, Editor EMBO Molecular Medicine
© European Molecular Biology Organization
26
EMBO Molecular Medicine
Peer Review Process File - EMM-2010-00562
***** Reviewer's comments ***** Referee #1: The authors have spent remarkable efforts to address the various reviewers' comments. In my opinion they have appropriately addressed all issues concerning experimental details, controls and additional mechanistic experimentation. Moreover, they have thoughtfully responded to the comments that concerned rather subjective opinions and speculations. In my opinion, the manuscript now presents major new insights into the role of LOXL2 in breast cancer progression. Despite the fact that certain details have been published before or have been reported with conflicting results, this manuscript sets a standard and provides novel, validated views on the scientific topic. Additional experimental systems and further functional analysis has been included. Notably, a thorough examination of a large number of antibodies has been performed, and even antibodies used in previous studies have been now included in the study. Overall, the report here is not only providing convincing new insights but also raising concerns on the significance of some of the recently published reports. As a result of the substantial revisions and additions, the manuscript is now rather lengthy and in parts too review-like (clearly not the fault of the authors but rather caused by the comments of the reviewers). Referee #2: The authors successfully addressed my concerns. Particularly they used an additional antiserum for IHC and increased the numbers of investigated human breast cancers, thereby confirming their previous data. Referee #4 (Comments on Novelty/Model System): As mentioned in my original comments & as they state in the paper - the results are not completely novel, which is why I ticked medium. The medical impact might be at the diagnostic level, but the group of basal or triple-negative cancers is still too diverse to be certain that LOXL2 staining will be useful - more needs to be done by the breast cancer community. Referee #4 (Other Remarks): The authors have attempted to answer all the comments of the 3 reviewers. If these reviewers are satisfied, I would agree since they also answered the comments that I made. The text should be shortened-in particular there is a lot of repitition in the introduction and the discussion. This would also make the paper easier to read for a nonspecialist.
2nd Revision - authors' response
06 June 2011
Please, find enclosed the new revised version of our ms. In the present version we have considered the recommendations of the reviewers as well as your suggested final amendments, as follows: 1. The GEO accession number (GEO29581) has been included in the corresponding section of Supplementary information (page 4). Please, notice that we include a complete Supplementary file in which some minor corrections have also been included. 2. As recommended by reviewers 1 and 4, the ms has been shortened, particularly in the Introduction and Discussion sections eliminating repetitions. Also, some paragraphs included in the Results section have been eliminated (regarding differences in tumorigenic
© European Molecular Biology Organization
27
EMBO Molecular Medicine
Peer Review Process File - EMM-2010-00562
behavior between our study and that of Barker et al, 2011) and moved to the Discussion section. 3. Regarding the Review Process File, we agree that all information, including Figures and Table, provided in the point-by-point response to the reviewers provided with the former revised version will be included and published on-line. We hope that this revised version fully satisfy all required improvements and the ms will be now formally accepted for publication. Thanking you once more for your consideration and support along the review process.
4th Editorial Decision
25 May 2011
We are pleased to inform you that your manuscript is accepted for publication and will be sent to our publisher to be included in the next available issue of EMBO Molecular Medicine if or once we have received your licenses (see below). Please see below for additional IMPORTANT information and instructions regarding your article, its publication, and the production process. Congratulations on your interesting work. Yours sincerely, Editor EMBO Molecular Medicine
© European Molecular Biology Organization
28