LNMAT1 promotes lymphatic metastasis of bladder cancer via CCL2 dependent macrophage recruitment

Tumor-associated macrophages (TAMs) are the most abundant inflammatory infiltrates in the tumor microenvironment and contribute to lymph node (LN) metastasis. However, the precise mechanisms of TAMs-induced LN metastasis remain largely unknown. Herein, we identify a long noncoding RNA, termed Lymph Node Metastasis Associated Transcript 1 (LNMAT1), which is upregulated in LN-positive bladder cancer and associated with LN metastasis and prognosis. Through gain and loss of function approaches, we find that LNMAT1 promotes bladder cancer-associated lymphangiogenesis and lymphatic metastasis. Mechanistically, LNMAT1 epigenetically activates CCL2 expression by recruiting hnRNPL to CCL2 promoter, which leads to increased H3K4 tri-methylation that ensures hnRNPL binding and enhances transcription. Furthermore, LNMAT1-induced upregulation of CCL2 recruits macrophages into the tumor, which promotes lymphatic metastasis via VEGF-C excretion. These findings provide a plausible mechanism for LNMAT1-modulated tumor microenvironment in lymphatic metastasis and suggest that LNMAT1 may represent a potential therapeutic target for clinical intervention in LN-metastatic bladder cancer.

This is an interesting study that addresses the role of long non coding RNAs in bladder cancer metastasis. The importance of the study is based on the relatively little that is known about the mechanisms underlying metastasis and in particular the role of RNAs. The current study identifies a long non-coding RNA that is associated with metastasis. The study is very interesting but some of the data are not clearly presented or entirely convincing.
First of all --why choose this long non coding RNA from among those identified? This should be clarified.
Should explain why the popliteal metastasis assay is relevant for these studies.
The connection to lymphagenesis is very difficult to evaluate.
Need to demonstrate that the LNMAT1 is specific expressed. There is limited data confirming the expression.     In the manuscript "Long noncoding RNA LNMAT1 promotes lymphatic metastasis of bladder cancer via CCL2-dependent macrophage recruitment" the authors found and characterized a novel long noncoding RNA (LMAT1) implicated in lymphatic dissemination of bladder cancer cells. Additionally, the authors propose a mechanism where LNMAT1 upregulates CCL2 in cancer cells leading to a higher recruitment of macrophages, which promotes lymphangiogenesis and metastasis.
The results are interesting and novel, the mechanism of CCL2 activation by LNMAT1 is wellcharacterized and the correlation with human data is remarkable. However, the manuscript seems incomplete, with a number of significant issues that need to be addressed, as described below.
Major comments: 1. Materials and methods information about certain experiments (e.g. wound healing) is missing. Moreover, "Methods" and "Supplementary Materials and Methods" seem to be exactly the same, whereas those 2 parts should complement each other.
2. The reviewer is not convinced about the cell isolation and differentiation protocols of human monocytes in macrophages. How is it possible to differentiate monocytes into macrophages without adding any differentiation factors (usually GM-CSF or M-CSF)? Furthermore, 24 hours of treatment with LPS or IL-4 are usually considered sufficient for macrophage polarization, so why did the authors decide to polarize them for 7 days? The authors also should mention the paper in which the protocol that they followed is reported. What kind of macrophages were used in experiments shown in Fig. 7 and Fig. 8? Why do the authors claim they are TAMs if they were not isolated from tumors? 3. Could the authors asses the influence of corresponding treatments (Fig. 7 C) on the expression of at least one more M2 marker and two chosen M1 markers (FACS)? For the assessment of capillary formation, could the authors use CM from TAMs sorted from murine tumors (overexpressing and silenced for LNMAT1)? 4. Except for microlymphatic vessel density, could the authors also assess the presence of big LVs in murine tumors sh-NC vs sh-LNMAT1 (Fig. 2H,I), as usually such bigger vessels participate in lymphatic metastasis? Moreover, the authors quantified signal of "LYVE-1 positive cells", and attributed it to lymphatic vessels. How do the authors know that they are not macrophages? Could the authors use double fluorescence staining (F4/80 and LYVE-1 or F4/80 and VEGFR3) in order to quantify LVs exclusively? 5. The authors demonstrate the presence of increased metastasis upon LNMAT1 overexpression in both tail vein and tumor inoculation experiments. Thus, does it observation relay more on increased tumoral lymphangiogenesis or increased extravasation of cancer cells? Could the authors assess the metastasis levels from primary tumors upon the blockage of lymphangiogenesis (e.g. with anti-VEGF-C)? Moreover, could the authors asses the migration of cancer cells (with and without LNMAT1 overexpression) on the invasion through lymphatic endothelium monolayer in vitro?
Minor comments: 1. The manuscript has a many grammar mistakes that need to be corrected (e.g. "lncRNAs regulate chemokine activation by interact with chromatin" should be "lncRNAs regulate chemokine activation by interaction with chromatin"). Please revisit the whole manuscript.
1. Could the authors indicate what was the LN status in high-grade muscle invasive bladder cancers (MIBC) used for NGS analysis (Fig. 1A).
2. The authors demonstrate the effects of LNMAT1 on LN metastasis ( Figure 4I). What are the effects of the corresponding treatments on distant organ metastasis, as this is a more clinically relevant question?
3. Could the authors asses the effect of CCL2 blockage in vitro (a migration or a wound healing assay with cancer cells)? 4. In Fig. 6G,H the cells lines were silenced for hnRNPLs, however in the text referring to this figure the authors state that the cells were transduced to express hnRNPLs. Which version is correct? Reviewer #3 (Expertise: lncRNA, epigenetics, Remarks to the Author): The manuscript by Chen et al. focused on the role of a long non-coding RNA LNMAT1 in lymphnode metastasis from bladder cancer. The authors showed evidence that LMNAT1 is associated with advanced grade of bladder cancer, poor prognosis and lymphatic infiltration. The authors further used a somewhat artificial tumor model, in which human bladder cancer cells were injected into the footpads of nude mice, to examine the function on lymphatic infiltration. They concluded that LMNAT1 enhanced lymphatic infiltration through a CCL2 dependent mechanism. Mechanistically, the authors demonstrate that LMNAT1 enhances CCL2 expression from cancer cells, which is regulated possibly through direct binding of LMNAT1 to CCL2 promoter, recruiting hnRNPL, leading to H3K4me3 mark deposition on CCL2 promoter. In the authors' model, the increase of CCL2 in cancer cells leads to increased macrophage recruitment, and possibly M2-like polarization, resulting in increased VEGF-c expression to enhanced lymphangiogenesis.
This study is an overall interesting story. The major novel contribution is the role of LNMAT1 in lymphatic infiltration by bladder cancer cells, and the mechanism through which LNMAT1 regulates CCL2. There are a lot of data ranging from human clinical specimens to mouse models to molecular mechanisms. The mechanistic studies, particularly the regulation of CCL2 expression, were well done, and convincing to me. The correlation between CCL2 and LNMAT1 in human cancer data is also supporting the mechanism and very nice. However, a few major concerns on the in vivo experiments exist that need to be properly addressed.
1. In the in vivo model, the authors showed that modulating LNMAT1 leads to changes in lymph node size and lymphatic infiltration. However, they did not show any data on the primary tumor size. An equally possible alternative explanation of their data is that LNMAT1 regulates primary tumor size, and larger primary tumor leads to increased lymphatic infiltration. Judging from the luciferase images (e.g. fig 4i), the primary tumor size seems to correlate with LNMAT1 modulation. 2. The critical experiment in which anti-CCL2 antibody treatment reduces lymphatic infiltration lacks key controls. This experiment is key to pinpoint a functional role of tumor-cell-derived CCL2 in the lymphatic infiltration process. Currently, anti-CCL2 is only used in mice with cells overexpressing LNMAT1. But it is equally possible that anti-CCL2 is inhibiting CCL2 from other non-cancer cell types in vivo-if true, one would predict that anti-CCL2 would be effective in control bladder cancer cells as well. 3. Follow up with the point 2 above, another experiment that can help pinpoint a role of tumor-cell derived CCL2 is to perform CCL2 knockdown in cancer cells, rather than using antibody-antibody cannot discriminate the source of CCL2. 4. The in vivo tumor model has limitations-it is human bladder cells injected in a nonphysiological site (foot pad), with the lack of effective adaptive immunity (using nude mice). I think the authors should at least discuss these limitations.
Minor: 1. The authors claim the use of "in vitro synthesized hnRNPL", but I couldn't find this info in the methods. Is this truly "synthesized", or it is purified recombinant protein expressed in bacteria or another host? 2. The histology pictures (e.g. Fig 2h, 2i, 7a, 7b) are too small with low resolution of details. Not very easy to see even after expanding on computer screen. Maybe including the large versions in supplement?

Reviewer #1:
This is an interesting study that addresses the role of long non-coding RNAs in bladder cancer metastasis. The importance of the study is based on the relatively little that is known about the mechanisms underlying metastasis and in particular the role of RNAs. The current study identifies a long non-coding RNA that is associated with metastasis. The study is very interesting but some of the data are not clearly presented or entirely convincing.

2:
Should explain why the popliteal metastasis assay is relevant for these studies. The connection to lymphagenesis is very difficult to evaluate.

Response:
The reviewer's point is well taken. As suggested by the reviewer, the relevance of popliteal metastasis model has been discussed in the revised manuscript as the following: In the clinic, the most common sites of lymph node metastases of bladder cancer are the pelvic lymph nodes, including the obturator nodes, the external iliac nodes and the common iliac lymph nodes. The lymphatic metastasis of bladder cancer is directional and the common iliac lymph nodes are considered as the next station to the obturator and external iliac nodes. In the current study, popliteal lymph node metastasis model was established in BALB/c-nude mice by injecting the footpads with bladder cancer cells. Lymph drainage in the footpad is unidirectional with the popliteal, which leads to lymph draining directly up to the popliteal lymph node and subsequently to the external iliac nodes and the common iliac lymph nodes, which are the common sites of lymph node metastases of bladder cancer. Therefore, the popliteal lymph node metastasis model may simulate the functional characteristics of lymphatic metastasis of bladder cancer typically seen in the clinic. Furthermore, the well-defined lymph drainage from footpad injections has enabled more sensitive and quantitative in vitro measurements of lymphatic metastasis. Thus, the popliteal lymph node metastasis model of injecting cells into mouse footpads has been widely used in many previously published studies and has proved reliable and suitable in many cases 1,2,3 . The  3: Need to demonstrate that the LNMAT1 is specific expressed. There is limited data confirming the expression.

Response:
We thank the reviewer for this suggestion and the reviewer's point is well taken.
The qRT-PCR and in situ hybridization (ISH) analysis showed that LNMAT1 expression was marginally detected in normal bladder tissues and slightly increased in non-LN-metastatic bladder cancer tissues but strongly upregulated in LN-metastatic bladder cancer, suggesting that LNMAT1 was more specifically overexpressed in LN-metastatic bladder cancer tissues ( Fig. 1j). Moreover, analysis of tumor profiles in TCGA database showed that LNMAT1 is significantly overexpressed in various types of human cancers, such as bladder, thyroid, prostate, kidney, lung and liver cancers ( Supplementary Fig. 2a-f). Consistently, analysis of tumor profiles in TCGA databases indicated that LNMAT1 is overexpressed in LN-metastatic human cancers, such as thyroid carcinoma and kidney cancer ( Supplementary Fig. 2g-h).
Importantly, LNMAT1 overexpression correlated with poor prognosis in many human cancers, such as thyroid carcinoma, kidney cancer, colon carcinoma and liver cancer ( Supplementary   Fig. 2i-p). These results suggest that LNMAT1 is specifically expressed in LN-metastatic human cancer tissues and may play an oncogenic role in the progression and development of various human cancer types. Figure 1: (1) Specify the datasets being used.

4:
(2) Panel J -the staining is not very evident.
(3) Panel C is too small to read.

Response:
We thank the reviewer for this suggestion and the reviewer's point are well taken: (1) We are sorry that we did not write this point clearly in our originally submitted manuscript and appropriate modifications have been made in the revised manuscript. All the bladder cancer tissues used in Fig. 1  (2) As suggested by the reviewer, the images in the Panel J of Fig. 1 have been replaced by the higher resolution image in the revised manuscript.
(3) As suggested by the reviewer, the images in the Panel C of Fig. 1 have been replaced by a larger image in the revised manuscript. Figure 2: (1) What is the advantage of using the popliteal metastasis assay. (2) Panel D is illegible. Whatever data is intended to be conveyed is not evident.

5:
(3) Panel F should have the actual p-values.
Response: (1) We appreciate the reviewer's important point. In the clinic, the most common sites of lymph node metastases of bladder cancer are the pelvic lymph nodes, including the obturator nodes, the external iliac nodes and the common iliac lymph nodes. The lymphatic metastasis of bladder cancer is directional, and the common iliac lymph nodes are considered as the next station of the obturator and external iliac nodes. In our study, since lymph drainage in footpad is unidirectional with popliteal, which leads to lymph draining directly up to the popliteal lymph node and subsequently to the external iliac nodes and the common iliac lymph nodes, which are the common sites of lymph node metastases of bladder cancer.
Thus, popliteal lymph node metastasis model may simulate the functional characteristics of lymphatic metastasis of bladder cancer typically seen in the clinic. Moreover, the well-defined lymph drainage from footpad injections has enabled more sensitive and quantitative in vitro measurements of lymphatic metastasis. Therefore, the popliteal lymph node metastasis model of injecting cells into mice footpads have been employed widely by numerous previously published studies, which has proved reliable and suitable 1,2,3 . The abovementioned descriptions have been added to the revised manuscript.
(2) The reviewer's point is well taken. As required by the reviewer, the images in the Panel D of Fig.2 have been replaced by a higher resolution image in the revised manuscript. (2) We thank for the reviewer's comment. As requested by the reviewer, we further confirmed the metastatic bladder cancer cells in the lung by IHC using an anti-luciferase antibody ( Fig. 3f-g). Consistently, the luciferase signals in the lungs of mice injected with LNMAT1 cells were much higher than those in mice injected with control cells, suggesting that LNMAT1 overexpression strongly increased lung colonization by bladder cancer cells.
The abovementioned results have been incorporated into the revised manuscript. (2)The reviewer's point is well taken. As requested by the reviewer, the Panel I has been enlarged and appropriate inset data have been incorporated into Panel I of Fig. 4 in the revised manuscript.
8: Figure 5: Data are too difficult to evaluate -not clearly legible. Data for supporting a triplex are not sufficiently convincing.
Response: (1) We thank for the reviewer's comments and we apologize for the unclear images in our originally submitted manuscript. Appropriate modifications have been made and the higher resolution images have been added to the revised manuscript (Fig. 5).
(2) We thank for the reviewer's comment. To examine whether LNMAT1 formed a DNA-RNA triplex with the CCL2 promoter, luciferase activity and ChIRP analysis were further performed. The promoter luciferase assay showed an obvious increase in the transcriptional activity of the construct contained region -200 to+121 bp than in the +1 to +121 bp region ( Fig. 5c and Supplementary Fig. 8c). Moreover, a chromatin isolation by RNA purification (ChIRP) assay, which determines the exact locations of lncRNA binding sites on the chromatin 1 , , revealed that LNMAT1 bound to -10 to -118 bp (referred to as CCL2-S2; Fig. 5d-f). Furthermore, analysis with LongTarget, a web-based tool designed to predict lncRNA-DNA binding motifs and binding sites via Hoogsteen and reverse Hoogsteen interactions 2 , identified six potential triplex-forming oligonucleotides (TFOs) within LNMAT1 and the corresponding triplex target sites (TTS) in the CCL2 promoter for possible pairing (Supplementary Table 5). Next, we synthesized the predicted TFO labeled with 5-carboxy tetramethyl-rhodamine (TAMRA) and TTS labeled with fluorescein amidite (FAM) and performed fluorescence resonance energy transfer (FRET) analysis. FRET showed an obvious increase in the TAMRA fluorescence intensity at 570-580 nm and a decrease in 6-carboxyfluorescein (6-FAM) at 520 nm in the LNMAT1 (462 to 479 nt)/CCL2 TTS4 (-60 to -43 bp) group compared with that of the control ssRNA/ CCL2 TTS group (Fig. 5g-h), which was similar to the FENDRR/PITX2-positive control group (Fig. 5i). These data indicate that

Reviewer #2:
In the manuscript "Long noncoding RNA LNMAT1 promotes lymphatic metastasis of bladder cancer via CCL2-dependent macrophage recruitment" the authors found and characterized a novel long noncoding RNA (LMAT1) implicated in lymphatic dissemination of bladder cancer cells. Additionally, the authors propose a mechanism where LNMAT1 upregulates CCL2 in cancer cells leading to a higher recruitment of macrophages, which promotes lymphangiogenesis and metastasis. The results are interesting and novel, the mechanism of CCL2 activation by LNMAT1 is well-characterized and the correlation with human data is remarkable. However, the manuscript seems incomplete, with a number of significant issues that need to be addressed, as described below. Furthermore, 24 hours of treatment with LPS or IL-4 are usually considered sufficient for macrophage polarization, so why did the authors decide to polarize them for 7 days? The authors also should mention the paper in which the protocol that they followed is reported.
What kind of macrophages were used in experiments shown in Fig. 7 and Fig. 8? Why do the authors claim they are TAMs if they were not isolated from tumors?

Response:
We appreciate the reviewer for raising these excellent concerns, and the reviewer's comments are well taken.
(1) We apologize for the improperly provided protocols for cell isolation and differentiation and thank the reviewer for pointing this out. The original protocols for cell isolation and differentiation of human monocytes in macrophages used in our study were according to previously published reports 1,2,3,4 . We are sorry for the inappropriate modifications that were made in the last version. The 7 days in the protocols is the total period of time needed for macrophage differentiation and polarization (6 days for differentiation plus 1 day for polarization). In brief, human monocytes were isolated from buffy coats prepared from healthy volunteer donors. Peripheral blood monocytes isolated by Ficoll-Hypaque for 50 min at 400g density gradient centrifugation were seeded at 2×10 6 /ml in 24-well plates in RPMI (2)We thank the reviewer for this comment and we are sorry that we did not clearly explain this point in our original manuscript. In Fig. 7 and Fig. 8, the macrophages treated with CM from LNMAT1-transduced bladder cancer cells but not those treated with CM from control group displayed the stretched and elongated morphology and exhibited a CD206 high /HLA-DR low phenotype ( Fig. 7e and Supplementary Fig. 11a)

3:
Could the authors assess the influence of corresponding treatments (Fig. 7C)  Response: We appreciate the reviewer for making these excellent suggestions and the reviewer's points are well taken. As requested by the reviewer, we assessed the expression of specific M2 marker (CD206 and CD163) and M1 markers (HLA-DR and CD86) using FACs.
As shown in Fig. 7e and Supplementary Fig. 11b-c, treatment of CM derived from LNMAT1/cells significant increased the expression of the specific M2 marker (CD206 and CD163) but decreased the expression of M1 markers (HLA-DR and CD86), compared to vector-control cells, which provided more evidence that LNMAT1 overexpression in cancer cells induces macrophage activation.
As requested by the reviewer, we further examined the effect of TAMs sorted from murine tumors on capillary formation. As shown in Supplementary Fig. 12a-b, treatment with CM derived from TAMs sorted from murine LNMAT1/tumors significantly increased lymphatic capillary formation compared with CM derived from TAMs sorted from murine vector/tumors, which further supports the notion that LNMAT1 overexpression in cancer cells promotes lymphangiogenesis. The abovementioned results have been incorporated into the revised manuscript.

4:
Except for microlymphatic vessel density, could the authors also assess the presence of big LVs in murine tumors sh-NC vs sh-LNMAT1 (Fig. 2H,I), as usually such bigger vessels participate in lymphatic metastasis? Moreover, the authors quantified signal of "LYVE-1 positive cells", and attributed it to lymphatic vessels. How do the authors know that they are not macrophages? Could the authors use double fluorescence staining (F4/80 and LYVE-1 or F4/80 and VEGFR3) in order to quantify LVs exclusively?

Response:
We thank the reviewer for raising these excellent suggestions and the reviewer's comments are greatly appreciated. Through the reassessment of IHC staining in the vector/tumors and LNMAT1/tumors, we did observe significant decreased big LVs in murine sh-LNMAT1/tumors compared with sh-NC/tumors in both the intratumoral and peritumoral regions (p < 0.05; p < 0.05) ( Supplementary Fig. 5a-b). Moreover, as suggested by the reviewer, the previous improperly prepared images presented with small in LVs have been replaced by images presented with big LVs in the revised manuscript.
As requested by the reviewer, we further examined macrophages with an anti-F4/80 antibody and lymphatic vessels with an anti-LYVE-1 antibody through double staining assays. As shown in Supplementary Fig. 5d, the F4/80 + -cells were distinct with LYVE-1 + -cells, which is consistent with previous reports, which LYVE-1 is considered as a common lymphatic marker that is extensively used for the detection of lymphatic vessels 1,2 .
The abovementioned results have been incorporated into the revised manuscript. Response: We appreciate the reviewer for raising these excellent suggestions and the reviewer's comments are greatly appreciated. As requested by the reviewer, the effect of VEGF-C inhibition by the anti-VEGF-C antibody on LNMAT1-induced LN metastasis was examined in vivo. As shown in Fig. 8f-g, blockade of the VEGF-C signaling using a VEGF-C neutralizing antibody (pV1006R-r) significantly decreased the intratumoral and peritumoral lymphatic vessel densities and reduced the lymphatic metastasis capability of LNMAT1-overexpressing cells, suggesting that blocking of VEGF-C signaling abrogated the

LNMAT1-induced lymphangiogenesis and lymphatic metastasis in vivo.
We further examined the effect of LNMAT1 overexpression on the invasion of cancer cells through a lymphatic endothelium monolayer in vitro. As shown in Supplementary Fig. 12c  Minor comments:

1:
The manuscript has a many grammar mistakes that need to be corrected (e.g. "lncRNAs regulate chemokine activation by interact with chromatin" should be "lncRNAs regulate chemokine activation by interaction with chromatin"). Please revisit the whole manuscript.

Response:
We thank the reviewer for these comments, which are well taken. We have carefully edited the entire manuscript and have had the revised manuscript corrected again by professional editors before this resubmission.

2:
Could the authors indicate what was the LN status in high-grade muscle invasive bladder cancers (MIBC) used for NGS analysis (Fig. 1A).

Response:
We thank the reviewer for this comment. Five high-grade muscle invasive bladder cancer tissues (MIBC) were used for NGS analysis in Fig. 1A Response: We thank the reviewer for this comment and the reviewer's point is well taken.
As suggested by the reviewer, we further examined the effect of CCL2 inhibition in vitro via a migration assay with cancer cells. Consistent with previous reports 1,2 , CCL2 inhibition significantly inhibited the migratory capability of bladder cancer cells ( Supplementary Fig.   6a-b). The abovementioned results have been incorporated into the revised manuscript.  Figure   1C). Beside the significant overexpression of LNMAT1 in figure 1E, could the authors also display the expression levels of the other 2 lncRNAs in the Supplementary figures?

Response:
We thank the reviewer for this comment and the reviewer's point is well taken.
As requested by the reviewer, the expression levels of BCAR4 and CTD-2231H16.1 have been incorporated into Supplementary Fig. 1b-c in the revised manuscript.

7:
The authors need to correct figure legend accuracy (e.g. no figure legend in Fig. 3G; Fig.   3D appears twice in the legend; what do the arrows in Fig. 4I indicate?).

Response:
We appreciate the reviewer's comments, and we thank the reviewer for pointing out the mistakes. All the figure legends have been carefully edited and appropriate corrections, including those for Fig. 3D and 3G, have been made in the revised manuscript.
We apologize that we did not explain what the arrows indicated in Fig.4I in our original manuscript. The arrows in Fig. 4I indicate the magnified inset area. To avoid confusion, we have changed the arrows to the red box, which is explained in the figure legend.
8: Some figures do not appear in the order introduced in the text.

Response:
We apologize for the mistakes and thank the reviewer for pointing these out. We have gone through the entire manuscript and have made appropriate corrections into the revised manuscript.

Reviewer #3:
The manuscript by Chen et al. focused on the role of a long non-coding RNA LNMAT1 in lymphnode metastasis from bladder cancer. This study is an overall interesting story. The major novel contribution is the role of LNMAT1 in lymphatic infiltration by bladder cancer cells, and the mechanism through which LNMAT1 regulates CCL2. There are a lot of data ranging from human clinical specimens to mouse models to molecular mechanisms. The mechanistic studies, particularly the regulation of CCL2 expression, were well done, and convincing to me. The correlation between CCL2 and LNMAT1 in human cancer data is also supporting the mechanism and very nice. However, a few major concerns on the in vivo experiments exist that need to be properly addressed.

1:
In the in vivo model, the authors showed that modulating LNMAT1 leads to changes in lymph node size and lymphatic infiltration. However, they did not show any data on the primary tumor size. An equally possible alternative explanation of their data is that LNMAT1 regulates primary tumor size, and larger primary tumor leads to increased lymphatic infiltration. Judging from the luciferase images (e. g. fig 4i), the primary tumor size seems to correlate with LNMAT1 modulation.

Response:
The reviewer's comment is well taken. We appreciate the reviewer for pointing out the improperly prepared Fig. 4i and we are sorry that we did not explain this clearly in have been added to the revised manuscript.

2:
The critical experiment in which anti-CCL2 antibody treatment reduces lymphatic infiltration lacks key controls. This experiment is key to pinpoint a functional role of tumor-cell-derived CCL2 in the lymphatic infiltration process. Currently, anti-CCL2 is only used in mice with cells overexpressing LNMAT1. But it is equally possible that anti-CCL2 is inhibiting CCL2 from other non-cancer cell types in vivo-if true, one would predict that anti-CCL2 would be effective in control bladder cancer cells as well.

Response:
We appreciate the reviewer's comments and the reviewer raises a very interesting and important question. After submission of our manuscript, we realized that our data were not sufficient to support our conclusion, which the tumor-cell-derived CCL2 was the key mediator for LNMAT1-induced lymphatic metastasis. Therefore, 2 sets of experiments were immediately started. First, we examined the inhibitory effect of CCL2 via anti-CCL2 antibody on the lymphatic metastasis of control bladder cancer cells, which was also suggested by the reviewer. The anti-CCL2 antibody only partially inhibited the LN-metastatic capability of the control bladder cancer cells (the ratio of metastatic LNs from 37.50% reduced to 25.00%) but strongly inhibited the LN-metastatic capability of Consistently, silencing CCL2 also only moderately reduced the LN-metastatic capability of vector-control bladder cancer cells (from 31.25% reduced to 18.75%) but remarkably inhibited the LN-metastatic capability of LNMAT1-overexpressing cells (from 81.25% to 12.50%; Fig. 4l-m and Supplementary Fig. 7d). Taken together, these results provided further evidence that tumor cell derived CCL2 is a key mediator for LNMAT1-induced lymphatic metastasis. These abovementioned results have been incorporated into the revised manuscript.

3:
The in vivo tumor model has limitations-it is human bladder cells injected in a non-physiological site (foot pad), with the lack of effective adaptive immunity (using nude mice). I think the authors should at least discuss these limitations.

Response:
We thank the reviewer for these comments and the reviewer's points are well taken. In this study, popliteal lymph node metastasis model was employed by injecting cells into mouse foot pads. Although it has been reported that footpad injection is a sensitive and quantitative way to measure lymphatic metastasis in vivo 1,2,3 , multiple limitations of this model need to be noticed. First, the microenvironment of the footpads is quite different from the bladder microenvironment, and thus the mechanism in which the formation of microlymphatic vessels within both the bladder cancer and bladder cancer-adjacent tissues in footpads is different than in the bladder. Moreover, the intratumoral interstitial fluid pressure (IFP) is commonly much higher than that in the surrounding host tissues, which influences the different lymphatic fluid reflux and filtration rates, causing some differences in the results of lymphatic metastasis 4 . Thus, it would be better to inoculate LNMAT1 +/-UMUC-3 cells or tumors formed by LNMAT1 +/-UMUC-3 cells directly into the bladder walls of mice to examine the effect of LNMAT1 induced lymph node metastasis. In addition, the BALB/c nude mouse model lacks effective adaptive immunity and is commonly used in studies to test the pharmacological treatment of human tumor xenografts 5 . Therefore, the effect of LNMAT1 induced-lymph node metastasis could also be analyzed in mouse cell lines using BALB/c mice with human immunity. The abovementioned descriptions have been added to the Discussion section in the revised manuscript.