Author's response to reviews Title:The orthotopic xenotransplant of human glioblastoma successfully recapitulates glioblastoma-microenvironment interactions in a non-immunosuppressed mouse model Authors: Celina Garcia Graduate student (celinagarcia@icb.ufrj.br) Luiz G Dubois Dr (dubois@icb.ufrj.br) Anna L Xavier Dr (alr.xavier@gmail.com) Luiz H Geraldo Undergraduate student (lh_geraldo@icb.ufrj.br) Anna CC Fonseca Dr (annafonseca@icb.ufrj.br) Ana H Correia Dr (anahcorreia@gmail.com) Fernanda GM Ferreira Dr (nandagmferreira@gmail.com) Grasiella Ventura Dr (grasiella@icb.ufrj.br) Luciana Romão Dr (romao@icb.ufrj.br) Nathalie HS Canedo Dr (nathaliecanedo@gmail.com) Jorge M de Souza Dr (jormarcondes@gmail.com) João RL Menezes Dr (jrlmenezes@gmail.com) Vivaldo Moura-Neto Dr (vivaldo@icb.ufrj.br) Fernanda T Moll Dr (tovarmollf@gmail.com) Flavia RS Lima Dr (flima@icb.ufrj.br) Version:4Date:25 October 2014 Author's response to reviews: see over
Universidade Federal do Rio de Janeiro Instituto de Ciências Biomédicas October 22 th, 2014 To: Dr. Amancio Carnero Associate Editor, BMC Cancer Ref: MS: 1719540798137228 Dear Editor Dr. Amancio Carnero, We revised the manuscript entitled The orthotopic xenotransplant mouse model of human glioblastoma successfully recapitulates glioblastoma microenvironment interactions to fully address the Reviewers comments. We appreciate the Referee s critical and constructive comments. We are sure that they strengthened the findings of our manuscript. We kindly request for your most careful consideration of this manuscript, since all points of criticisms were further elaborated. Sincerely, Dr. Flavia R. S. Lima Assistant Professor, ICB, UFRJ
Point-by-point responses to the reviewers are as follows: Reviewer's report 1: Title:The orthotopic xenotransplant mouse model of human glioblastoma successfully recapitulates glioblastoma-microenvironment interactions Version:1Date:30 August 2014 Reviewer:Rolf Mentlein Reviewer's report: Animal models of human glioblastomas often hamper from the lack of non-bordered invasion of tumor cells in the brain parenchyma as it is typically for the human brain tumors (and one of the reasons for the poor prognosis of the patients). The study of Garcia et al. aim to establish a model of invasion of human tumor cells into non-immunosuppressed mice. In fact, this is of considerable interest, as the tumor microenvironment is then more close to the human situation as in immunosuppressed nude or SCID mice. The study documents tumor growth and invasion by MRI and various (immuno)cytochemical methods. Experiments seem carefully performed, however, more clarity and experiments appear to me necessary: Major Compulsory Revisions (1) Title, clarity: As the use of normal mice is important, use non-immunosuppressed mice in title. We thank the Reviewer for the suggestion. The title was changed to: The orthotopic xenotransplant of human glioblastoma successfully recapitulates glioblastomamicroenvironment interactions in a non-immunosuppressed mouse model (lines 1-3). (2) Animals: Please provide more details about the strain. Inbred? Genotype? Identical to http://jaxmice.jax.org/strain/000686.html? Yes, we have used male Swiss mice (SWR/J), inbred strain (http://jaxmice.jax.org/strain/000689.html). These details were added in Methods section of the manuscript (page 6, lines 133,134).
(3) Human tumor cell line: Also here, I would like to have more information. Marker expression also in vitro (GFAP, Vimentin, Ki67-/proliferation-index; stem cell markers?) comparison with the human astrocytes used. Are both cells devoid of microglial cells? Yes, we have used several cell markers to characterize the GBM95 cell line in vitro. Immunocytochemistry for GFAP, vimetin and nestin were already shown in cultured GBM95 cells, see the Figure 2 from Faria et al., 2006 (DOI: 10.1111/j.1432-0436.2006.00090.x ). Studies about the biology of GBM95 cells in vitro are performed in our laboratory. See below a figure showing cultured GBM95 cells marked for GFAP, vimetin, nestin and SOX2. Figure. Characterization of GBM95 cell line by confocal microscopy. The cells were stained for neural and progenitors markers. A) GFAP (red) B) Vimentin (green) C) Nestin (yellow) and D) SOX2 (purple). The nuclei were labeled with DAPI (blue). Bar: 10 µm. We have also performed immunocytochemistry using a microglial marker (IB4) and did not find microglial cells in GBM95 cell cultures, see the figure below:
Figure: Culture of GBM95 cells is negative for IB4, but stains for Ki67. Cultured GBM95 cells were stained for IB4 (D, E, F) and Ki67 (G). Cultured microglial cells from cortex of newborn mice were used as positive control for IB4 (A, B, C). A, D = DAPI (blue); B, E = IB4 (red). C, F = merge DAPI/IB4. G= DAPI/Ki67. A-F Bar: 50 µm. G Bar: 10 µm. We cited these data in the new version of the manuscript (page 6, lines 142-144). The human astrocytes used in the works of Diniz et al. 2012 (DOI: 10.1074/jbc.M112.380824) and 2014 (DOI: 10.1002/glia.22713) and the human astrocytes used in our study had the same origin (more information in the Methods section, page 7, lines 146-168), and express GFAP, GLAST and HLA, attesting to their human and astrocytic nature (See in the Figure 1 from Diniz et al., 2012). We cited these data in the new version of the manuscript (page 7, lines 165-168). The human astrocytes used in our study were obtained as previously described by Diniz et al., 2012. Microglial cells are removed during steps of astrocyte culture (see details in the Methods section of our manuscript). In relation to cell proliferation, human astrocytes, as opposed to GBM cells (figure above, G), have limited proliferation on culture, they are inhibited by cell contact. (4) Antibodies: What about the cross-reaction between human and mouse proteins? Verified with mouse tissues? Yes, the rabbit anti-gfap antibody (Dako) that we have used recognizes both mouse and human cells (see Figure 3 C-F). On the other hand, the mouse anti-vimentin clone V9 antibody (Dako) recognizes only human cells. We note this difference in the figure 5: Vimentin was expressed only in the tumor mass produced from human GBM cells (Fig. 5D), whereas, in the contralateral hemisphere where there is not tumor, Vimentin was not expressed (Fig. 5A). Discretionary Revisions
(5) Invasion: Since glioma cells in human brain migrate as guerillia cells (distant from the tumor at vessels, at white matter tracts, around neurons and spread in the subpial region, did you analyze this, e.g with specific (human-directed) antibody? Yes, we did not detect human vimentin + cells outside of tumor region (Figure 5), at least using anti-vimentin antibody from Dako. As we described in the manuscript, the borders of tumor mass in the mouse brain were rather circumscribed and not infiltrative (Figure 2A). This difference may be due to a mismatch in cell surface recognition proteins, mainly MHC, between mice and humans. However, the fact that the tumor has well-defined border does not invalidate the diagnosis of a malignant neoplasia. Indeed, it resembles the so-called malignant glioneuronal tumor (MGNT) described by the team of Dr. Daumas-Duport as a more superficial and fairly well-defined tumor, although highly aggressive, causing recurrence and patient death (Varlet et al., 2004; PMID: 15574220). The MGNT is classified by the WHO as a GBM. This point was commented in the Discussion section of the manuscript (page 14, lines 317-329). Level of interest: An article of importance in its field Quality of written English: Acceptable Statistical review: No, the manuscript does not need to be seen by a statistician. Declaration of competing interests: I declare that I have no competing interests Reviewer's report 2: Title:The orthotopic xenotransplant mouse model of human glioblastoma successfully recapitulates glioblastoma-microenvironment interactions Version:1Date:23 September 2014 Reviewer:Uwe Himmelreich Reviewer's report: The authors have developed a human murine glioma model using immunocompetent mice and a human glioma cell line. A number of features typical to glioma have been identified in this model. The development of new experimental glioma models is highly relevant due to the limitations of currently existing murine models. Major comments:
(1) The main reason why immune-compromised mice are used in glioma models that use human glioma cells is the fact that the mouse immune system shows a reaction to human cell lines. The success of this GBM model using Swiss mice without immune suppression is somewhat puzzling. The authors should comment why no immune reaction was observed in their model. In fact, there is an immune reaction against tumor progression. The human GBM grown in the brain of Swiss mice, despite having all the features recommended by the WHO to be diagnosed as a GBM, it does not invade nearby tissue and grows in a certain circumscribed area, probably this difference may be due to a mismatch in extracellular recognition proteins, mainly MHC, between mice and humans. Despite it, this study can be a good model for investigating tumor progression. This comment is written with more details in the Discussion section of the manuscript (page 14, lines 317-329). (2) The authors have used human astrocytes as controls. However, no results on those experiments are shown. The authors should either show those results or omit those experiments. In my opinion, the engraftment of astrocytes can be of importance to show the presence/ absence of a reaction of the murine immune system towards human cells. Yes, we are in accordance with the Reviewer and we show these results in the manuscript (page 11, lines 354-356): injections of healthy human astrocytes did not induce cell mass development at 14 (Figure 2D) or 30 days (see Additional file 2 in the new version of the manuscript), suggesting the presence of immune reaction to eliminate the human astrocytes from mouse brain. For this experiment, we used the same cell number used in the injections of GBM cells. (3) The authors claim that their model represents all hallmarks of glioblastoma. However, glioblastoma also show a highly infiltrative growth pattern, which was not the case in their model (solid tumors). This is a limitation of most experimental glioma models and should also stressed more clearly in the manuscript. In particular for the assessment of novel therapeutic approaches, this is extremely important as it is one of the most frequent reasons why novel (and conventional) therapeutic approaches fail and result in tumor re-occurrence. In fact, this report was seen not only by us, but also by other groups that used in vivo glioma models (Pyonteck et al., 2013; Radaelli et al., 2009; Brehar et al., 2008). Particularly in our model, the human tumor in the brain of Swiss mouse does not invade nearby tissue, but grows in a certain circumscribed area, probably this occurs due to a mismatch in cell surface recognition proteins between mice and humans. However, the fact that the tumor has well-
defined border does not invalidate the diagnosis of a malignant neoplasia. Indeed, it resembles the so-called malignant glioneuronal tumor (MGNT) described by the team of Dr. Daumas- Duport as a more superficial and fairly well-defined tumor, although highly aggressive, causing recurrence and patient death (Varlet et al., 2004; PMID: 15574220). The MGNT is classified by the WHO as a GBM. Although nude animals are widely used, they represent a limited model to investigate the influence of immune cells during tumor progression. Tests with anti-tumor drugs using our model could have a better outcome than that obtained with nude animals. In this context, we have recently demonstrated that Equinatoxin II, a pore-forming toxin from sea anemones, potentiates the effects of Etoposide in the induction of GBM cell death (Kahn et al., 2012; DOI: 10.2174/1568026611212190006 Figure 6). These data were commented in the Discussion section, new references were also added (page 14, lines 317-329). (4) The group size (four animals per group) seems to be rather small. We did not think it was necessary to use a larger number of animals since the results were very similar, there were not expressive differences between the results of four animals per group. In addition, the Guide for the Care and Use of Laboratory Animals (published by the National Academy of Science, National Academy Press, Washington, D.C.) was strictly followed in all experiments. All efforts were made to minimize the number of animals used and their suffering. (5) It is not clear why proton-density images were used in this study. PD-MRI does not provide strong contrast (as seen in images of Fig. 1A-D). T2-weighted MRI (as also used by others working in the field and in the clinic) would have provided better contrast in particular for the assessment of tumor heterogeneity (necrosis/ hemorrhages). We thank the Reviewer for pointing this issue. The images were acquired on fast-spin-echo T2 weighted and also on fast-spin-echo proton density and also on spin-echo T1 weighted images before and after gadolinium injection. Tumor morphology and heterogeneity were inspected on all type of sequences. We have corrected the text accordingly (page 8, lines 182-202; page 11, lines 241-246; page 13, lines 310-311). (6) The authors should perform a quantitative analysis of the MR images instead of just showing the images for just one time point (14 days). This could be, for example, the analysis of
tumor volumes over time and the relative contrast enhancement after the administration of contrast (for example, analysis of relative signal intensities). This is a very relevant suggestion and we have now performed the tumor volume measures, as suggested. We have included the description of the methodology (page 8, lines 182-202), and provided a new figure (additional file 1) expressing tumor s growth. (7) Based on contrast enhancement in T1-weighted MRI, the authors conclude that hemorrhages and necrosis can be detected in the core of the tumor mass (page 10). First of all, contrast enhanced T1-weighted MRI can only visualize the breakdown of the BBB. All other conclusions based on image heterogeneity are indirect. Therefore, it would be important to confirm hemorrhages and/ or necrosis also by histology. Secondly, it looks like that this hypointense area was only present in one animal (Fig. 1J) and not in the others. Could the authors comment on that? We agree with the Referee. We have changed the sentences in order to be more careful and clear in the results description and interpretation (page 11, lines 241-246; page 13, lines 310-311). The tumor characteristics and heterogeneity, including areas suggestive of necrosis and hemorrhage, were evaluated on multiple sequences (T2, T1 and proton density weighted MRI) in sagittal, coronal and axial planes. The T1-weighted MRI was performed in order to investigate the breakdown of the BBB, as comment by the Reviewer. The Figure 1 shows data representative of four separate experiments. As also pointed by the Reviewer, importantly, later histological analysis confirmed the presence of necrosis within tumor mass (Figure 2A). We have added in the figure an asterisk indicating a necrosis area in the core of the tumor mass. (8) It would be important to discuss the findings of the authors in relation to other humanized brain tumor models in mice (for example the Hs683 or the hu373 model). We thank the suggestion of the Reviewer. We added text and references about these findings in the Discussion section (page 14, lines 330-337). (9) The last statement of the discussion is not clear ( Further studies should be done to strengthen this issue). To strengthen what issue?
This sentence refers to the report of the previous sentence and it is not essential, therefore we removed it from the text. Minor comment: (1) The authors state that the isolation of apparently healthy astrocytes from patients was in agreement with the Brazilian Ministry of Health Ethics Committee. However, were those procedures also reviewed by an Ethics Committee? Yes, the cortico-subcortical tissue, obtained from standard temporal lobectomies in the treatment of refractory epilepsy, was obtained under IRB consent, number CEP-HUCFF 060/05. This information was added in the Methods section (page 6, line 129; page 7, line 153). (2) In the materials and methods section, the authors claim to have acquired proton density images (what seems to be correct). However, in the legend of figure 1 (panel A-D), the authors mention T2-weighted MRI. Please, correct this discrepancy! We thank the Reviewer for pointing this issue and we have corrected the text in the Methods section (page 8, lines 182-202). (3) What was the in-plane resolution of the MR images? We have included the in-plane resolution for the images in the Methods section (page 8, lines 182-202). (4) What type of MRI sequences were used (spin-echo, gradient-echo, )? We have specified the type of the sequences in the Methods section (page 8, lines 182-202). (5) Materials and Methods; reagent section: Please, state the city of origin for all companies. It was done (page 5, lines 110-120).
(6) The statement Four animals were used for each experiment described below (page 8) is not clear. Do you mean that four animals per group (glioma, astrocyte engraftment) were used? Yes, we corrected this sentence in the manuscript (page 8, lines 178-179). (7) It is not clear from the legend to figure 1 that the four rows represent four different animals. The Figure 1 shows a representative data of one animal. It shows images (four slices) acquired with different sequences: T2 weighted sequence (first row), T1 weighted sequence (middles row) and T1 weighted sequence after gadolinium injection (lower row). This experiment was performed 4 times and we obtained similar results. We added this information in order to clarify this issue in the figure legend (page 22, line 605-611). Level of interest:an article whose findings are important to those with closely related research interests Quality of written English:Acceptable Statistical review:no, the manuscript does not need to be seen by a statistician. Declaration of competing interests: I declare that I have no competing interests.