Referee 3's review (Helmut Segner):

The Ms describes the development of a gonad-derived zebrafish cDNA microarray and the
use of this array to identify sexually dimorphic gene expression patterns in ovary, testis and brain of adult zebrafish. Importantly, the array is constructed using not only cDNAs from the gonads of adult zebrafish but also from gonads of developing zebrafish. This feature opens the possibility that the array could be used to study the developmental regulation of gonad differentiation (what is a most promising field for future research, as our current under-standing of the regulation of sex differention in zebrafish is very limited).

The results of this study on sexual dimorphic gene expression profiles of testis and ovary, and on their correlation to gene expression profiles in the brain are novel and gog beyond the present state of knowledge. The claims of the authors that this study (a) identifies genes with sexual dimorphic expression in both the gonads and the brain, and (b) is the first to provide evidence for a large-scale differential gene expression between male and female brains are well supported by the experimental findings. Still, the authors have to be careful in formu-lating the conclusions from their study. For instance, I found the formulation as given in the Abstract “.. thereby implicating these genes in sexual development” to be highly misleading. The observation that certain genes show sexually dimorphic expression in the brain or gonads is by no means an indication that these genes are involved in the regulation of sexual differentiation. This may be well illustrated from the recent discussion on a possible role of brain aromatase in guiding sexual differentiation of zebrafish (see Trant et al., 2001, Kalli-vretaki et al., 2007). The authors must be very clear with respect to the study subject of the present work: notwithstanding the fact that for construction of the array, data from developing zebrafish gonads were included, the results presented in the study deal only and exclusively with adult zebrafish. Thus, the results do not support any conclusion as to the regulation of sexual development in zebrafish. Confusing gene expression during develop-mental sex differentiation with gene expression in the already differentiated, adult stage is conceptually misleading.

The technical basis of this study appears to be sound (although I am not particularly ex-pertised in array technology) and the study is well placed into the context of previous studies; the existing literature on zebrafish sexual differentiation is very well covered. Given the use of zebrafish as vertebrate model in developmental biology, environmental toxicology, etc., this paper will be of interest to a broad readership. A weakness of the study is the presen-tation of the results, which is somewhat messy and not easy to follow (see also Specific Comments). A careful revision of the text would cleraly improve the readability and value of the Ms. Also, I wonder whether it would be advantageous to separate “Results” and “Discussion”.

Specific Comments

Abstract
In the “Background”, the authors mention the elucidation of the “regulation of zebrafish gonad development”, however, this is clearly not the topic of the presnet work (it appears to be the topic of ongoing research in the author’s laboratory, but it is not contained in this study). The presnet study deals with (a) the construction of the gonad-derived array, and (b) with de-scribing sexual dimrophic gene expression profiles in gonads and brain. Nothing else. In particular the study is not dealing with any regulation or differentiation processes. The authors must be very clear on this point to avoid misunderstandings and confusion.

Same critics apply to the sentence at the bottom of the Abstract page: “thereby implicating these genes in sexual development”

I feel uncomfortable with the conclusive sentence saying that “our array may provide information on genetic disorders affecting gonadal disroders and fertility in mammals”. Particularly reproductive processes and structures differ greatly between fish and mammals, and I doubt that fish can serve as model in this field of biological processes (although fish is well suitable as model in other aspects). Is it not a sufficiently justified result to provide finally an array that enables better understanding of a key biological process in the model species zebrafish ?

Introduction
well introduces the background/state of knowledge and study objectives

Results and Discussion
p. 8 ff (24 candidate genes): If my understanding is correct, in order to “validate” the UniClone microarray, the authors introduced 24 genes into their array that were known from previous studies to possess sexually dimorphic expression. Of these 24 genes, 18 genes showed significant expression differences between the analysed tissues. Among these 18 genes, 11 genese showed 1.5 fold higher expression in “comparison to the common reference target”. This part of the Ms leads to a number of questions. First, as a reader I would like to know the identity of the 24, 18 and 11 genes, respectively.The 24 genes are shown in the supplementary material, however, the identity of these genes need to be indicated in the text (not all the information on FB1, etc – this can stay in the supplementary material, but for the reader it is important to know what genese were used to “validate” the UniClone array). Further, in Table S3, the data source providing evidence for sexually dimorphic expression of these gens in zebafish needs to be given. it needs to be indicated. Second, I do not understand the difference between the 18 genes that were significantly different and the 11 genes that showed a > 1.5 fold expression difference ? Why making this distinction ? Third, the authors have to inform the reader which of the 24 genes are significantly differently epxressed and which of the 18 genes are >1.5fold differently expressed. The easiest way to do this would be to have a small table with the names of the 24 genes and an indication if they are significantly/1.5-fold differently expressed in which organs (no extensive numerical descriptions – this can really stay in Table S3 – but a comprehensive summary of the key information from table S3). Fourth, if these 24 genes were selected since they were reported in previous studies to show sexual dimorphic expression pattern, why only 11 (18?) of them show dimorphic expression in the UniClone array ? I did not find any discussion on this in the Ms. Finally – and this applies also to the other parts of the Results – what is this “common reference target” ? This is nowhere in the Ms described, although most comparisons are made to this reference target. The “common reference target” needs a substantially improved explanation and description.

The information provided by Figure 2 is enigmatic. I guess the authors want to illustrate their finding that “1366 and 881 genes showed ate least 1.5 fold higher expression in the adult ovary or testis”, however, how I can extract that information from Figure 2 is not clear to me. Thus, either the authors delete Figure 2 or they provide a reasonable explanatory legend.
Further, the authors should be stringent in their terminology, in order to avoid confusion: on p. 11,they say that genes showing 1.5 fold higher epxression in, e.g., ovary will be designated in the following as “ovary-up” genes, but most of the time they are designated as “ovary-enhanced” genes.

p. 12, line 5 from top: “Out of the 1366 ovary-enhanced transcripts, 46 % originated from the developing gonads”. I guess, this means that the sequence information originated from the developing gonads, but the expression was measured in the adult gonad ?
p. 13, line 10 from top: “All the ovary-expressed genes were found to be oocyte-specific”. This is unlikely to be true – I guess there are other genes expressed in the ovary which are not oocyte-specific (for instance, genes of the follicle cells). What the authors obviously mean is that “all the ovary.enhanced genes were found to be oocyte specific”. I agree, this may be considered to be semantics, but exactly these small inaccuracies in the description of the Results make it sometimes difficult to follow and understand. With more stringency in the data description and wording, the clarity of the Ms would clearly improve.

p. 19, line 5 from top: Here, the authors talk of 144 genes showing higher expression in ovary than testis, on p. 12, they talk of 1266 ovary-up genes – I guess this difference comes from the respective reference (p.12: the enigmatic “common reference”, p.19 – the direct ovary-testis comparison). However, exactly for this reason it is so important that this “common reference” is better explained. It makes a great deal of difference if the message from this study is that the difference of ovarain and testicular expression proifle consists of 144 or 1266 transcripts.

As said, probably a separation of Resulst and Discussion would improve the Ms presentation. Then the Results could contain “establishing the UniClone microarary”, “Validating the array”, “sexually dimorphic expression differences between tissues (addressing initially the overall differences and then dual differences, i.e. male-female brain, male-female gonad, male-female “rest of body”) and, finally, “co-expression analysis of brain and gonads”, providing a short summary of the main findings at the end of each section thereby making the key information more easily accessible to the reader. The Discussion could then focus on the patterns and on selected genes with dimorphic epxression/co-expression such as the septin family.

Methods
Explain abbreviations ! (what is wpf, etc.)
Cryostat: This is “Leica Microsystems GMBH, Nussloch, Germany”
Gonad dissection of juvenile zebrafish: It needs a better description of the agarose method – is the agarose filled into the body cavity or ist just used to surround the fish ? Further, to sex the juvenile fish, the authors used – as macrocopical differentiation of male/female gonads at the juvenile stage would be too arbitrary - transcription profiles determined by means of a former version of the UniClone array – how “sex-discriminative” is this former version ? How well can developing gonads – which are not yet fully dimorphic – be sexed by this array ?

**********
N.B. These are the comments made by the referee when reviewing an earlier version of this paper. Prior to publication the manuscript has been revised in light of these comments and to address other editorial requirements.