Overall Considerations: In this paper, Moon et al show that osteoblasts may synthetize testosterone from DHEA and therefore contribute to PCa progression in bone metastases. Identifying this new mechanism is off interest since it can be targeted by androgen therapies. The paper combines in vitro data (OB primary cells), three PCa cells line and mass spectrometry. The paper is clearly presented but several concerns need to be settled to make the paper appropriate for publication in JBMR.

Major point
The concentration of metabolites and T are so different in OB versus PCa cells that it is hard to compare and make statement concerning the AR signaling activation. The quantification of estrone/estradiol should be performed in all conditions combine with ER signaling activation. How the activation of AR signaling by OB-CM is as high as in C42b while the T concentration is 60x less (25/1500 pg/ml). If AR signaling is stimulated in C42b and LNCaP, what is the impact on tumor cells (proliferation test, invasion test?). If mice are treated by DHEA, do we see 3b-HSD-b7, 17b-HSD-b4 and b11, rdh11 and srd5a 1 and a3 expression stimulation in bone metastases induced by C42b injection (in tumor and OB?). Statistics should be shown in Fig.2-4 and some graphs in Fig.5. Some SD are visible, other no ( for instance Fig.6 (100nM DHEA CM+enz)).

Mat and Meth: Page 4 L32, what is the gender of the mice used for long bone osteoblast? If male, do they observe the same curve and concentration of androstenedione and androstenediol (Fig. 3) from female bone derived osteoblast?
Page 5, L1: bone and hypoxia should be moderated since depending where in the bone the osteoblasts are located O2 concentration may very much vary.

-Fig.2: L11-18 should be removed from results section. The data in Fig.2 should be clearly presented in the text to bring out the high level expression of 3b-HSD-b7, 17b-HSD-b4 and b11, rdh11 and srd5a 1 and a3 that are all expressed in A-C. Results from biopsies should be moderated since they are also containing other cell types than osteoblasts. They should also be presented by gender.
-Fig.3: Since no DHT is detectable, could we imagine that T and androstenedione are metabolized in estradiol and estrone respectively through the aromatase that is known to be expressed in OB?
-Fig.3/4: if we compare Fig.3 and 4, at 72h after 100nM DHEA treatment: increase of androstenedione: in OB (50 pg/ml), C42b (2500 pg/ml), 22rV1 (500pg/ml) and ARCaPm (50pg/ml); androstenediole : in OB (1500pg/ml), C42b (1000pg/ml) and 22rV1 /ARCaPm (8000/6000pg/ml) conducting to Testosterone concentration: OB (25pg/ml), C42b (1500pg/ml) and 22rV1 /ARCaPm (150/25pg/ml) which may explain the fact that DHT is only detectable in C42b.
-Fig.5: The results are quite surprising if we only focus on Testosterone concentration. If we look at Fig .3 the conc T after 100nM DHEA at 72h we can observed around 25pg/ml. Now if we look in Fig.4. at the conc T after 100nM DHEA treatment of C4-2B cells at 12h, we observed around 50pg/ml suggesting that at 12h (Fig.5), we have 2 times more T produced after direct DHEA treatment on tumor cells than from the OB-CM. In accordance with T concentration and AR signaling stimulation, how the authors reconcile these data? Plus, after 72h of treatment the conc of T is 1500pg/ml (Fig.4). Now what is the conc of T after 72h (on OB)(OB-CM) then +72h (on tumor cells) in Fig.5 since the level of AR activation signaling is quite high and almost similar to direct DHEA treatment on tumor cells ? Could we imagine that OB-CM may stimulate testosterone production by tumor cells? The panel C and F are hard to compare since the data on direct DHEA treatment in tumor cells were realized at 100nM and not 1microM in Fig.4.
Fig.7: this data may also suggest that T in OB (see previous comments on Fig.3) is mainly converted into Estrone and Estradiol and consequently can’t activate AR. Estrone and estradiol should be quantified. This question is off interest since ER signalling is involved into PCa progression. Now, we may not observe stimulation of AR because the concentration of T is low and therefore DHT is not detected. If we compare the Testosterone concentration from ARCaPm (25pg/ml) after 100nM of DHEA at 72h that correspond to the same concentration of T in OB after 100nM of DHEA at 72h. In both conditions DHT is not detected.

I would like first to thank the authors for the new version of the manuscript and for the consideration of some of my comments: expression of genes involved in T and DHT production in male versus female OB that clearly show that Female OB expressed way more Hsd3b and Hsd17b (Fig.2), the fact the authors try to address my concern on the estrogen pathway and for adding PSA data that comfort the data showing OB CM on AR signalling in PCa cells (Fig.7).

Having saying that, the proliferation data with DHEA-CM versus DHEA in Fig. 8 are not really convincing. If PSA data (obtained at 72h) are convincing, the data obtained at Day6 for proliferation are more questionable. What is the DHT concentration at that time (day 6)? PSA and proliferation data should be comparable and shown at the same time, meaning at 72h. Clearly at day 3, no difference is seen in proliferation. Again data in Fig.5 in C42b are still also very questionable based on the fact that DHEA alone conditions reach the same level or override the level observed in DHEA-CM condition. In fig.6, the reduction of the AR signalling is obtained because the authors treat the cells daily with ENZ.
Also no in vivo data are shown in the R1 and the paper is still only based on in vitro data and PCa cell lines.