Bone marrow adipocytes drive the development of tissue invasive Ly6Chigh monocytes during obesity

  1. Parastoo Boroumand
  2. David C Prescott
  3. Tapas Mukherjee
  4. Philip J Bilan
  5. Michael Wong
  6. Jeff Shen
  7. Ivan Tattoli
  8. Yuhuan Zhou
  9. Angela Li
  10. Tharini Sivasubramaniyam
  11. Nancy Shi
  12. Lucie Y Zhu
  13. Zhi Liu
  14. Clinton Robbins
  15. Dana J Philpott
  16. Stephen E Girardin
  17. Amira Klip

  • Curated by eLife

    eLife logo

    Evaluation Summary:

    Using mice fed with high fat diet (HFD), Boroumand et al. observed a link between bone marrow (BM) adipocyte whitening and the expansion of BM Ly6Chi monocytes and derived cells in the adipose tissue. By adopting an in vitro approach, they also show that BM conditioned medium is able to metabolically rewire Ly6Chi monocytes notably concerning mitochondrial fission/fusion gene expression. They conclude that early changes in the BM adipocytes induced by HFD drive the activation of monocytes and influence the outcome of the disease.

    This study is of interest to those investigating BM adaptations to lipid signals as well as macrophage biologists interested in macrophage recruitment and differentiation in the context of obesity and beyond in inflammation.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. The reviewers remained anonymous to the authors)

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

During obesity and high fat-diet (HFD) feeding in mice, sustained low-grade inflammation includes not only increased pro-inflammatory macrophages in the expanding adipose tissue, but also bone marrow (BM) production of invasive Ly6C high monocytes. As BM adiposity also accrues with HFD, we explored the relationship between the gains in BM white adipocytes and invasive Ly6C high monocytes by in vivo and ex vivo paradigms. We find a temporal and causal link between BM adipocyte whitening and the Ly6C high monocyte surge, preceding the adipose tissue macrophage rise during HFD in mice. Phenocopying this, ex vivo treatment of BM cells with conditioned media from BM adipocytes or bona fide white adipocytes favoured Ly6C high monocyte preponderance. Notably, Ly6C high skewing was preceded by monocyte metabolic reprogramming towards glycolysis, reduced oxidative potential and increased mitochondrial fission. In sum, short-term HFD changes BM cellularity, resulting in local adipocyte whitening driving a gradual increase and activation of invasive Ly6C high monocytes.

Article activity feed

  1. Version published to 10.7554/elife.65553 on eLife
  2. eLife

    Author Response

    Reviewer #1 (Public Review):

    1. In the future it will be interesting to determine how these changes in the bone marrow relate to the different subsets or recruited macrophages present in obese tissues. For example, whether monocytes in the bone marrow preferentially generate CD9+Trem2+ Lipid associated macrophages recently described in obese adipose tissue (Jaitin et al, Cell, 2019) or if they are equally capable of generating monocyte-derived tissue resident macrophages in obese tissues.

    We appreciate and concur with the Reviewer’s suggestion, as stated for future analysis, of how bone marrow monocytes compare with macrophages in adipose tissue. That is a long-term plan and will make the subject for a full new study of interest to the immunometabolism community. As preamble to that future study -and considering that Jaitin et al identified CD9 and Trem2 in lipid-laden macrophages- we have tentatively explored if bone marrow-derived macrophages (BMDM) from mice fed HFD and LFD differ in their expression of these markers. In these exploratory experiments, however, HFD did not statistically change the expression of either marker in the BMDM.

    1. The main strength of this paper is in the identification of the changes in the monocyte subsets abundance early after feeding a HFD and in uncovering the metabolic changes in and between these two monocyte subsets in obese mice. One concern regarding the data as a whole is that, while the authors have nicely indicated the number of samples/mice in each figure, there is no mention of how many times each experiment was performed.

    We have more explicitly and amply stated the number of times every experiment was performed and this information is also added to the figure legends.

    Additionally, the inclusion of the different gating strategies used particularly for the first figures would be advantageous to fully appreciate the findings being presented. This is particularly relevant for the identification of the Ly6Chi and Ly6Clo BM monocytes.

    We now present the gating strategy at the beginning of the Results as Figure 2 – figure supplement 1. In Figure 2 – figure supplement 1B, control flow experiments without anti-Ly6C or anti-Cd11b are shown gated for the Gr1(-) vs CD115(+) subset, confirming the proper positioning for the Ly6clo and Ly6chi gates. In Figure S1C, we illustrate the gating strategy shows that CX3CR1 segregates with the Ly6Clo sorted monocytes, and CCR2 segregates with the Ly6Chi, as expected. We hope that this complements the information on the identification of the Ly6Chi and Ly6Clo monocytes. In the future, a more complete analysis of the FACS-sorted Ly6Clo and Ly6Chi cells could be performed using RNAseq, which was however outside the realm of possibilities for this study.

    1. The alternative explanation (to Ly6Clo conversion to Ly6Chi monocytes) could be that there are some progenitors remaining in these cultures that give rise to Ly6Chi monocytes following exposure to the conditions media. .. It is important to confirm that the sorted cells are a pure population of Ly6Clo monocytes with no contamination from progenitors that are also Ly6Clo

    We appreciate the suggestion; to address this interesting possibility, in new experiments we used markers of myeloid progenitor cells (CD117+;Sca1-, followed by gating for CD16/32 vs. CD34 to identify GMP, CMP and MEP populations). The new findings show that GMP represent the majority of progenitors that are present in the FACS-identified Ly6Clo and Ly6Chi monocyte populations derived from complete bone marrow cells. In this analysis, we find the GMP are present at low abundance (ranging 140 GMP per 1000 Ly6Clo, in results from 3 separate mice). (NEW Figure 7E). This finding complements our original observation that there are relatively few progenitor cells in the in vitro-generated monocyte samples, detected by FACS analysis (Figure S5). Despite their relatively low abundance, we cannot discount that some could become Ly6Chi cells. However, the 18-hour duration of exposure of FACS-sorted Ly6Clo cells to WATA-CM would have allowed for only about one doubling event of precursor cells. If so, the progenitors could not fully account for the entirety of the change in proportion of Ly6Clo in favour of Ly6Chi.

    Supporting this argument, when treating in vitro-generated monocytes with WATA-CM, the slight increase in CMP progenitors did not manifest as an increase in the downstream GMP progenitor numbers (now in Figure 7 – figure supplement 1), which are upstream of the Ly6C monocyte lineage.

    To more directly explore the growth potential of progenitor cells, we have now used the Colony Forming Assay to determine the ability of progenitors to give rise to more differentiated monocyte precursor colonies upon incubation with the various conditioned media. In vitro-generated bone marrow-derived monocytes were exposed to control-, WATA- and BATA- CM, The new results, shown in the new Figure 7C,D, indicate that whereas white adipocyte media (WATA-CM) did not expand colonies from CMP (GEMM), GMP (CFU-G and CFU-M) or MEP (BFU-E) progenitors, brown adipocyte media (BATA-CM) slightly expanded colonies derived from CMP/GEMM and skewed GMP cell differentiation toward CFU-M (which lead to monocytes) from CFU-G (which lead to granulocytes and ‘neutrophil’ like monocytes, Yáñez et al, 2017). The new text on pages 15-16, lines 341-365 reads as follows:

    “To buttress the above results, we also assessed the colony forming potential of in vitro-generated monocytes that received pre-treatment of WATA-CM or BATA-CM, to assess the potential for expansion of progenitor cells present in the samples. Colonies were identified as BFU-E (giving rise to erythroid cells); CFU-GEMM (giving rise to large mixed cultures of granulocyte, erythroid, macrophage, megakaryocyte; also, known as CMP); CFU-G (giving rise to granulocytes) or CFU-M (giving rise to macrophages). BATA-CM promoted growth of CMP/CFU-GEMM cultures over control Media (** p<0.01 Two-way ANOVA, Tukey’s multiple comparisons, Figure 7C) and biased granulocyte/macrophage progenitors (widely known as GMP) towards macrophage over granulocyte differentiation, relative to control Media (** p<0.01) or WATA-CM (*p<0.05 or ***p<0.001, Figure 7C). The total numbers of colonies that grew after 7-10 days of culturing of each pre-treated cohort of monocytes were not different across the three treatments, although trended upwards with BATA-CM (Figure 7D). These results indicated that while BATA-CM promoted expansion of selected populations (CMP/CFU-GEMM and CFU-M), consistent with the increase in BrdU incorporation shown above, WATA-CM was without effect relative to control Media. The above findings suggest that while the proportional increase in Ly6Clow monocytes induced by BATA-CM involves cell proliferation, the proportional increase in Ly6Chigh monocytes induced by WATA-CM does not. As a complementary approach, BM cells were analyzed by flow analysis for the presence of monocyte progenitors within the Ly6Clow or Ly6Chigh monocyte subsets. GMP progenitor cells were essentially the only progenitors detected by this approach in the Ly6Clow monocyte subset, and they represented 140 GMP per 1000 Ly6Clow cells (Figure 7E). During the incubation time of 18 h with conditioned medium, we anticipate the progenitors could theoretically undergo only one doubling and therefore unlikely to account for the full changes in Ly6Clow cell numbers produced by WATA-CM. Collectively, the results in Figures 7C-E indicate that WATA-CM treatment did not result in an appreciable expansion of progenitor cells or colony formation. Therefore, alternative mechanisms were explored that might contribute to the WATA-CM induced shift towards Ly6Chigh monocyte preponderance, particularly the possible conversion of one subset into the other.”

    Altogether, we feel that these previous and new results do not endorse the possibility that the brunt of the Ly6Chi cells increase is due to progenitor differentiation in response to WATA-CM. We therefore lean towards the interpretation that Ly6Clo cells convert to Ly6Chi but agree that this potential mechanism will require further additional analysis in the future.

  3. Version published to 10.1101/2020.12.07.414862v2 on bioRxiv
  4. eLife

    Reviewer #1 (Public Review):

    In this study, Boroumand et al investigate abundance and metabolic phenotype of Ly6Chi and Ly6Clo monocytes in the bone marrow (BM) following feeding a HFD for 3, 8 and 18 weeks compared with a control diet. The authors suggest that upon accumulation of white adipocytes in the BM (8 weeks of feeding), monocytes are skewed towards the Ly6Chi subset, which have been shown to give rise to many macrophage subsets in obese tissues. The authors further demonstrate metabolic changes in Ly6Clo monocytes which may contribute towards this phenotype. Finally, through a series of in vitro and ex vivo cultures, the authors suggest that the increase in Ly6Chi monocytes is due to conversion of Ly6Clo monocytes into Ly6Chi monocytes as a result of the increased prevalence of white adipocytes in the bone marrow.

    Overall the findings of this work are interesting to the field and in the future it will be interesting to determine how these changes in the bone marrow relate to the different subsets of recruited macrophages present in obese tissues. For example, whether these monocytes preferentially generate CD9+Trem2+ Lipid associated macrophages recently described in obese adipose tissue (Jaitin et al, Cell, 2019) or if they are equally capable of generating monocyte-derived tissue resident macrophages in obese tissues.

    The main strength of this paper is in the identification of the changes in the monocyte subsets abundance early after feeding a HFD and in uncovering the metabolic changes in and between these two monocyte subsets in obese mice. One concern regarding the data as a whole is that, while the authors have nicely indicated the number of samples/mice in each figure, there is no mention of how many times each experiment was performed. Including this would greatly aid in an understanding of the reproducibility of the results. Additionally, the inclusion of the different gating strategies used particularly for the first figures would be advantageous to fully appreciate the findings being presented. This is particularly relevant for the identification of Ly6Chi and Ly6Clo BM monocytes.

    The conclusions made regarding the role of white adipocytes in skewing the monocyte subsets and particularly regarding the conversion of Ly6Clo monocytes to Ly6Chi are however less convincing. The authors use a culture strategy where they grow BM monocytes in vitro for 5 days. They then culture these 'monocytes' for a further 18 hours with conditioned media from BM adipocytes from control or HFD fed mice. They show that culture with 8 & 18 week conditioned media results in the increased abundance of Ly6Chi monocytes. The authors later claim this is not through proliferation of the existing Ly6Chi monocytes but conversion from Ly6Clo monocytes. However, the alternate explanation could be that there are some progenitors remaining in these cultures that can give rise to Ly6Chi monocytes following exposure to the conditioned media. To further validate these claims, it would be beneficial to sort Ly6Chi monocytes and culture them with the conditioned media to demonstrate the numbers do not increase. Moreover, it is important to demonstrate that there are no progenitors left in these cultures when the conditioned media is added. Indeed, later in the manuscript, when Ly6Clo monocytes are sorted and cultured with media from EWAT or BAT, it would be important to confirm that the sorted cells are a pure population of Ly6Clo monocytes with no contamination from progenitors that are also Ly6Clo that could give rise to Ly6Chi monocytes without going through the Ly6Clo monocyte stage.

    In a similar vein, the authors suggest no conversion of Ly6Chi monocytes to Ly6Clo monocytes, but that Ly6Clo monocytes would convert into Ly6Chi monocytes (fig. 7). As this is a rather controversial claim, additional data in support of this conclusion would be beneficial. For example, after 18 hours of culture it is possible that if the authors are sorting Ly6Chi monocytes on the basis of Ly6Chi expression, that the antibody staining may be maintained for 18 hours. Similarly, after culture, it is possible that the cells are less healthy and hence non-specific binding should also be ruled out. Alternatively, qPCR for gene expression associated with Ly6Chi and Ly6Clo monocytes could be utilised to further substantiate the claims. For example, Spn expression for Ly6Clo monocytes, Ly6c2 expression for Ly6Chi monocytes.

    Thus overall, this manuscript nicely demonstrates changes in the BM monocyte subsets and their metabolism, however some additional controls are required to further validate the claim that Ly6Chi monocytes are increased due to Ly6Clo monocyte conversion to Ly6Chi monocytes.

  5. eLife

    Reviewer #2 (Public Review):

    The paper presented by Boroumand et al. aims to delineate the impact of bone marrow resident adipocytes on the phenotype, development, and metabolism of murine monocyte subsets during diet-induced obesity and leanness. The paper provides an interesting analysis of the metabolic state and phenotype of mitochondria in murine monocytes during high-fat diet feeding. Furthermore, it provides some insight on the crosstalk between bone marrow resident adipocytes and different monocytes.

    The paper will help to further delineate the response of monocytes during obesity, however, the impact the paper will have on the field of mononuclear phagocytes biology and our understanding of myelopoiesis during low-grade inflammation is limited.

    Several claims should be more thoroughly addressed, such as the phenotypes of macrophages found within the adipose tissues and a more fine-grained analysis of the mononuclear phagocyte progenitors within the bone marrow. Furthermore, a central claim of the paper is that Ly6clow monocytes convert to Ly6chigh monocytes. If the authors would like to hold that claim it needs some experiments which are supportive of that hypothesis.

  6. eLife

    Evaluation Summary:

    Using mice fed with high fat diet (HFD), Boroumand et al. observed a link between bone marrow (BM) adipocyte whitening and the expansion of BM Ly6Chi monocytes and derived cells in the adipose tissue. By adopting an in vitro approach, they also show that BM conditioned medium is able to metabolically rewire Ly6Chi monocytes notably concerning mitochondrial fission/fusion gene expression. They conclude that early changes in the BM adipocytes induced by HFD drive the activation of monocytes and influence the outcome of the disease.

    This study is of interest to those investigating BM adaptations to lipid signals as well as macrophage biologists interested in macrophage recruitment and differentiation in the context of obesity and beyond in inflammation.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. The reviewers remained anonymous to the authors)

  7. Version published to 10.1101/2020.12.07.414862v1 on bioRxiv