Enhancement of hematopoiesis and lymphopoiesis indiet-induced obese miceMark D. Trottier, Afia Naaz1, Yihang Li2,3, and Pamela J. Fraker4

Biochemistry and Molecular Biology Department, Michigan State University, East Lansing, MI 48824

This contribution is part of the special series of Inaugural Articles by members of the National Academy of Sciences elected in 2007.

Contributed by Pamela J. Fraker, March 27, 2012 (sent for review March 2, 2012)

A rodent model of diet-induced obesity revealed that obesitysignificantly altered hematopoietic and lymphopoietic functionsin the bone marrow and thymus. C57BL/6 mice were fed a mixedhigh-fat diet (HFD) of 45% fat or 10% fat diet (lean controls) for180 d. A sustained increase in the numbers of cells found in bonemarrow and thymus of HFD mice occurred from day 90 to day 180.However, with the exception of a 10–18% increase in the propor-tion of lymphocytes, the composition of monocytes, granulocytes,erythrocytes, and mixed progenitor lineages remained normal inthe marrow. Likewise, thymuses of HFD mice increased 30–50%in size comparedwith controls, with analogous increases in thymo-cyte numbers. The overall thymus cellular composition remainednormal. Although increased blood and lymphatic volume in obesemicewould play a role in increased hematopoiesis, therewere largeand disproportionate increases in blood leukocytes of HFD mice,indicating that homeostasis was notmaintained. Leptin, which pro-motes lymphopoiesis and myelopoiesis, reached 100 ng/mL in serafrom HFDmice. Moreover, a three- to sixfold increase in adipocytesin marrow resulted in spiked leptin mRNA expression in bones ofHFDmice comparedwith lean controls. Other cytokines and growthfactors did not show any increases in obese marrow. The substan-tial increase in lymphopoietic and hematopoietic processes in HFDmice indicates that the primary tissues are another facet of theimmune system dysregulated by obesity, which was perhaps fos-tered by higher amounts of leptin in marrow and serum.

The bone marrow, which is located throughout the skeletalmass, is one of the largest and most active tissues of the body.

Through complex differentiation pathways originating with hema-topoietic stem cells, the marrow produces billions of new leuko-cytes and erythrocytes each day (1–4). Osteoblasts, chondrocytes,myocytes, adipocytes, etc., also arise daily from marrow mesen-chymal stem cells. Because it is a highly vascular tissue, themarrowprovides a highway for the constant transport of newly producedcells to the systemic circulation. Thus, the breadth of divergentevents occurring daily in the marrow is remarkable.Given the complex operations occurring in the bone marrow

each day, it would seem logical to assume that it would be alteredby disease states, changes in metabolism and metabolic diseases.However, the roles that the marrow might play in the etiology ofdisease have been largely overlooked. It is known, however, thatincreased proliferation of granulocyte percursors is observed inmice in response to inflammation and infection (5, 6). Malnu-trition, including zinc deficiency and protein calorie deficiencies,profoundly alters the primary immune tissues, especially thebone marrow (7, 8). Stress, which may create elevated levels ofendogenous glucocorticoids, initiates marked apoptosis amongpre-B and pre-T cells while enhancing myelopoiesis in the mar-row (9, 10). In dextran sodium sulfate-induced colitis, the mar-row exhibits marked declines in erythropoiesis and acceleratedproduction of monocytes and granulocytes, thereby exacerbatingthe anemia and inflammation of the gut that accompanies thisdisease. These examples make clear that hematopoietic processesin the marrow are actively responding to new and diverse phys-iological processes and not simply maintaining homeostasis.

In addition to metabolic changes, obesity is also associated withchronic low-grade inflammation that leads to the increased pro-duction of adipokines, proinflammatory cytokines, leptin, etc.,in blood (11). These factors play a role in the development ofseveral comorbidities associated with obesity, such as cardio-vascular disease, diabetes, hypertension, arthritis, and stroke (12,13). Inflammation, along with documented changes in T-cell andmacrophage function, indicates that immune dysregulation hasoccurred in the obese (14, 15). Considering the system-widechanges occurring during obesity, it is possible that immune celldevelopment in the primary immune tissues would be affectedby obesity-associated inflammation, increased adiposity, and/orhormonal changes. Indeed, leptin-deficient obese ob/ob micehave a 40% reduction in the number of bone marrow cells andsmaller thymuses compared with lean mice (16). These samemice also have reduced lymphocyte numbers in peripheral blood(17). Similarly, obese db/db mice, which have an inactive leptinreceptor, have a deficit in lymphocyte progenitors as well as re-duced B- and CD4+ T-cell numbers in peripheral blood (18).Provision of recombinant leptin to ob/ob mice promoted bothlymphopoiesis and myelopoiesis (16). Thus, it seemed that lep-tin, which is substantially elevated in the obese, might alter he-matopoiesis. Comparatively little was known about the effectsof diet-induced obesity (DIO) on hematopoiesis. It is known thatadipose tissue in DIO mice contains increased numbers of proin-flammatory macrophages (adipose tissue macrophages) that arebone marrow-derived as well as a large number of stem cells (19–22). Similarly, obesity in humans has been associated with in-creased white blood cell counts (23–25), but the role of themarrow in these changes was unclear.Considering all of the aforementioned issues, it seemed plausible

to assume that other obesity-related changes would occur in he-matopoietic and lymphopoietic processes in bone marrow and thethymus. In this study, we investigated the effects of diet-inducedobesity in young mice on hematopoietic and lymphopoietic func-tion. Profound changes in the primary immune tissues were notedin the obese mice that persisted throughout 180 d of this study.

ResultsWeight Gain and Leptin. Mice were fed either a 10% (control) or45 Kcal percent fat (mixed fat; HFD) diet starting at 6–7 wk ofage. Diet consumption, body weight, and serum leptin were

Author contributions: M.D.T. and P.J.F. designed research; M.D.T., A.N., and Y.L. per-formed research; M.D.T. and A.N. contributed new reagents/analytic tools; M.D.T., A.N.,Y.L., and P.J.F. analyzed data; and M.D.T. and P.J.F. wrote the paper.

The authors declare no conflict of interest.1Present address: ResearchGroupDivision, KytheraBiopharmaceuticals, Calabasas, CA91301.2Present address: Geriatric Research, Education and Clinical Center (GRECC), VeteransAffairs Palo Alto Health Care System, Palo Alto, CA 94304.

3Present address: Division of Endocrinology, Gerontology, and Metabolism, Departmentof Medicine, Stanford University, Palo Alto, CA 94305.

4To whom correspondence should be addressed. E-mail: fraker@msu.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1205129109/-/DCSupplemental.

7622–7629 | PNAS | May 15, 2012 | vol. 109 | no. 20 www.pnas.org/cgi/doi/10.1073/pnas.1205129109

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

020

monitored from day 7 to approximately day 180. HFD miceconsumed the same amount of food in grams as mice fed controldiet, indicating a lack of hyperphagia (Fig. 1A). Because of thehigher caloric density of HFD (4.73 vs. 3.85 kcal/g for HFD vs.control, respectively), HFD mice consumed ∼25% more caloriesthan control mice (Fig. 1A). Over time, this increased calorieconsumption resulted in significantly elevated body weights inHFD mice throughout the experiment starting as early as day 7(Fig. 1B). HFD mice had body weights that were increased 10–15% by day 30, 20–30% by day 60, 40% by day 90, and 50–60%by later time points compared with controls. As expected, thisweight gain was accompanied by significant increases in theamount of epididymal fat. Fig. S1 shows epididymal fat weightincreased 1.8- to 3.0-fold in HFD mice. HFD mice also hadsignificantly elevated serum leptin throughout the experiment(Fig. 1C), ranging from a threefold increase at day 33 to a five-fold increase by day 162. Given the substantial increase in epi-didymal fat, this finding would be expected, because adipocytesare the primary producers of leptin. This significant and earlyrise in leptin would be important, because leptin is known topromote lymphopoiesis and granulopoiesis (16), which will beshown to be enhanced in HFD mice. More modest increases inweight, fat, and serum leptin with age were noted in mice oncontrol diet.

Effect of HFD on Bone Marrow Classes. To assess the effects of HFDon bone marrow cellularity and its hematopoietic composition,flow cytometry was used to analyze the proportion and totalnumber of bone marrow cells in both control and HFD mice. Asystem that delineates the major subclasses of marrow cells ofthe hematopoietic system (e.g., erythrocytes, lymphocytes, gran-ulocytes, monocytes, and progenitors) was used. Flow cytometricprofiles and secondary antibodies used to validate these majorhematopoietic populations (Fig. S2) have been used many timesby our laboratory and others (8, 9, 26). Examination of early,

intermediate, and late time points after initiation of diet revealedthat HFD mice showed little or no significant changes in cellularcomposition compared with control mice (Fig. 2). Only starting atday 90 and continuing thereafter were modest increases in theproportion of marrow lymphocytes observed in HFD mice. Thelymphocyte precursors increased at days 90, 142, and 162, rangingfrom 10% to 18% (Fig. 2). Overall, HFD produced only modestchanges in the composition of marrow, with the exception notedfor the increase in the proportion of lymphocytes starting at day 90.Although the relative proportion of marrow cell subclasses in

HFD mice remained mostly unchanged, total marrow cell num-bers tell a different story. Indeed, HFD mice had a significantlyincreased number of nucleated cells in the bone marrow startingat day 90 compared with lean controls. This increase in totalmarrow nucleated cells continued through day 184 of the study(Fig. 3) and represented a 20–25% increase in total cell numbersin marrow at each time point. This finding translated into in-creased numbers of cell subclasses in HFD mice compared withcontrols (Fig. 4). In particular, total marrow lymphocyte numbershowed the largest increases in total numbers in HFD mice, withincreases of ∼40% at day 90 and thereafter (Fig. 4A). Marrowgranulocytes and monocytes were elevated in total number inHFD mice but were not significant except for at the earliest timepoint (day 33) (Fig. 4 B and C). Of note, Fig. 3 shows a drop inmarrow cell numbers in lean mice between days 33 and 60. Thisdrop is believed to be an artifact of the experimental model,

Fig. 1. Characteristics of mice fed HFD. Mice fed a 45% mixed fat diet(white bars) consumed similar amounts of food in grams as mice fed a 10%fat diet (black bars) but ingested significantly more calories (A). Data shownare average food consumption per day for the entire 180-d experiment. (B)Mice fed a 45% fat diet (○) gained significantly more weight than mice feda 10% fat diet (●). (C) Serum leptin in mice fed a 45% fat diet (white bars)was significantly elevated compared with mice fed a 10% fat diet (blackbars; n = 7–8 mice per group). **P < 0.01.

Fig. 2. Distribution of bone marrow hematopoietic cell subpopulations inmice fed 10% or 45% fat diet. Bone marrow cells were labeled with anti-CD31 (clone ER-MP12) and anti–Ly-6C (clone ER-MP20), with cell sub-populations determined as described in Fig. S2, including specific markers toconfirm the identity and size of each marrow subpopulation. Data show thepercent of total nucleated marrow cells for each cell type at various timesafter the start of diet. n = 7–8 mice per group. *P < 0.05, **P < 0.01.

Trottier et al. PNAS | May 15, 2012 | vol. 109 | no. 20 | 7623

IMMUNOLO

GY

INAUGURA

LART

ICLE

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

020

because modest variations in cell counts are often observed be-tween different groups of mice.The observed increase in total numbers of bone marrow cells

represents a very large overall increase given the large size of theskeletal mass in the body. Similar results were observed when theB-cell subpopulations of marrow were analyzed. Increased totalmarrow cell number translated to significantly increased totalnumbers of large and small pre-B cells, immature B cells, andmature B cells (Table S1).

Effect of HFD on Thymus and Thymocytes. The thymus is oftena bellwether of immunological changes, altering its size in responseto various stimuli, including stress, infection, nutritional status, etc.(27–29), along with progressive atrophy with aging in both humansand mice (30, 31). Because lymphocyte numbers were significantlyelevated in the bone marrow of HFDmice, the state of the thymuswas also analyzed. As can be seen in Fig. 5A, thymuses in HFDmice were 30–50% larger in size (wet weight) than thymuses fromcontrols starting at 120 d on diet, and they remained significantlylarger for more than 180 d. This size difference was caused by anactual increase in the overall size of thymuses in HFDmice ratherthan any significant decrease in the size of thymus because of agingamong control mice.Representative time points were chosen for analysis of the

thymus for total numbers of cells. As expected based on the lackof change in thymus size up to day 90, no difference was ob-served in total thymocyte number between HFD mice and con-trols. However, by days 142, 162, and 184, when thymuses weresignificantly larger in HFD mice, thymic total cell counts werealso 30%, 42%, and 55% higher, respectively (Fig. 5B). Thus,increased thymus size in HFD mice resulted in increased totalthymocyte numbers. This finding was also a rather remarkableincrease from normal, and it indicates that there is active pro-motion of lymphopoiesis in both primary tissues.Flow cytometry was then used to determine cellular compo-

sition in HFD and control mice. Representative time points wereused, namely days 33, 60, and 90 for early time points and day162 for times when thymus size was larger and bone marrow cellcounts were increased in HFD mice compared with controls. Thedistribution of cells in thymuses from control and HFD mice wasnot significantly different when analyzing CD4/CD8-labeledsubpopulations at any time point tested, including when thethymuses from HFD mice were substantially increased in size(Fig. 6 A and B). Thus, HFD did not significantly alter the dis-tribution of CD4/CD8 cells in mice to include CD4+, CD8+, andCD4+CD8+ subsets.Double negative (CD4−/CD8−) thymocytes can be divided into

four distinct subsets based on CD44 and CD25 expression(named DN1–4, respectively), which comprise sequential early

T-cell precursors that make up a minority of the total thymocytepopulation. The proportion of double negative subpopulations ofthymocytes was unchanged in HFD mice throughout the exper-iment, with the exception of modest changes at day 162 when theproportion of DN3 cells was reduced ∼15%, whereas the pro-portion of DN4 cells was increased by ∼14% (Fig. S3A). Theproportion of DN1 and DN2 subpopulations was not altered byHFD at any time point. (Fig. S3A). These results indicate thatonly very modest alterations in the proportion of thymocytesubclasses were caused by HFD.When total numbers of these thymocyte subpopulations were

analyzed, no significant differences were observed between HFDand control mice up to day 90 on diet. On day 162, when thymussize was larger in HFD mice, double positive (CD4+/CD8+) anddouble negative (CD4−/CD8−) thymocytes both exhibited an∼35% increase in total numbers in HFD mice compared withcontrols (Fig. 6 C and D). Total numbers of mature CD4+ andCD8+ T cells were not elevated in HFD mice compared withcontrols at any time point (Fig. 6E).

Fig. 3. Increased numbers of nucleated cells in bone marrow of obese mice.Mice fed the 45% fat diet (○) showed increased bone marrow cell numberscompared with mice fed the 10% fat diet (●). Data shown are the totalnumbers of nucleated marrow cells per two femurs. n = 7–8 mice per group.*P < 0.05, **P < 0.01.

Fig. 4. Total numbers of different bone marrow cell subclasses. Mice feda 45% fat diet (white bars) had increased numbers of lymphocytes (CD31+/Ly-6C−/B220+) in their bone marrow compared with mice fed a 10% fat diet(black bars) starting at day 90 on diet (A). Total numbers of (B) marrowgranulocytes (CD31−/Ly-6C+/Ly-6G+/Gr-1+) and (C) monocytes (CD31+/Ly-6Chi/Ly-6G−/Gr-1+) were only modestly affected. Data represent total numbers ofeach marrow cell type in two femurs. n = 7–8 mice per group. *P < 0.05.

7624 | www.pnas.org/cgi/doi/10.1073/pnas.1205129109 Trottier et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

020

The total cell counts of double negative thymocyte sub-populations DN1, DN2, and DN4 were also significantly in-creased in HFDmice by 24%, 31%, and 53%, respectively, at thistime point (day 162) (Fig. S3B), whereas the numbers of DN3thymocytes remained unchanged. These results indicate increasesin early stages of T-lymphopoiesis and also fit with the increasedtotal numbers of B cells and their progenitors in marrow. To-gether, these results show a general increase in lymphopoiesis inprimary immune tissues in diet-induced obese mice.

Changes in Leukocytes in Blood Because of HFD. To determine ifincreased cell numbers in bone marrow and thymus of HFD miceaffected peripheral sites, blood was collected from mice at twodifferent time points (day 142 and day 171); at this time, thymuseswere enlarged, and bone marrow cell counts were elevated inobese mice. Analysis of complete blood counts revealed 70–125%increases in total white blood cell counts at both time points aswell as 75–120% increases in total blood lymphocyte number(Table 1). In addition, on day 171 whenWBC numbers more thandoubled, a threefold increase in blood neutrophil numbers wasalso observed. These results indicate that the changes occurring inthe bone marrow and thymus of HFD mice are also manifest inblood, particularly with the observed increases in lymphopoiesisin HFD mice.

Expression of Growth Factors and Inflammatory Molecules in BoneMarrow. The reason for increased cell counts in bone marrow ofHFD mice is not known, but it may have involved increasedproduction of growth factors, cytokines, and proinflammatorymolecules that have been shown to have a role in modulatinghematopoiesis. To assess this possibility, quantitative RT-PCRwas performed onRNAof bonemarrow cells from lean and obesemice to determine relative mRNA expression levels of a variety of

growth factors and proinflammatory molecules as well as thehormone leptin. Stem cell factor was investigated, because it playsa key role in maintenance of hematopoiesis and expansion of avariety of hematopoietic progenitors (32). Specific growth factorstested included the colony stimulating factors granulocyte colony-stimulating factor (G-CSF), macrophage colony-stimulating fac-tor (M-CSF), and granulocyte macrophage colony-stimulatingfactor (GM-CSF), which are known to play a role in the differ-entiation and proliferation of granulocyte progenitors, monocyteprogenitors, and granulocyte–monocyte progenitors, respectively(33, 34). The expression of IL-7 mRNA was investigated becauseof its role in the development and proliferation of marrow lym-phocytes and their progenitors, especially precursor B and T cells(35). Leptin was also a focus of this gene expression analysis be-cause of its known role in hematopoiesis, especially lymphopoiesisand myelopoiesis (16, 36). In addition, the expression of in-flammatory factors IL-6, IL-1β, and TNF-α was investigated be-cause chronic inflammation is a characteristic of obesity, andsome inflammatory molecules have been implicated in increasedgranulopoiesis andmonopoiesis (6, 28). The expression of patternrecognition receptor Toll-like receptor 4 (TLR4), which plays arole in bacterial endotoxin signaling, was investigated, becauseTLR signaling has been shown to play a role in myeloid cell dif-ferentiation (37). When isolated bone marrow cells were used assource for total RNA, no significant increases were observed inthe expression of mRNA of any of the factors tested in HFDmice

Fig. 5. Thymus size and cell number were elevated in HFD mice. Mice feda 45% fat diet (○) had increased thymus size (A) and total thymocytenumber (B) beyond 120 d on diet compared with mice fed a 10% fat diet (●).n = 7–8 mice per group. *P < 0.05.

Fig. 6. Thymus cellular composition and total numbers in HFD mice. Thy-muses from mice fed a 45% fat diet (○) were analyzed by flow cytometry forthymus cell subpopulations and compared with mice fed a 10% fat diet (●).The composition of thymocyte subclasses as defined by CD4 and CD8 expres-sion was not significantly altered by obesity (A and B). When total numbers ofthymocytes were assessed, significant increases in (C) DN (CD4−/CD8−) and (D)DP (CD4+/CD8+) subtypes were observed at day 162 on diet, when thymus sizewas significantly larger in obese mice compared with controls. No significantalterations were seen in total numbers of mature single positive CD4+/CD8−

and CD4−/CD8+ thymocytes (E). n = 7–8 mice per group. *P < 0.05.

Trottier et al. PNAS | May 15, 2012 | vol. 109 | no. 20 | 7625

IMMUNOLO

GY

INAUGURA

LART

ICLE

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

020

compared with controls (Fig. 7). Leptin mRNA was not detectedin isolated marrow cells. Because isolation of bone marrow cellsfrom bone requires removal of red blood cells and other pro-cessing, adipocytes within the bone marrow do not survive theharvesting process. To assess the expression of mRNA from bonemarrow containing adipocytes, whole bones (femurs) from con-trol and HFDmice were processed for isolation of RNA, and RT-PCR was performed. Similar to what was observed in isolatedbone marrow cells, growth factor and proinflammatory mRNAexpression were not changed in whole-boneRNA fromHFDmicecompared with control mice. In contrast, a more than eightfoldincrease in leptin mRNA expression was observed in HFD micecompared with controls. These results suggest that leptin playsa role in promoting hematopoiesis in the absence of elevatedexpression of other growth factors and cytokines.

Analysis of Bone Marrow Adipocyte Number. Bone marrow is alsoknown to contain adipocytes that are the primary producers ofleptin. Total RNA from whole bone of HFD mice showed aneightfold increase in leptin mRNA expression compared withwhole bone from control mice, whereas isolated marrow cellsshowed no detectable leptin mRNA expression; therefore, it washypothesized that an increase in the number of marrow adipo-cytes might be the cause of the increase in leptin mRNA ex-pression observed in bones of HFD mice. To assess the numbersof adipocytes in bone marrow, femurs were removed and pro-cessed for sectioning and immunohistochemical staining, which

allows visual identification of adipocyte outlines or ghosts. Theadipocytes in marrow were quantified by visual counting. Fig. 8shows the average number of adipocytes counted per femursection from HFD and control mice. As can be seen in Fig. 8,a significant increase in adipocyte number is seen in HFD micecompared with controls—sixfold increase at day 162 and three-fold increase at day 176, indicating that increased leptin mRNAexpression in marrow of HFD mice probably resulted from in-creased adipocyte numbers.

DiscussionThe collective data herein clearly show that diet-induced obesityin mice substantially increased the number of nucleated cells in themarrow and thymus for the 180-d period studied. This finding aloneis in marked contrast to the effects of aging, inflammatory boweldisease, stress, etc., that cause depletions in the primary immunetissues (3, 38, 39). It suggests a complex interplay among themetabolic factors, cytokines, and adipokines that accompany obe-sity, and therefore, the end result is a promotion of lymphopoiesisand myelopoiesis. Because the marrow is located throughout theskeletal mass of the body, the observed 20–25% increase inmarrownucleated cells in HFDmice by 90 d is a noteworthy rise in cellularmass (Fig. 3). Within the marrow, total cell numbers of the eryth-rocytic, monocytic, and granulocytic lineages increased, but pro-portions were unchanged (Figs. 2 and 3). Only cells of thelymphocytic lineages increased 10–18% above normal (Fig. 2). Thistrend was mimicked by the greatly enlarged weight of the thymusof HFD mice (30–50% increase), which included increases in thenumber of CD4+CD8+ thymocytes andmature single positive (SP)

Table 1. Blood differential counts in mice fed HFD

Diet Days on diet WBC (number of cells)* Neutrophils Lymphocytes Monocytes

10% fat 142 4.7 ± 1.2 0.70 ± 0.23 3.8 ± 0.94 0.15 ± 0.0545% fat 142 7.9 ± 1.3† 0.86 ± 0.28 6.7 ± 1.2† 0.26 ± 0.1510% fat 171 6.1 ± 2.2 0.90 ± 0.39 4.8 ± 1.7 0.45 ± 0.2845% fat 171 13.7 ± 1.5‡ 2.7 ± 0.63‡ 10.5 ± 1.8† 0.45 ± 0.26

n = 5 at each time point and diet group.*Data shown are number of cells (×103) per microliter blood ± SD.†P < 0.01.‡P < 0.001.

Fig. 7. Analysis of expression of various genes in isolated bone marrow cellsthat would not contain adipocytes because of processing method and wholebone that would contain adipocytes from HFD mice. Total RNA was isolatedbone marrow cells (white bars) and whole bone (black bars) and subjected toRT-PCR for relative quantitation of mRNA expression. Data shown representexpression of each mRNA in obese mice relative to expression in control leanmice. Data represent the average of two experiments for bone marrow andwhole bone. For marrow, cells from days 142 and 165 on diet were used. Forwhole bone, femurs from days 157 and 165 were used. n = 8 per group ateach time point. –, not detected.

Fig. 8. Quantification of adipocytes in bone marrow of HFD mice. Femursfrom mice fed 10% or 45% fat diet were sectioned and stained with H&E forvisualization of adipocytes. Adipocyte number was determined by countingthe number of adipocytes per femur section using a light microscope (twosections per animal). Each circle represents one mouse, and black bars in-dicate mean for that group. n = 7–8 mice per group. *P < 0.05.

7626 | www.pnas.org/cgi/doi/10.1073/pnas.1205129109 Trottier et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

020

cells compared with controls (Figs. 5 and 6). In sum, these sub-stantial changes in the primary immune tissues add to the growinglist of immune dysregulation known to be associated with obesity.Thus, the question arises as to why obesity, with its numerous

metabolic changes and presumed state of inflammation, wouldpromote hematopoiesis and enlarge the marrow and thymus. Amundane explanation would be that a 40% increase in bodyweight in HFD mice by day 90, rising to a 50–60% increasebeyond day 120, requires a larger blood and lymphatic volume tosupport this extra weight (Fig. 1). This result, no doubt, is part ofthe needed change, but a substantial amount of the excess weightfound in obese mice includes epididymal fat, which was fromtwo- to fourfold greater than the amount found over time in thelean controls (Fig. S1). In Fig. S1 as well as Figs. 1 and 4, one cansee that control mice also underwent modest increases over timein body weight and adipose tissue weight as well. However, theseincreases in lean mice did not translate to increases in the totalnumber of lymphocyte, monocyte, or granulocyte precursors intheir marrow. Moreover, over time, there was an actual declinein the lymphocyte lineages in the marrow and thymus of leanmice—the very lineages that increased the most in HFD mice(Figs. 4 and 5). Finally, the distribution of WBC was changed inHFD mice (Table 1), being almost double the normal number inmice on day 171 with double the proportion of lymphocytes.These increases are out of proportion, being beyond what mightbe needed for an expanding blood volume, and they correlatewith observed increases in WBC in obese human subjects, in-cluding the morbidly obese (23, 24).Another explanation for the cellular increase in the primary

tissues may be the two- and fourfold increases in adipose tissuethat include substantial increases in adipocytes, which are theprimary producers of leptin. Fig. 1C shows a fivefold increase inserum leptin in HFD mice over controls, with up to 100 ng/mLbeing similar to the increase noted in obese humans. Moreover,there were also three- and sixfold increases in the adipocytesfound in the marrow itself by day 162 (Fig. 8). This findingtranslated into a more than eightfold increase in the expressionof leptin mRNA in the bones of HFD mice by day 162 comparedwith control mice (Fig. 7). Whereas leptin was initially known forits role in controlling food intake, it has been shown to havemany immunological properties, with many cells of the immunesystem bearing leptin receptors.Recent studies have shown that, in the obese ob/ob mouse,

leptin promoted hematopoietic processes to include lympho-poiesis and myelopoiesis when recombinant exogenous leptinwas used to rejuvenate the depleted primary tissues of the leptin-deficient ob/ob mice (16). The ob/ob mice produce a non-functioning leptin and despite their large size, had atrophiedthymuses and depleted hematopoietic compartments (16, 28).Compared with same-aged lean controls, the lymphocytic line-ages were depleted by 60%, with 40% and 25% depletions notedin the granulocytic and monocytic compartments, respectively(16, 28); 7 d of provision of recombinant leptin nearly restoredthe monocytic and granulocytic marrow compartments, whereas12 d of supplement were required to rejuvenate lymphopoiesis.Others have noted accelerated thymopoiesis in ob/ob mice pro-vided leptin (28). Moreover, the provision of exogenous leptinprotected mice from thymic atrophy when subjected to starvationor glucocorticoid treatment (40, 41).Althoughmanymature cells of the immune system are known to

have leptin receptors, there is disagreement over whether thy-mocytes have significant numbers of them (42, 43). Studies ofwhole marrow indicate the presence of leptin receptors, but reli-able quantitative information on the expression of these receptorson early lineages of cells is not yet available. Even less informationis available on whether early cell lineages become leptin-resistant.Interestingly, neutrophils frommorbidly obese subjects adequatelyperformed key functions, such as phagocytosis, chemotaxis, and

production of superoxide anion, despite being in an environmentwith high levels of leptin and inflammatory cytokines (25). Howhigh levels of leptin impact immune defense in the obese may wellvary among classes of cells.Interestingly, the enhanced number of WBC precursors in

marrow and mature versions circulating in the blood does nottranslate into better immune defense. Whereas it was recentlyfound that neutrophils even in morbidly obese patients couldreadily carry out phagocytosis, chemotaxis, and production ofsuperoxide (25), this finding is not normal for obese subjects.More infections, poorer responses to influenza vaccination, and in-creased production of inflammatory cytokines by T cells and mac-rophages have been noted in obese humans and mouse models(14, 44). Thus, enhanced numbers of hematopoietic and thymiccells could be benign, or they could contribute to increased mal-function and inflammation. How these observations fit into the bigpicture of immune status in obesity awaits additional under-standing of the status of the immune system as a whole.One emerging problem is the variance in outcomes from

laboratories essentially testing the same immune functions,which is clearly noted in a recent review (14). Within the humanpopulation, not enough consideration is given to how variance inbody mass index, diabetic status, age, and sex may impact out-comes. With regard to mouse models, there is growing evidencethat the feeding of 45% fat vs. 60% fat generates variances inoutcomes. It is well-known that a 60% fat diet creates an in-tensified state of hyperlipidemia, causing a much higher degreeof fatty liver compared with a 45% fat diet (45). In addition,discrepancies between 45% and 60% fat diets have been ob-served in mice when it comes to weight gain, insulin resistance,and fat mass in various mouse models (21, 46, 47). There is ev-idence that 60% fat diets may also alter observed immune out-comes as well. In experiments where control C57BL/6 mice werefed chow (soy- or alfalfa-based) with 4.5% fat and others werefed a casein-based dietary formulation containing 60% fat, obesemice after 13 mo (390 d) on the diet showed an increase inperithymic fat, with reduced numbers of thymocyte subsets andreduced numbers of peripheral naïve T cells (48). These changeswere not noted at 3 mo (90 d) like in our study (Fig. 5). Whetherthe subsequent decline in thymic size and cell numbers wasa product of aging, the HFD, or both is not clear (48). It, how-ever, might be of benefit to the field if the 45% fat diet was usedwith the complementary low-fat control diet (both with casein asprotein source) to provide more unified data to generate a modelmore analogous to high-fat human nutrition.In sum, obesity based on the DIO mouse model used herein

seems to be among the growing list of diseases and conditionsthat alter marrow and thymic function. Given the level ofproinflammatory cytokines and metabolic changes that accom-pany obesity, it was somewhat surprising to find enhancednumbers of WBC precursors in marrow and thymus and WBC inblood of HFD mice. Although not the only factor, the largeincreases in leptin in the blood and bone have a high probabilityof contributing to the observed results. It is acknowledged thatleptin may not have been the only contributing factor and maysimply stand in for other factors that accelerated lymphopoiesisand myelopoiesis. Additional experiments are required to ad-dress this matter and determine if there are any deleteriouseffects on the host created by the increased output of the marrowand thymus.

Materials and MethodsMice and Diet. Male C57BL/6 mice (4 wk) were purchased from Jackson Labsand maintained according to guidelines approved by the University Labo-ratory Animal Research Committee at Michigan State University in a facilitymaintained at 25 °C with 12-h light and dark cycles. Starting at ages 6–7 wk,weight-matched mice were fed either a low-fat diet (control; 10% mixedfats) or HFD (45% mixed fats) ad libitum. Diets were purchased from

Trottier et al. PNAS | May 15, 2012 | vol. 109 | no. 20 | 7627

IMMUNOLO

GY

INAUGURA

LART

ICLE

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

020

Research Diets (10% mixed fats Diet D12450B and 45% mixed fats DietD12451). Food was weighed, and diet was replaced two times per week.Mice were weighed one time per week.

Harvesting and Processing of Tissues. Mice were anesthetized with isoflurane,blood was collected by cardiac puncture, and serum was collected for ad-ditional analysis. In some experiments, blood was transferred to EDTAmicrotainers (BD Biosciences) for analysis of total blood cell counts. Bonemarrow was harvested from both femurs of each mouse into Harvest Buffer(HBSS; Invitrogen), 10 mM Hepes, 4 mM sodium bicarbonate, and 4% heat-inactivated FBS (pH 7.2; Atlanta Biologicals) as described previously (42). Afterred blood cell lysis, bone marrow cells were resuspended in Label Buffer(HBSS, 10 mM Hepes, 4 mM sodium bicarbonate, 2% heat-inactivated FBS,0.15% sodium azide, pH 7.2) for counting and flow cytometry analysis.Thymuses were removed, weighed, and processed by mincing and passingthrough a 100-gauge stainless steel screen. Thymocytes were washed andresuspended in Label Buffer for counting and flow cytometry analysis aspreviously described (10). Epididymal fat was removed and weighed at eachtime point.

Determination of Serum Leptin. Serum leptin was determined using themouseLeptin Quantikine ELISA kit (R&D Systems).

Flow Cytometry. All antibodieswere fromeBiosciences exceptwhere indicated.For analysis of major classes of cells within the bone marrow, the antibodiesbiotinylated rat anti-CD31 (clone ER-MP12; AbD Serotec) and FITC-conjugatedrat anti–Ly-6C (clone ER-MP20; AbD Serotec) were used. Five bone marrowsubpopulations were identified: erythrocytes (CD31−/Ly-6C−), lymphocytes(CD31+/Ly-6C−), granulocytes (CD31−/Ly-6Cmed), monocytes (CD31+/−/Ly-6Chi),and mixed progenitors (CD31+/Ly-6Cmed). To verify bone marrow subpop-ulations, antibodies allophycocyanin (APC)-conjugated Ter119, R-phycoery-thrin (PE)-conjugated rat anti-CD45R/B220, PE-conjugated rat anti–Ly-6G, andPE-conjugated rat anti–Ly-6G and Ly-6C (Gr-1) were used to identify eryth-rocytes, lymphocytes, and myeloid cells, respectively, as described previously(8, 26) (Fig. S2).

Composition of B-cell lineages in the marrow was analyzed using CD43-FITC, CD45R-PerCP-Cy5.5, IgM-PE, and biotinylated IgD detected withstreptavidin-PECy5 as previously described (8, 49). Gr-1–APC was used todetect and eliminate myeloid cells from analysis. Single-color positive andthree-color isotype samples were used for fluorochrome spectral compen-sation and correction for false-positive staining. All B220+, Gr-1− cells weregated as B lineage-restricted cells. Pro-B cells were defined as CD43highIgM−,Pre-B cells were defined as CD43neg-lowIgM−, and immature–mature B cellswere defined as CD43neg-lowIgM+. CD19 was used to distinguish betweenPre–Pro-B (CD19−) and late Pro-B cells (CD19+). Forward scatter (FSC) wasused to distinguish between early and late Pre-B cells. IgD was used to dis-tinguish between immature (IgD−) and mature B cells (IgD+) (8, 49).

For labeling, bone marrow cells (1 × 106) in Label Buffer were incubatedwith antibody for 25 min on ice. Biotinylated antibodies were detected bythe addition of streptavidin–PE-Cy5 conjugate in a second incubation step.

Thymocytes (1 × 106) in Label Buffer were labeled with anti-CD4 (PE-Cy5–conjugated) and anti-CD8 (FITC-conjugated) to determine major subpopu-lations of thymocytes. To determine subsets of double negative (DN; CD4−/CD8−) thymocytes, cells were labeled with anti-CD25 (PE-Cy5–conjugated),anti-CD44 (FITC-conjugated), c-kit (PE-conjugated), anti-CD4 (PE-Cy7–conju-gated), and anti-CD8–conjugated (PE-Cy7) antibodies. Cells were gated onCD4−/CD8− cells and delineated as follows: DN1, CD44+/CD25−/c-kit+; DN2,CD44+/CD25+/c-kit+; DN3, CD44−/CD25−/c-kit−; DN4, CD44−/CD25+/c-kit−.

Isotype control antibodies used were PE-conjugated rat IgG2a, bio-tinylated rat IgG2a, FITC-conjugated rat IgG2a, biotinylated rat IgG2b, PE-conjugated rat IgG2b, and APC-conjugated rat IgG2b.

After labeling, cells were kept in 1.5% formaldehyde in PBS (pH 7.2–7.4)until ready for analysis. Flow cytometry was performed on either an FACSVantage flow cytometer or an LSRII flow cytometer (BD) using a 488-nmlaser. Data were analyzed using WinList 6 software (Verity Software House).

Isolation of Whole-Bone RNA, Bone Marrow RNA, and Quantitative RT-PCR.Isolation of bone marrow cells results in the total loss of marrow adipocytes.Because adipocytes in marrow express certain genes that may play a role inhematopoiesis, total RNA was isolated from both whole bone (with adipo-cytes) and marrow cells (without adipocytes). Mouse femurs were snap-frozen and crushed, and cells were lysed in TRIzol Reagent (Invitrogen).Separately isolated bone marrow cells were lysed in TRIzol Reagent, withtotal RNA isolated according to the manufacturer’s recommendations, andthey were further purified using the RNeasy Mini Kit (Qiagen) contaminat-ing genomic DNA removed by RNase-free DNase set (Qiagen) for on-columnDNase digestion according to the manufacturer’s recommendations.

Reverse transcription of total bone marrow RNA was performed using theAffinityScript QPCR cDNA Synthesis kit (Agilent), and quantitative PCR wasperformed using the Brilliant II QPCR Master Mix (Agilent) according to themanufacturer’s recommendations. Primers are described in Table S2 andwere verified to have efficiency of >90% as estimated by a standard curve ofpurified amplicon. Analysis of quantitative PCR data was done using the−ΔΔCT method of comparison.

Immunohistochemical Staining of Bone. Femurs were fixed in 10% formalin for24 h and then transferred to 70% ethanol. Fixed samples were decalcified inEDTA, embedded in paraffin, and sectioned at 5 μm. Slides were stained withH&E, and adipocytes were enumerated by visual inspection.

Statistics. Data were analyzed using the Student t test. Statistical significancewas set at P < 0.05. Unless otherwise stated, means ± SEM are reported. Thenumber of mice per treatment group was n = 7–8 unless otherwise stated.All experiments were performed two or more times unless noted.

ACKNOWLEDGMENTS. We thank Gerald Dekker, Brent Campbell, and ZacharyPena for assistance with mouse feeding and weighing. This work was sup-ported by the College of Natural Science and AgBioResearch at MichiganState University.

1. Hardy RR, et al. (2000) B-cell commitment, development and selection. Immunol Rev

175:23–32.2. Rolink AG, Schaniel C, Andersson J, Melchers F (2001) Selection events operating at

various stages in B cell development. Curr Opin Immunol 13:202–207.3. Fraker PJ, King LE (2004) Reprogramming of the immune system during zinc de-

ficiency. Annu Rev Nutr 24:277–298.4. Lodish H, Flygare J, Chou S (2010) From stem cell to erythroblast: Regulation of red

cell production at multiple levels by multiple hormones. IUBMB Life 62:492–496.5. Ueda Y, Kondo M, Kelsoe G (2005) Inflammation and the reciprocal production of

granulocytes and lymphocytes in bone marrow. J Exp Med 201:1771–1780.6. Ueda Y, Cain DW, Kuraoka M, Kondo M, Kelsoe G (2009) IL-1R type I-dependent

hemopoietic stem cell proliferation is necessary for inflammatory granulopoiesis and

reactive neutrophilia. J Immunol 182:6477–6484.7. Fraker P, King L (1998) Changes in regulation of lymphopoiesis and myelopoiesis in

the zinc-deficient mouse. Nutr Rev 56:S65–S69.8. King LE, Fraker PJ (2002) Zinc deficiency in mice alters myelopoiesis and hematopoi-

esis. J Nutr 132:3301–3307.9. Trottier MD, Newsted MM, King LE, Fraker PJ (2008) Natural glucocorticoids induce

expansion of all developmental stages of murine bone marrow granulocytes without

inhibiting function. Proc Natl Acad Sci USA 105:2028–2033.10. King LE, Osati-Ashtiani F, Fraker PJ (2002) Apoptosis plays a distinct role in the loss of

precursor lymphocytes during zinc deficiency in mice. J Nutr 132:974–979.11. Hotamisligil GS, Shargill NS, Spiegelman BM (1993) Adipose expression of tumor

necrosis factor-alpha: Direct role in obesity-linked insulin resistance. Science 259:

87–91.

12. Kern PA, Ranganathan S, Li C, Wood L, Ranganathan G (2001) Adipose tissue tumor

necrosis factor and interleukin-6 expression in human obesity and insulin resistance.

Am J Physiol Endocrinol Metab 280:E745–E751.13. Hotamisligil GS (2006) Inflammation and metabolic disorders. Nature 444:860–867.14. Karlsson EA, Beck MA (2010) The burden of obesity on infectious disease. Exp Biol

Med (Maywood) 235:1412–1424.15. Smith AG, Sheridan PA, Harp JB, Beck MA (2007) Diet-induced obese mice have in-

creased mortality and altered immune responses when infected with influenza virus. J

Nutr 137:1236–1243.16. Claycombe K, King LE, Fraker PJ (2008) A role for leptin in sustaining lymphopoiesis

and myelopoiesis. Proc Natl Acad Sci USA 105:2017–2021.17. Faggioni R, et al. (2000) Leptin-deficient (ob/ob) mice are protected from T cell-me-

diated hepatotoxicity: Role of tumor necrosis factor alpha and IL-18. Proc Natl Acad

Sci USA 97:2367–2372.18. Bennett BD, et al. (1996) A role for leptin and its cognate receptor in hematopoiesis.

Curr Biol 6:1170–1180.19. Weisberg SP, et al. (2003) Obesity is associated with macrophage accumulation in

adipose tissue. J Clin Invest 112:1796–1808.20. Xu H, et al. (2003) Chronic inflammation in fat plays a crucial role in the development

of obesity-related insulin resistance. J Clin Invest 112:1821–1830.21. Lumeng CN, Deyoung SM, Bodzin JL, Saltiel AR (2007) Increased inflammatory

properties of adipose tissue macrophages recruited during diet-induced obesity. Di-

abetes 56:16–23.22. Ye J, Gimble JM (2011) Regulation of stem cell differentiation in adipose tissue by

chronic inflammation. Clin Exp Pharmacol Physiol 38:872–878.

7628 | www.pnas.org/cgi/doi/10.1073/pnas.1205129109 Trottier et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

020

23. Nanji AA, Freeman JB (1985) Relationship between body weight and total leukocytecount in morbid obesity. Am J Clin Pathol 84:346–347.

24. Pratley RE, Wilson C, Bogardus C (1995) Relation of the white blood cell count toobesity and insulin resistance: Effect of race and gender. Obes Res 3:563–571.

25. Trottier MD, Naaz A, Kacynski K, Yenumula PR, Fraker PJ (2011) Functional capacity ofneutrophils from class III obese patients. Obesity, 10.1038/oby.2011.354.

26. de Bruijn MF, et al. (1994) Distinct mouse bone marrow macrophage precursorsidentified by differential expression of ER-MP12 and ER-MP20 antigens. Eur J Im-munol 24:2279–2284.

27. Fraker PJ, King LE (2001) A distinct role for apoptosis in the changes in lymphopoiesisand myelopoiesis created by deficiencies in zinc. FASEB J 15:2572–2578.

28. Hick RW, Gruver AL, Ventevogel MS, Haynes BF, Sempowski GD (2006) Leptin selec-tively augments thymopoiesis in leptin deficiency and lipopolysaccharide-inducedthymic atrophy. J Immunol 177:169–176.

29. Wang SD, Huang KJ, Lin YS, Lei HY (1994) Sepsis-induced apoptosis of the thymocytesin mice. J Immunol 152:5014–5021.

30. Dorshkind K, Swain S (2009) Age-associated declines in immune system developmentand function: Causes, consequences, and reversal. Curr Opin Immunol 21:404–407.

31. Nikolich-Zugich J (2008) Ageing and life-long maintenance of T-cell subsets in the faceof latent persistent infections. Nat Rev Immunol 8:512–522.

32. Broudy VC (1997) Stem cell factor and hematopoiesis. Blood 90:1345–1364.33. Hamilton JA (2008) Colony-stimulating factors in inflammation and autoimmunity.

Nat Rev Immunol 8:533–544.34. Burgess AW, Metcalf D (1980) The nature and action of granulocyte-macrophage

colony stimulating factors. Blood 56:947–958.35. von Freeden-Jeffry U, et al. (1995) Lymphopenia in interleukin (IL)-7 gene-deleted

mice identifies IL-7 as a nonredundant cytokine. J Exp Med 181:1519–1526.36. Fantuzzi G, Faggioni R (2000) Leptin in the regulation of immunity, inflammation,

and hematopoiesis. J Leukoc Biol 68:437–446.

37. Nagai Y, et al. (2006) Toll-like receptors on hematopoietic progenitor cells stimulateinnate immune system replenishment. Immunity 24:801–812.

38. Aspinall R, Andrew D (2000) Thymic involution in aging. J Clin Immunol 20:250–256.

39. Scollay RG, Butcher EC, Weissman IL (1980) Thymus cell migration. Quantitative as-pects of cellular traffic from the thymus to the periphery in mice. Eur J Immunol 10:210–218.

40. Howard JK, et al. (1999) Leptin protects mice from starvation-induced lymphoid at-rophy and increases thymic cellularity in ob/ob mice. J Clin Invest 104:1051–1059.

41. Trotter-Mayo RN, Roberts MR (2008) Leptin acts in the periphery to protect thymo-cytes from glucocorticoid-mediated apoptosis in the absence of weight loss. Endo-crinology 149:5209–5218.

42. Gruver AL, Ventevogel MS, Sempowski GD (2009) Leptin receptor is expressed inthymus medulla and leptin protects against thymic remodeling during endotoxemia-induced thymus involution. J Endocrinol 203:75–85.

43. Velloso LA, SavinoW, Mansour E (2009) Leptin action in the thymus. Ann N Y Acad Sci1153:29–34.

44. Karlsson EA, Sheridan PA, Beck MA (2010) Diet-induced obesity impairs the T cellmemory response to influenza virus infection. J Immunol 184:3127–3133.

45. Ha S-K, Chae C (2010) Inducible nitric oxide distribution in the fatty liver of a mousewith high fat diet-induced obesity. Exp Anim 59:595–604.

46. Weisberg SP, et al. (2006) CCR2 modulates inflammatory and metabolic effects ofhigh-fat feeding. J Clin Invest 116:115–124.

47. Jiang L, et al. (2009) Leptin contributes to the adaptive responses of mice to high-fatdiet intake through suppressing the lipogenic pathway. PLoS One 4:e6884.

48. Yang H, et al. (2009) Obesity accelerates thymic aging. Blood 114:3803–3812.49. Hardy RR, Hayakawa K (2001) B cell development pathways. Annu Rev Immunol 19:

595–621.

Trottier et al. PNAS | May 15, 2012 | vol. 109 | no. 20 | 7629

IMMUNOLO

GY

INAUGURA

LART

ICLE

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

020