To assess the postprandial regulation of serum LCN2, we used data from four separate studies where healthy individuals with normal weight, overweight, obesity, and severe obesity were challenged with a meal after an overnight fast. In Study 1 with young healthy lean women (BMI: 21.8 ± 0.6 Kg/m²; Supplementary file 1A), analysis revealed a tendency for increase of circulating LCN2 levels with time (F_{7, 70}=3.07, p=0.065; Figure 1A), although not significant. LCN2 serum concentration at baseline (t = 0 min) trended to differ from the one at t = 30 min (F_{1, 10}=3.8, p=0.080), t = 45 min (F_{1, 10}=4.6, p=0.058), t = 60 min (F_{1, 10}=4.87, p=0.052), and t = 90 min (F_{1, 10}=3.9, p=0.076), a similar magnitude of postprandial upregulation to what was previously reported (Paton et al., 2013). Interestingly, postprandial LCN2 serum levels (mean concentration at each timepoint) were robustly inversely correlated with hunger scores (mean hunger scores at each timepoint) of the challenged subjects (Spearman r = −0.98, p=0.0004) after the consumption of the liquid meal (Figure 1B). Serum LCN2 levels peaked at 45 min after meal ingestion, increasing by 16% (Supplementary file 1A, Figure 1—figure supplement 1, A).

Figure 1 with 3 supplements see all

Download asset Open asset

Figure 1—source data 1

Serum LCN2 levels are postprandially increased in individuals with normal weight and overweight but not in individuals with obesity or with severe obesity.

https://cdn.elifesciences.org/articles/58949/elife-58949-fig1-data1-v1.xlsx

Download elife-58949-fig1-data1-v1.xlsx

Similarly, postprandial circulating LCN2 levels were significantly altered with time (F_{2, 16}=27.87, p=0.002) in a separate, second study of young healthy lean women (BMI: 20.8 ± 0.5 Kg/m²; Supplementary file 1A). Specifically, serum LCN2 concentration at t = 60 min (F_{1, 8}=59.64, p=0.002) and t = 105 min (F_{1, 8}=15.36, p=0.009; Figure 1C) were significantly increased from baseline. Here serum LCN2 levels peaked at 60 min increasing by 54% (Supplementary file 1A, Figure 1—figure supplement 1, G).

The third study included 47 subjects, 28 women and 19 men, with overweight and/or obesity (BMI: 28.7 ± 0.5 Kg/m²). The whole cohort consisted of 30 subjects (18 women and 12 men) with overweight (BMI = 26.4 ± 0.3 Kg/m²) and 17 subjects (10 women and 7 men) with obesity (BMI = 32.7 ± 0.4 Kg/m²). Contrary to lean groups, LCN2 significantly decreased after the meal challenge (F_{5, 215}=2.61, p=0.026; Figure 1D).

Interestingly, based on their postprandial LCN2 response, this initial cohort could be divided into two subgroups (Figure 1E). The first group (n = 25) included responders (R), that is, subjects that had a ‘positive’ postprandial LCN2 response with elevated LCN2 levels in multiple timepoints after the meal. The second group (n = 22) consisted of non-responders (NR), that is, subjects that had a ‘negative’ postprandial LCN2 response with decreased LCN2 levels after the meal challenge.

Responders showed a trend, though not statistically significant, toward a 12% increase in serum LCN2 levels 60 min after meal consumption (F_{1, 21}=3, 24, p=0.086; Figure 1E; Supplementary file 1B; Figure 1—figure supplement 2A). The inverse correlation between postprandial LCN2 serum levels and hunger scores was attenuated compared to that of Study 1 (Spearman r = −0.66, p=0.33; Figure 1F). On the other hand, non-responders showed decreased LCN2 levels at all timepoints examined, reaching a nadir 60 min after the meal, with a 19% reduction (F_1,21=37.08, p<0.0001; Supplementary file 1B). Non-responders trended to have a significantly larger waist circumference (Supplementary file 1B). Higher values for BMI, body fat, diastolic blood pressure, and fasting serum glucose and LCN2 levels were also observed in the non-responders but did not reach statistical significance (Supplementary file 1B).

When the 47-subject mixed cohort was analyzed by sex, subjects could again be divided into responders and non-responders, based on their postprandial LCN2 response. Women (Figure 1G; Figure 1—figure supplement 2, B) and men (Figure 1J; Figure 1—figure supplement 2, C) responders showed a trend, though not statistically significant, toward a 10% and 15% increase in serum LCN2 levels 60 min after meal consumption, respectively (Supplementary file 1B). The inverse correlation between postprandial LCN2 serum levels and hunger sensation found in women with normal weight (in Study 1), was not present in these overweight/obese groups of women (Figure 1H; Figure 1—figure supplement 2, E) or men (Figure 1K; Figure 1—figure supplement 2, F). On the other hand, non-responders showed a significant LCN2 decrease postprandially at all timepoints examined, reaching a nadir at 60 min for women (F_1,12=36.9, p<0.0001; Figure 1G) and at 90 min for men (F_1,8=25.21, p=0.001; Figure 1J).

Whereas women NRs had significantly higher waist circumference (Figure 1I) and showed a trend toward higher BMI, body fat, serum glucose, diastolic blood pressure (Supplementary file 1B), men NRs did not show any major differences in BMI, waist circumference (Figure 1L) or any other parameters (Supplementary file 1B).

The Study 4 included individuals with severe obesity, studied before and after Roux-en-Y gastric bypass surgery. The initial BMI of 47.4 ± 1.9 kg/m² was reduced to 29.6 ± 1.8 kg/m² one year after the surgery (Stano et al., 2017). Baseline fasting LCN2 levels were marginally decreased after surgery (Supplementary file 1C, Figure 1—figure supplement 3A,C). Postprandial levels of serum LCN2 were rather decreased before surgery (F_{10, 109}=1.4, p=0.253) and trended to be significantly increased after surgery (F_{10, 107}=1.97, p=0.079), suggesting re-sensitization of these subjects after normalization of BMI. Similar to the overweight and obese non-responders of the previous study, pre-surgery postprandial circulating levels of LCN2 showed a 19% decrease from baseline at 90 min (F_1,11=6.54, p=0.026) after the ingestion of the liquid meal (Supplementary file 1C). Interestingly, post-surgery, postprandial concentrations of LCN2 changed to the opposite direction showing a 42% increase at 15 min (F_{1, 10}=7.54, p=0.023) and a trend, though not statistically significant, toward 59% increase from baseline at 90 min (F_{1, 11}=4.32, p=0.065) after ingestion of the meal (Supplementary file 1C). Furthermore, Roux-en-Y gastric bypass significantly affected (F_{1, 220}=5.89, p=0.024) the observed difference in LCN2 levels at 90 min before and after surgery (Figure 1M). Of note, while postprandial LCN2 concentration did not correlate with hunger score before surgery (Spearman r = −0.18, p=0.64; Figure 1N; Figure 1—figure supplement 3B), there was an association, albeit non-significant after surgery (Spearman r = −0.64, p=0.096; Figure 1O; Figure 1—figure supplement 3D).

In order to place in context the regulation of postprandial LCN2 serum levels and its association with hunger, to those of other feeding-regulating hormones, we measured glucagon-like peptide 1 (GLP-1) and insulin circulating concentrations. In the normal-weighted cohorts (1st and 2nd Study) LCN2 showed a postprandial response similar in magnitude to that of GLP-1 (Figure 2A–D and Figure 1—figure supplement 1A–B, G–H). In both studies circulating insulin showed higher postprandial upregulation than LCN2 (Figure 2A,C and Figure 1—figure supplement 1C,I). However, the total response of LCN2 was significantly lower than GLP-1 (p=0.035; Figure 2B and Figure 1—figure supplement 1A–B) in Study 1, but not in study 2 (p=0.385; Figure 2D and Figure 1—figure supplement 1G–H). In study 1, LCN2 was the postprandial protein with the highest inverse correlation with hunger score (Figure 1—figure supplement 1D–F); GLP-1 was also inversely correlated with hunger, yet less strongly (Figure 1—figure supplement 1E). We did not find any correlation between insulin levels and hunger in this cohort (Figure 1—figure supplement 1F). For this reason, we more closely compared total responses of LCN2 and GLP-1.

Figure 2

Download asset Open asset

Figure 2—source data 1

Similar postprandial regulation of serum LCN2 and GLP-1 levels in subjects with normal weight but not in subjects with overweight or obesity.

https://cdn.elifesciences.org/articles/58949/elife-58949-fig2-data1-v1.xlsx

Download elife-58949-fig2-data1-v1.xlsx

In study 3, the total GLP-1 response was significantly higher than LCN2 (p<0.0001). For consistency purposes, we also analyzed GLP-1 and serum levels in responders and non-responders, although segregation in these two groups was based on LCN2 serum levels. The response of LCN2 was significantly different between responders and non-responders (p=0.014; Figure 2H) and this was more pronounced in females (Figure 2I), than males (Figure 2J). GLP-1 or insulin response between responders and non-responders was not in the opposite direction, as in the case of LCN2 (Figure 2E-G; Figure 1—figure supplement 2, G–I,M–O). Within the responders, an inverse correlation with hunger was present for insulin (Figure 1—figure supplement 2, P–R) but not for LCN2 (Figure 1—figure supplement 2, D–F) or GLP-1 (Figure 1—figure supplement 2, J–L,).

In study 4, the improvement of body weight and BMI after gastric bypass surgery was accompanied by a large increase in postprandial GLP-1 and insulin concentrations and to a lesser extent in LCN2 concentration (Figure 2K-L). GLP-1 response was higher than that of LCN2, both before (pre) and after (post) bariatric surgery (Figure 2K-L). While the GLP-1 response was significantly increased after the surgery, LCN2 only showed a tendency for increase (Figure 2M). GLP-1 and insulin showed a strong inverse correlation with hunger both before (Figure 1—figure supplement 3, F,J, respectively) and after the bariatric surgery (Figure 1—figure supplement 3, H,L, respectively). In contrast to LCN2, which did not correlate with hunger scores before surgery (Figure 1—figure supplement 3, B), tended to inversely correlate with it after the surgery (Figure 1—figure supplement 3, D), although not significantly.

Combined, our studies in humans show a postprandial increase in circulating LCN2 levels in humans with normal weight, which notably correlates with a drop in hunger sensation in the same individuals. Furthermore, subjects with overweight or obesity lose postprandial regulation of LCN2 and this may be a new mechanism of resistance that contributes to obesity.

Next, we examined whether the mechanism of action of LCN2 is conserved in primates. As a first approach we evaluated whether [¹²⁴I] rh-LCN2 crosses the blood-brain barrier in non-human primates. Combined analysis of MRI and PET representative images of vervet monkey brain demonstrated an initial peak of activity throughout the brain during the first 30 s after the end of intravenous administration of [¹²⁴I] rh-LCN2 that is characteristic of BBB permeability. The sagittal, coronal, and axial MRI T1-weighted template images show a volume of interest (VOI) in the anatomical area of the hypothalamus where there is an indication of tracer binding although it may partially be spillover from an adjacent area outside the brain that also shows substantial tracer uptake (Figure 3A-I, Figure 3—figure supplement 1A–M).

Figure 3 with 1 supplement see all

Download asset Open asset

Figure 3—source data 1

LCN2 crosses the blood-brain barrier of vervet monkeys.

https://cdn.elifesciences.org/articles/58949/elife-58949-fig3-data1-v1.xlsx

Download elife-58949-fig3-data1-v1.xlsx

PET acquisition was repeated using a chase/blocking paradigm to determine whether there is specific binding of [¹²⁴I] rh-LCN2 in the hypothalamus. The results from the chase experiment—standard uptake values at every timepoint—were compared to those of the no-chase experiment, by using the same procedure, software and atlas. Infusion of the MC4R receptor ligand, α-MSH, 15 min after [¹²⁴I] rh-LCN2 did not seem to affect the tracer uptake in the thalamic region (Figure 3J and L), whereas it did reduce uptake compared with the no-chase condition in the hypothalamus (Figure 3K and M). We observed a 6.3% and 5.7% difference in the standard uptake value (SUV) in the left and right thalamus respectively, and a 49.8% and 51.2% reduction in the left and right hypothalamus. These results indicate that [¹²⁴I] rh-LCN2 penetrates the BBB and shows specific binding defined by displacement with α-MSH in the hypothalamus but not in the thalamus.

To further prove that LCN2 can bind to the hypothalamic feeding center of primates and to also exclude the possibility, inherent to PET studies, that a spillover signal from outside the brain may confound the findings, we examined LCN2 binding to brain sections where no such potential confounder exists. Rhesus macaque, baboon, and human brain sections containing the hypothalamus were incubated with either [¹²⁵I] rh-LCN2 alone or in the presence of excess unlabeled LCN2 or α-MSH to assess the specificity of binding. [¹²⁵I] rh-LCN2 binding was observed in the hypothalamic area of the baboon (Figure 4A; Figure 4—figure supplement 1B,F) and the rhesus macaque (Figure 4—figure supplement 1, C–D).

Figure 4 with 1 supplement see all

Download asset Open asset

Figure 4—source data 1

LCN2 binds to the hypothalamus of primates.

https://cdn.elifesciences.org/articles/58949/elife-58949-fig4-data1-v1.xlsx

Download elife-58949-fig4-data1-v1.xlsx

In the baboon, the specificity of binding was confirmed by the use of unlabeled LCN2 which blocked part of the [¹²⁵I] rh-LCN2 binding (Figure 4B). Specific binding was observed in the paraventricular nucleus of the hypothalamus (PVN) and both the dorsomedial (DM) and ventrolateral (VL) nuclei of the human hypothalamus, all areas where MC4R is expressed (Figure 4C, and Figure 4—figure supplement 1, A). Unlabeled LCN2 blocked part of the binding of labeled LCN2 (Figure 4E; Figure 4—figure supplement 1, E), indicating specific binding. Similarly, unlabeled a-MSH also blocked some LCN2 binding to the hypothalamus, indicating that LCN2 binds to MC4R (Figure 4D; Figure 4—figure supplement 1, E). That a-MSH blocked less [¹²⁵I] rh-LCN2 binding than non-radiolabeled LCN2 may suggest that, at least in primates [¹²⁵I] rh-LCN2 has a higher binding affinity for MC4R than α-MSH.

To enhance the rigor of the autoradiography experiments, we also examined LCN2 binding using immunofluorescence in baboon (Figure 4F, G) and human brain sections (Figure 4H) containing the hypothalamus. Binding was again shown in both human and baboon brain sections and quantified as the average of LCN2-positive cells (21.3 ± 1.3% and 17.6 ± 1.8%, respectively) and it was specific since it was competed by non-biotinylated LCN2 (Figure 4G, I). The reduction in binding was approximately 75%. If the concentration of the blocking agent is insufficient then the block may be incomplete and explain why 25% nonspecific or non-displaceable binding is observed even when the tracer and blocking drugs are almost the same. It is also possible that a slightest difference in structure may mean differences in nonspecific binding or off-target high-affinity binding (Hamill et al., 2005). Of note, bar graphs were obtained from a single brain section and therefore depict qualitative representations of binding. Overall, we observed a consistent and comparable degree of binding in the hypothalamus, among the three species examined (Figure 4—figure supplement 1A–F), which indicates that the PET findings are evidence of specific binding in the non-human primate hypothalamus and supports the premise of a common interspecies target of action for LCN2.

Having established that rh-LCN2 is able to cross the BBB of vervets and localize to the hypothalamus, we then sought to examine whether a daily treatment of lean monkeys with intravenously administered rh-LCN2 would lead to appetite suppression. As described in Materials and methods, this was a cross-over study with two treatment weeks and one washout period in between (Figure 5A). The LCN2 dose was extrapolated from our studies in mice (Mosialou et al., 2017). In the mouse hypothalamus, the amount of naturally occurring LCN2 is 28 pg/mg and in the adult mouse and human serum, it is on average 100–150 ng/mL. In mice, the administration of LCN2 by intraperitoneal injection of 150 ng/g daily crosses the blood-brain barrier and suppresses appetite. Using interspecies conversion per m², we calculated the monkey dose to be 0.0375 mg/kg. This dose is equivalent to the amount used to treat mice and it is calculated based on body surface area; it takes into account the interspecies variation in several physiological parameters including oxygen utilization, caloric expenditure, basal metabolism, and blood volume (Reagan-Shaw et al., 2008).

Figure 5 with 1 supplement see all

Download asset Open asset

Figure 5—source data 1

Rh-LCN2 administration suppresses food intake in vervets within 5 days of treatment.

https://cdn.elifesciences.org/articles/58949/elife-58949-fig5-data1-v1.xlsx

Download elife-58949-fig5-data1-v1.xlsx

Digestion of an LCN2 aliquot with N-glycanase showed that the recombinant protein is pure and N-glycosylated (Figure 5—figure supplement 1). In addition, each of the non-human primates received a maximum endotoxin of 0.03 EU/kg/hr, a dose well below the endotoxin limit (5 EU/kg/hr) defined as acceptable by the U.S. Food and Drug Administration (Office of Regulatory Affairs, 2014).

During the first treatment week, the LCN2-treated group of monkeys showed a 27% decrease in food intake compared to baseline (food intake before treatment) and a 25% decrease compared to the saline-treated monkeys (Figure 5B). Following the first week of treatment recombinant LCN2 was washed out by allowing the monkeys some days to recover. The following week, baseline food intake was measured (baseline 2). As shown in Figure 5C, both saline and LCN2 groups returned to baseline levels of food intake before the start of the second treatment week, where groups were assigned to the opposite treatment scheme (crossover). Similarly, during the second treatment week, the monkeys that received rh-LCN2 showed a 29% reduction in food intake compared to baseline and a 15% decrease compared to the saline-treated group (Figure 5D). When data from both treatment weeks were combined, the LCN2-treated group had a 28% reduction in food intake compared to baseline and a 21% reduction compared to the saline-treated group (Figure 5E). It should be noted that in Figure 5B and D the results show a trend toward reduction of food intake in the LCN2-treated animals; they do not reach statistical significance due to the small number of animals per group (N = 3). However, a robust and statistically significant reduction in food intake was observed when data from the two treatment weeks were combined as shown in Figure 5E (N = 6 animals/group).

Notably, the administration of hLCN2 to monkeys led to a 1.7-fold increase in circulating levels (combined monkey and human 15.9 ng/mL) as compared to baseline levels (9.4 ng/mL) within 4 hr after injection (Figure 5F). It should be noted that even though the combined human and monkey LCN2 levels approximately doubled following the administration of recombinant human LCN2, this increase was not statistically significant (p=0.35), most probably due to the small number of monkeys used in each group.

To examine if the observed anorexigenic effect of LCN2 affected body weight or adiposity, we measured body weight, serum leptin, as a marker of adiposity proportional to fat stores size, and serum triglycerides, as a marker of ingested fat load and lipid metabolism. Due to the persisting anorexigenic effect of LCN2 in the second week of treatment, we focused our analysis of body weight and adiposity on the first week of treatment. We found that even during this short period of time rh-LCN2 reduced body weight (Figure 6A, D), serum leptin (Figure 6B, E), and triglyceride levels (Figure 6C, F) in treated animals, even though without a statistical significance, probably due to the small sample size (n = 3 per group).

Figure 6

Download asset Open asset

Figure 6—source data 1

Body weight, leptin, and triglyceride show a tendency for decrease in vervet monkeys after rh-LCN2 treatment.

https://cdn.elifesciences.org/articles/58949/elife-58949-fig6-data1-v1.xlsx

Download elife-58949-fig6-data1-v1.xlsx

To determine whether the rh-LCN2 treatment causes major side effects we measured several biomarkers in the serum of the monkeys, both before and after treatment. As shown in Table 1, serum samples from two timepoints were collected before the first week of treatment, whose values were averaged for each subject and indicated as ‘baseline 1’ (n = 6 subjects). Furthermore, we collected serum samples from the end of the first and second treatment week. ‘Saline’ and ‘LCN2’ groups consisted of n = 6 subjects/serum samples, three from the first and three from the second treatment week.

First, we examined whether our recombinant protein preparation caused any acute-phase response to the treated monkeys. To test this, we measured serum CRP levels before and after the treatment. As shown in Table 1, while CRP levels appear to increase during the time of LCN2 injections, the effect was mostly driven by the extreme value for one animal in the first week of treatment. Rhabdomyolysis is indicated as a rare cause of CRP increase, and may account for the high levels of CRP in that subject (Landry et al., 2017). It is possible that the animal did not tolerate well the daily intramuscular injections of dexmetodimine, which is also evident by the AST data (the value for this animal was the highest). As we mention in the next paragraph, high levels of AST may indicate muscle inflammation or injury. Furthermore, a subject of the saline-treated group was difficult to dose throughout the study, and this stress may account for the high CRP concentrations in this subject at each time point. Nevertheless, serum CRP levels did not change with LCN2 treatment. Similarly, saline-treated monkeys had normal CRP levels compared with baseline.

Quantification of liver enzyme levels in the serum that would indicate liver injury or toxicity, showed no significant elevation with treatment, except for aspartate aminotransferase AST (SGOT). AST was elevated both in saline-treated and LCN2-treated subjects to a similar extent (Table 1). Higher levels of AST compared to ALT, as in this study, may indicate muscle inflammation due to myocyte injury following vigorous exercise, toxins, or drugs use (Cobbold et al., 2010). In this study, monkeys had to be injected daily with a mild sedative drug, dexmetodimine, which was given intramuscularly. It is possible that the daily intramuscular injections, led to myocyte injury that was in turn responsible for the rise in AST levels.

The kidney function and possible injury before and after treatment were assessed by measuring serum creatinine and blood urea nitrogen levels (BUN). Serum creatinine was elevated in the LCN2-treated group compared to baseline but this rise was within the normal range and did not differ significantly from the saline-treated group. Similarly, even though BUN was elevated with LCN2 treatment, this rise was again found within physiological levels and did not change significantly from the saline-treated group. In addition, the BUN/Creatinine ratio did not change with treatment ruling out substantial acute renal injury or dysfunction (Table 1).

Additional serum parameters, such as serum total protein, albumin, globulin, and calcium were analyzed (Table 1) but did not change with treatment. Serum phosphorus was elevated in the LCN2-treated monkeys but this rise was within physiological levels and did not differ significantly from the saline-treated monkeys.

An examination of metabolic parameters showed a decrease in serum glucose and triglycerides and no significant changes in serum cholesterol, amylase, and lipase levels. Regarding glucose, a significant decrease was observed in both treatment groups when compared to baseline, whereas in triglycerides the reduction was borderline significant only in the LCN2-treated group. Serum insulin was not altered by LCN2 treatment. Of note, monkey LCN2 levels remained stable, both with saline and LCN2 treatment ruling out the possibility of regulation of the endogenous monkey LCN2 by exogenous administration of the recombinant human protein.

Data from four separate studies were used to assess the relationship between LCN2 levels and postprandial regulation in healthy, overweight, obese, and severely obese humans. Study 1 was performed at the Department of Medicine-Endocrinology of Columbia University Medical Center (CUMC), Study 2 at the INSERM UMR 1033 institute of the University of Lyon, Study 3 at the Department of Nutritional Sciences of Rutgers University and Study 4 at New York Obesity Nutrition Research Center of CUMC.

An intravenous catheter was placed in the antecubital vein and fasting-baseline blood sample was taken (t = 0). A normal saline (0.9%) flush was used after each blood draw. Following ingestion of the test meal, blood samples were taken at 15, 30, 45, 60, 90, 120, and 180 min for Study 1, at 60 and 105 min for Study 2, at 30, 60, 90, 120, and 180 min for Study 3 and at 15, 30, 45, 60, 90, 120, 180, 240, 300, and 360 min for Study 4. Each blood draw consisted of a total of 10 mL of blood and was collected into serum-separator tubes (Becton Dickinson, Franklin Lakes, NJ). Blood samples were left to clot at room temperature for 30 min and were subsequently spun at 3400 rpm for 15 min at 4°C. In Study 4, blood was collected in chilled EDTA tubes with aprotinin (500 kallikrein inhibitory units/mL of blood) and dipeptidyl peptidase-4 inhibitor (10 μL/mL of blood; Millipore, Burlington, MA). Serum or plasma was then aliquoted and stored at −80°C in cryovials until assayed.

Total RNA from human osteoblasts was extracted using standard protocols. The coding sequence of human LCN2 was amplified and the amplified insert was cloned into the πα-SHP-H vector, kindly provided by Prof. Shapiro. The correct clone was verified by sequencing. Before proceeding to a full-scale production of the protein, we tested the expression in HEK-293 (ATCC) cells by qRT-PCR.

To express the rh-LCN2, 1.56 mL of PEI max transfection reagent (Polysciences Inc, Warrington, PA) was added to 25 mL Opti-MEM medium (Life Technologies, Carlsbad, CA) and incubated for 5 min at room temperature (RT). Meanwhile, 500 μg of the πα-SHP-H-LCN2 vector was added to 25 mL of Opti-MEM medium in another tube. The 25 mL Opti-MEM medium containing 1.56 mL of PEI max was then mixed with 25 mL Opti-MEM medium containing 500 μg of the πα-SHP-H-LCN2 plasmid, incubated for 15 min at RT, and added to 800 mL of Expi293 cells (Life Technologies,) at 0.9–1 million cells/mL. The transfected cells were cultured in a shaker incubator at 120 rpm, 37°C, 9% CO₂ overnight. The cell lines used for these assays were purchased from commercial sources and were free of mycoplasma contamination.

Four days after transfection, supernatants were harvested and purified over 13 mL nickel-charged resin (Ni-NTA Agarose; QIAGEN, Venlo, the Netherlands) in columns. Isolated rh-LCN2 was eluted with imidazole buffer (500 mM NaCl, 10 mM Tris pH = 8.0, 250 mM imidazole). Rh-LCN2 was then buffer exchanged in aqueous buffer, pH = 8.0 (150 mM NaCl, 10 mM Tris pH = 8.0) by dialysis and adjusted to a concentration of 1 mg/mL and filtered (0.22 μm).

The purity and integrity of recombinant human LCN2 stocks was assessed by SDS–PAGE followed by Coomassie blue staining. To verify proper N-glycosylation of the protein, 15 ug of the protein were digested with N-glycanase (Sigma-Aldrich, St. Louis, MO) for 16 hr at 37°C, according to the manufacturer’s instructions. In addition, in the same stocks, we measured endotoxin levels by a chromogenic LAL assay (Genscript, Piscataway, NJ). We then used an endotoxin removal kit (Genscript) to further lower the endotoxin levels.

The radiolabeling of rh-LCN2 was achieved by following a modified protocol by Dong (Dong et al., 2002). Rh-LCN2 (200 µL, 1.0 mg/mL solution) was added to a yellow-capped iodogen tube (Pierce Iodination Tube; Thermo Fisher Scientific, Waltham, MA). This was then diluted in 100 µL of Dulbecco's phosphate buffer saline (DPBS, 1×, with calcium chloride and magnesium chloride, pH 7.0–7.2; Gibco-Thermo Fisher Scientific) to bring to a calculated 0.75 mg/mL concentration. Twenty-five microliters (25 μL) of radioactive sodium iodide (Na¹²⁵I in 0.1 M NaOH; Perkin Elmer, Waltham, MA) was added to this solution (assay 635 µCi). The reaction mixture was sealed and incubated for 60 min at room temperature. After 60 min, the reaction mixture was passed through a preconditioned (with 100 mL of DPBS, 1× solution) size-exclusion column (PD-10 Sephadex G-25M column, Part No. 17085101OL/AG; GE Healthcare, Chicago, IL) with the use of DPBS. Fractions were collected in 1.0 mL increments. The resulting eluent (~2 mL) was assayed to give 150 µCi (24% rcy) of [¹²⁵I]-I-rh-LCN2. TLC: 1 µL of [¹²⁵I]-I-rh-LCN2 sample was spot on the iTLC-SG plate and developed in 70:30 MeOH: H2O. ‘Free’ (i.e. unlabeled) iodine moved with the solvent up the iTLC-SG plate, while the desired I-125 labeled complex stuck to the baseline. The strip was then cut into thirds, placed in tubes for gamma counter analysis (Hidex Automatic Gamma Counter; Hidex, Turku, Finland) and run on an open window.

Based on our previous mouse studies where we observed an 18% reduction in food intake in wild-type lean mice that were treated with recombinant LCN2, we calculated that in order to have an 18% reduction (SD=+/-5 g), and error a = 0.05 and power of analysis 0.90, we needed n = 6 monkeys per group (Brant, 2020). To minimize the number of animals we employed a cross-over design, in which each animal served as its control. There were two treatment weeks with one washout period of 9 days in between. In the first treatment week, three monkeys were treated with saline and three with rh-LCN2. After the washout period monkeys were assigned to the opposite treatment group (Figure 4A).

Six male vervet/African green monkeys (AGM; Caribbean-origin Chlorocebus aethiops sabaeus) used in this study were housed at the Vervet Research Colony at Wake Forest School of Medicine. The age of the subjects ranged from 4.7 to 16.6 years. Vervets were pair-housed in standard four cage racks (81 × 71 × 81 cm per compartment) that contained sitting perches and enrichment items. Animals had unlimited access to commercial monkey chow (LabDiet 5038; Labdiet, St Louis, MO) and water via automated lixit. They were fed supplemental fruits and vegetables 5 days/week.

Vervets were initially anesthetized with ketamine (15 mg/kg, intramuscularly-i.m.), intubated, and maintained on a combination of sevoflurane (2.0–4.0%) and isoflurane (1.5–2.0%). A catheter was surgically placed in the right saphenous vein with a vascular access port (Swirl Phantom P-SPH; Access Technologies, Skokie, IL) located subcutaneously in the right thigh. Monkeys were allowed to recover for 14 days. After recovery, they were re-acclimatized to pair-housing and food intake measurements for 7 days. At the end of the acclimatization period, the baseline food consumption for 5 days was measured. Animals were injected with rh-LCN2 or saline in the morning and then pairs were separated and fed individually for 4 hr. After the 4 hr feeding periods, all food was removed and weighed and then pairs were reunited. This means that animals were essentially fasted each night. Some enrichment foods (small pieces of fruit or vegetables) were provided in the afternoon, but those accounted to be a low percentage of their daily caloric intake. On blood sampling days, the animals had samples collected in the afternoon, after the 4 hr feeding period. Three animals were injected daily, via vascular access port, with saline and three with rh-LCN2 for five 5 days. To facilitate daily IV injections via the vascular access port, animals were lightly sedated with dexmedetomidine (0.035 mg/kg, i.m.) and then administered reversal agent atipamezole (0.35 mg/kg) following access port injections. For afternoon blood samples at the end of the week, they were anesthetized with ketamine.

Food consumption was measured daily as described above. At the end of day 5, a blood sample was collected for analyses, under ketamine anesthesia (10 mg/kg). A washout period of 9 days followed and then monkeys were subjected to a second baseline food consumption for 5 days. Then animals were switched to the opposite group. Again three animals were injected daily with saline and three with rh-LCN2 for 5 days. Food consumption was measured daily as described above. At the end of day 5, a small blood sample was collected for analyses. After the end of the second treatment week, monkeys were left to washout for 10 days. Two out of the six monkeys underwent PET imaging with a no-chase or chase design, as described below. At the end of all the studies, monkeys were anesthetized as described before and vascular access ports were surgically removed and the animals were left to recover for 14 days.

PET scans were performed in two of the six vervets using a GE 64-slice PET/CT Discovery VCT scanner (General Electric). The day prior to the PET scan, animals were administered a solution (0.5 mL) containing 4% potassium iodide and 2% iodine via oral gavage under ketamine anesthesia (10 mg/kg, i.m.). For each scan, the fasted animal was initially immobilized with ketamine (15 mg/kg, i.m.) and maintained on 1.5–2.0% isoflurane anesthesia via an endotracheal tube. The core temperature was kept constant at 37°C with a heated water blanket. An intravenous infusion line with 0.9% NaCl was maintained during the experiment and used for hydration and radiotracer injection. The head was positioned at the center of the field of view, and a 10 min transmission scan was performed before the tracer injection. [¹²⁴I] rh-LCN2 radiotracer was injected (407 ± 80 MBq) as an intravenous bolus over 30 s, and emission data were collected for 120 min in three-dimensional mode.

In a subsequent chase experiment, α-MSH—ligand of the MC4 receptor—was given 15 min after the administration of [¹²⁴I] rh-LCN2 to block non-specific binding in the areas of interest. The same procedure, software and atlas, as in the no-chase experiment, were used to extract time-activity curves (TACs) in the thalamus and hypothalamus, bilaterally. The TACs from the chase experiment were then compared to the corresponding TACs from the first, no-chase experiment in the same regions. We compared the average standard uptake value (SUV) between the two experiments, using the formula: 100*[SUV (no-chase) – SUV (chase)]/SUV (no-chase), to calculate the percent difference. Animals had MRI scans performed 1 week before PET scans using similar anesthesia procedures. MRI scans were performed using a Siemens MAGNETOM Skyra 3T MRI Scanner (Siemens, Munich, Germany) with the following sequences: large fov localizer, small fov localizer 128 mm, MPRAGE Axial_ND, and MPRAGE Axial (TE = 3.39 ms, TR = 2700 ms, TI = 880 ms, resolution = 0.5 × 0.5 × 0.5 mm; FOV = 128 mm × 128 mm; matrix size = 256 × 256). All PET images were coregistered to the corresponding MRI (see volume of interest (VOIs) and outcome variable of SUV on PET).

PET scans were reconstructed using the iterative ordered-subset expectation-maximization (OSEM) algorithm, which were corrected for attenuation, scatter, and dead time. The radioactivity was corrected for physical decay to the injection time and rebinned to 23 dynamic PET frames of 256 (left-to-right) by 256 (nasion-to-inion) by 000 (neck-to-cranium) voxels. The frame schedules were two 30 s, three 60 s, five 120 s, four 240 s, and nine 600 s for 120 mins. The final spatial resolution is expected to be less than 00-mm-full width half maximum in three directions.

For spatial normalization, tissue segmentation, and anatomical labeling, PMod software (version 3.7; PMod Technologies Ltd, Zürich, Switzerland) was used with INIA 19 rhesus high-quality template for non-human primate brains (Rohlfing T, 2012). INIA 19 template was created from high-resolution, T1-weighted magnetic resonance (MR) images of 19 rhesus macaque (Macaca mulatta) animals (Collaboratory, 2020), including > 100 brain regions per side and transferred to individual animal’s magnetic resonance imaging (MRI) using MRI-to-MRI spatial normalization (target: INIA 19 monkey MRI). Then VOIs were transferred to the PET scan of the animal using the MRI-to-PET co-registration module, and minimally adjusted for radioactivity distribution of the scan. Averaged SUV images and time-activity curves (TACs) were generated.

Human serum LCN2, insulin, and GLP-1 were quantitated using commercially available ELISA kits and following the manufacturer’s instructions (#DY1757, R and D Systems, Minneapolis, MN;#90095, CrystalChem, Inc, Elk Grove Village, IL,; #EZGLP1T-36K, Merck, Burlington, MA, respectively). Monkey LCN2 and insulin were assayed with ELISA kits from LifeSpan BioSciences (#LS-F38530 and LS-F10306 respectively; Lifespan Biosciences, Inc, Seattle, WA) and leptin with an ELISA kit from Cusabio (#CSB-E14936Mk, Cusabio, Houston, TX). Circulating levels of primate CRP were determined using a commercially available ELISA kit using the manufacturer’s instructions (911CRP01P-96; Helica Biosystems, Inc, Santa Ana, CA). Blood chemistry was performed using a Heska Element DC Veterinary Chemistry Analyzer (Heska, Loveland, CO) at Columbia University’s Institute of Comparative Medicine Diagnostic Laboratory.

Analyses were performed with SAS 9.4 (SAS Institute, Inc, Cary, NC) and GraphPad Prism 8 (GraphPad, San Diego, CA), with a level of significance at p<0.05. Since serum LCN2 concentration for each sample was measured multiple times over time, a one- or two-way Analysis of Variance (ANOVA) repeated measures design was appropriate for examining whether statistically significant differences in means exist in each experiment. Data were examined for normality using the Shapiro-Wilk test. Variables were log-transformed. Variables were log-transformed using the natural logarithm (ln) when appropriate, and nonparametric tests were used when necessary. Using the F-ratio statistic, we examined whether the repeated measured variable, time, had an overall statistically significant effect, implying that the corresponding values of the outcome that is examined in each experiment is statistically different from values obtained at other points of time. To correct for any violations of the assumption of sphericity (ϵ), degrees of freedom for each F value were adjusted according to the estimated epsilon obtained in each analysis. For values of ϵ < 0.75, the Greenhouse-Geisser Epsilon correction was used, while in cases when ϵ > 0.75, the less conservative Huynh-Feldt Epsilon correction was preferred. Using the CONTRAST option in PROC General Linear Model (GLM) in SAS, we investigated whether and which mean values at each time point were statistically different from baseline, at t = 0. Pearson’s correlation coefficient, or Spearman’s correlation for data not normally distributed, were used. For non-repeated observations either two-tailed Student’s t-tests, Mann-Whitney or rank-sum Wilcoxon non-parametric tests were used depending on the normality of distribution. For the vervet studies, comparisons were made with paired or unpaired t-tests as needed. Data are reported as mean ± SEM. */^# indicates p<0.05, ‡ indicates p<0.06, and † indicate p<0.1. n indicates the number of human subjects or animals tested in each experiment.

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Human Study #1 and the non-human primate study were supported by the National Center for Advancing Translational Sciences, National Institutes of Health, through Grant Number UL1TR001873 to Stavroula Kousteni and Mishaela Rubin. Research reported in this publication was supported by the National Heart, Lung, And Blood Institute of the National Institutes of Health under Award Number T32HL120826. Additional support was provided from the National Institute on Aging through grant number P01AG032959 awarded to Stavroula Kousteni. The content of this publication is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. Human Study #2 was supported by a research grant from Roche-Chugai to Cyrille Confavreux. Human Study #3 was supported by OmniActive Health Technologies and USDA-NIFA hmg (NJAES - 0153866). Human Study #4 was supported by grants from the NIH (R01 DK067561, P30 DK26687-30, and P30 DK063608) and by the National Center for Advancing Translational Sciences, National Institutes of Health, through Grant Number UL1 TR000040.

Additional support for the vervet/African green monkey colony was provided by P40-OD010965 and UL1-TR001420. We thank Dr. Heather DeLoid for performing vascular access port surgeries and Stacey Combs, Kortne Hudick, Eric McCloud, Justin Herr, Edison Floyd, and Stephanie Rideout Danner for technical assistance.

We also thank Columbia University’s Institute of Comparative Medicine Diagnostic Laboratory for blood chemistry analyses of the non-human primate study. Furthermore, we would like to thank Dr. Franck Oury for helpful discussions on the PET study.

Human subjects: The Institutional Review Boards of Columbia University (protocol numbers for human studies 1 and 4 respectively: IRB-AAAQ9254 and IRB-AAA03051), University of Lyon (Comité de Protection des Personnes SUD-EST IV - France, protocol number for human study 2: RCB-2014-A00237-40) and Rutgers University (protocol number for human study 3: IRB-15-524M) approved these studies and informed written consent and consent to publish was obtained from each subject. All samples were obtained in accordance with the Health Insurance Portability and Accountability Act (HIPAA). All clinical investigation has been conducted according to Declaration of Helsinki principles.

Animal experimentation: Animal studies were performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All of the animals were handled according to the approved institutional animal care and use committee (IACUC) protocol (#A18-067) of Wake Forest School of Medicine. All surgery was performed under sodium pentobarbital anesthesia, and every effort was made to minimize suffering.

1. Received: May 15, 2020
2. Accepted: October 23, 2020
3. Version of Record published: November 24, 2020 (version 1)

© 2020, Petropoulou et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.