Dipeptidyl Peptidase 3 Activity as a Promising Biomarker of Bone Fragility in Postmenopausal Women

The dipeptidyl peptidase 3 (Dpp3) is a ubiquitous zinc-dependent aminopeptidase, participating in the activation or degradation of signaling peptides and in the Keap1–Nrf2 antioxidant pathway. The absence of Dpp3 in the Dpp3 knockout mouse model causes increased osteoclast activity, altered osteogenic function, sustained oxidative stress in the bone tissue, and bone loss. We aimed to assess the association of Dpp3 activity with bone fragility in postmenopausal osteoporosis and the impact of denosumab on enzymatic activity. We conducted a two-phase study including 69 postmenopausal women with severe osteoporosis and 36 postmenopausal women without osteometabolic conditions, as controls (cross-sectional phase). Subjects with severe osteoporosis were assessed at baseline and 14 days after the first denosumab administration (prospective phase). The results showed significant reduction in serum Dpp3 activity (expressed as nmoles of formed product/mg proteins/min) in patients vs. controls (0.791 ± 0.232 vs. 1.195 ± 0.338; p < 0.001), and significant association with bone mass at the femoral neck (r = 0.28, p = 0.02) in patients prior to treatment. We found a negative correlation between C-terminal telopeptide (CTX) or N-terminal pro-peptide of type 1 procollagen (P1NP) levels and Dpp3 activity (respectively, r = −0.29, p = 0.012; and r = −0.2572, p = 0.033). Dpp3 activity did not change after denosumab injection. Our findings support a critical role played by Dpp3 in bone homeostasis as a potential bone protective factor. Additional clinical studies in larger cohorts might explore the implementation of Dpp3 assessment as a biomarker of bone health status.


Introduction
The dipeptidyl peptidase 3 (Dpp3) is a ubiquitous zinc-dependent aminopeptidase, highly expressed in human cells and conserved among higher vertebrates [1]. It cleaves dipeptides from the N-terminus of oligopeptides, thereby participating in the activation or degradation of signaling peptides, such as angiotensin 2 and enkephalins, involved, respectively, in the renin-angiotensin system and blood pressure regulation, and in nociception [2]. Moreover, Dpp3 is a regulator of the cellular oxidative stress response through the Keap1-Nrf2 antioxidant pathway; in this context, binding of Dpp3 to Keap1 releases the transcription factor Nrf2, which thus escapes degradation by the 26S proteasome and migrates into the nucleus, where it drives the expression of various antioxidant enzymes [3]. Dpp3 antioxidant activity has been particularly associated with inflammation and carcinogenesis [4][5][6], and recently with bone metabolism in a mouse model [7]. In fact, Menale et al. demonstrated that adult mice lacking Dpp3 (Dpp3 knockout (KO) mice) presented bone mass loss owing to increased osteoclast activity, as well as altered osteogenic function secondary to sustained oxidative stress in the bone tissue, and that, in an estrogen deprivation model, Dpp3 KO female mice had a greater decrease in bone mass, compared with wild-type animals. This pointed to a non-redundant role of Dpp3 in the maintenance of bone homeostasis.
Fragility fractures are prevalent in postmenopausal women resulting in bone-associated morbidities, reduced quality of life, increased mortality, and healthcare costs. Accordingly, owing to the estimated number of people affected by osteoporosis (about 200 million worldwide) and the considerable social and healthcare costs associated, this disease clearly constitutes a global (though often overlooked) public health problem [8].
Bone mineral density (BMD) evaluated via dual-energy X-ray absorptiometry (DXA) is the most common parameter to identify patients at risk of fragility fractures. A T-score (i.e., standard deviation difference from mean values of sex-matched young, healthy individuals) equal to or below −2.5 standard deviation (SD) is considered diagnostic for osteoporosis; however, most individuals that sustain fragility fractures are above this cutoff, which highlights an incongruity with serious drawbacks. To overcome this issue, other imaging techniques have been developed, but they are mostly employed as research tools [9].
Considering this scenario, the identification of an early biomarker of bone impairment to be easily assessed in the routine clinical practice is expected to improve patients' management, prevent fractures, preserve the quality of life, and alleviate healthcare costs.
Dpp3 is primarily a cytosolic protein; however, it is also detectable extracellularly in several biological fluids, most importantly in human plasma and serum in pathophysiological conditions [10][11][12]. In the framework of bone pathology, there are no clinical studies aimed at investigating circulating Dpp3 activity as a marker of bone loss in postmenopausal women with fragility fractures. Hence, in our research, we investigated whether Dpp3 had a role in skeletal fragility in humans. To answer this question, we measured serum Dpp3 activity in a cohort of postmenopausal women with severe osteoporosis. Furthermore, we evaluated if a powerful antiresorptive agent can affect Dpp3 activity.
With respect to bone turnover markers, we found a negative correlation between CTX or P1NP levels and DPP3 activity (respectively, r = −0.29, p = 0.012; and r = −0.2572, p = 0.033; Figure 2B).  We observed a significant negative correlation between both age and years from menopause and DPP3 activity in controls (respectively, r = −0.3521, p = 0.0352; and r = −0.3532, p = 0.0346) but not in the patient population ( Figure 3). Finally, we found no significant correlation between vitamin D, PTH, calcium, and phosphate levels and DPP3 activity in the sera of osteoporotic patients. Furthermore, no significant association was revealed with the number of fractures in postmenopausal women (Supplementary Table S1).
We observed a significant negative correlation between both age and years from menopause and Dpp3 activity in controls (respectively, r = −0.3521, p = 0.0352; and r = −0.3532, p = 0.0346) but not in the patient population ( Figure 3). We observed a significant negative correlation between both age and years from menopause and DPP3 activity in controls (respectively, r = −0.3521, p = 0.0352; and r = −0.3532, p = 0.0346) but not in the patient population ( Figure 3). Finally, we found no significant correlation between vitamin D, PTH, calcium, and phosphate levels and DPP3 activity in the sera of osteoporotic patients. Furthermore, no significant association was revealed with the number of fractures in postmenopausal women (Supplementary Table S1).

Prospective Evaluation
As expected, we found a significant reduction in CTX (0.370 ± 0.180 ng/mL vs. 0.097 ± 0.122 ng/mL, p < 0.001) and P1NP (50.51 ± 24.57 ng/mL vs. 37.80 ± 18.21 ng/mL; p < 0.001) 14 days after injection of the monoclonal antibody against receptor activator of NF-kB ligand (RANKL) denosumab ( Figure 4A). On the other hand, DPP3 activity did not change after denosumab treatment (0.791 ± 0.232 vs. 0.815 ± 0.207; p = 0.105) ( Figure 4B). Finally, we found no significant correlation between vitamin D, PTH, calcium, and phosphate levels and Dpp3 activity in the sera of osteoporotic patients. Furthermore, no significant association was revealed with the number of fractures in postmenopausal women (Supplementary Table S1).

Discussion
Despite the attention gained by DPP3 since its discovery at the end of the 1960s several aspects related to its function remain elusive [14]. For example, a po investigated topic is DPP3 function in the context of bone biology, which has addressed only recently by Menale et al., who showed that the absence of DPP3 in mouse model Dpp3 KO caused bone loss owing to increased osteoclast activity impaired bone quality, particularly in female mice after ovariectomy [7]. For the hu counterpart, the present report is the first study that evaluated DPP3 activity in a co of subjects with severe osteoporotic bone fragility. In this context, we found significa lower DPP3 levels in patients, compared with a control population with no osteometa disorders, in agreement with previous evidence in mice.
Moreover, we found a positive correlation between FN-BMD and DPP3 activi severe osteoporosis; this is relevant since it has been well demonstrated that FN-BM the most accurate DXA parameter to estimate the fracture risk [15]. The finding o inverse correlation between TH-BMD and DPP3 activity in the patients compared controls is puzzling, and at present, we cannot provide an explanation. Expanding study to a larger population might correct or confirm this evidence.
The negative correlation between standard bone turnover markers and D enzymatic activity in osteoporotic women prior to antiosteoporotic treatm strengthened DPP3 association with bone metabolism. A single administration of a po antiresorptive agent (denosumab) caused a significant reduction of serum CTX and P levels two weeks after treatment, as expected, while did not significantly affect D activity in patients. We cannot exclude changes in DPP3 levels at later time points drug administration or after prolonged treatment. It would be also interesting to as changes in DPP3 levels after treatment with different antiresorptive as wel osteoanabolic drugs [16].
Overall, our data indicate that DPP3 preserves bone health, which sounds reason based on evidence of reduced Nrf2 and HO-1 expression and increased lipid peroxida in the bone tissue lacking Dpp3 and on the well-known detrimental effect of susta inflammation and oxidative stress on bone homeostasis [17].
The negative correlation between age and years from menopause, and DPP3 act in the control population suggests that DPP3 activity progressively decreases with ag healthy individuals and approaches the (already low) levels measured in the patient

Discussion
Despite the attention gained by Dpp3 since its discovery at the end of the 1960s [13], several aspects related to its function remain elusive [14]. For example, a poorly investigated topic is Dpp3 function in the context of bone biology, which has been addressed only recently by Menale et al., who showed that the absence of Dpp3 in the mouse model Dpp3 KO caused bone loss owing to increased osteoclast activity and impaired bone quality, particularly in female mice after ovariectomy [7]. For the human counterpart, the present report is the first study that evaluated Dpp3 activity in a cohort of subjects with severe osteoporotic bone fragility. In this context, we found significantly lower Dpp3 levels in patients, compared with a control population with no osteometabolic disorders, in agreement with previous evidence in mice.
Moreover, we found a positive correlation between FN-BMD and Dpp3 activity in severe osteoporosis; this is relevant since it has been well demonstrated that FN-BMD is the most accurate DXA parameter to estimate the fracture risk [15]. The finding of an inverse correlation between TH-BMD and Dpp3 activity in the patients compared with controls is puzzling, and at present, we cannot provide an explanation. Expanding the study to a larger population might correct or confirm this evidence.
The negative correlation between standard bone turnover markers and Dpp3 enzymatic activity in osteoporotic women prior to antiosteoporotic treatment strengthened Dpp3 association with bone metabolism. A single administration of a potent antiresorptive agent (denosumab) caused a significant reduction of serum CTX and P1NP levels two weeks after treatment, as expected, while did not significantly affect Dpp3 activity in patients. We cannot exclude changes in Dpp3 levels at later time points after drug administration or after prolonged treatment. It would be also interesting to assess changes in Dpp3 levels after treatment with different antiresorptive as well as osteoanabolic drugs [16].
Overall, our data indicate that Dpp3 preserves bone health, which sounds reasonable based on evidence of reduced Nrf2 and HO-1 expression and increased lipid peroxidation in the bone tissue lacking Dpp3 and on the well-known detrimental effect of sustained inflammation and oxidative stress on bone homeostasis [17].
The negative correlation between age and years from menopause, and Dpp3 activity in the control population suggests that Dpp3 activity progressively decreases with age in healthy individuals and approaches the (already low) levels measured in the patients. In addition, the absence of correlation with age in the patients, compared with the controls, suggested that osteoporosis owing to estrogen withdrawal had a greater effect on Dpp3 activity, compared with the impact of aging. Indeed, as mentioned above, in the Dpp3 KO mouse model, estrogen withdrawal aggravated bone loss, compared with the wild-type counterpart. Accordingly, an important upregulation of Dpp3 expression by estrogens has been demonstrated in the liver of wild-type mice treated with 17β-estradiol [18].
Notably, we revealed here that lower circulating Dpp3 enzymatic levels are associated with a worse skeletal phenotype; conversely, in sepsis, cardiogenic shock, heart failure, and acute kidney injury the correlation goes in the opposite direction, with higher levels resulting in more severe disease and reduced survival rate [19][20][21][22][23][24][25]. In these conditions, angiotensin 2, the primary effector of the renin-angiotensin system (RAS), is recognized as the key substrate of Dpp3 activity and clinical evidence has been interpreted based on the modulation of the RAS by Dpp3, and the resulting impact on hemodynamics and on the physiology of the cardiovascular system. By contrast, the specific substrate(s) of Dpp3 enzymatic activity in the bone is not known yet, and the skeletal phenotype in the Dpp3 KO mice has been essentially related to sustained oxidative stress. In fact, the Keap1-Nrf2 antioxidant system, which comprises also Dpp3 among its activators, is known to exert both protective and detrimental effects, in cancer [4] as well as in bone biology, acting both on the osteoclast and on the osteoblast lineage [26].
All these data suggest that Dpp3 participates in a complex and pleiotropic network that requires to be better elucidated. Dissection of basic Dpp3-related pathophysiological mechanisms will likely have implications for the diagnosis and treatment of human diseases.
Our study has some limitations-namely, the cross-sectional design, the limited number of controls, and the short follow-up period for the prospective evaluation; on the other hand, it constitutes the starting point for investigation in a larger cohort of osteoporotic patients. In parallel, it would be interesting to test the hypothesis of a bone-protective role of Dpp3 in other pathological conditions resulting in bone deterioration. As a perspective, serum Dpp3 activity measurement might be easily implemented as a biomarker of bone fragility in routine clinical practice.

Study Population and Design
We performed a two-phase study with a cross-sectional and a prospective evaluation, overall including 105 Caucasian subjects. The cross-sectional phase was conducted on 69 postmenopausal women affected by severe osteoporosis and 36 postmenopausal women with either osteopenia [27] or normal BMD, as controls. The prospective phase of the study included 69 subjects with severe osteoporosis. In detail, after baseline evaluation, all subjects started denosumab (Prolia ® , Amgen Europe; 60 mg subcutaneously, every six months) as a treatment for severe osteoporosis. Fourteen days after the first denosumab administration, subjects underwent blood drawing for evaluation of bone turnover markers and Dpp3 activity.
Participants were consecutively enrolled at Fondazione Policlinico Campus Bio-Medico of Rome. Inclusion criteria for the cases were postmenopausal status, lumbar spine and/or non-dominant total hip and/or femoral neck T-score <2.5, at least one vertebral fragility fracture, and absence of ongoing or previous antiosteoporotic treatments. Inclusion criteria for controls were postmenopausal status, normal or osteopenic BMD at L1-L4 and non-dominant TH and FN, and absence of osteoporotic fractures.
Exclusion criteria for all participants were the presence of any condition that can affect bone and calcium metabolism-namely, early menopause, chronic kidney disease (glomerular filtration rate, GFR, <30 mL/min/1.73 m 2 ), a history of possible high-energy vertebral fractures, metabolic bone diseases, sarcoidosis, inflammatory bowel diseases, rheumatic diseases, musculoskeletal disorders other than osteoporosis and medications such as bisphosphonates, teriparatide, estrogens, corticosteroids, aromatase inhibitors, or other drugs that could interfere with bone metabolism.

Biochemical Analysis
In the morning, fasting blood samples were collected for the evaluation of serum total calcium, phosphate, creatinine, and 25 OH vitamin D, using automated methods.

Dpp3 Activity Measurement
Dpp3 enzymatic activity in the sera of patients and controls was measured as previously reported [7], with minor modifications. Briefly, serum protein concentration was determined using the DC Protein Assay kit (Bio-Rad, Hercules, CA, USA), according to the manufacturer's instructions. Then, a volume of serum corresponding to about 50 µg of total proteins was assayed with 0.04 mM Arg-Arg-β-naphthylamide (Sigma-Aldrich, St. Louis, MO, USA) in Tris-HCl, pH 8.5, at 37 • C. The reaction was stopped by adding 2 M acetate buffer, pH 4.5, containing 10% Tween and 1.5 mg/mL Fast Blue B Salt (all chemicals from Sigma-Aldrich). The absorbance of the released product (β-naphthylamine) was measured at 530 nm, using a Synergy TM H4 Microplate Reader. The enzymatic activity was expressed as nmol of β-naphthylamine (2-NA)/mg proteins/min).

Dual-Energy X-ray Absorptiometry (DXA)
BMD was measured by DXA at L1-L4, TH, and FN (Hologic Discovery QDR Instrument, Marlborough, MA, USA). We reported data for absolute BMD. All scans were performed according to the International Society for Clinical Densitometry (ISCD) guidelines [28]. Fractured vertebrae and vertebrae with structural changes (T-score difference with the adjacent vertebra >1.0) were excluded from the analysis.

Ethics
The study was conducted in compliance with the Declaration of Helsinki and the International Conference on Harmonization Principles of Good Clinical Practice. The local ethics committee approved the research protocol, and all participants gave informed consent prior to inclusion in the study.

Statistical Analysis
Data were described as mean and standard deviation. Data distribution adherence to Gaussian distribution was evaluated using the Kolmogorov-Smirnov test. The difference between case and control groups was explored with an unpaired Student's t-test if the distribution was approximately Gaussian, or with Mann-Whitney test otherwise. For Student's t-test, variance equality was determined according to Levene's test. Correlations between Dpp3 activity and CTX, P1NP, calcium phosphate, PTH, 25 OH-Vitamin D, FN-, TH-, and L1-L4 BMD were tested using Pearson's correlation. The association of Dpp3 activity and case and control was explored with linear regression, singularly considered and age corrected. Results were then expressed as a regression coefficient with a 95% confidence interval (95% CI). A p value lower than 0.05 was considered significant.

Conclusions
The results of the present clinical study, together with previous evidence in a relevant mouse model, point to a critical role played by Dpp3 in bone homeostasis as a bone protective factor. Additional clinical studies in larger cohorts of osteoporotic women, as well as in patients affected by different bone pathologies, might extend further the range of diseases in which Dpp3 is involved and its possible exploitation and implementation as a biomarker.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/molecules27123929/s1, Figure S1: Pearson correlation analysis between BMD at hip or spine and circulating Dpp3 activity, Table S1: Correlation analysis between cDPP3 enzymatic activity in our cohort of postmenopausal women with severe osteoporosis and biochemical clinical parameters that did not reach statistical significance after linear regression and Pearson's correlation.