Differential solubilization of osteoblastic alkaline phosphatase from human primary bone cell cultures

Application of machine learning algorithms to identify people with low bone density

Background

Osteoporosis is becoming more common worldwide, imposing a substantial burden on individuals and society. The onset of osteoporosis is subtle, early detection is challenging, and population-wide screening is infeasible. Thus, there is a need to develop a method to identify those at high risk for osteoporosis.

Objective

This study aimed to develop a machine learning algorithm to effectively identify people with low bone density, using readily available demographic and blood biochemical data.

Methods

Using NHANES 2017-2020 data, participants over 50 years old with complete femoral neck BMD data were selected. This cohort was randomly divided into training (70%) and test (30%) sets. Lasso regression selected variables for inclusion in six machine learning models built on the training data: logistic regression (LR), support vector machine (SVM), gradient boosting machine (GBM), naive Bayes (NB), artificial neural network (ANN) and random forest (RF). NHANES data from the 2013-2014 cycle was used as an external validation set input into the models to verify their generalizability. Model discrimination was assessed via AUC, accuracy, sensitivity, specificity, precision and F1 score. Calibration curves evaluated goodness-of-fit. Decision curves determined clinical utility. The SHAP framework analyzed variable importance.

Results

A total of 3,545 participants were included in the internal validation set of this study, of whom 1870 had normal bone density and 1,675 had low bone density Lasso regression selected 19 variables. In the test set, AUC was 0.785 (LR), 0.780 (SVM), 0.775 (GBM), 0.729 (NB), 0.771 (ANN), and 0.768 (RF). The LR model has the best discrimination and a better calibration curve fit, the best clinical net benefit for the decision curve, and it also reflects good predictive power in the external validation dataset The top variables in the LR model were: age, BMI, gender, creatine phosphokinase, total cholesterol and alkaline phosphatase.

Conclusion

The machine learning model demonstrated effective classification of low BMD using blood biomarkers. This could aid clinical decision making for osteoporosis prevention and management.

Membrane-bound alkaline phosphatase from ectopic mineralization and rat bone marrow cell culture

Cells from rat bone marrow exhibit the proliferation-differentiation sequence of osteoblasts, form mineralized extracellular matrix in vitro and release alkaline phosphatase into the medium. Membrane-bound alkaline phosphatase was obtained by method that is easy to reproduce, simpler and fast when compared with the method used to obtain the enzyme from rat osseous plate. The membrane-bound alkaline phosphatase from cultures of rat bone marrow cells has a MW(r) of about 120 kDa and specific PNPP activity of 1200 U/mg. The ecto-enzyme is anchored to the plasma membrane by the GPI anchor and can be released by PIPLC (selective treatment) or polidocanol (0.2 mg/mL protein and 1% (w/v) detergent). The apparent optimum pH for PNPP hydrolysis by the enzyme was pH 10. This fraction hydrolyzes ATP (240 U/mg), ADP (350 U/mg), glucose 1-phosphate (1100 U/mg), glucose 6-phosphate (340 U/mg), fructose 6-phosphate (460 U/mg), pyrophosphate (330 U/mg) and beta-glycerophosphate (600 U/mg). Cooperative effects were observed for the hydrolysis of PPi and beta-glycerophosphate. PNPPase activity was inhibited by 0.1 mM vanadate (46%), 0.1 mM ZnCl2 (68%), 1 mM levamisole (66%), 1 mM arsenate (44%), 10 mM phosphate (21%) and 1 mM theophylline (72%). We report the biochemical characterization of membrane-bound alkaline phosphatase obtained from rat bone marrow cells cultures, using a method that is simple, rapid and easy to reproduce. Its properties are compared with those of rat osseous plate enzyme and revealed that the alkaline phosphatase obtained has some kinetics and structural behaviors with higher levels of enzymatic activity, facilitating the comprehension of the mineralization process and its function.

Chick embryo anchored alkaline phosphatase and mineralization process in vitro

Bone alkaline phosphatase with glycolipid anchor (GPI-bALP) from chick embryo femurs in a medium without exogenous inorganic phosphate, but containing calcium and GPI-bALP substrates, served as in vitro model of mineral formation. The mineralization process was initiated by the formation of inorganic phosphate, arising from the hydrolysis of a substrate by GPI-bALP. Several mineralization media containing different substrates were analysed after an incubation time ranging from 1.5 h to 144 h. The measurements of Ca/Pi ratio and infrared spectra permitted us to follow the presence of amorphous and noncrystalline structures, while the analysis of X-ray diffraction data allowed us to obtain the stoichiometry of crystals. The hydrolysis of phosphocreatine, glucose 1-phosphate, glucose 6-phosphate, glucose 1,6-bisphosphate by GPI-bALP produced hydroxyapatite in a manner similar to that of beta-glycerophosphate. Several distinct steps in the mineral formation were observed. Amorphous calcium phosphate was present at the onset of the mineral formation, then poorly formed hydroxyapatite crystalline structures were observed, followed by the presence of hydroxyapatite crystals after 6-12 h incubation time. However, the hydrolysis of either ATP or ADP, catalysed by GPI-bALP in calcium-containing medium, did not lead to the formation of any hydroxyapatite crystals, even after 144 h incubation time, when hydrolysis of both nucleotides was completed. In contrast, the hydrolysis of AMP by GPI-bALP led to the appearance of hydroxyapatite crystals after 12 h incubation time. The hydroxyapatite formation depends not only on the ability of GPI-bALP to hydrolyze the organic phosphate but also on the nature of substrates affecting the nucleation process or producing inhibitors of the mineralization.

Different distributions of human bone alkaline phosphatase isoforms in serum and bone tissue extracts

BACKGROUND

In vitro, bone alkaline phosphatase (BALP) is released from the osteoblast membrane with its glycosylphosphatidylinositol (GPI) anchor still attached (i.e., in an anchor-intact form); however, in vivo, BALP circulates as a variable mixture of anchorless isoforms, which can be identified by high-performance liquid chromatography (HPLC). Previous studies have shown that the relative abundance of these BALP isoforms in serum may be clinically useful for the diagnosis and management of metabolic bone disease.

METHODS

In the current studies, we describe a method for the determination of anchorless BALP isoforms in extracts of bone and we present novel data on the conversion of anchor-intact to anchorless BALP by incubation with endogenous circulating GPI-specific phospholipase D (GPI-PLD).

RESULTS

A 72-h extraction with 0.1% Triton X-100 released approximately 90% of the BALP activity from powdered bone. An average of 19% of this activity was anchorless, but essentially all of the activity could be converted to the anchorless form by incubation with partially purified GPI-PLD from human serum. Using HPLC, we detected four BALP isoforms (B/I, B1x, B1, and B2) in these GPI-PLD-treated extracts of bone. An additional BALP fraction was also detected in the samples during the initial phase of GPI-PLD treatment.

CONCLUSIONS

The abundance of the BALP isoforms differed between bone and serum, particularly for the B/I isoform, which comprised, on average, 18% of the BALP in GPI-PLD-treated extracts of healthy bone tissue, but only 6% of the total BALP activity in serum from healthy individuals. Based on our recent finding of differences in the number of sialic acid residues between the BALP isoforms, we hypothesize that this difference between BALP isoforms in serum and extracts of bone is due to the different patterns of glycosylation, which results in different biological half-lives in the circulation. A preliminary application of our method to the extraction of BALP isoforms from a small number of human bone samples suggests that the method should be useful for studies of human skeletal site-specific and metabolic bone disease-specific differences in the amounts and distributions of the BALP isoforms in bone.

Detergent-mediated reconstitution of a glycosyl-phosphatidylinositol-protein into liposomes

A three-step detergent-mediated reconstitution has been applied to the incorporation of a glycosyl-phosphatidylinositol-protein into liposomes. The protein studied was alkaline phosphatase from bovine intestine. Liposomes prepared by dialysis were treated with various amounts of two detergents, either n-octyl beta-D-glucoside or Triton X-100. At different steps of the solubilization process, protein was added and the detergent was removed by hydrophobic resins. The most efficient reconstitutions were obtained with an octyl glucoside concentration corresponding to the onset of liposome solubilization and with a Triton X-100 concentration leading to partial solubilization of the liposomes. The involvement of the glycosyl-phosphatidylinositol anchor in alkaline phosphatase reconstitution was demonstrated by the inability of phosphoinositol-specific phospholipase-C-hydrolysed alkaline phosphatase to incorporate into liposomes. Between 70-85% of the protein associated with liposomes were anchored in the outer leaflet of the bilayer, oriented towards the outside of the liposome. The remainder was trapped within the lumen of the liposomes.