Calcium, Phosphate and Bone

About 99 percent of the calcium in an adult human body is in bone, yet the parathyroid glands read only a sliver of the remaining 1 percent: the free calcium ions in plasma. That sliver stays in a narrow range while bone is dissolved and rebuilt and the kidneys filter about ten grams of calcium a day. How?

Bone is a living mineral bank

Muscle, bone and movement gave the first look, including why vertebrates carry an endoskeleton and arthropods an exoskeleton. In brief: osteoblasts lay down matrix, osteoclasts resorb it, and osteocytes, former osteoblasts walled into the matrix, stay linked to one another by fine processes. The matrix is collagen stiffened by a calcium phosphate mineral, hydroxyapatite. Compact bone is built of osteons, cylinders of matrix around a central canal; spongy bone is a lattice of struts called trabeculae. Bone ends at most synovial joints are covered by hyaline articular cartilage, which has no blood vessels and reduces friction. Ligaments join bone to bone and tendons join muscle to bone. By structure, joints are fibrous, cartilaginous or synovial.

Think of bone as a vault and plasma as your wallet. The vault holds about 99 percent of body calcium and roughly 85 percent of its phosphate. Remodeling, osteoclasts resorbing a patch and osteoblasts refilling it, keeps traffic moving between the two.

The wallet has three pockets. In adult human plasma, roughly half of total calcium is free ionized calcium, about 40 percent is bound to protein, mostly albumin, and about 10 percent is complexed with citrate, phosphate and bicarbonate. Only the ionized fraction acts on cells and is sensed and regulated. In alkalosis albumin binds more calcium, so ionized calcium falls while total calcium stays the same.

Total calcium (mg per dL) and ionized calcium (by ion-selective electrode, mmol per L) are different measurements. Formulas that adjust total calcium for albumin do not reliably reproduce ionized calcium: in 22,658 Alberta adults with total and ionized calcium measured in the same samples (albumin in 10,659 of them), albumin-adjusted calcium generally did no better than unadjusted total calcium (Desgagnés and colleagues, 2025).

Osteoclasts are told what to do: RANKL and OPG

Osteoblast-lineage cells and osteocytes display RANKL, a cytokine of the tumor necrosis factor family. Its receptor, RANK, sits on osteoclast precursors, and engaging it drives their formation and activation. Osteoprotegerin (OPG) is a soluble decoy receptor: it binds RANKL and keeps it off RANK. Picture a key, a lock and a dummy lock that swallows keys. RANKL relative to OPG is a main control on how many osteoclasts form. Precursor supply, M-CSF and other signals also matter.

The pieces came from three papers, not one. Simonet and colleagues (Cell, 1997) made transgenic mice that expressed OPG from the liver; they developed profound but nonlethal osteopetrosis, an excess of dense bone, and OPG blocked ovariectomy-associated bone loss in rats. Lacey and colleagues (Cell, 1998) identified the ligand OPG binds as a cytokine that drives osteoclast differentiation. Hsu and colleagues (PNAS, 1999) showed, in mouse bone-marrow progenitors, that it acts through RANK. Other laboratories described the same molecules independently under other names.

Which cell’s RANKL matters? When Nakashima and colleagues (Nature Medicine, 2011) deleted RANKL in mice with a Dmp1-Cre transgene aimed at osteocytes, the mice developed severe osteopetrosis. That Cre also acts in some osteoblasts and lining cells, so Xiong and colleagues (PLoS ONE, 2015) repeated the deletion with a Sost-Cre that spares osteoblasts and lining cells, though it is also active in some marrow hematopoietic cells, and saw a threefold fall in osteoclast number in cancellous bone.

Parathyroid hormone (PTH) reaches osteoclasts mainly through this system. Its receptors sit on osteoblast-lineage cells, including osteocytes, and PTH raises RANKL relative to OPG. Direct PTH effects on osteoclast-lineage cells reported in culture remain an open question in the body. Continuous PTH excess causes net bone loss even though formation also rises; intermittent exposure behaves differently and should not be lumped with it. The ovariectomized-rat result bridges to estrogen and bone in the ovarian cycle.

Four hormones, three organs, one sensor

The sensor. The parathyroid glands, usually four, sit on the back of the thyroid. Their chief cells carry the calcium-sensing receptor (CaSR), a G protein-coupled receptor (G proteins) cloned from bovine parathyroid by Brown, Hebert and colleagues (Nature, 1993). A rise in ionized calcium activates CaSR and inhibits PTH secretion, steeply: a small change in calcium moves secretion a long way. CaSR in the thick ascending limb also lowers calcium reabsorption when calcium is high. (Calcium signaling covers calcium inside cells; this lesson is about the calcium outside them.)

PTH, a peptide, acts through a G protein-coupled receptor that raises cAMP (cAMP and PKA). On the kidney it:

• raises active calcium reabsorption in the distal nephron;
• retrieves the sodium-phosphate cotransporter NaPi-IIa from the proximal brush border, so phosphate is lost in urine (Traebert and colleagues, 2000, rat kidney in vivo and isolated perfused mouse tubules);
• stimulates 1-alpha-hydroxylation of vitamin D; in thyroparathyroidectomized rats on a low-calcium diet, production of active vitamin D appeared to depend on the parathyroid (Garabedian and colleagues, 1972).

With bone resorption added, plasma calcium rises and plasma phosphate falls: the phosphaturia outweighs the phosphate released from bone.

Calcitriol. Vitamin D is made in skin (skin and human heat), 25-hydroxylated in the liver (absorption and the liver) and 1-hydroxylated in the kidney, giving calcitriol, 1,25-dihydroxyvitamin D, which acts through the vitamin D receptor (nuclear receptors). In mice lacking that receptor, a diet that restored ionized calcium prevented rickets and the rise in PTH, which points to impaired intestinal calcium absorption as the cause (Li and colleagues, 1998). In wild-type mice, two days of calcitriol raised active jejunal phosphate transport while the paracellular route was unchanged (Hernando and colleagues, 2021). In rats, calcitriol inhibits PTH gene transcription (Silver and colleagues, 1986).

FGF23 is a hormone made by bone, mainly by osteocytes and osteoblasts. It acts on the kidney through an FGF receptor bound to the co-receptor Klotho (Urakawa and colleagues, 2006), lowering phosphate reabsorption and calcitriol: in rodents, injected FGF23 lowered renal mRNA for 1-alpha-hydroxylase and raised it for 24-hydroxylase, the enzyme that inactivates vitamin D metabolites. In normal mice, injected calcitriol raised serum FGF23 within four hours. Calcitriol raises FGF23 and FGF23 lowers calcitriol: a closed loop.

Calcitonin comes from the parafollicular C cells of the thyroid (thyroid hormone). In osteoclasts isolated and cultured on glass or plastic, it stops lamellipodial motility within minutes (Chambers and Magnus, 1982), yet its role in normal adult human calcium control appears modest.

The classic experiment: a hormone that lowers calcium

By the early 1960s, PTH alone explained calcium control: when calcium rose, PTH secretion fell. Copp, Cameron, Cheney, Davidson and Henze (Endocrinology, 1962) isolated the thyroid-parathyroid tissue of fasted, anesthetized dogs and perfused it through its own circulation. If high calcium only switched PTH off, perfusing the glands with high-calcium blood should let systemic calcium fall slowly, as existing PTH was cleared. Perfusion with blood containing 12 mg of calcium per 100 mL lowered systemic plasma calcium, and the authors argued the fall was too fast for PTH withdrawal. They inferred a calcium-lowering factor from the glands and named it calcitonin.

That ruled in an active, calcium-lowering signal released in response to high calcium, in this dog preparation, and ruled out withdrawal of PTH as the only response. It did not locate the source: Copp’s group placed it in the parathyroid, and later work placed it in the thyroid’s C cells. Nor did it show that calcitonin matches PTH in importance in adult humans.

The kidney sets the daily balance

GFR is in glomerular filtration and clearance in the kidney. Protein-bound calcium is not filtered; the ionized and complexed fractions, roughly 55 to 60 percent of total, are. Roughly 60 to 70 percent of filtered calcium is reabsorbed in the proximal tubule and 20 to 25 percent in the thick ascending limb, largely between cells. The distal convoluted and connecting tubules take back roughly 5 to 10 percent through cells, the step where PTH does most of its fine-tuning; PTH and the calcium-sensing receptor also act upstream. Phosphate is reabsorbed mainly in the proximal tubule, the step PTH and FGF23 turn down. Why thiazides lower urinary calcium and loop diuretics raise it is in tubular transport and diuretics.

Worked problem: calcium and phosphate in a day

Made-up round numbers for a reference adult at steady state. Assume bone formation equals resorption, no sweat losses, and a fixed filterable calcium fraction of 60 percent.

filtered load = GFR x filterable plasma concentration; fractional excretion = excreted / filtered load

1. Filtered load. GFR 125 mL per min = 0.125 L per min x 1,440 min per day = 180 L per day. Total plasma calcium 10 mg per dL = 100 mg per L, 60 percent filterable: 60 mg per L. Filtered load = 180 x 60 = 10,800 mg per day.
2. Fractional reabsorption. Urinary calcium is 200 mg per day. Fractional excretion = 200 / 10,800 = 0.0185 (1.85 percent); 98.15 percent is reabsorbed.
3. Balance. With bone in balance, net intestinal absorption equals urinary excretion, 200 mg per day. With an intake of 1,000 mg per day, fecal calcium = 1,000 - 200 = 800 mg per day.
4. Sensitivity. If fractional reabsorption falls from 98.15 to 97.0 percent with the same filtered load, as when PTH falls, excretion = 10,800 x 0.030 = 324 mg per day, 1.62 times the original. A fall of about 1.15 percentage points in reabsorption moves excretion by 62 percent, which is why the small, regulated distal step matters.
5. Phosphate. Made-up spot values from a physiology study in which plasma phosphate is below its usual range: plasma phosphate 2.0 mg per dL, urine phosphate 40 mg per dL, plasma creatinine 1.0 mg per dL, urine creatinine 100 mg per dL. Assume creatinine clearance approximates GFR, phosphate is treated as freely filtered, and both come from the same samples, so urine flow cancels.

FE(P) = (U_P x P_Cr) / (P_P x U_Cr)

FE(P) = (40 x 1.0) / (2.0 x 100) = 40 / 200 = 0.20, or 20 percent; 80 percent is reabsorbed. A kidney conserving phosphate at low plasma phosphate would excrete only a small fraction of its filtered load, so a fifth points to renal phosphate wasting, for example from a phosphaturic signal such as PTH or FGF23. The calculation cannot tell which cause.

GFR 125 mL per min Filterable plasma calcium 60 mg per L Fractional reabsorption 98.15 percent

Bone in balance (formation equals resorption) Net bone release 0 mg per day (used only when bone is not in balance)

Filtered load = GFR × 1.44 × filterable calcium

Urinary excretion = filtered load × (1 − fractional reabsorption)

Net intestinal absorption needed = excretion − net bone release

In this sketch: made-up round numbers for a reference adult at steady state, starting from the lesson’s worked problem (GFR 125 mL per min, filterable calcium 60 mg per L, 98.15 percent reabsorbed, bone formation equal to resorption, no sweat losses). The 1.44 is 1,440 minutes per day divided by 1,000 mL per L. The slider ranges, the 0 to 200 mg per day of net bone release are made up for the sketch. Each slider is moved on its own and filterable calcium is held at the value you set; no hormones are simulated, so the sketch does not show how the body would set or respond to any of these values. A fall of about 1.15 percentage points in reabsorption, from 98.15 to 97.0 percent, moves excretion by 62 percent in the lesson’s example. At whole-body balance, net intestinal absorption equals urinary loss plus net bone accretion (net bone release is negative accretion).

Rule to keep: filtered load = GFR × filterable calcium, and excretion = filtered load × (1 − fractional reabsorption).

The modern experiment: mice without FGF23

FGF23 was first tied to rare human disorders in which too much FGF23 activity makes the kidney lose phosphate. Is it needed normally? Shimada and colleagues (Journal of Clinical Investigation, 2004) deleted Fgf23 in mice. If FGF23 is a normal phosphaturic hormone, kidneys without it should keep more phosphate and make more calcitriol. Compared with littermates, the null mice had high serum phosphate, increased renal phosphate reabsorption and elevated calcitriol, with abnormal bone mineralization. In the same group’s injection study, recombinant FGF23 lowered serum phosphate even in parathyroidectomized rats, so that effect did not need PTH.

That ruled in endogenous FGF23 as necessary for normal phosphate and vitamin D regulation in the mouse. A lifelong, whole-body deletion cannot separate direct renal effects from those of high calcitriol or development, and says nothing quantitative about humans.

Predicting the four classic disorders

Start with untreated primary hyperparathyroidism and normal kidney function. A parathyroid source secretes PTH that ionized calcium no longer suppresses normally, so PTH is high or inappropriately normal for the calcium (endocrine axes and testing). Bone resorption rises through RANKL. Distal calcium reabsorption rises, yet urinary calcium can still be raised, because plasma calcium and so the filtered load rise. NaPi-IIa leaves the brush border, so plasma phosphate is often low-normal or low. Calcitriol synthesis rises, and so does gut calcium absorption. Net: calcium up, phosphate down or low-normal, PTH up, calcitriol up or high-normal.

The table adds the other disorders, untreated, in adults: directions, not criteria, which stage, treatment and other disease can change.

Setting	Calcium	Phosphate	PTH	Calcitriol
Primary hyperparathyroidism	High	Low-normal or low	High or inappropriately normal	High or high-normal
Vitamin D deficiency, early	Often normal	Often normal	High	Normal, or even raised
Vitamin D deficiency, later	Usually falls	Often falls	High	May fall
Early chronic kidney disease (CKD)	May stay normal	May stay normal	Often rising	May begin to fall
Advanced CKD	Normal or tending low	Usually high	High	Low
Hypoparathyroidism	Low	High	Low or inappropriately low	Low or low-normal

In early vitamin D deficiency, high PTH stimulates 1-alpha-hydroxylation, so calcitriol can be normal or high while vitamin D stores are low. In CKD, the baseline analysis of 3,879 adults in the Chronic Renal Insufficiency Cohort, nearly all in CKD stages 2 to 4, found FGF23 elevated while mean phosphate and median PTH were still normal (Isakova and colleagues, 2011), a cross-sectional finding that fits FGF23 rising first without proving the order. Later, the kidney retains phosphate, calcitriol falls and PTH rises. In hypoparathyroidism, losing PTH removes its boost to distal calcium reabsorption, its phosphaturia and its drive to calcitriol.

Design problem

Osteoblast-lineage cells, osteocytes and other cells can all make RANKL in culture. Design an experiment in mice to find which cell’s RANKL is needed for normal bone remodeling in adults, and say what each outcome would mean.

One way to do it

Cross mice carrying a floxed RANKL allele with Cre lines active in osteocytes and, separately, in earlier osteoblast-lineage cells, with floxed-only and Cre-only controls. Confirm specificity with RANKL mRNA and protein in sorted cells and a Cre reporter. Read out bone mass by micro-CT, osteoclast number on bone surfaces by histology, and serum markers. Use an inducible Cre switched on in adults to separate development from adult remodeling, then add a challenge that requires resorption, such as sustained PTH excess or unloading, to ask whether the source changes with the stimulus.

Osteopetrosis after osteocyte deletion, as Nakashima and colleagues found in 2011, means osteocyte RANKL is necessary in that setting, not that osteocytes are the only source at every age, site or in humans. No change after a deletion means that cell’s RANKL is dispensable there, or that another source compensates.