Alkaline Phosphatase Knock-Out Mice Recapitulate the Metabolic and Skeletal Defects of Infantile Hypophosphatasia

Authors


  • Presented in part at the 18th Annual Meeting of the American Society for Bone and Mineral Research, Seattle, WA, U.S.A. (September 7–11, 1996) and the 46th Annual Meeting of the American Society of Human Genetics, San Francisco, CA, U.S.A. (October 30–November 2, 1996).

Abstract

Hypophosphatasia is an inborn error of metabolism characterized by deficient activity of the tissue-nonspecific isoenzyme of alkaline phosphatase (TNSALP) and skeletal disease due to impaired mineralization of cartilage and bone matrix. We investigated two independently generated TNSALP gene knock-out mouse strains as potential models for hypophosphatasia. Homozygous mice (–/–) had < 1% of wild-type plasma TNSALP activity; heterozygotes had the predicted mean of ∼50%. Phosphoethanolamine, inorganic pyrophosphate, and pyridoxal 5′-phosphate are putative natural substrates for TNSALP and all were increased endogenously in the knock-out mice. Skeletal disease first appeared radiographically at ∼10 days of age and featured worsening rachitic changes, osteopenia, and fracture. Histologic studies revealed developmental arrest of chondrocyte differentiation in epiphyses and in growth plates with diminished or absent hypertrophic zones. Progressive osteoidosis from defective skeletal matrix mineralization was noted but not associated with features of secondary hyperparathyroidism. Plasma and urine calcium and phosphate levels were unremarkable. Our findings demonstrate that TNSALP knock-out mice are a good model for the infantile form of hypophosphatasia and provide compelling evidence for an important role for TNSALP in postnatal development and mineralization of the murine skeleton.

INTRODUCTION

Hypophosphatasia is a heritable form of rickets or osteomalacia characterized biochemically by subnormal activity of the tissue-nonspecific isoenzyme of alkaline phosphatase (TNSALP).(1,2) To date, 70 different mutations in the TNSALP gene have been identified in patients.(3,4) In this disorder, three phosphocompounds (phosphoethanolamine [PEA], inorganic pyrophosphate [PPi], and pyridoxal 5′-phosphate [PLP]) accumulate endogenously and seem therefore to be natural substrates for TNSALP.(2) Six clinical types of hypopophosphatasia in humans encompass extremely variable disease expression that ranges from intrauterine death with profound skeletal hypomineralization (perinatal form) to dental manifestations alone in children or in adults (odontohypophosphatasia).(1,2) In the infantile form, patients seem well as neonates, but then present with failure to thrive and rachitic disease before 6 months of age. Skeletal deterioration is common thereafter, and it has been estimated that 50% of these infants die several months later.(5)

There is no established medical therapy for hypophosphatasia.(6) Enzyme replacement therapy by intravenous infusion of alkaline phosphatase (ALP) and additional treatment approaches have generally not been helpful.(1) Preliminary experience involving one severely affected infant includes transient clinical and radiographic improvement, but no biochemical changes, following bone marrow transplantation.(7) Accordingly, an animal model for hypophosphatasia is needed to further understand the physiological role of TNSALP and to evaluate potential treatments.

As reported in 1995 and 1997, two laboratories independently disrupted the mouse TNSALP gene generating null mutant strains.(8,9) The initial biochemical, radiographic, and histopathological findings suggested that these animals were a promising model for hypophosphatasia. Here we describe in detail the metabolic and skeletal disease in each type of knock-out mouse. We report that phenotypic differences exist between the two strains, but that both represent a good model for the severe infantile form of this inborn error of metabolism.

MATERIALS AND METHODS

Mice and husbandry

Animal treatment:

All studies were conducted in accordance with the highest standards of humane animal care and in compliance with federal, state, and local laws and institutional regulations.

Mice:

The two TNSALP knock-out mouse colonies were created in two independent laboratories. The initial descriptions were published elsewhere.(8–10) The colony developed in the laboratory of MacGregor and coworkers,(8,10) now at Emory University, is designated EM in this paper; the colony generated in the laboratory of Millán and colleagues in La Jolla(9) is designated LJ.

Pyridoxal treatment:

As published by Waymire et al.,(8) prolonged survival of homozygous TNSALP knock-out mice (–/–) requires parenteral administration of pyridoxal (PL) to prevent lethal seizures that seem to be due to low PL levels. PL was injected subcutaneously each day for mice up to 12 days of age. Subsequently, daily intraperitoneal (ip) injections were used. The PL dose was adjusted to body weight (∼200 μg/g) and according to observed seizure control. Radiographic assessment indicated that daily PL dosing for 18 days did not cause skeletal disease in +/+ mice (n = 4).

Feeding:

The –/– mice were fed a soft diet mixture containing Purina Mouse Chow, Enfamil with iron (Mead Johnson & Co., Evansville, IN, U.S.A.), and corn oil. In addition, it was necessary to give supplemental gastric tube feedings daily with Nutri-Cal (Evsco Pharmaceuticals, Buena, NJ, U.S.A.) beginning at ∼7 days of age.

Urine collection:

Urine was collected under anesthesia when necessary. At sacrifice, specimens were removed from resected bladders with a tuberculin needle and syringe.

Biochemical determinations

Alkaline phosphatase:

Enzymatic activity was quantitated as previously described using the Tris-buffered saline preparation at pH 9.6 and 37°C with a saturating amount (1.5 mM) of 4-methylumbelliferyl phosphate substrate.(11) Units are reported as nanomoles per minute per milliliter. This nonphysiological ALP substrate was selected because of its detectability by fluorescence, yet it is hydrolyzed similarly to the putative natural TNSALP substrates PEA and PPi.(11)

Calcium:

Calcium concentrations were determined reproducibly using as little as 0.5 μl of plasma or 0.25 μl of urine by our modification of the o-cresolphthalein complex ion method(12,13) (Procedure no. 575; Sigma Chemical Co., St. Louis, MO, U.S.A.). Calcium working solution was altered by combining calcium-binding reagent and calcium buffer in a 1.25:1 ratio. The biological sample (13.75 μl), or sample plus dH2O, was combined with 112.5 μl of calcium working solution. Absorbance was read at 570 nm on a microtiter plate reader (MR5000; Dynatech, Chantilly, VA, U.S.A.) after 3 minutes.

Phosphate:

Plasma inorganic phosphate (Pi) concentration was determined using as little as 0.5 μl of plasma by a modification of the ammonium molybdate method(14) (Procedure no. 360-UV; Sigma Chemical Co.). Briefly, 10 μl of plasma plus saline was mixed in a microtiter plate with 250 μl of 400 μM ammonium molybdate, and absorbance was read at 410 nm using a microtiter plate reader (MR5000; Dynatech). The urine phosphorus concentration, measured reproducibly using as little as a 0.25 μl sample, was determined by a modification of the Fiske and SubbaRow method(15) (Procedure no. 670; Sigma Chemical Co.). Generally, 0.75 μl of urine was trichloracetic acid (TCA) precipitated (8% final) in a total volume of 75 μl by a 10 minute, room temperature incubation and centrifuged at 10,000g for 10 minutes. To 80 μl of supernatant (or supernatant + 8% TCA), 120 μl of dH2O, 40 μl of acid molybdate, and 10 μl of Fiske and SubbaRow reducer was added. After a 10 minute, room temperature incubation, A630 was measured on a microtiter plate reader (MR5000; Dynatech).

Creatinine:

The urine creatinine (Cr) concentration was determined in microtiter plates by an acid picrate modification of the method by Jaffé(16) (Procedure no. 55; Sigma Chemical Co.) using 21 μl of urine (or urine + dH2O) and 210 μl of alkaline picrate solution. Absorbance was read at 490 nm (MR 5000; Dynatech). Acid reagent (10 μl) was added to samples and absorbance (A490) was read again.

Phosphoethanolamine:

Urinary PEA levels were determined in the Metabolic Genetics Laboratory of the Department of Pediatrics, Washington University School of Medicine, St. Louis, MO, U.S.A. using a Beckman 7300 analyzer and reagents and procedures provided by the manufacturer (Beckman, Palo Alto, CA, U.S.A.).

Inorganic pyrophosphate:

Urinary PPi assays were conducted using a modification of the radiometric uridine diphosphoglucose pyrophosphorylase method with established sensitivity in the picomole range.(17)

B6 vitamers:

Plasma concentrations of PLP and PL were determined as described using cation-exchange high-performance liquid chromatography.(18)

Radiography

Radiographic survey of the skeleton was performed with highly sensitive mammographic procedures (using magnification and adjusting the mass and the X-ray energy level to optimize skeletal detail). Single emulsion, high-density film was used with extended processing time. More than 80 animals (total) were studied; a minimum of 3 mice were examined from each group at given ages.

Skeletal histology

Sample preparation:

Because hypophosphatasia features impaired mineralization of the skeletal matrix, we used established histologic techniques for nondecalcified bone specimens.(19) Skeletal samples (after fixation in 10% Pi-buffered formalin) were first embedded in a water-soluble methacrylate (glycol derivative, JB-4; Polysciences, Inc., Warrington, PA, U.S.A.). Tissue sections were then obtained using a Zeiss Rotary Microtome with a glass triangular knife. A minimum of four transverse slices (assuring inspection at different depths into the growth plate) were examined in more than 20 mice.

Tetracycline labeling in vivo was accomplished by injecting mice (30 days of age or older) ip with 30 μg of oxytetracycline/g of body weight on day –8, and again on day –3. Mice were sacrificed on day 0. After removing the distal end of the tibia, the proximal end was processed for nondecalcified histology(20) using Millonig's fixative for 20–24 h, followed by successive dehydrations and embedding in methyl methacrylate.

Staining procedures:

A triple stain (for ALP activity, Fast Green, and Alizarin Red) was devised. Skeletal sections were placed on charged glass slides (Fisher Scientific, Pittsburgh, PA, U.S.A.) and fixed in citrate/acetone (85–4C; Sigma) for 3 minutes according to the manufacturer. Slides were then incubated in NBT/BCIP (#34041; Pierce Chemical Co., Rockford, IL, U.S.A.) at room temperature for 30 minutes to 3 h in the dark until optimal color was achieved. Next, the slides were rinsed with dH2O and counterstained with 2% Fast Green for 5–15 minutes, followed by 3 × 10 s rinses in 1% acetic acid. The slides were then stained with Alizarin Red as described.(21) Samples were dehydrated in successive steps of 95% and 100% ethanol, and then with 100% xylene (2 × 2 minutes). Toluidine blue and modified Masson trichrome stains were performed as described.(22)

RESULTS

Matings of TNSALP +/− breeding pairs produced viable homozygous offspring with approximately the 25% frequency predicted for Mendelian inheritance. Genotypes of the pups were established by ALP assay using plasma from 1 μl of tail blood. As illustrated in Fig. 1, knock-out mice (–/–) showed ∼1% of wild-type (+/+) plasma ALP activity; heterozygotes had a mean of ∼50%. Previous PCR experience with both mouse colonies confirmed the precision of genotyping by plasma ALP activity quantitation.(8–10) TNSALP knock-outs had a smaller body weight (and length; data not shown) at birth, which persisted with age. This deficit was more marked in the EM mice (Fig. 2).

Figure FIG. 1.

Plasma ALP activity determined < 24 h after birth. PCR analyses previously demonstrated(8–10) that genotype is the basis for the three clusters of ALP activity reflecting knock-out (–/–), heterozygote (+/–), and control (+/+) mice. Individual values, means (circle in a square, circle in a circle), and SEM are depicted.

Figure FIG. 2.

Body weight versus age in EM versus LJ mice. Knock-out mice have lighter body weights despite rescue with soft diets and PL treatment (see Materials and Methods). The weight deficiency is greater for the EM knock-outs.

Substrate accumulation

As in patients with hypophosphatasia, there was a striking elevation of urinary PPi levels in the knock-out mice. Illustrated in Fig. 3 are values found at 25 (EM) and 6 (LJ) days of age. Hence, TNSALP in mice appears to function physiologically as a PPi-phosphatase. Also shown in Fig. 3 are elevated levels of urinary PEA and a striking accumulation of plasma PLP in the knock-outs. Plasma levels of the dephosphorylated PLP product, PL, were low in the EM knock-outs (days 10–14) despite PL therapy. This is in contrast to all but a subset of especially severe hypophosphatasia patients.(23,24) PL levels in the LJ mice were normal (without vitamin supplementation) at 6 days of age. We previously reported that low PL levels in the EM knock-out mice could explain the decreased γ-aminobutyric acid (GABA) levels in their central nervous system and their lethal seizures.(8) These seizures (which are prevalent in both the EM and LJ mice) can be prevented (and the life span of –/– mice increased) by daily ip injections of PL (see Materials and Methods). Despite PL injections, however, death occurs prematurely in the knock-outs, although the mechanism is not fully understood.(8–10)

Figure FIG. 3.

Substrate accumulation. Marked elevations of the TNSALP substrates found in hypophosphatasia patients occur in the EM and LJ knock-out mice (note log scale). EM mice were examined at 25 days of age for levels of urinary PPi (+/+, n = 6; –/–, n = 5), and PEA (+/+, n = 5; –/–, n = 7). Plasma PLP and PL data are shown for 10- to 14-day-old mice (+/+, n = 13; –/–, n = 9) as previously reported.(8) LJ mice were examined at 6 days of age for levels of urinary PPi (+/+, n = 14; –/–, n = 8), urinary PEA (+/+, n = 6; –/–, n = 7), and plasma PLP and PL (+/+, n = 6; –/–, n = 4). Urinary data were normalized to grams of creatinine (Crt).

Radiography

Radiographic studies demonstrated no evidence of skeletal disease in knock-out mice during the first 8 days of life. As shown in Fig. 4A (EM), the +/+ and –/– mice had indistinguishably mineralized skeletons at 6 days of age with obvious metaphyses. Secondary growth centers (epiphyses) had not appeared at the knees.

Figure FIG. 4.

Radiographs of the lower quadrants or of the knees. Findings are shown at the indicated ages. (A, B, E, and F) EM mice. (C and D) LJ mice. Although knock-out mice appear normal at 6 days of age, delayed appearance of epiphyses or secondary ossification centers (white arrows), fractures (open arrow), osteopenia, and bone deformities are noted with aging.

By day 9 (EM) (Fig. 4B), however, when secondary ossification centers at the knees were just beginning to appear in the control tibia (and forepaws), they were absent in the knock-outs. Similar observations were made for the LJ mice (Fig. 4C); epiphyses in the LJ +/+ mice were well mineralized by day 10, but absent in the –/– mice.

At subsequent ages (in contrast to controls), both the EM and the LJ knock-outs had indistinct metaphyses, hypomineralized condyles, and persistent absence of the secondary ossification centers of the tibias. By day 23, fractures and skeletal deformities were apparent, especially in LJ mice (Fig. 4D). On day 37, osteopenia was noted and the femurs and the tibias were now markedly shorter in the knock-outs (EM; Fig. 4E). The secondary ossification centers in the feet and tail were well formed in the +/+ mice but remained inapparent in the knock-outs.

Although the knock-outs often died before puberty, extreme skeletal disease was present in those few mice who survived to 6 months of age (shown at day 62; Fig. 4F).

Figure 5 depicts in greater detail a longitudinal radiographic examination of secondary ossification centers at the metaphyses in the LJ mice. The minimum number of animals examined in each group at given ages was two. Nevertheless, the findings were invariant and in keeping with the progressive changes noted over time.

Figure FIG. 5.

Detailed radiographs of the hind limbs of LJ mice. There is delayed appearance of secondary ossification centers and frayed metaphyses in the knock-out mice at the ages indicated.

Although we previously reported macroscopic abnormalities in EM knock-out teeth,(8) dental and skull radiographs appeared normal (data not shown). Carrier (+/–) mice (also not shown) did not have radiographic evidence of skeletal disease.

Plasma and urine biochemistry

Despite their skeletal disease and profound deficiency of ALP activity, plasma and urine calcium and phosphate levels (data not shown) were similar in –/– versus +/+ EM and LJ mice (assessed at 18 and 25 days of age, respectively). These results essentially excluded other causes of rickets (see below).

Histology

Histologic study of nondecalcified sections of long bones showed normal appearances at birth through 8 days of age in both the EM and LJ knock-out mice, consistent with their unremarkable radiographs at this early age. Figure 6A illustrates a toluidine-stained, proximal tibia of a 1-day-old control mouse next to a knock-out littermate. In both mice, there was a well organized growth plate with the characteristic reserve, proliferative, and hypertrophic zone of chondrocytes. Mineral spicules were found between the stacks of lower hypertrophic cells and extended into the metaphyseal bone in both types of mice. Osteoblasts lined the mineralized bone and osteoidosis (excessive unmineralized matrix) was not apparent in the knock-outs. Figure 6B (similar sections) illustrates staining results using alizarin red for mineral, and histochemical exposure of ALP activity. In the secondary ossification centers, no mineral was seen in the newborn EM or LJ control or knock-out mice (previously noted radiographically). Mineral staining was also similar at the lower hypertrophic zones of their femurs and tibias. At higher magnification (Fig. 6C), histochemistry showed abundant ALP staining in the lower proliferative zones and in the hypertrophic zones in the control animals. However, in the knock-out mice, there was virtual absence of ALP activity.

Figure FIG. 6.

Histology of proximal tibias of 1-day-old EM mice. (A) Toluidine staining. Epiphyses (E) containing undifferentiated chondrocytes, growth plates (GP), and bone (B). Development is similar in control and knock-out mice. (B) ALP histochemistry (purple) and mineral staining (alizarin red). Normal mineralization at this early age is seen in the knock-out mouse, despite the complete absence of ALP staining. (C) ALP histochemistry and mineral staining at higher magnification. The +/+ control shows characteristic ALP staining which is nearly absent in the resting zone (RZ), but gradually increases in intensity in the proliferative zone (PZ) and the hypertrophic zone (HZ). Strong staining of osteoblasts and periosteal cells is also noted. No ALP staining is observed in the knock-out. In both mice, red-staining mineral is observed between the columns of hypertrophic cells (mineralization front) and was observed in the cortical bone.

By 18 days of age in the knock-outs, histopathology was evident. Illustrated in Fig. 7A, tibial growth plates had markedly smaller lower hypertrophic zones with widened vascular invasion channels. The chondrocytes in this region remained nested.

Figure FIG. 7.

Histology of older mice. (A) At 18 days of age, toluidine-stained proximal tibial growth plate of LJ mice. In the –/– knock-out mouse, a blunted lower hypertrophic zone with widened vascular invasion channels is seen; the lower hypertrophic zone chondrocytes are nested. (B) At 25 days-of-age, modified Masson trichrome-stained proximal tibial growth plate of EM mice. In the knock-out, there is an enlarged upper hypertrophic zone (UHZ); the lower hypertrophic zone (LHZ) is blunted, and blue-staining mineral aberrantly extends between the columns of chondrocytes. (C) At 25 days of age, modified Masson trichrome-stained diaphyseal shaft of EM mice. The diaphysis of the knock-out mouse shows a widened osteoid seam (pink-staining: O) that extends into the blue-staining mineralized bone. (D) At 101 days of age, modified Masson trichrome stain of trabecular bone tissue of EM mice. Gross osteoidosis (pink-staining material: O) is seen in the knock-out mouse.

By 25 days of age (Fig. 7B) in the knock-outs, the upper hypertrophic zone was enlarged. The lower hypertrophic zone remained blunted, but showed blue-staining mineral (Masson trichrome stain) extending between the chondrocyte stacks (in contrast to controls).

In the controls, the cortical bone of the diaphyses contained narrow and well delineated osteoid seams (Fig. 7C). In the knock-outs, however, these seams were widened and there were large islands of osteoid within the mineralized bone.

By 101 days of age in the knock-outs, the cancellous bone was composed almost entirely of unmineralized osteoid (Fig. 7D). Tetracycline fluorescence was not apparent.

In knock-outs at all ages, ALP staining remained negative (data not shown). There was no evidence of peritrabecular fibrosis, osteoclastosis, or numerous osteoblasts. These observations suggested that secondary hyperparathyroidism was not present.

In +/+ and heterozygous +/– mice, dual tetracycline labeling at 3–4 months of age resulted in two fluorescent bands, revealing normal and indistinguishable mineral appositional rates. In contrast, –/– mice had at most a single diffuse tetracycline label (confirming a mineralization defect).

DISCUSSION

Patients with hypophosphatasia manifest the following major pathophysiological disturbances: hypophosphatasemia due to global deficiency of TNSALP activity; endogenous accumulation of the putative ALP substrates PEA, PPi, and PLP; and impaired skeletal matrix mineralization leading to rickets or osteomalacia.(1) In general, the clinical severity of this inborn error of metabolism is correlated inversely with age of onset of the skeletal disease and with the circulating age-appropriate levels of TNSALP activity, and directly with plasma PLP concentrations.(1) We now document in TNSALP knock-out mice these same principal biochemical and skeletal aberrations, and a clinical course that resembles especially the infantile form of the disease.

Biochemical disturbances

Hypophosphatasemia in EM and LJ knock-out mice strains at all ages examined is similar in severity to the ALP deficiency encountered in patients with the perinatal or infantile forms of hypophosphatasia (< 2%).(1,2) Predictable from clinical experience, absence of TNSALP in the knock-out mice is lethal early in life.(1,2)

Our previous investigation of a relatively large group of patients with more mild TNSALP deficiency (childhood hypophosphatasia) confirmed accumulation of PEA in the urine, with values ranging from 0.5 mmol/g to 1.5 mmol/g of Cr (control mean 0.15 mmol/g of Cr).(25) The EM and LJ knock-out mice have much greater urinary PEA concentrations (1670 ± 254 and 34 ± 5 mmol/g of Cr) that were, respectively, 1400- and 29-fold elevated compared with control means. This difference may reflect the severity of the TNSALP deficiency in the knock-out mice and/or indicate that other phosphatases present in humans (but absent in mice) have some physiological specificity toward PEA. PLP levels in severely affected patients with hypophosphatasia are elevated in plasma to about 5000 nM.(1) Similar levels were noted in the EM and LJ –/– mice (mean 3170 ± 1290 and 1200 ± 65 nM vs. control values of 133 ± 48 and 183 ± 104 nM, respectively). In childhood hypophosphatasia, elevated urinary PPi levels range from 500 μmol/g to 1000 μmol/g Cr (the control mean was ∼200 μmol/g of Cr(25)). In the EM and LJ –/– mice, urinary PPi levels are in this range (833 ± 99 μmol/g and 386 ± 5 μmol/g of Cr, respectively), perhaps helping to explain the delayed onset of the skeletal disease (see below).

As we reported in the EM knock-out mouse,(8) TNSALP deficiency in the animals is associated with low plasma levels of PL and epileptic seizures (70 ± 61 nM vs. 874 ± 354 nM in controls). Seizures also occur in the LJ knock-out mice.(9) However, here we found that plasma PL levels in unsupplemented 6-day-old LJ knock-outs were normal at the time when seizures first occurred. It may be that PL levels diminish with age in the LJ mice, or that this finding represents the genetic differences between the two strains (see below). In 1985, we reported that plasma levels of PL were normal in five patients with severe forms of hypophosphatasia.(23) However, we subsequently encountered a few severely affected patients who had low plasma PL levels(24) and some had epilepsy.(26) Thus, this variability in plasma PL levels encountered in genetically heterogeneous patients seems to be reflected in the EM and LJ knock-outs which differ in genetic background.(8–10) We have speculated that the small amount of ALP activity remaining in most seizure-free patients is sufficient to dephosphorylate PLP to PL, the metabolite of vitamin B6 that is permeable to plasma membranes.(1) Intracellularly, PL is rephosphorylated by PL kinase to PLP, the cofactor form of vitamin B6. Perhaps in the knock-out mice, complete absence of TNSALP activity results in plasma PL levels that are too low to sustain vitamin B6–dependent cellular homeostasis, including glutamate decarboxylase activity and the synthesis of GABA in the brain. Evidence that low levels of GABA may be responsible for the seizures in EM knock-outs includes GABA quantitation, and rescue by PL administration.(8) Alternatively, we have reported anatomic disturbances in the CNS itself that perhaps account for the epilepsy.(9) However, there is some anecdotal evidence that PEA may be epileptogenic in hypophosphatasia patients.(27) The especially great elevations of endogenous PEA levels in the TNSALP knock-out mice could cause this complication, but would not be expected to respond to PL administration.

Other biochemical disturbances noted in hypophosphatasia patients include hypercalcemia in severely affected infants, and hyperphosphatemia in the childhood and adult forms of the disease.(1) The knock-out mice had normal calcium and Pi levels in blood and urine. This observation is significant because it excludes aberrations in extracellular mineral concentrations in the pathogenesis of their severe skeletal disease.

Skeletal defects

Radiographic findings:

In the perinatal form of hypophosphatasia, skeletal disease manifests in utero in affected neonates with < 2% serum ALP activity. In the EM and LJ knock-out mice, radiographs show normal bones at birth. Subsequently, rachitic disease (that is complicated by gradual generalized skeletal demineralization) develops in a pattern that is typical of patients with the infantile form of hypophosphatasia.(1) Accordingly, it is unclear why skeletal disease is not more severe in the knock-out mice having virtually no endogenous TNSALP activity during gestation. Perhaps maternal TNSALP or other fetal phosphatase activities expressed in utero in mice can compensate for the virtual absence of TNSALP. In fact, other investigators have noted two partially characterized murine Mn2+-dependant phosphatases that could provide some protective activity.(28) We detected such activity in the plasma of these mice by its Mn2+ dependence and sensitivity to the TNSALP-specific inhibitor l-tetramisole and found that this activity accounted for ∼5% of the total plasma ALP activity in wild-type mice. This Mn2+-dependent ALP isoenzyme is encoded by the Akp1 locus on mouse chromosome 1, which is distinct from the Akp2 locus that encodes TNSALP on mouse chromosome 4.(29) Targeted mutagenesis of the Akp2 TNSALP gene should have no direct effect upon expression of the Mn2+-dependent isoenzyme of ALP. Thus, it is possible that any residual ALP activity found in serum of TNSALP knock-out animals is contributed to by this Mn2+-dependent isoenzyme. Issues of genetic redundancy in murine ALP function could be addressed by investigation of mice lacking both Akp2 (TNSALP) and Akp1 (Mn2+-dependent ALP).

Histopathology–Bone:

Histopathological changes are essentially limited to hard tissues in patients with hypophosphatasia.(1) In all but the mildest cases, nondecalcified sections of bone reveal evidence of defective skeletal matrix mineralization, consistent with rickets or osteomalacia.(30,31) The degree of rickets or osteomalacia in patients generally reflects the clinical severity of their disorder.(31)

TNSALP knock-out mice also have defective skeletal mineralization, documented at 18 days of age by widened osteoid seams, which progresses to gross osteoidosis by age 3 months. Features of secondary hyperparathyroidism (present in most types of rickets or osteomalacia associated with hypocalcemia) are absent in TNSALP knock-out mice as in patients with hypophosphatasia.(31) This observation documents that TNSALP itself plays a critical role in mineralization of the mouse as well as the human skeleton.

Although a pivotal role for TNSALP in skeletal matrix mineralization is established in humans and now in mice, the mechanism remains uncertain.(2) We recently reported that skeletal tissue from infants who succumbed to hypophosphatasia is replete with matrix vesicles that contain normal amounts of mineral having characteristics of hydroxyapatite.(32) However, there appeared to be no further enlargement of these crystals after the matrix vesicles had ruptured. Possibly, these observations are explained by increased extracellular levels of PPi within the hypophosphatasia skeleton, because PPi would be expected to block hydroxyapatite crystal growth by surface mineral adsorption.(1,2) These results from hypophosphatasia patient studies indicate that TNSALP is not required for the initial phase of skeletal mineralization, but is necessary for hydroxyapatite crystal growth.(32) Our current study of TNSALP knock-out mice documenting PPi accumulation is consistent with this hypothesis.

Histopathology–Growth plates:

As recently reviewed, mammalian physeal chondrocytes differentiate through a series of maturational stages, establishing different histologic zones, with many secreted signaling molecules which activate different types of chondrocytes.(33–35) The extracellular matrix of growth plate cartilage is initially synthesized by the prechondrocytes and is rich in type II collagen (which provides structural integrity) and aggrecan (which confers resiliency).(33) Chondrocytes in the resting zone are the target for 24,25-dihydroxyvitamin D3, which induces differentiation into proliferative zone cells, with concomitant changes in membrane fluidity, Ca2+ ion fluxes, phospholipid metabolism, as well as protein kinase C and metalloproteinase activity.(33) During elongation of the growth plate, the level of mRNA expression increases several-fold in chondrocytes, which then further differentiate into large hypertrophic cells and initiate expression of type X collagen.(33,34) Bone morphogenetic proteins appear to play key roles.(35)

In several select patients with severe hypophosphatasia, we have noted delayed appearance of some secondary ossification centers (unpublished observation). We now find that delayed formation or complete failure to mineralize epiphyses is also an important feature of TNSALP knock-out mice. In both the EM and LJ knock-outs, we report here a developmental arrest of chondrogenesis. This defect includes not only a failure to form secondary ossification centers, but also to develop the lower hypertrophic zone in growth plates. In our TNSALP knock-out mice, epiphyseal development is disrupted and there are only a few hypertrophic chondrocytes. Instead, these cells remain nested, indicating a defect in cartilage matrix production. In vitro, this developmental step can be blocked by antisera directed against avb3 integrin, the receptor for fibronectin and osteopontin (Kayath MJ, Tu C-L, Damsky C, Erickson HP, Strewler GJ 1997 Cell matrix interactions in chondrogenesis. Keystone Symposia on Molecular and Cellular Biology. “Bone, Cartilage and Collagen: Growth and Differentiation.” Santa Fe, NM, U.S.A.). It has been postulated that osteopontin functions during cell adhesion; binding to the avb3 receptor can be modulated by the phosphorylation state of osteopontin.(36,37) In a preliminary communication, we reported that TNSALP-deficient fibroblasts from hypophosphatasia patients produce matrix containing excessively phosphorylated osteopontin with aberrant ligand-binding activities.(38) Accordingly, the dyschondrogenesis in TNSALP knock-out mice may result from abnormal developmental signals due to deficient dephosphorylation of matrix phosphoproteins.

Differences between the two mouse colonies

Each of the major findings in this study was observed in the knock-out mice from the EM and the LJ colonies. However, the severity of the hypophosphatasia phenotype did differ somewhat between them. In the EM –/– mice, reductions in body weight, growth velocity, and plasma PL concentrations as well as increased urinary PEA levels were greater than in the LJ knock-outs. However, fractures seemed more prevalent in the LJ mice. Nevertheless, the ALP gene inactivation procedures themselves were not expected to underlie differences between the knock-out mice, because they both resulted in undetectable levels of TNSALP message expression (using reverse transcriptase polymerase chain reaction against each of the exons). The mutant alleles of TNSALP have been confirmed to be null for ALP activity in both mouse models(8–10) (and data not shown). Thus, it is likely that differences in the background strain composition of the two colonies are responsible for the variability in the mutant phenotypes. Although the ES cell lines used to transmit the targeted mutation through the germline were similar (129/Sv vs. 129/J for EM and LJ, respectively), the chimeric animals were bred with C57BL/6 females (EM) or 129/SvJ females (LJ). Accordingly, the genetic constitutions of the resulting mouse strains were different. The strain composition (genetic background) of the EM breeding colony was ∼87% C57BL/6, 13% 129/Sv. For the LJ colony, it was 25% C57BL/6, 75% 129/J. In fact, several examples show that genetic background can have a marked influence on phenotype after gene disruption in mice.(8,39–44) Thus, for example, it is likely that the greater incidence of fractures and more severe skeletal deformities in the LJ knock-out mice resulted either from their higher proportion of 129/J and/or reduced C57BL/6 strain composition. In support of this hypothesis, we have observed that the phenotype of EM knock-out animals becomes more severe after backcrossing with 129/J mice. Indeed, such subtle differences in parental allele combinations could also underlie the variability observed in clinical expression of hypophosphatasia in affected siblings.(1,2)

Acknowledgements

Skilled technical assistance was provided by Fikret Terzic, Michelle Meyer, Robert Henry, and Teresa Tolley (Washington University School of Medicine, St. Louis, MO, U.S.A.), J. Dennis Mahuren (Fort Wayne State Developmental Center, Fort Wayne, IN, U.S.A.) performed the PLP and PL analyses. Histology was reviewed by Dr. Steven L. Teitelbaum and Dr. Robert B. Kimble (Washington University School of Medicine) and Dr. H. Clarke Anderson (University of Kansas Medical Center, Kansas City, KS, U.S.A.). Darlene Harmon (Shriners Hospital for Children, St. Louis, MO, U.S.A.) provided expert secretarial help. This work was supported in part by grants #15963 and #15958 from the Shriners Hospitals for Children; The Clark and Mildred Cox Inherited Metabolic Bone Disease Research Fund; grants CA42595, AR 38656, PO1-AG-13918–01, DE12889, and HD-36437 from the National Institutes of Health; grant 95–37200–1703 from the USDA NRICGP; and the Swedish Medical Research Council.

Ancillary