Molecular basis of canine muscle type phosphofructokinase deficiency.

Muscle type phosphofructokinase (M-PFK) deficiency is a rare inherited glycogen storage disease in humans that causes exertional myopathy and hemolysis. The molecular basis of canine M-PFK deficiency, the only naturally occurring animal homologue, was investigated. Lack of M-PFK enzyme activity was caused by a nonsense mutation in the penultimate exon of the M-PFK gene, leading to rapid degradation of a truncated (40 amino acids) and therefore unstable M-PFK protein. A polymerase chain reaction-based test was devised to identify M-PFK-deficient and carrier animals. This represents one of only a few inborn errors of metabolism where the molecular defect has been identified in a large animal model which can now be used to develop and assess novel therapeutic strategies.

Phosphofructokinase deficiency, due to a lack of muscle type phosphofructokinase (M-PFK) 1 subunits or activity, has been described in humans as a rare autosomal recessive trait. This disorder, also known as glycogenosis type VII or Tarui-Layzer syndrome, causes a metabolic myopathy and compensated hemolytic disorder (1). Mutations in the M-PFK cDNA have been described in Ashkenazi Jewish, Japanese, French Canadian, and Swiss patients, but little information is available regarding these defects at the protein level (2)(3)(4).
A naturally occurring animal model of M-PFK deficiency has been described in English springer spaniels (5,6). The affected dogs have a chronic compensated hemolytic disorder and exertional myopathy, as is seen in human patients. However, these animals most often present with hemolytic crises due to the high capacity of the dog for aerobic work and the intrinsic alkaline fragility of erythrocytes (7). Dogs with M-PFK deficiency have 6 -22% of normal erythrocyte PFK activity and 1-4% of normal muscle PFK activity (5)(6)(7)(8)(9). The normal canine M-PFK cDNA has been sequenced recently (10), making it possible to elucidate the underlying molecular mechanisms of canine M-PFK deficiency. We identified a single nonsense mutation that leads to rapid degradation of an unstable truncated M-PFK protein.

EXPERIMENTAL PROCEDURES
Animals-Unless otherwise indicated, tissue and blood samples for PFK studies were obtained from a colony of English springer spaniels, founded and maintained at the School of Veterinary Medicine, the University of Pennsylvania. The M-PFK-deficient animals used to create this colony originated from Wisconsin, Maryland, and Indiana. The pedigrees of the three founding families have no common ancestor for at least six generations (11). All animals were maintained in accordance with the guidelines of the Institutional Animal Care and Use Committee of the University of Pennsylvania.
Muscle samples for PFK activity assays, protein studies, and RNA isolation were obtained by surgical biopsy of pentobarbital-and halothane-anesthetized dogs or by sterile excision post-mortem. These tissues were snap frozen in liquid nitrogen and stored at Ϫ70°C. Blood samples anticoagulated with EDTA were kept on ice until used for PFK enzyme activity assays and DNA extraction. Animals were considered to be PFK-deficient (affected), carriers (heterozygous asymptomatic), or normal based on erythrocyte and muscle PFK activities and pedigree analysis (9).
Immunoblots-Immunoblots of M-PFK were prepared as described previously (12). Briefly, 100 g of total protein was run on a 7% denaturing polyacrylamide gel and transferred to a 0.45-m nitrocellulose membrane. The membrane was blocked with 5% milk powder and was probed with guinea pig anti-rabbit M-PFK antiserum (12).
Metabolic Labeling-Primary canine myoblasts were obtained and cultured from a normal and an M-PFK-deficient dog, as described previously (13). The myoblasts were plated at high density (approximately 1.25 ϫ 10 3 /mm 2 ) and allowed to differentiate and form myotubes in Dulbecco's modified Eagle's medium supplemented with 10% horse serum. The medium was removed, and the adherent cells were washed with phosphate-buffered saline. Minimum essential medium (Eagle's medium-deficient) (Sigma), supplemented with L-leucine, L-lysine, and sodium bicarbonate, but without methionine, was added for 30 min. The medium was removed and deficient medium (above) supplemented with [ 35 S]methionine (Ͼ1000 Ci/mmol, Amersham Corp.) at 5.5 Ci/ml was added for 18 h. The cells were washed with phosphate-buffered saline and either harvested or Dulbecco's modified Eagle's medium was added for periods of up to 24 h. M-PFK was then immunoprecipitated with rabbit anti-dog M-PFK antibody as described and visualized by 7% SDS-polyacrylamide gel electrophoresis (PAGE) followed by autoradiography (12,14).
RNA Blots-Total muscle RNA was prepared from a PFK-deficient English springer spaniel and a normal beagle by the method of Chomczynski and Sacchi and was subjected to formaldehyde-agarose gel electrophoresis (15). The RNA was transferred by capillary action onto a nylon membrane (Hybond N, Amersham Corp.) and fixed to the membrane in a UV light cross-linking apparatus (Stratolinker, Strat-agene, La Jolla, CA). Blots of total muscle RNA were probed with the full-length canine M-PFK cDNA (PCRPFKd4) (10) for 18 h at 55°C, washed at 60°C in 1 ϫ SSC, and exposed to x-ray film.
cDNA Library-Total muscle RNA prepared as described above was used to make a cDNA library. Poly(A) ϩ RNA was isolated using an oligo(dT)-cellulose column (Collaborative Research, type III). A cDNA library was constructed using this mRNA in the Lambda ZAP II vector (Stratagene) following the manufacturer's protocol. Approximately 1 ϫ 10 5 plaques constituting the entire primary plating of the M-PFKdeficient canine cDNA library were lifted on nitrocellulose membranes and screened using a full-length human M-PFK cDNA, HMPFK2, or the 5Ј or 3Ј EcoRI fragments of HMPFK2 (10). Positive clones were plaque purified and excised in vivo into the plasmid form, according to the manufacturer's instructions.
RT-PCR Cloning-First strand cDNA from an affected male dog was synthesized using 2 g of total skeletal muscle RNA prepared as for the Northern blot, or mRNA from the same preparation that was used to construct the cDNA library (15). Primers with the sequences gaGCG-GCCGCtcatgacccatgaagagc (PFKNot) containing the start codon (underlined, lower case, bold), and cgCTCGAGcacagtgaccagttggcat (PFKXho) were designed with homology to M-PFK (lower case, bold) and NotI or XhoI restriction sites (upper case). These primers were used to amplify oligo(dT)-primed first strand cDNA with a reaction profile of 40 cycles of 1 min at 95°C, 1 min at 50°C, and 3.5 min at 72°C. The PCR product was purified by agarose gel electrophoresis, electroeluted, serially digested with NotI and XhoI, and ligated into NotI/XhoI-digested Bluescript II phagemid (Stratagene).
DNA Sequencing-Sequencing reactions were performed on doublestranded and single-stranded DNA templates according to a modification of the Sanger dideoxy nucleotide chain termination method, both manually and with an automated sequencer (Applied Biosystems, Inc. model 373A automated sequencer at the DNA Facility, the School of Veterinary Medicine, of the University of Pennsylvania) (16). Both strands of M-PFK were completely sequenced and in many cases from multiple primers. In areas covered by PCR-derived clones, sequence from clones representing three separate PCR reactions was obtained.
Allele-specific PCR-based M-PFK Test-Genomic DNA was harvested from 100 l of EDTA-anticoagulated blood as described (17). Amplification of 5-10% of this DNA was performed using primers with the sequence ctggggatgcgtaagagggctctgg (PFKexon21) and gaggatgggcctcagcttcaggcac (PFKBan) which correspond to bases 2137-2161 of the coding sequence and bases 2229 -2253 of the non-coding strand, respectively. The predicted size of the product of amplification with the PFKexon21/PFKBan primer set was 326 base pairs, based on 116 base pairs in exons 21 and 22 and the rabbit muscle intron length of 210 base pairs (18,19). Following a PCR amplification profile of 9 min at 95°C, 5 cycles of 1 min at 95°C and 1 min at 55°C, and 35 cycles of 1 min at 95°C and 1 min at 60°C, the product of predicted size was purified by sequential phenol and chloroform:isoamyl alcohol (24:1) extractions. After ethanol precipitation, one-half of the product was digested with the restriction endonuclease BanI and electrophoresed on a 5% acrylamide, 1 ϫ TBE gel. The PCR product was visualized on a UV transilluminator after staining with ethidium bromide. Human genomic DNA was not amplified by this primer set under these conditions. RESULTS SDS-PAGE of skeletal muscle extracts from normal, carrier, and M-PFK-deficient English springer spaniels, as determined by total erythrocyte PFK activity, were prepared and stained for total protein. A band of ϳ85 kDa, which was present in normal dogs, was absent in affected dogs (Fig. 1A, lanes 1 and  3). An 85-kDa band of reduced (approximately one-half) intensity was seen in muscle from carrier dogs (Fig. 1A, lane 2).
To determine whether the 85-kDa band corresponded to M-PFK, similar gels were transferred electrophoretically to nitrocellulose membranes and hybridized with a guinea pig antirabbit M-PFK antibody known to cross-react with M-PFK subunits from a variety of mammalian species (20). An 85-kDa protein was present in muscle extracts from normal dogs (Fig.  1B, lane 1). In contrast, no band was detected in affected individuals (Fig. 1B, lane 3), and a band of approximately half-normal intensity was seen in carriers (Fig. 1B, lane 2). Pulse-chase labeling of primary muscle cell cultures followed by immunoprecipitation with an M-PFK-specific antibody indicated that protein was present for at least 24 h (half-life, ϳ18 h) in normal dogs (Fig. 1C). No immunoprecipitated protein was seen in myotubes from affected dogs, even immediately after labeling (Fig. 1C, lane 7). A second pulse-chase labeling with 20 times as much label added for only 30 min resulted in a faint band being visible at chase times up to 2 h in affected dogs.
A Northern blot of total RNA was performed on muscle samples taken from a normal and an affected dog (Fig. 2). The M-PFK transcript from both normal and affected dogs measured ϳ2.8 kilobases, and the amount of M-PFK transcript in affected muscle appeared to be greater than normal. The amount of mRNA was not quantified due to the large number of possible variables affecting gene expression in muscle, such as age, physical condition, and breed.
Based on this information, canine M-PFK deficiency was thought to be most likely due to a missense or nonsense mutation within the coding sequence, which would account for the absence of the M-PFK protein and activity and the rapid turnover of the protein. To determine the genetic basis of this disease, a skeletal muscle cDNA library was created from an affected male dog using the Lambda ZAP II vector. Clone PFKd13a was the longest clone isolated from the M-PFK-deficient library, starting at a position corresponding to base 1426 of the human sequence and containing a polyadenylation site and poly(A) tail. Rescreening of this cDNA library failed to yield any additional clones containing the sequence between the first base and the start of PFKd13a. M-PFK clones containing the remaining 5Ј region were generated using PCR of affected muscle cDNA and sequenced.
Comparison of the sequences of the open reading frames from normal and affected dogs revealed a single difference at position 2228. This alteration, from a guanosine in the normal dog to an adenosine in the affected dog, resulted in an amber codon (UAG) replacing the normal tryptophan codon (UGG) and caused the termination of translation 40 amino acid residues before the normal termination codon (Fig. 3). The mutant canine sequence contained the termination codon TAA at the same location as the normal canine, rabbit, and human stop The mutation in the affected dog neither created nor abolished a known restriction enzyme recognition sequence. For this reason, an allele-specific test was designed for the mutation using the normal base in combination with a PCR primer (PFKBan) that introduced a mismatch to create a restriction enzyme recognition site (Fig. 4A). Genomic DNA from dogs whose PFK status was known from enzyme assay was amplified with the primer set PFKexon21/PFKBan, digested with BanI, and evaluated by polyacrylamide gel electrophoresis. The PCR product from normal dogs was completely digested, and it migrated in a position (310 base pairs) ϳ25 nucleotides smaller than the undigested product (Fig. 4B, lane 5). In contrast, BanI-digested PCR product from affected individuals showed no change in electrophoretic mobility (Fig. 4B, lane 2). Carriers had bands of reduced intensity at both positions, when compared with normal or affected dogs (Fig. 4B, lanes 1, 3, 4, and  6). Thus, the PFKexon21/PFKBan assay allowed the number of normal alleles at position 2228 to be determined, thereby permitting carrier animals to be distinguished from normal and affected dogs.
Twenty-one of the dogs in our colony for which enzyme activities were available were evaluated utilizing the PFKexon21/PFKBan primer set. Since this colony was derived from three separate families of dogs with no relatives in common for the previous six generations, segregation of this mutation with the disease would strongly support its causative role. Five normal dogs, eight carrier dogs, and eight affected dogs were evaluated, and the correlation between enzyme activity and the number of normal alleles was complete. Fig. 4B shows PCR results from a typical mating between two carriers which produced one affected puppy, one normal puppy, and two carrier puppies.
Because M-PFK deficiency is a common inherited disorder in the English springer spaniel (5-7), we examined 35 samples from animals outside of the colony for M-PFK deficiency with the allele-specific PCR-based test. Of these dogs, 23 were English springer spaniels, 3 were American cocker spaniels, and 9 were breeds other than spaniels from which blood samples were submitted as normal controls for PFK enzyme activity assays. Twenty-two of these samples were submitted because the animal had been suffering from chronic or recurrent regenerative anemia and/or exercise intolerance, for breeding evaluation, or because the animal was related to an affected dog. Five English springer spaniels were shown to be homozygous, and three were heterozygous for the G to A mutation. An American cocker spaniel dog, with hemolytic anemia, was examined by this technique and found to have no normal M-PFK alleles (21). Subsequent sequencing confirmed that the disease in this American cocker spaniel was due to the same mutation that causes the disease in the English springer spaniel breed (data not shown).

DISCUSSION
Canine M-PFK deficiency represents the only disease model of this autosomal recessive inborn error of metabolism in humans. The clinicopathological similarities of the exertional myopathy and erythroenzymopathy in both species have been reported (7,9). Our studies characterize the molecular basis of M-PFK deficiency in dogs and suggest that canine PFK deficiency is an excellent disease homologue to further investigate pathogenesis and novel therapeutic approaches. When compared with the previously reported cDNA sequence from normal dogs (10), the coding sequence from affected dogs showed only a single base substitution of an A to G at position 2228 which changes a tryptophan codon to a stop codon, thereby truncating the M-PFK subunit by 40 amino acid residues. No other alterations were found in the protein-coding region compared to normal M-PFK. PCR testing of the region of the  1, 3, 4, and 6). The pedigree indicates the relationship of the animals tested. mutation established an absolute correlation between the number of mutant alleles and normal, intermediate (carrier), and deficient PFK activity in all individuals tested, further supporting the causative nature of this mutation. Thus far, only very few canine and other large animal models have been defined at the molecular genetic level.
From among the Ͼ50 M-PFK-deficient humans, several mutations have been identified, but none have exactly the same genetic defect as in dogs. A splicing defect that results in recognition of a cryptic splice site within exon 15 and an in frame 75-base pair deletion in the mRNA has been found in a Japanese family (2). In Ashkenazi Jews, the most commonly reported mutations are at the splice donor site of intron 5, leading to a precise deletion of exon 5, and a single base deletion resulting in a frameshift mutation in exon 22 and a premature stop codon (3), which is similar to the one we found in deficient dogs. In addition, missense mutations have been reported in French Canadian and Swiss patients (4).
The information regarding the enzyme defect at the protein level is very limited in human patients. Generally, M-PFK enzyme activity is nearly completely lacking in muscle tissue and reduced to one-half in erythrocytes, due to the contribution of L-PFK. The amino acid substitution in M-PFK found in a French Canadian patient resulted in a completely inactive protein, whereas the missense mutations in a double-heterozygote Swiss patient only modestly changed the enzyme activity and thermal stability as shown in a yeast expression system (4). No structure-function studies have, thus far, been performed with normal and mutant canine M-PFK.
Our data for the dog show that the 40-amino acid deletion caused by the mutation, representing only 5% of the protein, has a profound effect on the assembly and stability of the molecule. The lack of this region apparently leads to the rapid degradation of the entire subunit, since the half-life of the mutant M-PFK was drastically shortened. The region deleted by this mutation corresponds to a sequence that forms an ␣ helix in Bacillus stearothermophilus (22), as determined by x-ray crystallography, and is highly conserved in canine (10) and other mammalian (20, 23) M-PFKs. In Escherichia coli, this region has been shown to be critical to the stability of the tetramer and the individual subunit (24). When 1 of 4 subunits was truncated by 40 amino acids, the tetramer was no longer stable in the absence of fructose 6-phosphate. The incorporation of two truncated subunits resulted in the loss of allosteric regulation (24). Chou-Fasman analysis of this sequence of human, rabbit, and canine M-PFKs for ␣-helical content indicates that the mutation interrupts a region which is highly likely to form an ␣ helix in mammalian M-PFK (25,26). No amino acid residues in this terminal ␣ helix correspond to amino acids known to be involved in the active site or effector site (27)(28)(29)(30). Kyte-Doolittle hydrophobicity analysis of the normal canine amino acid sequence demonstrates that this region is hydrophilic (31). The combination of the potential of this region for ␣-helical structure and its hydrophilicity and terminal location suggest that it may be involved in subunit recognition or interaction in the formation of tetramers or in subunit stability.
In contrast to muscle, partially purified extracts of PFK from the erythrocytes and brains of affected dogs showed very small amounts of a lower molecular weight band that was hypothesized to be the truncated M-PFK subunit (14), which has an apparent molecular mass of 84 kDa (32,33). This is 3.3 kDa larger than the predicted molecular weight from the DNA sequence; however, the mobility of proteins in polyacrylamide gels may be affected by factors other than molecular weight, such as glycosylation (34). The reason for the apparent greater stability of the truncated subunit in these tissues than in muscle is unclear; however, the lack of enzymatic M-PFK activity appears to be unaffected by the tissue source. One possible explanation for this apparent difference in stabilities is that erythrocytes and brain contain other subunits, while muscle expresses only M-PFK subunits, and the mutant M-PFK molecular may be partially stabilized by complexing with Land P-PFK subunits in erythrocytes and brain cells. The low residual PFK activity found in muscle (1-4%) could be caused by the mutant M-PFK. Alternatively, there is expression of other PFK isoforms in skeletal muscle (9,14), because in vitro experiments with C2-C12 myogenic cell lines have shown that L-PFK and P-PFK are expressed in myoblasts and at low levels even after myotube formation, indicating that a low expression of these other isoforms in muscle may account for the residual PFK activity (35).
To screen for this M-PFK mutation in dogs, we designed an antisense primer with a mismatched nucleotide, located downstream from the mutation site, that created a restriction enzyme cleavage site for BanI in the PCR fragment from the normal, but not in that derived from the mutant sequence. The same mutation was identified in the three English springer spaniels that originated from different parts of the United States and were not related over six generations. Affected offspring from these dogs were homozygous for this mutation, whereas clinically normal animals with half-normal enzymatic PFK activity in muscle and erythrocytes had one normal and one mutant allele. The complete concordance of this mutation in affected and carrier animals and absence in the normal population confirm that it is in fact the deleterious mutation and not a polymorphism. The availability of this simple PFK-PCR test eliminates the need for enzyme activity assays and allows population screening. The frequency of this disease in the English springer spaniel breed is likely underestimated because the clinical manifestations of myopathy and hemolysis are intermittent and thus may be mistakenly presumed to be due to acquired diseases.