Originally published In Press as doi:10.1194/jlr.M700145-JLR200 on May 24, 2007
Journal of Lipid Research, Vol. 48, 1857-1872, August 2007
Copyright © 2007 by American Society for Biochemistry and Molecular Biology
The pathologies associated with functional titration of phosphatidylinositol transfer protein 
activity in mice
James G. Alb, Jr.*,
Scott E. Phillips*,
Lindsey R. Wilfley
,
Benjamin D. Philpot,
,** and
Vytas A. Bankaitis1,*
* Department of Cell and Developmental Biology, Lineberger Comprehensive Cancer Center, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-7090
Department of Cell and Molecular Physiology, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-7090
Neuroscience Center, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-7090
** Neurobiology Curriculum, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-7090
Published, JLR Papers in Press, May 24, 2007.
1 To whom correspondence should be addressed. e-mail: vytas{at}med.unc.edu
 |
ABSTRACT
|
|---|
Phosphatidylinositol transfer proteins (PITPs) bind phosphatidylinositol (PtdIns) and phosphatidylcholine and play diverse roles in coordinating lipid metabolism/signaling with intracellular functions. The underlying mechanisms remain unclear. Genetic ablation of PITP
in mice results in neonatal lethality characterized by intestinal and hepatic steatosis, spinocerebellar neurodegeneration, and glucose homeostatic defects. We report that mice expressing a PITP selectively ablated for PtdIns binding activity (PitpT59D), as the sole source of PITP, exhibit phenotypes that recapitulate those of authentic PITP nullizygotes. Analyses of mice with graded reductions in PITP activity reveal proportionately graded reductions in lifespan, demonstrate that intestinal steatosis and hypoglycemia are apparent only when PITP protein levels are strongly reduced (
90%), and correlate steatotic and glucose homeostatic defects with cerebellar inflammatory disease. Finally, reconstitution of PITP expression in the small intestine substantially corrects the chylomicron retention disease and cerebellar inflammation of Pitp0/0 neonates, but does not rescue neonatal lethality in these animals. These data demonstrate that PtdIns binding is an essential functional property of PITP in vivo, and suggest a causal linkage between defects in lipid transport and glucose homeostasis and cerebellar inflammatory disease. Finally, the data also demonstrate intrinsic neuronal deficits in PITP-deficient mice that are independent of intestinal lipid transport defects and hypoglycemia.
Supplementary key words signaling • phospholipids • lipoproteins • neurodegeneration
Abbreviations: CRD, chylomicron retention disease; ER, endoplasmic reticulum; EtOH, ethanol; fEPSP, field excitatory postsynaptic potential; GFAP, glial fibrillary acidic protein; MAMA, mismatch amplification mutation assay; PAC, P1 artificial chromosome; PITP, phosphatidylinositol transfer protein; PtdCho, phosphatidylcholine; PtdIns, phosphatidylinositol; TG, triglyceride
|
INTRODUCTION
|
|---|
The physical compartmentation of signal transduction to specific microdomains of membrane surfaces is a common property of all eukaryotic cells, and various mechanisms are applied toward achieving this end (1–5). One such mechanism involves the spatially and temporally regulated metabolic channeling of phosphatidylinositol (PtdIns) and/or phosphatidylcholine (PtdCho) monomers by phosphatidylinositol transfer proteins (PITPs). This PITP-mediated channeling is itself linked to specific physiological outcomes (6–8). Prime among these are the roles of PITPs in coordinating various interfaces between lipid metabolism and membrane trafficking reactions. PITPs fall into two unrelated structural classes: the Sec14-like PITPs and the so-called class 1 metazoan PITPs (9–12). The Sec14-like PITPs are better understood because detailed functional analyses have been more accessible for these proteins. The metazoan PITPs are poorly characterized from a functional point of view.
Mammals express four known class 1 PITP forms: PITP, PITPß, an alternative spliceoform of PITPß, and rdgBß (13–18). Data that address in vivo functions for PITP and PITPß are building. These proteins are encoded by distinct genes but are highly similar, sharing 77% and 95% primary sequence identity and similarity, respectively. The extensive homology notwithstanding, available data suggest that PITP and PITPß execute distinct roles in mammals. First, these proteins exhibit distinct intracellular distributions. PITP represents a cytosolic/nuclear protein, whereas PITPß is localized predominantly to the trans-Golgi network of mammalian cells [as is its alternative spliceoform (17)]. Second, PITPß binds the ceramide-based phospholipid sphingomyelin, whereas PITP does not (15, 17, 19). Finally, the consequences of the genetic ablation of each protein are different (20, 21). The functional nonredundancy of PITP and PITPß conforms to the general theme that PITPs are tightly coupled to specific physiological functions in cells (6–8). How PITPs couple individual PtdIns and PtdCho binding activities to biological functions remains to be determined.
Genetic ablation of the PITP structural gene in mice has interesting consequences. Pitp0/0 progeny develop to term, but expire within several days after birth from a complex disease course that includes defects in dietary fat and -tocopherol transport from the small intestine [chylomicron retention disease (CRD)], liver steatosis, hypoglycemia, inflammation of cerebellum and hindbrain, and spinocerebellar neurodegenerative disease (21). These pathologies raise a number of questions. First, which biochemical properties of PITP are relevant for in vivo function (i.e., are the PtdIns and PtdCho binding activities differentially used in vivo, or are these themselves coupled?). Second, which of these pathologies are causally linked, and which derive from independent insults? It is suggested the intestinal defects and hypoglycemia are linked, and that these defects provoke the cerebellar and hindbrain neurodegenerative diseases (21). Resolution of these questions is key to an understanding of PITP function in mammalian systems.
Here, we show that PtdIns binding is an essential functional property of PITP in mammals. It is further demonstrated that intestinal steatosis and hypoglycemia require manifest reductions in PITP activity, and that these pathologies correlate with cerebellar inflammatory disease. Finally, the data indicate that reconstitution of PITP expression in the small intestine alleviates the CRD and cerebellar inflammatory disease of Pitp0/0 neonates, but fails to effectively rescue the neonatal lethality of these animals. This failure is accompanied by exacerbations in microvesicular steatosis of the liver, a result consistent with the idea that PITP regulates the trafficking of luminal lipid cargos.
|
MATERIALS AND METHODS
|
|---|
Mouse genotyping
Wild-type and Pitpa::NEO alleles were diagnosed from tail clip genomic DNA templates using a three-primer PCR assay described previously (21). To screen for the vibrator allele (Pitpvb), a hypomorphic mutation that reduces the expression of wild-type PITP by 
80%, an intracisternal A particle retrotransposon three-primer PCR assay was used (22). The allele-specific forward primers for the wild-type and vibrator alleles were Ham-For (5'-AGTGATCGGGACTTTTGTTTCC-3') and intracisternal A particle retrotransposon-1 (5'-AACGCGTCTAATAACACTTGTG-3'), respectively. A common reverse primer was used, Ham-Rev (5'-CCAAAAGGACTGCCCAGTCATTTTTCCCC-3'). The vibrator primer sequences were provided courtesy of Bruce Hamilton (University of California at San Diego).
To generate a PITP transgene whose expression is limited to the duodenum, plasmid pNSE-EX4 (23) (a gift of Greg Cox, Jackson Laboratories, Bar Harbor, ME) was cleaved with XhoI to remove the existing neuronal enolase promoter, and the intestinal fatty acid binding promoter (PIFABP) (24) was introduced in its stead. The PITP open reading frame cDNA was generated by PCR as a 5'-HindIII-EcoRI-3' fragment and correspondingly subcloned into the PIFABP-pNSE-EX4 vector to yield plasmid pFabpiPITP. The resultant plasmid was linearized by digestion with restriction endonuclease SalI and injected into C57BL/6JxC3H hybrid pronuclei at the University of North Carolina Animals Model Core to generate transgenic murine founder lines. Transgenic founders were identified with a PITP cDNA-specific set of primers: Alpha-For (5'-GCAAAGTTCCCACGTTTGTTCG-3') and Alpha-Rev (5'-CTCTTGCTTATGTATAAAGTTTTCC-3'). These primers support the synthesis of a diagnostic 450 bp PCR product. Three independent founders were crossed to C57BL/6J PITP+/0 mice to generate Pitp+/0;Tg(PIFABP::PITP) lines.
Generation of a PtdIns binding-deficient PITP transgene
Escherichia coli strain DY380 was transformed with P1 clone 4232 (a gift from Bruce Hamilton, University of California at San Diego), which contains the entire PITP genomic locus in a P1 artificial chromosome (PAC) vector harboring a kanamycin resistance selection marker, to generate E. coli strain DY380-Tg(PITP). DY380 is permissive for homologous recombination when electroporated linear double-stranded DNA is introduced into this E. coli host (25, 26). The T59D missense substitution was incorporated into exon 3 of the PITP gene as follows. A pair of 100mer oligonucleotides precisely complementary to each other in their respective 3' 20 nucleotides was constructed. The 20 bp complementary regions spanned the threonine 59 codon, which was converted during oligonucleotide synthesis from the wild-type ACA nucleotide sequence to GAG (and the corresponding reverse complement). Oligonucleotides were annealed at 42°C for 1 h, and the remaining 3' recessed ends were filled in with Klenow fragment of DNA polymerase I (Promega, Madison, WI). The resultant 200 bp linear double-stranded DNA fragment represented the targeting cassette for incorporating the T59D missense mutation into exon 3 of the PITP structural gene.
The 200 bp target fragment was electroporated into DY380-Tg(PITP) cells prechallenged at 42°C for 15 min (to induce the recombineering machinery), transformants were selected in Luria broth supplemented with kanamycin (50 µg/ml), survivors were serially diluted into five 96-well culture plates to a density of 10 colony-forming units per well, and the individual pools were expanded. Aliquots from each well were collected and corresponding DNA templates were prepared for screening using a mismatch amplification mutation assay (MAMA) PCR. MAMA-PCR used a forward primer (5'-GGCGAGAAAGGCGAGTACGAG-3') specific for the mutant T59D codon (underlined) and reverse primer (5'-CTCACATGCCACACACTCC-3') that hybridizes to sequences within the PITP gene not represented in the original 200 bp targeting fragment. Aliquots yielding a 300 bp PCR product diagnostic of PitpT59D clones identified the wells of interest, and the corresponding well cultures were subjected to limit dilution and plated for single colonies. Individual colonies derived from well pools of interest were rescreened by MAMA-PCR. Positive clones yielded candidate purified P1 Tg(PitpT59D clones. Identities were verified by nucleotide sequence analysis of PITP exon 3 of individual candidates. We estimate a 1:12,000 ratio of Tg(PitpT59D) to Tg(PITP) in the final pools.
Immunoblotting and ELISA
Whole mouse brains or duodenal sections were solubilized in RIPA buffer (50 mM Tris-HCl, pH 7.2, 1.0% Nonidet P-40, 0.1% SDS, 0.15 mM NaCl, and 1x protease inhibitor cocktail tablet; Roche, Indianapolis, IN). Protein was estimated using BCA (Pierce, Rockford, IL) for purposes of normalization in immunoblot and ELISA experiments. For immunoblotting experiments, 75 µg of brain lysate was resolved by SDS-PAGE and transferred to nitrocellulose membrane (Bio-Rad Laboratories, Hercules, CA). Membranes were subsequently decorated with PITP-specific antibody as described previously (20), and with a ß-actin specific antibody (Oncogene, Carpinteria, CA) for purposes of normalization. For ELISA analyses, serial amounts of brain lysate protein for each experimental sample (1 µg, 100 ng, and 10 ng) were coated, in duplicate, on 96-well plastic plates. Samples were dried overnight, washed in 0.5% Tween 20, blocked with 2% BSA, incubated with the appropriate primary antibody overnight at room temperature, washed in PBS, incubated with secondary antibody, and finally developed using o-phenylenediamine reagent (Sigma, St. Louis, MO).
Blood chemistry
Whole blood was collected from mice immediately after decapitation or via heart puncture (in 2 mM EDTA, pH 7.2, final concentration for plasma), clotted, and clarified by low-speed centrifugation. Serum glucose was determined by the Trinder assay (Sigma). Plasma triglyceride (TG) measurements were performed on 30 µl plasma aliquots at the Animal Clinical Chemistry and Gene Expression Laboratories at the University of North Carolina-Chapel Hill. Agarose electrophoresis of whole serum used the Titan Gel Lipoprotein Kit (Helena Laboratories, Beaumont, TX) according to the manufacturer's specifications. Resolved samples were developed by staining of gels in 0.1% Fat Red 7B (w/v) in 95% methanol. Chylomicron densities were determined by scanning dried Titan gels using a Bio-Rad VersaDoc imaging system (525 nm) and quantified by the Quantity One software package.
Whole carcass analyses
Carcass compositional analyses were performed as described (21). Briefly, eviscerated carcasses were weighed and body water content determined after drying. Dried carcasses were ground to homogeneity and extracted with petroleum ether in a Soxhlet apparatus to determine fat mass and fat-free dry mass. Carcass ash was measured after limit incineration of fat-free dry mass material.
Histological methods
Mice were prepared for all analyses by anesthesia with 1.25% Avertin (Sigma) and fixed by heart perfusion with sequential phosphate-buffered saline and 4% paraformaldehyde solutions. Intestines, brains, and livers were harvested, intestines were flushed with fixative, and all tissues were postfixed in 4% paraformaldehyde for 24 h. When dehydration was required, tissues were embedded in paraffin, and 5 µm thick sections were cut and mounted onto slides. When frozen samples were required, tissues were embedded in OCT, and 12 µm thick sections were cut and mounted.
For intestinal and liver staining with osmium (27), paraffin sections were rehydrated in a step series starting with xylene and terminating with 50% ethanol (EtOH) (i.e., xylene, 95% EtOH, 90% EtOH, 80% EtOH, 70% EtOH, 60% EtOH, and 50% EtOH), stained for total lipid by incubation with 5% potassium dichromate and 2% OsO4 for 8 h, and counterstained with toluidine blue O. Samples were then dehydrated in a 50% EtOH-to-xylene series that was the reverse of the rehydration series described above. Dehydrated sections were mounted in Permount (Fisher Scientific, Atlanta, GA). Oil Red O staining was used to identify neutral lipid (27). Frozen sections mounted onto slides were incubated in 85% propylene glycol, stained with Oil Red O, counterstained with hematoxylin, and mounted in glycerol.
For whole brain analyses, fixed samples were washed for several days in phosphate-buffered saline. Brains were then sliced in half along the sagittal plane and embedded in paraffin, and 5 µm-thick sections were cut. Sections were rehydrated as described above and incubated with anti-glial fibrillary acidic protein (GFAP) antibody (EMD Biosciences CalBiochem, San Diego, CA). Decorated slices were developed with the Vectastain ABC kit (Vector Laboratories, Burlingame, CA), followed by counterstaining with toluidine blue. Samples were finally dehydrated in xylene/EtOH (50%) and mounted in Permount (Fisher Scientific).
-Tocopherol measurements
Whole brains were harvested from decapitated mice, and -tocopherol was extracted with chloroform as described (21). Samples were resolved by HPLC using a Perkin-Elmer Life Sciences model LC200 gradient pump with an AS 200 Autosampler and an LC 295 ultraviolet-visible light detector.
Liver glycogen analysis
Whole livers were extracted from euthanized animals and homogenized, and glycogen was recovered by acid extraction (28). Recovered material was quantitatively hydrolyzed to glucose by amyloglucosidase, and released glucose was quantified by Trinder assay (Sigma).
Hippocampal field excitatory postsynaptic potential recordings
Age-matched (P4–P6) PITP+/+ and Pitp0/0 mice were anesthetized with an overdose of pentobarbital barbiturate and decapitated upon the disappearance of corneal reflexes, in compliance with U. S. Department of Health and Human Services and University of North Carolina guidelines. Brains were dissected immediately, and dorsal hippocampi were cut in 400 µm coronal slices in ice-cold oxygenated dissection buffer (75 mM sucrose, 87 mM NaCl, 2.5 mM KCl, 1.25 mM NaH2PO4, 26 mM NaHCO3, 10 mM glucose, 7 mM MgCl2, 0.5 mM CaCl2, and 1.3 mM ascorbic acid). Hippocampal slices were equilibrated for 30 min in a submersion chamber at 35°C filled with artificial cerebrospinal fluid (124 mM NaCl, 3 mM KCl, 1.25 mM Na2PO4, 26 mM NaHCO3, 1 mM MgCl2, 2 mM CaCl2, and 20 mM D-glucose, saturated with 95% O2 and 5% CO2; 315 mOsm and pH 7.25) and then held at room temperature until use. For recordings, hippocampal slices were placed in an interface chamber, maintained at 31°C, and perfused with oxygenated artificial cerebrospinal fluid. Synaptic responses were evoked by stimulating the Schaffer collaterals with a bipolar stimulating electrode (FHC, Inc., Bowdoin, ME), and field excitatory postsynaptic potentials (fEPSPs) were recorded in stratum radiatum of the CA1 region. Input-output curves were generated by systematically varying the stimulation intensity and measuring the fEPSP slope. Fiber volley amplitude was compared with the fEPSP slope to relate presynaptic to postsynaptic responses. The paired-pulse facilitation of the fEPSP slope and amplitude was examined by adjusting the stimulation intensity to the half-maximal response and systematically varying the interstimulus interval between two stimulations. Statistical analyses used a multivariate two-way ANOVA, with significance placed at P < 0.05.
|
RESULTS
|
|---|
A murine strain that exclusively expresses a PtdIns binding-deficient PITP
PITP binds/exchanges PtdIns and PtdCho monomers between membrane bilayers in vitro. It is not yet established what each individual activity contributes to PITP action in cells. The availability of mutant PITP derivatives with selective defects in PtdIns binding (29), and the ability to generate PITP nullizygous (Pitp0/0) mice with outstanding phenotypes (21), permit experimental assay of the contribution of PtdIns binding to PITP function in the mammal. To this end, the mutant PitpT59D was used to assess the in vivo consequences of inactivating PITP PtdIns binding activity. Residue threonine 59 is required for coordination of the inositol headgroup of PtdIns, and the introduction of any residue bulkier than threonine at this position evokes a selective steric incompatibility with PtdIns binding (Fig. 1
) (10, 29, 30). Because choline is less bulky than inositol, PtdCho binding is unaffected in PitpT59D.
The strategy used for generating mice that express only PitpT59D involved recombineering-mediated site-directed mutagenesis of a 76 kDa PAC containing the entire PITP genomic locus. This PAC [henceforth referred to as Tg(PITP)] was shown previously to phenotypically complement the hypomorphic PITP vibrator mutation (22). As a control for all subsequent experiments, we tested whether Tg(PITP) also rescues Pitp0/0 lethality. Indeed, Tg(PITP) efficiently complements all Pitp0/0-associated phenotypes. Pitp0/0;Tg(PITP) animals are long-lived (the oldest animals are 16 months of age at present and healthy), are fertile, and are phenotypically PITP+/+ in all obvious respects (data not shown). Thus, the Tg(PITP) vector is a suitable precursor for the expression of mutant forms of PITP in the mouse for functional studies.
The PITP PAC mutagenesis scheme is outlined in Fig. 2A
. The PitpT59D PAC [referred to as Tg(PitpT59D)] was injected into C57BL/6JxC3H hybrid pronuclei to generate founder mice. Standard mating strategies were used to generate PITP+/0;Tg(PitpT59D) parental lines from which Pitp0/0;Tg(PitpT59D) progeny were derived. The presence of Tg(PitpT59D) was diagnosed by probing for DNA sequences unique to the bacterial plasmid backbone of this vector (Fig. 2B) (see Materials and Methods). However, because an intact 76 kDa PITP locus was used as the mutagenesis substrate, diagnostic assays for Tg(PitpT59D) were more complicated. Additional methods were required to independently confirm mouse genotypes (see below). To that end, a robust MAMA-PCR strategy that distinguishes the endogenous PITP locus from ectopic Tg(PitpT59D) was developed for purposes of diagnostic genotyping of transgenic mice (see Materials and Methods) (Fig. 2B). Demonstration that this assay reliably detects PitpT59D in genomic DNA preparations is shown in Fig. 2C.
View larger version (31K):
[in this window]
[in a new window]
|
Fig. 2. Generation of mice expressing a PtdIns binding-deficient PITP. A: P1 artificial chromosome (PAC) mutagenesis strategy. Tg(PITP+) represents P1 clone 4232, which contains the entire 76 kb PITP+ structural gene. The PitpT59D missense substitution was introduced into this PAC by recombineering-driven site-directed mutagenesis, as described in Materials and Methods. dNTP, deoxynucleoside triphosphate. B: Diagnosis of Tg(PitpT59D) in mouse genomic DNA. The mutant transgene was detected in candidate transgenic mouse founder lines using a PCR strategy screening for the presence of the PAC vector backbone. Arrows indicate the positions of forward and reverse diagnostic PCR primers that yield a 350 bp product detected exclusively in genomic DNA prepared from founder mice carrying Tg(PitpT59D). C: Confirmation of Tg(PitpT59D) in candidate transgenic founder lines. Arrows indicate primers used in mismatch amplification mutation assay (MAMA)-PCR analysis (which yields a 300 bp PCR product; left panel) to specifically reveal the PitpT59D allele. D: Tg(PitpT59D) is actively transcribed. The upper panel illustrates the RT-PCR strategy used to detect transcripts in Tg(PitpT59D) mice. Solid arrows indicate MAMA-PCR primers used to distinguish endogenous PITP+ mRNA (WT59-5) transcripts from those of PitpT59D (TD59-5). In both cases, the diagnostic PCR products are 350 bp. Dashed arrows indicate the forward and reverse PCR primers (PITP-5 and PITP-3) used in RT-PCR assays to reveal full-length PITP transcripts (820 bp product) without distinguishing PITP+ transcripts from those of PitpT59D. The lower panel demonstrates this method. Mouse genotype is indicated at the top and RT-PCR products are identified at the bottom of individual lanes. E: Confirmation that PitpT59D is properly expressed. Nucleotide sequence analysis of the RT-PCR product generated by the PITP-5/PITP-3 primer combination. The left panel shows the cDNA sequence readout of relevant codons from a PITP+/+ mouse. The middle and right panels are the corresponding readouts from PITP+/0,Tg(PitpT59D) and Pitp0/0,Tg(PitpT59D) mice, respectively. Genotypes are indicated at the top, and relevant sequences are identified at the top of the red boxes. Corresponding amino acids are listed: Asp, aspartate; Thr, threonine. Note that the PITP+/0,Tg(PitpT59D) reactions reveal sequences derived from both PITP+ from PitpT59D alleles.
|
|
That Tg(PitpT59D) is transcribed was verified by two lines of evidence. First, reverse transcription PCR using the same primers used for MAMA-PCR detected Tg(PitpT59D) mRNA only in Tg(PitpT59D) mice (Fig. 2D). Second, DNA sequence analysis of the entire PITP cDNA coding sequence generated by reverse transcription of brain mRNA from PITP+/0;Tg(PitpT59D) animals identified a single polymorphism. This polymorphism resulted in the detection of both threonine and aspartate codons at PITP codon 59 [from PITP+ and PitpT59D alleles, respectively (Fig. 2E, left and middle panels)]. The similar magnitudes with which each polymorphism was recorded indicate that Tg(PitpT59D) supports transcriptional expression at levels comparable to those of endogenous PITP.
PtdIns binding/transfer is required for PITP function in mammals
Standard matings between PITP+/0;Tg(PitpT59D) males and females generated the full complement of expected genotypes. As all PITP-sufficient progeny exhibited phenotypes indistinguishable from the wild type, irrespective of their Tg(PitpT59D) genotype, we restricted our attention to comparisons of PITP+/+;Tg(PitpT59D), Pitp0/0, and Pitp0/0;Tg(PitpT59D) littermates. Lifespan-monitoring experiments demonstrated that Tg(PitpT59D) expression does not rescue the neonatal and perinatal lethality of otherwise PITP nullizygous offspring. For Pitp0/0;Tg(PitpT59D) progeny, the rapid onset of postnatal mortality characteristic of PITP nullizygous mice was on display (Fig. 3A
). Comparison of Pitp0/0 and Pitp0/0;Tg(PitpT59D) progeny at P3 and P6 revealed that both strains exhibited the wasting phenotype typical of PITP nullizygotes. This wasting was associated with dramatic reductions in the contribution of subcutaneous fat to total body mass (data not shown). Additional analyses of Pitp0/0;Tg(PitpT59D) mice demonstrated that these also exhibited the striking hypoglycemia and intestinal and liver microvesicular steatosis disorders that accompany PITP nullizygosity (Table 1
). Phenotypic manifestations of the robust neurodegenerative disease characteristic of Pitp0/0 neonates were also apparent (data not shown). In sum, the Pitp0/0 and Pitp0/0;Tg(PitpT59D) phenotypes were indistinguishable based on criteria of severity and rate of onset.
View larger version (43K):
[in this window]
[in a new window]
|
Fig. 3. PitpT59D is not a biologically functional protein. A: Lifespans for PITP+/+ (open circles), Pitp0/0 (closed circles), and Pitp0/0,Tg(PitpT59D) (open squares) mice are plotted. The data were obtained from a dedicated pool of 176 mice with a genotypic distribution of 118 PITP+/+, 26 Pitp0/0, and 32 Pitp0/0,Tg(PitpT59D). B: Images of PITP+/+, Pitp0/0, and Pitp0/0,Tg(PitpT59D) mice taken at the indicated ages. C: Tg(PitpT59D) protein is expressed at physiological levels. Immunoblot analysis of PITP immunoreactive species from whole brain (WBr), duodenum (Dnm), and liver, collected from P6 littermates of the indicated genotypes, using a PITP-specific C-terminal polyclonal antibody (top panels). The bottom panels depict the corresponding blots that were stripped and reprobed with a ß-actin-specific antibody for purposes of normalization.
|
|
Failure of Tg(PitpT59D) to rescue lethality, or the other various pathologies of Pitp0/0 mice, revealed an intrinsic biological nonfunctionality of PitpT59D. Lack of activity was not a consequence of reduced protein expression or stability. Immunoblotting for PITP in total brain, duodenum, and liver extracts culled from appropriate mouse strains showed that PitpT59D levels are comparable to those of PITP in vivo (Fig. 3C). For final confirmation, mRNA was purified from brains of Pitp0/0;Tg(PitpT59D) progeny, and the total brain mRNA fraction was used to template PITP cDNA synthesis. The templating strategy sustained unbiased synthesis of both PITP+ and PitpT59D cDNAs. The open reading frame sequence of PITP cDNA was then determined. These analyses confirmed PitpT59D as the sole detectable PITP species expressed in Pitp0/0;Tg(PitpT59D) mice (Fig. 2E, right panel). We conclude from these experiments that PtdIns binding is essential for PITP biological activity.
An allelic series for titrating PITP expression in mice
PITP-deficient mice elaborate multiple phenotypes. Manifestation of some of these phenotypes is sensitive to amounts of PITP protein (21, 22). These findings suggest that titration of PITP function in mice is a viable approach for discerning relationships between the complex phenotypes that accompany PITP insufficiencies. To this end, we took advantage of the availability of Pitp0 and Pitpvb alleles to create a series of murine strains with graded levels of PITP expression. The Pitpvb allele reduces the expression of wild-type PITP as a result of the insertion of an IAP element into intronic sequences of the PITP structural gene (22). Because the Pitp0 allele was originally fixed in the 129SVJ genetic background, and the hypomorphic Pitpvb phenotype is sensitive to genetic modifiers (22), Pitp0 was incorporated into the C57BL/6J genetic background by serial backcross (see Materials and Methods). The C57BL/6J background elicits the most severe phenotype for the Pitpvb homozygote mice and is minimized for the Pitpvb genetic modifiers detected in other murine strain backgrounds (22). Comparisons of the morbidity of Pitp0/0 offspring as a function of genetic background (C57BL/6J vs. 129SVJ) demonstrated that the lifespans of nullizygotes were very similar in both genetic background contexts (50% mortality at 4.2 ± 1.2 vs. 4.8 ± 0.7 days, respectively).
Standard mating schemes were used to generate C57BL/6J-derived mice harboring Pitp+/+, Pitp+/0, Pitpvb/vb, Pitpvb/0, and Pitp0/0 allelic combinations. Relative levels of endogenous PITP in whole brain cytosol extracts were prepared from mice of each genotype and compared by immunoblotting and quantitative ELISA (Fig. 4A
, B). Again, in both experiments, antibodies specific for PITP were used to avoid confounding issues presented by the highly similar PITPß isoform in these mice. In that regard, PITP+/+, PITP+/vb, PITP+/0, Pitpvb/vb, Pitpvb/0, and Pitp0/0 mice express 100, 94, 61, 18, 13, and <3% of wild-type PITP levels, respectively (Fig. 4B). These data indicate that the titration range for PITP was a satisfactorily broad one. For purposes of these studies, we focus on PITP+/+ and Pitp0/0 strains as positive and null controls, respectively, and Pitpvb/vb and Pitpvb/0 mice as experimental variables.
Phenotypes of mice with graded reductions in PITP expression
Analyses of the allelic series emphasized the close correlation between the primary phenotypes of PITP-deficient animals and levels of PITP expression. From the perspective of viability, Pitp0/0 offspring exhibited a steady mortality rate in the P0 to P11 window. Pitpvb/0 animals survived until P19 and then expired precipitously over the next 3 days (Fig. 5A
). Similarly, and in agreement with previous reports (22, 31), Pitpvb/vb juveniles persisted until P28 and then expired during the next 5 days. No deaths were scored for PITP+/+ animals during this period. Indeed, both PITP+/vb and PITP+/0 heterozygotes were phenotypically normal in every obvious way throughout their normal lifetime of 24 months (data not shown). The intermediate lifespan of Pitpvb/0 offspring, relative to Pitp0/0 and Pitpvb/vb counterparts, extended to other phenotypes. For example, Pitpvb/vb juveniles were afflicted with a whole body tremor whose onset was observed at P18. This onset advanced to P12 for Pitpvb/0 mice. Body mass analyses revealed a similar pattern. Whereas a clear distinction in mass was already apparent between PITP+/+ and Pitp0/0 littermates by P5 (3.3 ± 0.4 vs. 1.9 ± 0.5 g; P < 0.003) (Fig. 5B), a significant difference between the average body mass of PITP+/+ and Pitpvb/0 animals was not scored until P11 (Fig. 5B, C). This difference in mass coincided with onset of the whole body tremor in the Pitpvb/0 animals, and it exaggerated rapidly. By P15, Pitpvb/0 juveniles embarked on a swift wasting process that persisted until death (Fig. 5B).
View larger version (35K):
[in this window]
[in a new window]
|
Fig. 5. Phenotyping of mice with graded levels of PITP protein. A: Lifespan plots for PITP+/+ (open circles), Pitpvb/vb (closed squares), Pitpvb/0 (open squares), and Pitp0/0 (closed circles) are shown. B: Average body masses of PITP+/+ (open circles), Pitpvb/vb (closed squares), Pitpvb/0 (open squares), and Pitp0/0 (closed circles) neonates are plotted in the P5 through P21 age window. Values represent averages over lifespan for progeny derived from four independent litters for each genotype. No Pitp0/0 neonates survived past P9 in these dedicated pools of animals. C: Images of P9, P14, and P19 PITP+/+ and Pitpvb/0 mice (left and right animals, respectively, in each panel).
|
|
Quantitative relationship between PITP activity and steatosis of intestine and liver
A hallmark pathology of Pitp0/0 neonates is a microvesicular steatosis of duodenal enterocytes and liver. The intestinal steatosis reflects the accumulation of dietary fat in enterocytes as a result of defective transport of that material from enterocytes into the circulation. These steatotic phenotypes are proposed to reflect defects in chylomicron biogenesis in enterocytes and lipid processing in the liver, respectively (21). Steatosis is amenable to histological examination by osmium staining of thin tissue sections, and intestinal steatosis is also diagnosed by analytical measurements of postprandial plasma TGs and brain -tocopherol. Although osmium staining of duodenal and liver tissue sections revealed the expected steatosis in Pitp0/0 offspring, clear reductions in enterocytic and hepatic steatosis were observed in Pitpvb/0 siblings. No overt steatotic pathologies were apparent in Pitpvb/vb animals relative to wild-type sibling controls (Fig. 6A
, B).
The histological data were consistent with measurements of postprandial plasma TGs and brain -tocopherol. Regarding plasma TG content, Pitp0/0 neonates exhibited 4-fold reductions in circulating plasma TG relative to PITP+/+ siblings, whereas Pitpvb/0 mice showed only an 2-fold reduction in the same parameter. Pitpvb/vb offspring presented normal circulating plasma TG (Fig. 6C). Similarly, the brain -tocopherol deficit in Pitp0/0 offspring (0.69 ± 0.31 vs. 2.71 ± 0.45 ng/mg for PITP+/+ siblings) was corrected in Pitpvb/0 and Pitpvb/vb brain (2.60 ± 0.32 and 2.51 ± 0.56 ng/mg, respectively; Fig. 6D). These data indicate that the intestinal and liver steatosis pathologies manifest themselves only in the face of large (90%) reductions in PITP expression.
Reconstitution of PITP in small intestine of Pitp0/0 mice
The involvement of PITP in chylomicron trafficking from nullizygous enterocytes is puzzling, as the affected organelle [endoplasmic reticulum (ER)] is the major intracellular site of lipid synthesis. Moreover, it remains unknown whether this pathology reflects a tissue-autonomous defect and what CRD contributes to the complex disease phenotypes of Pitp0/0 neonates. To this end, a transgene was engineered in which duodenum-specific expression of PITP cDNA is driven by the intestinal fatty acid binding protein enhancer [Tg(PIFABP::PITP)] (Fig. 7A
) (see Materials and Methods). Immunoblotting experiments of brain and intestine culled from PITP+/+, Pitp0/0, and Pitp0/0;Tg(PIFABP::PITP) offspring confirmed that the transgene supports a satisfactory tissue-specific expression of physiologically relevant levels of PITP (Fig. 7A).
Osmium staining demonstrated that Pitp0/0;Tg(PIFABP::PITP) offspring were relieved of the intestinal steatosis observed for their Pitp0/0 siblings (Fig. 7B). The rescue-of-steatosis phenotype exhibited tissue specificity, as liver steatosis was not alleviated by Tg(PIFABP::PITP) (Fig. 7C). In fact, this liver pathology was, if anything, exacerbated. Thin-section electron microscopy confirmed this effect and indicated that the accumulated lipid in Pitp0/0;Tg(PIFABP::PITP) liver is divided between what appears to be a pool that resides in the ER lumen and a pool that partitions into cytoplasmic lipid droplets (data not shown). Rescue of intestinal steatosis, as scored by histological staining methods, was accompanied by improved transport of chylomicron into the circulation. Pitp0/0;Tg(PIFABP::PITP) animals exhibited 59.1 ± 10.6% of the serum chylomicron load of their age-matched PITP+/+ controls. Relative to Pitp0/0 animals, whose circulating chylomicron load was just 24.7 ± 8.8% of wild-type levels (Fig. 8A
), this represented a marked improvement.
View larger version (28K):
[in this window]
[in a new window]
|
Fig. 8. Chemical analyses of Pitp0/0 mice expressing PITP in the duodenum. A: Postprandial serum chylomicron levels. A Titan lipoprotein agarose gel is shown at left. Genotypes of P5 littermates are indicated, and bands are identified as follows: Chyl, chylomicron; LDL, low density lipoprotein; HDL, high density lipoprotein. The right panel represents quantification (by densitometry) of chylomicron bands from a total of five animals of each indicated genotype. Tg indicates the Tg(PIFABP::PITP+) allele. PITP+/+ values were uniformly set at 100%, and Pitp0/0 and Pitp0/0,Tg(PIFABP::PITP+) values were related to that benchmark. B: Postprandial plasma TG levels as a function of age. Plasma TGs from three age-matched littermates (as indicated) were measured. Avg. indicates the average of all sample measurements for that particular genotype [PITP+/+, open circles; Pitp0/0, closed circles; Pitp0/0,Tg(PIFABP::PITP+), open triangles]. C: Whole brain -tocopherol measurements. Four age-matched littermates of each indicated genotype were analyzed. Averaged values with SD are reported. Tg indicates Tg(PIFABP::PITP+).
|
|
Plasma TG and brain -tocopherol enter the circulation via the chylomicron pathway, and measurements of these components were congruent with the chylomicron data. Pitp0/0;Tg(PIFABP::PITP) neonates exhibited 59.2 ± 5.9 mg/dl relative to 39.2 ± 6.1 and 138.1 ± 7.9 mg/dl plasma TG for Pitp0/0 and PITP+/+ siblings, respectively (Fig. 8B). Substantial rescue of brain -tocopherol deficit was similarly registered in Pitp0/0;Tg(PIFABP::PITP) offspring (3.04 ± 0.72 vs. 1.19 ± 0.22 and 3.99 ± 0.68 ng/mg brain tissue for Pitp0/0 and PITP+/+ littermates, respectively; Fig. 8C). Thus, intestinal steatosis is a tissue-autonomous pathology in Pitp0/0 mice.
Lifespan and body mass measurements indicated that duodenal PITP expression levied modest improvement in these parameters relative to Pitp0/0 siblings. Although 75% of the Pitp0/0 offspring expired within the first 3 days after birth, some 80% of their Pitp0/0;Tg(PIFABP::PITP) siblings survived past this benchmark (Fig. 9A
). A 75% mortality standard was not reached for the Pitp0/0;Tg(PIFABP::PITP) neonates until 8 days after birth. Again, although modest, this represents an effective doubling of the Pitp0/0 lifespan. Body mass measurements also showed measurable increases in weight gain differential for Pitp0/0;Tg(PIFABP::PITP) and Pitp0/0 neonates (Fig. 9B). For instance, at P7, the respective mass differential was 3.1 ± 0.3 versus 2.3 ± 0.5 g (P = 0.026).
View larger version (29K):
[in this window]
[in a new window]
|
Fig. 9. Characterization of Pitp0/0,Tg(PIFABP::PITP+) mice. A: Lifespan profiles for PITP+/+ (open circles), Pitp0/0 (closed circles), and Pitp0/0,Tg(PIFABP::PITP+) (open triangles) are compared. The data were obtained from a dedicated pool of 148 mice with a genotypic distribution of 100 PITP+/+, 21 Pitp0/0, and 27 Pitp0/0,Tg(PIFABP::PITP+). B: Visual characterization and body mass measurements of Pitp0/0,Tg(PIFABP::PITP+) mice. The upper panel shows images of P4 (left) and P7 (right) mice. For each panel, genotypes of mice (from left to right) are PITP+/+, Pitp0/0, and Pitp0/0,Tg(PIFABP::PITP+). The lower panel charts body mass as a function of age for PITP+/+ (open circles), Pitp0/0 (closed circles), and Pitp0/0,Tg(PIFABP::PITP+) (open triangles). Data were obtained from the same dedicated pool of 148 mice used for A. Body mass values were averaged for each genotypic group. SD values are shown.
|
|
PITP expression and glucose homeostasis
Pitp0/0 neonates are hypoglycemic. This condition is suggested to contribute to the neurodegenerative pathologies and morbidity of mutant animals. Hypoglycemia coincides with pancreatic islet defects, reduced proglucagon expression, and inefficient glycogenolysis (21). Consistent with previous findings, serum glucose was reduced by 4-fold for Pitp0/0 neonates compared with wild-type littermate controls (2.0 ± 0.1 vs. 8.3 ± 0.2 mM; Fig. 10A
). By contrast, Pitpvb/0 neonates exhibited serum glucose levels that were 2.8-fold higher than those of Pitp0/0 mice (Fig. 10A). This increase was accompanied by the restoration of efficient glycogenolysis in Pitpvb/0 animals (data not shown). Although an improvement, those levels were still 2-fold lower than those of PITP+/+ siblings. PITPvb/vb animals exhibited no significant serum glucose deficits relative to wild-type siblings (8.2 ± 0.9 vs. 8.3 ± 0.2 mM, respectively; Fig. 10A). Thus, as was found for the microvesicular steatosis of duodenum and liver, hypoglycemias associated with PITP defects require manifest reductions in PITP activity (90%).
View larger version (12K):
[in this window]
[in a new window]
|
Fig. 10. PITP levels and glucose homeostasis. A: Average serum glucose values measured from age-matched (P4–P6) PITP+/+, Pitp0/0, Pitp0/0,Tg(PIFABP::PITP+), Pitpvb/0, and Pitpvb/vb mice. The genotypes are indicated at bottom. B: Serum glucose levels as a function of age. PITP+/+ (open circles), Pitp0/0 (closed circles), Pitp0/0,Tg(PIFABP::PITP+) (open triangles), Pitpvb/0 (open squares), and Pitpvb/vb (closed squares) serum glucose were determined at the indicated ages. No Pitp0/0 or Pitp0/0,Tg(PIFABP::PITP+) measurements were available beyond P8, as no animals survived past this threshold. Data were obtained from triplicate determinations from each of three mice of each genotype at each time point. The serum glucose values were quite constant for Pitp0/0 and Pitp0/0,Tg(PIFABP::PITP+) mice in the P3–P8 time window (2.3 ± 0.5 vs. 3.8 ± 0.6 mM, respectively), and these differences were significant (P < 0.0002).
|
|
Specific reconstitution of PITP expression in the duodenum of Pitp0/0 animals [Pitp0/0;Tg(PIFABP::PITP)] also led to an increase in serum glucose, but this too was incomplete. Circulating glucose values approached, but did not meet, those measured for Pitpvb/0 mice (3.5 ± 0.1 vs. 5.6 ± 0.8 mM; Fig. 10A). Analyses of serum glucose as a function of age were also informative. All neonates, regardless of genotype, exhibited low blood glucose for the first few days after birth. Although PITP+/+ and Pitpvb/vb animals rapidly overcame this condition, Pitp0/0 progeny did not (Fig. 10B). Pitpvb/0 and Pitp0/0;Tg(PIFABP::PITP) neonates exhibited intermediate increases in serum glucose with age but failed to meet the levels measured for PITP+/+ and Pitpvb/vb animals.
PITP expression and cerebellar neurodegeneration
A hallmark phenotype of Pitp0/0 mice is a robust hindbrain and cerebellar inflammatory disease characterized by reactive gliosis in these tissues (21). This pathology is scored by staining for GFAP, a marker for activated astrocytes (Fig. 11A
). It is suggested that the Pitp0/0 cerebellum is particularly sensitive to damage wrought by a systemic energy crisis, such as that precipitated by the CRD and hypoglycemia of Pitp0/0 neonates (21). Consistent with this proposal, cerebellar inflammatory disease was substantially alleviated in both Pitpvb/vb and Pitp0/0;Tg(PIFABP::PITP) neonates, as indicated by dramatic reductions in GFAP-positive cells in tissue derived from those mice. By contrast, hindbrain inflammation was not alleviated in either Pitp0/0;Tg(PIFABP::PITP) or Pitpvb/0 neonates (Fig. 11B). The persistent hindbrain inflammation presumably reflects some intrinsic fragility/deficit in Pitp0/0 hindbrain neurons.
PITP nullizygosity and efficacy of synaptic transmission
Phosphoinositide signaling plays integral roles in regulating neuronal membrane trafficking (32, 33). This raised the possibility that PITP regulates the synaptic cycle and that the neurodegener
TOP
ABSTRACT
INTRODUCTION
