Requirement for the lp_A1 lysophosphatidic acid receptor gene in normal suckling behavior

Mouse 129SvJ genomic clone isolation, characterization, and subcloning were described previously (17). A 5.7-kb EcoRI/NotI genomic subclone containing exon 3 was used as a source for fragments to insert into the pFlox vector (18). Cloning took place in three steps: (i) the 3.5-kb EcoRI/XbaI 3′ “long arm” from clone vzg5.5 was ligated into the SalI/XbaI sites of pFlox, (ii) the 1.0-kb BglII/BamHI from clone BX1.2 containing exon 3 was ligated into the BamHI site, and then (iii) the 1.1-kb SalI/SalI fragment from clone BN1.1 was ligated into the XhoI site. After checking this plasmid by PCR, Southern blotting, and sequencing, a 10-kb NotI/NotI linear fragment was isolated, and 20 μg were electroporated into R1 embryonic stem cells. Both homologous recombination and Cre-mediated deletion were detected by PCR initially and confirmed by Southern blot analysis. Of 228 G418-resistant colonies screened, six were positive for homologous recombination (efficiency 1:38 or 2.6%). Two of these six were electroporated with a Cre-expression plasmid. Of 72 gancyclovir-resistant clones analyzed, 10 had deleted both exon 3 and neomycin-resistant/thymidine kinase genes (type I deletion). One type I clone (28–5#11) was confirmed to be euploid and was injected into blastocysts, resulting in chimeric mice backcrossed with C57BL/6J females. Intercrosses were begun immediately with these mice, as well as with mice generated from one additional backcross.

To genotype mice, Southern blots were made using BglII-digested genomic DNA (gDNA) isolated from embryonic or P21 tails. The 650-bp PX650 probe was obtained from the XX1.5 plasmid by digestion with PvuII and XbaI (17). The exon 3 probe was generated by PCR from a cDNA plasmid using primers 513 M and 513T (17). For PCR genotyping of mice, 0.4 μl (100–200 ng) of gDNA was added as template in hot-start 20 μl of PCR, then cycled 35 times (95°C, 30 s; 56°C, 30 s; and 72°C, 2 min) with a final incubation of 72°C for 5 min before cooling to 4°C. The PCR mix consisted of 60 mM Tris (pH 8.5)/15 mM (NH₄)₂SO₄/2.0 mM MgCl₂/0.5 mM each oligonucleotide/0.25 mM each dNTP/0.4 U Taq. Primers in the PCR mix were as follows: 513QL, 5′-GCCAATCCAGCGAAGAAGTC-3′; vzg.il1, 5′-GGTATTCTTAATTCTAGAGGATCAGC-3′; vzg.is2, 5′-TATAGGAGTCTTGTGTTGCCTGTCC-3′.

Northern blotting and RT-PCR conditions have been described previously (16, 19). The number of cycles for all of the PCR shown were adjusted so product was not maximized, as compared with quantities obtained with excess product as template. All cDNA template amounts were standardized to yield identical quantities of β-actin product, which was amplified with primers that do not amplify any pseudogenes from gDNA (20). All reactions used primers located in different exons, eliminating the possibility of gDNA contamination. Primers used were the following: lp_A1, 513C/513T (17); lp_A2, edg6f (5′-GACCACACTCAGCCTAGTCA-3′) and edg6e3c′ (5′-ACATCCAGAGATAACACAGTAA-3′); lp_B1, edg1p (5′-CCGTCAGTCGCCGACAACAA-3′) and edg1b (5′-GTAGAGGATGGCGATGGAAA-3′).

Cortical cultures were made essentially as described (14, 15). Briefly, the cerebral cortices from embryonic (E) day 12–13 mice were dissected, and meninges were carefully removed. Tissues were transferred to culture medium and gently triturated with a fire-polished Pasteur pipette into small clusters (<100 cells per cluster). Cells were seeded onto glass coverslips (50–200 clusters on 12-mm coverslips or 5–20 clusters on the center of 40-mm coverslips) precoated with 2 μg/cm² of Cell-Tak (Becton Dickinson) and cultured at 37°C under 5% CO₂ for 18 h. For time-lapse video-enhanced differential interference contrast microscopy, the coverslips were mounted onto a heat-controlled perfusion chamber (BioOptiks) set at 37°C and observed with an inverted microscope (Axiovert 135, Carl Zeiss). Either vehicle (0.1% fatty acid-free BSA) or LPA was added for 15 min. The differential interference contrast images were detected with a cooled charge-coupled device color camera (DEI-47, Carl Zeiss). Cluster area was determined using Scion Image software, with changes calculated as percentage of initial area (t = 0 min). To detect proliferation, cultures were treated with or without LPA for 18 h and pulsed with 40 μM BrdUrd for the last 30 min, followed by staining with α-BrdUrd antibody (Boehringer Mannheim). Only cells in clusters were counted.

For apoptotic DNA in situ end-labeling plus (ISEL⁺; ref. 21), sciatic nerves (eight nerves from four different animals of each genotype) were isolated from young adult mice and rapidly frozen in Tissue Tek (Miles). Cryostat sections were cut at 10 μm and then fixed, processed, and labeled as described (21). Microscope fields (×100 objective) were counted along the entire length of each nerve section (>16,000 cells total for each genotype). For electron microscopy, sciatic nerves from adult mice were removed, and ≈2- to 3-mm pieces were immersed in 3% glutaraldehyde in 0.1 M cacodylate buffer at 4°C for 24 h. These were embedded in Scipoxy 812 resin (Energy Beam Sciences, Agawan, MA), and 80-nm sections were cut and stained with saturated ethanol/uranyl acetate and bismuth subnitrate. Grids were examined and photographed using a Zeiss electron microscope.

Mouse lp_A1 consists of at least 5 exons, with 68% of the coding region (including transmembrane domains I–VI) located in exon 3 (17). We therefore targeted exon 3 for deletion, using the Cre-lox system, which allows embryonic stem cells to be generated without a residual selectable marker gene and its associated constitutive promoter (18). In the event that exons 2 and 4 were spliced together, all coding sequence contributed from exon 4 would be out of frame. Mutant embryonic stem cell clones were obtained by homologous recombination of a replacement vector containing loxP sites flanking both exon 3 and the neomycin-resistant/thymidine kinase genes (Fig. 1 a and b). Expression of Cre recombinase in one of these clones allowed deletion of exon 3 and neomycin-resistant/thymidine kinase (Fig. 1c). After injection into blastocysts, chimeric mice were mated to C57BL/6J mice and progeny intercrossed. We found that some lp_A1^(−/−) mice could survive and reproduce, as determined by verification of homozygosity through Southern blot and PCR analyses of gDNA (Fig. 1 d and e). Northern blot analysis of lp_A1^(−/−) adult brain RNA confirmed the lack of exon 3 expression (Fig. 1f). Because the up-regulation of related LPA receptor genes might compensate for the absence of lp_A1, we determined whether transcript levels of lp_A2/edg-4 (10, 17, 19, 22, 23), lp_A3/edg-7 (refs. 10, 23, and 24; data not shown), or lp_B1/edg-1 (25, 26) were altered in lp_A1^(−/−) adult brain. No changes were observed in expression levels of any of these genes (Fig. 1f).

Juvenile and adult lp_A1^(−/−) mice could always be distinguished from their nonhomozygous siblings by their shorter snouts, more widely spaced eyes, and smaller size (Fig. 2 a and b). A quantitative assessment of craniofacial deformity in adults using a measure independent of overall head size (eye-to-nose tip length/interocular distance) supported qualitative impressions (lp_A1^(−/−): 1.04 ± 0.16 vs. lp_A1^(+/−): 1.47 ± 0.05; P < 0.05; unpaired t test). X-ray analysis of skeletal structures in adult lp_A1^(−/−) mice revealed only relative size differences in the frontal area rather than missing bones, cleft palate, or other abnormalities (data not shown). The underlying cause of the dysmorphism is likely related to loss of lp_A1 that is normally expressed within developing facial bone tissue (C. McGiffert and J.C., unpublished observation). An additional factor may be frontal hematomas observed in a small fraction (4 of 160) of embryonic and neonatal lp_A1^(−/−) mice (Fig. 2 c and d) because neonatal pups with such hematomas showed more pronounced craniofacial dysmorphism later in life. Hematomas might be caused by defects in LPA-induced endothelial cell monolayer barrier formation (27) or a localized role in angiogenesis similar to that hypothesized for LP_B receptors (28). The second obvious gross phenotype of adult lp_A1^(−/−) mice, smaller size, was found to be significant from several days after birth and into adulthood for both males and females, with up to a 30% reduction relative to control siblings (Fig. 2e). This primarily reflected size differences in all structures because of decreased postnatal growth rate (Fig. 2b). However, lower relative amounts of fat also contributed to the total mass differences (data not shown), possibly because of defective LPA signaling in adipose cell development (29).

To examine whether there might be a defect in Schwann cell survival in vivo, we examined young adult sciatic nerve sections for cells undergoing apoptosis using ISEL⁺ that labels fragmented DNA (21). These analyses demonstrated that there was an approximately 80% increase in the percentage of ISEL⁺-labeled cells in lp_A1^(−/−) mice compared with wild-type mice (Fig. 3 a and b). The presence of apoptotic cells was confirmed by observing condensed nuclei using electron microscopy (Fig. 3c). In addition to increased apoptosis of Schwann cells in vivo, fewer cells were also apparent in the sections (Fig. 3a). In support of an LPA signaling defect as the primary cause of the increased apoptosis rates, we determined that cultured Schwann cells had severely diminished LPA responses in a morphological assay (J.A.W., N.F., and J.C., unpublished observation). Despite these defects, we did not observe grossly abnormal movement in lp_A1^(−/−) mice, which might be expected from severe myelination defects. This may reflect a relatively minor extent of Schwann cell/myelin loss because the actual increase in Schwann cell apoptosis (18% in lp_A1^(−/−) mice as opposed to 10% in wild-type mice) still leaves the majority of Schwann cells intact.

Genotyping litters from both lp_A1^(+/−) × lp_A1^(+/−) and lp_A1^(+/−) × lp_A1^(−/−) crosses demonstrated that about half of both male and female lp_A1^(−/−) mice died sometime before 3 weeks of age (Table 1). This effect was not caused by a homozygous maternal phenotype because the semilethality was similar for both lp_A1^(+/−) and lp_A1^(−/−) mothers (Table 1 and data not shown). To determine whether pups were dying embryonically, we genotyped litters before birth, E12–18. Here, the expected ratios were present, indicating pups were being lost sometime between late embryonic development and 3 weeks of age (Table 1). During this analysis, we observed a small fraction (3 of 61) of lp_A1^(−/−) embryos with exencephaly (Fig. 4a). Exencephaly is a neonatal lethal phenotype that is also found in a small fraction of other null-mutant mice (30). Although we have no explanation for this effect, it does suggest that LPA signaling may have a previously unrecognized role in neural tube closure. Whereas one might speculate that the lp_A1^(−/−) lethality is related to frontal hematomas, we observed that several newborn pups with frontal hematomas survived, with hematomas dissipating after several days.

To determine the primary cause of lp_A1^(−/−) deaths, 10 litters from lp_A1^(+/−) × lp_A1^(−/−) crosses were observed from just after birth through weaning. A survival plot is presented in Fig. 4b. Of the 38 lp_A1^(−/−) pups genotyped [approximately the expected number based on 37 lp_A1^(+/−) littermates], six were dead immediately after birth. Two of these were partially cannibalized by the time of observation (possibly related to exencephaly), whereas the other four did not commence normal breathing (but were otherwise grossly normal). Twelve additional lp_A1^(−/−) pups died within the first 4 postnatal days, and all of them lacked milk in their stomachs (e.g., Fig. 4c). Observation of these pups before their death indicated that they were not suckling. The craniofacial dysmorphism was not apparent in the dead lp_A1^(−/−) pups (nor any neonatal pups) and thus was not causally related to the deaths. To determine whether the impaired suckling phenotype was present in all lp_A1^(−/−) pups rather than a subset, milk quantity was correlated with genotype at two postnatal ages (P0 and P3). The majority of lp_A1^(−/−) pups had little or no milk in their stomachs, unlike control lp_A1^(+/−) littermates, which normally had full stomachs (Fig. 4 c and d). Thus, the suckling defect was present in nearly all lp_A1^(−/−) pups. This result can explain the decreased size of surviving lp_A1^(−/−) pups relative to control littermates and is consistent with results from olfactory bulbectomized pups or mouse olfactory mutants (e.g., G_olf and cyclic nucleotide-gated channel mutants) that together show suckling defects, decreased size, and partial lethality (31–33).

Neonatal suckling in rats and mice is a complex behavior that depends on several brain structures as well as nerves and muscles required to extract milk from the nipple (31, 34, 35). Olfactants/pheromones present on the nipple are the primary sensory cue used in orientation and localization, although tactile sensation is also required. Once a nipple is located, a rooting reflex program is activated, which involves rhythmic mouth suckling movements and swallowing. Structures involved in the overall behavior include the olfactory epithelia, olfactory bulb, various cranial ganglia, the nerves/muscles involved in rhythmic mouth contractions and swallowing, and perhaps the vomeronasal organ and olfactory cortex (entorhinal/pyriform cortex and amygdala). To determine more precisely what behavioral defect impaired suckling in lp_A1^(−/−) pups, we observed their nipple searching and tactile reflexes. All lp_A1^(−/−) pups exhibited a normal rooting reflex in response to manual stimulation of their mouth region, indicating normal tactile sensation and motor control. Although we found that lp_A1^(−/−) pups searched for nipples normally, they often did not find them, similar to the phenotype observed with G_olf and cyclic nucleotide-gated channel mutant mice (32, 33) (data not shown). These results indicated that the impaired suckling was induced by a lack of olfactant detection and/or processing.

Olfactant detection and processing involves olfactory epithelia, bulb, and possibly cortex. To determine whether general histological abnormalities in these structures might explain the impaired behavior, we perfusion-fixed neonatal lp_A1^(−/−) pups that had little milk in their stomachs and stained coronal head sections with cresyl violet. These studies did not reveal obvious abnormalities in olfactory/vomeronasal epithelia, olfactory bulb, or cortex, except occasionally, slightly smaller cortical width (data not shown). Expression of lp_A1 in the embryonic brain is most prominent in the dorsal cerebral cortex (7) but also extends into the anterior aspect of the dorsal telencephalon that in part gives rise to the olfactory bulb (8). Although no clear differences between lp_A1^(+/+) and lp_A1^(−/−) brain histological sections assayed for cells in S-phase or apoptosis were observed, several embryonic lp_A1^(−/−) brains/olfactory bulbs analyzed showed reduced wall thickness (data not shown).

To investigate the possible mechanism of a growth defect in embryonic neuroblasts, we examined whether LPA signaling was normal. Additionally, there has been some debate regarding the identity of the lp_A genes as encoding endogenous LPA receptors, especially in view of a report implicating a molecularly dissimilar LPA receptor gene (PSP24) (36). Primary cultures from both lp_A1^(−/−) and lp_A1^(+/+) embryonic (E12–13) cerebral cortex were examined for LPA morphological and proliferative responses. Proliferating cells in these cultures existed predominantly in small clusters (Fig. 5a) and were immunoreactive for the intermediate filament protein, nestin (data not shown), consistent with being neuroblasts (37). Wild-type cultures responded to 100 nM LPA with cluster compaction (quantifiable as a reduction in cluster area) (15), reflecting both cell migration toward, and cell rounding within, clusters (Fig. 5 a and b). By comparison, lp_A1^(−/−) clusters exposed identically to LPA were severely attenuated in this response, with quantified area changes not significantly different from vehicle-treated controls (Fig. 5 a and c). A second LPA response, increased cell proliferation, was analyzed by treatment with vehicle solution or 100 nM LPA for 18 h followed by labeling with BrdUrd for the last 30 min. Although cells from lp_A1^(+/+) embryos responded to LPA with a 25% increase in BrdUrd incorporation relative to vehicle treatment, lp_A1^(−/−) cells did not respond (Fig. 5d). Although both lp_A1^(+/+) and lp_A1^(−/−) cultures exhibited proliferative responses to bFGF stimulation as expected (38), lp_A1^(−/−) cultures showed increased responsivity (Fig. 5d). These data indicate that LP_A1 mediates endogenous LPA effects in cerebral cortical neuroblasts and suggest that loss of normal LPA-dependent proliferation in lp_A1^(−/−) cortex may in part be compensated for by increased proliferative responses to bFGF.

The existence of two homologous high-affinity receptors (LP_A2 and LP_A3) and a putative low-affinity (LP_B1/Edg-1) LPA receptor (39) suggests that each may mediate distinct LPA-dependent functions. Thus, we examined expression of germane lp receptor genes in the mutant embryonic cerebral cortex and cortical cell cultures. Interestingly, despite the loss of LPA-dependent cell proliferation in these cultures, we detected relatively abundant transcript levels of lp_A2 and lp_B1 and low levels of lp_A3 (Fig. 5 e and f). The expression of these genes suggests that other LPA responses might still be found in some lp_A1^(−/−) cells, and in fact, we have observed LPA-induced, distinct morphological changes in other cell subtypes from lp_A1^(−/−) brain (N.F. and J.C., unpublished observation). Such differing effects mediated by LPA receptor subtypes are supported by comparative expression studies of LP receptors in neuronal cell lines (23). There was no obvious change in expression level of any of the examined receptors in lp_A1^(−/−) cells (Fig. 5 e and f). These data indicate that specific, nonredundant LPA effects are mediated by the LP_A1 receptor.

We have demonstrated using genetics that lp_A1 is required for normal neonatal suckling behavior. The defect in suckling is responsible for both the decreased size of the mice and the majority of lethality associated with the homozygous lp_A1 null allele. Enriched expression of lp_A1 in the dorsal telencephalic proliferative zone and loss of LPA responsivity in cells from this region suggest that altered development of the cerebral cortex and/or olfactory bulb may be responsible for the suckling defect. Increases in Schwann cell apoptosis within the intact sciatic nerve of lp_A1 null animals lend further support for LPA signaling in the survival of myelinating cells. Overall, our results provide the first evidence that endogenous LPA signaling is receptor mediated, that a single lp receptor gene can have nonredundant biological roles, and that receptor-mediated LPA signaling is required for normal developmental processes.