Systematic expression analysis of plasticity-related genes in mouse brain development brings PRG4 into play

Background: Plasticity-related genes ( Prgs /PRGs) or lipid phosphate phosphatase-related proteins (LPPRs) comprise five known members, which have been linked to neuronal differentiation processes, such as neurite outgrowth, axonal branching, or dendritic spine formation. PRGs are highly brain-specific and belong to the lipid phosphate phosphatases (LPPs) super-family, which influence lipid metabolism by dephosphorylation of bioactive lipids. PRGs, however, do not possess enzymatic activity, but modify lipid metabolism in a way that is still under investigation. Results: We analyzed mRNA expression levels of all Prgs during mouse brain development, in the hippocampus, neocortex, olfactory bulbs, and cerebellum. We found different spatio-temporal expression patterns for each of the Prgs, and identified a high expression of the uncharacterized Prg4 throughout brain development. Unlike its close family members PRG3 and PRG5, PRG4 did not induce filopodial outgrowth in non-neuronal cell lines, and does not localize to the plasma membrane of filopodia. Conclusion: We showed PRG4 to be highly expressed in the developing and the adult brain, suggesting that it is of vital importance for normal brain function. Despite its similarities to other family members, it seems not to be involved in changes of cell morphology; instead, it is more likely to be associated with intracellular signaling.


| INTRODUCTION
The five known plasticity-related genes (Prgs/PRGs), also named lipid phosphate phosphatase related proteins (LPPRs), form their own brain-and vertebrate-specific subgroup within the lipid phosphate phosphatase (LPP) protein superfamily. 1,2 LPPs are surface-located, membrane-spanning proteins with six transmembrane regions and three extracellular loops. These loops contain conserved ecto-enzymatic active sites that dephosphorylate bioactive lipid substrates such as lysophosphatidic acid (LPA) or sphingosine-1-phosphate (S1P). [3][4][5] This regulates the affinity of these extracellular lipids to their receptors and thereby modulates the associated intracellular signaling processes.
The PRGs are homologs of LPPs, but they lack the characteristic ecto-phosphatase activity because of nonconserved substitutions in the respective catalytic domains. 2,4,5 Regardless of the absent phosphatase activity, PRGs can still influence bioactive lipids and their signaling pathways, but the underlying mechanisms are currently under investigation. [6][7][8] and PRG2 have about 400 amino acid long C-termini, whereas PRG3, PRG4, and PRG5 have rather short ones, consisting of only around 50 amino acids. 9 They differ considerably in their unique intracellular C-termini, which therefore might play a role in mediating their specific functions. For PRG5, for example, a C-terminal binding to phosphorylated phosphatidylinositols (PtdInsPs) has been demonstrated, and its functionality has been linked to this specific interaction. 7 Previous studies have demonstrated a widespread involvement of PRGs in molecular mechanisms of neuronal differentiation. PRG1 was the first PRG identified in a screening of a cDNA library of the lesioned murine hippocampus, 1 and remains to date the most-studied PRG. It was shown to be expressed during axonal outgrowth after a lesion, and to attenuate LPA-mediated axon collapse; hence, promoting regenerative sprouting processes. 1,10 PRG2 has the greatest resemblance to PRG1, and it was functionally correlated to axonal growth and branching by its stabilizing of phosphoinositides in the plasma membrane during brain development. 8 It was also shown to be important for axonal guidance of thalamocortical fibers by its arbitrating of the axonal sensitivity to LPA. 11 The high homolog PRG3 and PRG5 have both been linked to morphological changes, more precisely to the induction of membrane protrusions such as filopodia. [12][13][14] In primary cultured neurons, PRG3 overexpression leads to increased neurite outgrowth and neurite shaft protrusion, and was associated with regeneration after spinal cord injury in adult mice. 15,16 PRG5, however, is assumed to be involved in the formation of dendritic spines, as well as in the morphology, stabilization, and proper function of excitatory synapses. 7 PRG4 has not been specifically analyzed and no functional data is yet available. In a screen of drug-resistant human melanoma cell lines, it was found to be upregulated, but a direct link to lipid phosphate signaling was not observed. 17 The existing information on PRG function implies a central role of this protein class in neuronal development, as well as in neuronal reorganization processes. Furthermore, this can be of use in understanding pathogenesis and associated repair mechanisms. It is therefore of vital importance to further investigate PRG expression and function, including the previously uninvestigated PRG4 protein.
All PRGs show vertebrate-and brain-specific expression. Most previous studies focused mainly on single PRGs, and expression analysis was limited to the corresponding research subject. Here, we aimed for a systematic expression analysis of all known PRGs. We focused on mouse brain development, from late embryonic stages to adulthood, and on different brain-cell types. Our analysis revealed dynamic expression patterns for Prg1, Prg2, Prg3, and Prg5. By contrast, Prg4 showed high expression levels throughout brain development, raising questions about its specific role in fundamental neuronal processes. In contrast to the closely related PRG3 and PRG5, we found that the overexpression of PRG4 does not induce membrane protrusion and filopodia formation, suggesting a role in a different pathway.
2 | RESULTS 2.1 | Prgs are dynamically expressed in different brain areas during development A previous study of the developmental mRNA expression of Prg3 and Prg5 in the hippocampus demonstrated a dynamic expression regulation for both Prgs. 7,16 To our knowledge, no systematic gene expression analysis of all Prgs during mouse brain development has yet been carried out. Therefore, in this study, we aimed for a complete comparison of the gene expression of all five known Prgs in different brain areas. We analyzed mRNA expression levels in the hippocampus, neocortex, olfactory bulbs, and the cerebellum at developmental stages E14 to P30 by quantitative real-time polymerase chain reaction (qRT-PCR). We found all of the Prgs expressed in all analyzed brain regions and developmental stages, but with different dynamic expression patterns (Figure 1, A, C, E, and G). The previously published Prg3 and Prg5 expression in the hippocampus is included in Figure 1A in shaded bars as a matter of completeness. For direct F I G U R E 1 Legend on next page. comparison of pre-and postnatal expression dynamics, we normalized expression levels to P0 expression and plotted changes of all Prgs together over time for each brain region (Figure 1, B, D, F, and H).
Our results show complementary expression levels for Prg1 and Prg2 during brain development, especially in the hippocampus, neocortex, and olfactory bulbs. Expression differences in adulthood were particularly high in hippocampus and neocortex samples.

| Temporally shifted expression of Prg3 and Prg5
In situ hybridization of Prg3 mRNA in the developing rat brain showed Prg3 expression from E16, mainly in the hippocampal anlage, thalamus, and the olfactory bulbs. 18 Our qRT-PCR experiments in different areas and different developmental stages of the mouse brain revealed high Prg3 mRNA expression from developmental stage E14 on in all investigated areas. Expression peaked at late embryonic stages and around birth and decreased to low expression levels in adult stages ( Figure 1; E14 vs P0: hippocampus ***P < .0001, neocortex ***P = .01, olfactory bulbs ***P = .0036, P0 vs P30: all areas ***P < .0001). Only in the cerebellum, we found higher Prg3 expression in early embryonic development and expression was absent in adult stages ( Figure 1G; E14 vs P0: not significant, P0 vs P30: ***P < .0001). In the olfactory bulbs, Prg3 expression remained slightly elevated in adult stages compared to the other brain areas. This was also observable for Prg5, were expression in the olfactory bulbs did not decrease towards adult stages ( Figure 1E; P0 vs P30: not significant). However, Prg5 expression peaked later, in early postnatal stages, resulting in a slightly shifted expression pattern of these genes (E14 vs P0: all areas ***P < .0001; P0 vs P30: hippocampus, neocortex, and cerebellum ***P < .01). Our results show similar expression patterns for Prg1 and Prg5 genes, as well as for Prg2 and Prg3 genes, albeit at different expression levels ( Figure 1B, D, F, and H).

| Continuously high expression of Prg4 throughout brain development
At the time of this study, Prg4 expression in mouse brain tissue had not been documented in the literature. Here, we show its mRNA expression in developing mouse brain areas for the first time ( Figure 1). We found Prg4 to be ubiquitously expressed throughout all developmental stages and all brain areas examined. We found only one significant expression peak, in the neocortex around birth ( Figure 1C; E14 vs P0 *** P < .0001) and significant expression decreases towards adulthood in the neocortex and the cerebellum (Figure 1 G; P0 vs P30: neocortex ***P < .0001, cerebellum ***P = .0006). In comparison to other Prg family members, only very minor, changes could be detected, and in the hippocampus and the olfactory bulbs, its expression remained stable throughout development ( Figure 1B, D, F, and H; E14 vs P30: hippocampus and olfactory bulbs, not significant).
F I G U R E 1 Expression of Prg genes during mouse brain development. Analysis of Prg1-5 mRNA expression in murine hippocampus, neocortex, olfactory bulbs, and cerebellum between E14 and P30 by quantitative real-time polymerase chain reaction (qRT-PCR). (A, C, E, and G) Relative mRNA expression normalized to Gapdh and was shown for each Prg over developmental stages. Prg3 and Prg5 expression during hippocampus development, which was published in Velmans et al 16 and Coiro et al. 7 is shown in (A) in shaded bars. Statistical analysis was performed using a one-way ANOVA followed by Bonferroni's multiple comparisons test. Data are shown as mean ± SD and were considered significant for P ≤ .05 (*P ≤ .05, ***P ≤ .01, ns = not significant). P-values are listed in Table 1 2.5 | Prg expression in different primary brain cells Previously, we investigated the expression of Prg3 and Prg5 in different primary brain cells (neurons, astrocytes, and microglia) and demonstrated their mainly neuronal expression (Figure 2A, shaded bars). 7,16 Analysis of Prg1, Prg2, and Prg4 in these cultured primary neuronal cells also revealed a predominantly neuronal expression. Only Prg4 additionally showed increased expression in astrocytes, whereas Prg1, Prg2, and Prg5 were only weakly expressed in this cell type. All Prgs are only poorly expressed in microglia, with Prg2 showing higher expression than the other family members (Figure 2A).
We found all Prgs to be expressed in primary cultured oligodendrocytes, with high expression levels of Prg1, Prg3, Prg4, and Prg5. Strikingly, Prg3 was highly expressed in immature oligodendrocytes, but decreased F I G U R E 2 Expression of Prg genes in primary cultured brain cells. (A) Analysis of Prg1-5 mRNA expression in primary cultured hippocampal neurons (DIV7), astrocytes (DIV14-16), and microglia (DIV14-16) by qRT-PCR. Prg3 and Prg5 mRNA expression were published in Velmans et al 16 and Coiro et al 7 and are included for comparison (shaded bars). Statistical analysis was performed using a oneway ANOVA followed by a Bonferroni's multiple comparisons test. Data are shown as the relative expression normalized to Gapdh as mean ± SD and were considered significant for P ≤ .05 (*P ≤ .05, ***P ≤ .01). Neurons: n = 3; N = 21 to 30. Astrocytes and microglia: n = 3; N = 9 to 12. (B) Analysis of Prg1-5 mRNA expression in immature (DIV3, clear bars) and mature (DIV6, patterned bars) primary cultured oligodendrocytes. Data are shown as the relative expression normalized to Gapdh as mean ± SD. Statistical analysis was performed using a one-tailed t-test. Data were considered significant for P ≤ .05 (*P ≤ .05; n = 3; N = 9). n, number of independent preparations; N, number of total animals T A B L E 1 Significance and P-values for one-way ANOVA of Prg gene expression during mouse brain development (*P ≤ .05, ***P ≤ .01, ns = not significant) during oligodendrocyte maturation ( Figure 2B; immature vs mature *P = .0294). Prg1, Prg4, and Prg5, on the other hand, exhibited a high expression in both immature and mature primary cultured oligodendrocytes. Prg2 was the least expressed Prg in oligodendrocytes, with a slight, but not significant decrease towards maturation ( Figure 2B).  Table 2). Due to their high aa sequence identity, we aligned murine PRG3, PRG4, and PRG5 aa sequences to localize differences within the proteins (Figure 3). We found the largest differences in the intracellular N-and C-termini, the first intracellular loop (ICL), and the second extracellular loop (ECL) of the proteins. The intracellular N- terminus of PRG5 is very short and consists of five aa less than that of PRG3, and six aa less than the N-terminus of PRG4. PRG3, and PRG4 N-termini are longer, but do not share sequence similarities. The second ECL of PRG4 consisted of seven aa more than that of PRG3 and PRG5, and thus differs distinctly. All three PRGs share an N-glycosylation side in the loop that is required for membrane insertion, as shown for PRG3 by Velmans et al. 16 PRG4 has a seven aa longer Cterminus than both PRG3 and PRG5. Domain similarities at the C-terminus of the three proteins can only be seen at the passage from the transmembrane region 6 (TM6) to the intracellular C-terminus. The additional aa in the N-and C-termini and the second ECL of PRG4 make it, with 343 aa, the longest of these three PRGs. Despite the about 50% identity in the aa sequence, PRG4 shows distinct differences in protein areas that might be crucial for its specific function. To analyze possible signal peptides in PRG sequences, we used the SignalP 5.0 online tool (https://services.healthtech.dtu.dk/service.php?SignalP-5. 0). 19 Interestingly, none of the PRGs is likely to have a known signal peptide at its N-terminus, or first transmembrane region. Hence, insertion of PRGs into the membrane might underlie a different, as yet unknown, mechanism.

|
2.7 | Unlike PRG3 and PRG5, PRG4 does not induce filopodial outgrowth in nonneuronal cells Overexpression of PRG3 and PRG5 has been demonstrated to lead to similar phenotypes in neuronal and non-neuronal cell lines. [12][13][14]18 These phenotypes are marked by the intense formation of actin-rich membrane protrusions that were identified as filopodia. Due to their high aa sequence identity, we investigated whether PRG4 induces a comparable phenotype in non-neural cell lines. We identified strong filopodial formation after PRG3 and PRG5 overexpression in HEK293H cells ( Figure 4) and therefore chose this cell line for the analysis of PRG4 overexpression. We know of no reliable antibody against PRG4 and therefore used N-and C-terminal enhanced green fluorescent protein (eGFP) fusion proteins and Cterminal FLAG-tagged proteins for transfection experiments. All PRG4 constructs were unable to induce the pronounced formation of membrane protrusions and filopodia-like structures seen for PRG3 and PRG5 eGFP and FLAG-tagged constructs ( Figure 4). We observe a mainly intracellular localization of the PRG4 fusion proteins, with only a few areas of plasma membrane localization, mainly restricted to central areas of the cell ( Figure 4B). Quantification of filopodia formation, by determining the number of filopodia per cell, supported the immunostaining results and showed significant differences between PRG3 and PRG5 compared to PRG4 overexpressing cells ( Figure 4C).
Surprisingly, the N-terminal fusion of eGFP to PRG5 disabled the induction of filopodial formation by PRG5. This was not the case for PRG3, where N-and C-terminal fusion proteins were able to promote filopodial outgrowth. For PRG4, no differences between fusion sites were observed ( Figure 4B).

| Cooperative expression of PRG3, PRG4, and PRG5 in HEK293H cells
Using co-immunoprecipitation and co-localization analysis, Yu et al showed cooperative interactions of PRG3 with its family members PRG1, PRG2, and PRG5. They demonstrated that the co-expression of PRG3 and PRG5 facilitated their localization to the plasma membrane, particularly to membrane protrusions and filopodia, and also increased membrane protrusion outgrowth. 13 We examined whether the localization of PRG4 to the filopodia plasma membrane is dependent on an interaction with its close family members PRG3 and PRG5, and if this interaction might be necessary for a possible phenotype of PRG4 overexpression. We therefore generated a PRG4-FLAG construct, which showed equivalent results to the eGFP constructs before, and when co-transfected PRG4 with PRG3 or PRG5 in HEK293H cells. Figure 5 shows HEK293H cells, co-transfected with the PRG4-FLAG and either PRG3-eGFP (A), PRG5-eGFP (B) or a CFP-MEM vector as a control membrane protein (C). Both family members failed to increase the localization of PRG4 to the filopodia plasma membrane ( Figure 5, white arrowheads), which mainly remained localized to intracellular structures and central plasma membrane areas. Both PRG3 and PRG5 showed intracellular colocalization with PRG4. The PRG3-and PRG5-specific phenotypes of increased filopodial outgrowth were not altered by additional PRG4 overexpression and could still be observed after co-transfection ( Figure 5A, B). The same results were obtained with co-transfection of PRG4-eGFP with PRG3-or PRG5-FLAG constructs (data not shown), and co-transfection with CFP-MEM control vector did not affect the overall distribution patterns ( Figure 5C).

| DISCUSSION
In this study, we examined the gene-expression profiles of all Prg family members during mouse brain development, in the hippocampus, neocortex, olfactory bulbs, and the cerebellum, from embryonic stage E14 to maturation at P30. Our results showed different temporally and spatially dependent dynamic expression patterns for each of the Prgs. Our data fit well to previous functional studies on Prgs, and can also be used for further analysis, as expression patterns can be linked to specific developmental processes at distinct stages.
Functional studies on PRG1 indicate its involvement in axonal outgrowth 1,10 and glutamatergic transmission at the postsynapse. 6,20 These processes are associated with late embryonic and early postnatal development, where we found the highest Prg1 expression levels.
Our results on Prg2 expression are in line with recent results on developmental protein expression in total rat brain lysates, with an expression peak around birth and low expression levels in mature stages. 8 In addition, we showed that Prg2 remains at elevated levels in the adult olfactory bulbs. Prg2 has been correlated to axonal and F I G U R E 4 PRG4-eGFP fusion proteins did not induce filopodial outgrowth in HEK293H cells, and the GFP fusion site plays a role in PRG5 function. Representative images of HEK293H cells transfected with C-terminal eGFP fusion proteins (A) and N-terminal eGFP fusion proteins (B) of PRG3, PRG4, and PRG5 (green). Cell nuclei were stained with DAPI (blue). Scalebars 20 μm. (C) Boxplot of number of filopodia per cell after overexpression with C-terminal FLAG-tagged PRG3, PRG4, or PRG5 proteins and the membrane marker CFP-MEM in HEK293H cells. Statistical analysis was performed using a Kruskal-Wallis test followed by a Dunn's multiple comparison test and was considered significant for P ≤ .05 (*P ≤ .05, ***P ≤ .01). n = 3 individual transfection experiments with at least 10 cells per experiment; N = number of total cells dendritic branching processes of cortical neurons in vitro by inhibition of PTEN and the resulting stabilization of PI(3, 4, 5)P 3 . 8 We detected Prg2 mRNA expression in embryonic and early postnatal stages that are linked to neuronal migration and branching processes, especially of axonal projections. Elevated Prg2 expression in the olfactory bulbs could be explained by the olfactory bulb's regenerative and re-wiring capacity in adulthood, which might require PRG2/PTEN interaction in axonal branching processes.
Several studies of Prg3 expression and function exist. Savaskan et al analyzed Prg3 mRNA expression in rat brain and found similar expression patterns to the current study, with high levels in late embryonic stages and an expression decline after birth. In adult stages, they reported high Prg3 expression in the hippocampus; additionally, we found high expression in adult olfactory bulbs. 18 Wang and Molnar analyzed Prg3 expression in mice, specifically in the cerebral cortex, and found differential expression between cortical layers. They describe a decline in differential expression after P10, which matches the general drop in Prg3 expression of our analysis. 21 The PRG3 protein was reported to influence cell morphology by inducing membrane protrusions in different cell lines, and Prg3 knock-down decreases the number of neurites in primary cultured neurons. 14,16,18 In line with these findings, stages of high Prg3 expression can be linked to neuronal branching processes during the development.
The closely related PRG5 protein has also been linked to cell morphological changes in different cell lines. Overexpression in primary cultured neurons leads to the premature formation of spine-like structures, and to an increased spine density and altered spine morphology in F I G U R E 5 Co-transfection of PRG3 and PRG5 with PRG4 did not increase its localization to the filopodia plasma membrane. Representative images of HEK293H cells co-transfected with a PRG4-FLAG construct (magenta) and either PRG3-eGFP (A) or PRG5-eGFP (B) constructs or CFP-MEM (C) as a control membrane protein (green). White arrow heads indicate absence of PRG4 in filopodia and membrane protrusions in higher magnification and orthogonal views. Scalebars 20 μm mature neurons. 7 We found Prg5 expression in stages of synapse formation and maturation, around birth and during early postnatal development. We also identified Prg5 expression as remaining elevated in regions of high neuronal plasticity: the hippocampus and the olfactory bulbs. This indicates a possible participation of the PRG5 protein in adult neuronal plasticity, which includes morphological changes in synaptic connections.
No expression studies for Prg4 have been conducted so far, and to date, almost no functional data is available. The only published data, from Tanic et al, identified Prg4 overexpression in a screening of differentially expressed genes in drug-resistant melanoma cells. 17 We found Prg4 to be highly expressed during the development of the hippocampus, neocortex, olfactory bulbs, and the cerebellum. Expression levels were remarkably stable during the explored time period and were not linked to specific developmental processes. This suggests a role of Prg4 in basic neuronal or cellular functionality, and makes its further investigation, including under pathological conditions, particularly interesting.
All Prgs are most highly expressed in primary cultured neurons, compared to astrocytes and microglia. We showed that all Prgs are also expressed in oligodendrocytes with different dynamic regulations during oligodendrocyte maturation. Especially, Prg3 showed a significant expression decrease in mature oligodendrocytes compared to immature ones. Prg4 and Prg5 on the other hand show strong and stable expression levels. Previous studies showed axonal expression of PRG1, PRG2, and PRG3, with links to axonal growth or branching processes. 1,8,16 These processes might be supported by additional PRG activity in oligodendrocytes. PRG3 has been shown to counteract neurite growth inhibitors, and to reduce myelin-and RhoAmediated axon collapse. 15 Our data show that further investigation of PRG function in oligodendrocytes could be of specific interest, especially for PRG3, which may be involved in the lack of an axonal regeneration capacity in the mammalian central nervous system.
Amino acid sequence analysis showed the highest similarities of PRG4 to PRG3 and PRG5, but the sequence alignment of these PRGs reveals pronounced differences at the N-and C-termini, and at the first ICL and the second ECL, regions that could be essential for protein functionality. We therefore compared morphological changes induced by overexpression of these PRGs. PRG4 did not show the filopodial formation as seen in PRG3 and PRG5, and we did not identify any other morphological changes after PRG4 overexpression. In addition, we did not find a localization of PRG4 to the filopodia plasma membrane, but instead mainly to intracellular membranes and plasma membrane areas of the cell body. This was also not affected when PRG4 was co-transfected with PRG3 or PRG5, whereas it was shown that PRG5 facilitated plasma membrane localization of PRG3 and further increased the induction of membrane protrusions. 13 Also, co-immunoprecipitation results of Yu et al support direct interactions of at least PRG2, PRG3, and PRG5, but did not include PRG4. 13 Our results do not support a direct functional interaction of PRG4 with its close family members PRG3 and PRG5, but this needs further examination. Notably, the lack of specific and reliable antibodies against PRG proteins impedes functional experiments and interaction studies, and complicates the analysis of native PRGs in tissue. The possibility that the protein tag interferes with the sorting or functionality of the original PRG4 protein is unlikely, as we analyzed PRG4 fusion proteins with different N-and C-terminal protein tags. Also, co-immunostaining with the endoplasmic reticulum (ER) marker calnexin revealed that PRG4 fusion constructs are not retained in the endomembrane system (Data not shown), and thus, indicating proper protein folding and transport.
Our results indicate that PRG4 has a different function to PRG3 and PRG5. PRG5 was found to interact with PtdInsPs via its C-terminus, and this is required for the induction of membrane protrusions. 7 We found PRG4 to be localized to intracellular membranes and only much less to the plasma membrane. This suggests a prospective involvement of PRG4 with intracellular lipid metabolism rather than an interaction with plasma membrane lipids. The functional role of the PRG4 C-terminus in protein sorting, and in interactions with phospholipids has to be evaluated.
Interestingly, we found that the site of a protein tag at a PRG can affect its functionality. The N-terminal fusion of eGFP to PRG5 prevented its localization to the plasma membrane and, consequently, the induction of filopodial formation. We did not observe this for PRG3 constructs, where both the N-and C-terminal eGFP had no influence on its observable function. The N-terminus, particularly, differs between PRG3, PRG4, and PRG5, and PRG5 has a very short N-terminus, only five amino acid long. Because of this very small size, it is possible that a tag could disturb its proper insertion into the membrane. Nevertheless, a specific role of each N-terminus cannot be ruled out, and remains to be individually elucidated. Our results show that protein tags can influence protein function or interaction, and that this should be considered when analyzing PRG function.

| CONCLUSION
During mouse brain development, Prgs show specific and highly dynamic expression levels. Expression levels can be directly linked to existing functional data of each PRG, and the expression pattern can be of help in further functional analysis. We demonstrated that the uncharacterized Prg4 is regulated differently, and is highly expressed during mouse brain development, and in all analyzed brain areas. Although PRG4 displays a high degree of similarity to PRG3 and PRG5, it is retained within the secretory pathway and does not induce morphological changes like filopodial outgrowth. The localization of PRG4 to intracellular membranes indicates a possible involvement in intracellular lipid metabolism.

| Animals
To obtain qPCR samples of staged brain development and primary neurons, astrocytes, and microglia, timedpregnant, postnatal, and adult C57BL/6 mice were obtained from the Charité-Universitätsmedizin central animal facility (FEM). Postnatal BALB/c mice were used for oligodendrocyte preparations and were obtained from the central animal facility of the University Medical Center in Rostock. Mice were kept under standard laboratory conditions (12-hour light/dark cycle; 55% ± 15% humidity; 22 C ± 2 C room temperature, and water ad libitum, enriched and grouped) in accordance with German and European guidelines (2010/63/EU) for the use of laboratory animals. Approval of experiments was obtained from the local ethics body of Berlin (LAGeSO: T0108/11) and Mecklenburg-Vorpommern (LALLF). For primary cell culture preparation, the day of the vaginal plug following mating was assigned as embryonic day 0.5 (E0.5). Experiments were performed on E16, E18, and E19 embryos and perinatal pubs (postnatal day 0, P5, P10, P15, P20, and P30). Sample and animal numbers were defined as "n" for the number of independent preparations carried out and "N" for the total amount of animals used.

| RNA extraction, cDNA synthesis, and qRT-PCR
Postnatal and adult mice were sacrificed by cervical dislocation, and early postnatal and embryonic mice by decapitation. Neocortex, hippocampus, cerebellum, and olfactory bulbs of the developmental stages embryonic day (E) 14, E16, E19, and of postnatal days (P) 0, P5, P10, P15, P20, and P30 were dissected out and immediately snap-frozen. For embryonic stages, tissue samples of all embryos of one litter (7-10 embryos; N = 21-30, sex not specified) were pooled, and three independent litters (n = 3) were used. Sexes of embryos were not determined. Postnatal tissue samples were pooled from six mice of both sex and three independent preparations (n = 3) were carried out, resulting in a total of 18 animals for each postnatal developmental stage (N = 18). Primary hippocampal neurons at DIV7, astrocytes and microglia cells at DIV14 to 16, immature oligodendrocytes at DIV3, and mature oligodendrocytes at DIV6 (approximately 3 Â 10 5 cells for each experiment from three independent preparations of three pregnant mice) were scraped in Â1 PBS, centrifuged for 5 minutes at 900g at 4 C. Cell pellets and tissues were homogenized in TRIzol (Thermo Fisher Scientific, Waltham, MA, USA), and total RNA extraction was performed following the manufacturer's instructions. RNA concentrations were determined by spectroscopy using an ultraviolet-visible spectroscopy (UV/Vis) spectrometer (BioSpectrometer basic, Eppendorf, Hamburg, Germany). For cDNA synthesis, 2.5 μg of total RNA was used for reverse-transcription to single-stranded cDNA with the high capacity cDNA reverse transcription kit (Thermo Fisher Scientific) following the manufacturer's protocol. A control reaction was performed without Multi-Scribe reverse transcriptase and the quality of the amplified cDNA (with and without MultiScribe reverse transcriptase) was tested using beta-actin (Actb) PCR.
Quantitative-RT-PCR was performed using the TaqMan Fast Universal PCR Master Mix (2X), No AmpErase UNG (Thermo Fisher Scientific), and 96-well optical reaction plates from Applied Biosystems (Thermo Fisher Scientific), or hard-shell 96-Well PCR plates from Bio-Rad Laboratories (Hercules, CA, USA). Reactions were prepared according to the manufacturer's protocols. The TaqMan expression assays used are listed in Table 3. Reactions contained a TaqMan probe (5 μM), a forward primer (18 μM), and a reverse primer (18 μM). TaqMan probes were tagged at the 5 0 -end with the reporter dye FAM (6-carboxyfluorescein) and with a minor groove binder (MGB) attached to a nonfluorescent quencher at the 3 0 -end. QRT-PCR reactions were carried out with the ABI PRISM 7700 sequence detection system (Thermo Fisher Scientific), the ViiA 7 real-time pcr system (Thermo Fisher Scientific), or the CFX96 touch real-time PCR detection system (Bio-Rad Laboratories) using the following cycling parameters: 95 C for 20 seconds, 95 C for 1 second and 60 C for 20 seconds, for 45 cycles. Expression data of three independent preparations with duplicates of each reaction were calculated using the ΔCt method, with normalization to Gapdh and Actb as housekeeping genes. 22 Results with both housekeeping genes revealed similar expression patterns and therefore only data normalized to Gapdh is shown in the results. Primary cultured cell samples were validated by qRT-PCR with the following cell-type-specific marker genes: neuroblast-specific class III β-tubulin (Tuj1) for neurons, glial fibrillary acidic protein (Gfap) for astrocytes, ionized calcium-binding adaptor molecule 1 (Iba1) for microglia, neuron-glial antigen 2 (Ng2) for immature oligodendrocytes, and myelin basic protein (Mbp) for mature oligodendrocytes. Oligodendrocyte maturation was verified by statistical analysis of Ng2 and Mbp expression using an unpaired, two-tailed t-test (Ng2: *P = .0407; Mbp: **P = .0003).

| Sequence alignment
The alignment of amino acid sequences for the analysis of sequence similarities was accomplished with the basic local alignment online search tool provided by BLAST. Figures of aligned sequences were generated with CLC Sequence Viewer 8 software (Qiagen, Hilden, Germany).

| Primary cell cultures
For primary neuron culture, hippocampi of 7 to 10 E18 (±0.5 days) mouse embryos from one pregnant mouse were dissected and cultures prepared as previously described. 23 Neurons were plated onto poly-l-lysine coated plastic ware at a density of 2.1 Â 10 4 cells/cm 2 in minimal essential medium (Thermo Fisher Scientific) supplemented with 0.6% glucose, 10% horse serum, and 100 U/mL penicillin/streptomycin (PAN-Biotech, Aidenbach, Germany). Media were changed after 3 to 4 hours to neurobasal A media supplemented with 2% B27, 0.5 mM glutamine (all from Gibco, Thermo Fisher Scientific), and 100 U/mL penicillin/streptomycin. Three independent preparations were carried out (n = 3; N = 21-30). Sample and animal numbers were defined as "n" for the number of independent preparations carried out and "N" for the total amount of animals used.
For preparation of murine astrocytes and microglia, three to four P0 to P2 mouse pups from the same litter were sacrificed, their neocortex and cerebellum dissected and washed in Dulbecco's Modified Eagle's Medium (DMEM, 4.5 g/L glucose, pyruvate, Thermo Fisher Scientific) containing 10% fetal bovine serum (FCS), 200 mM L-glutamine, and 100 U/mL penicillin/streptomycin. Sexes of pubs were not determined. Astrocyte and microglia preparation and cultivation were performed as previously described 24 and three independent preparations were carried out (n = 3; N = 9 to 12). Cells were harvested after 14 to 16 days in vitro (DIV).
Oligodendrocyte cultures were prepared from cerebral hemispheres of three P5 mouse pups from the same litter as described in Suckau et al. 24 Sexes of pubs were not determined. After seeding of purified oligodendrocytes, the medium was not further changed and immature oligodendrocytes were harvested after 3 days, mature ones after 6 days. Three independent preparations for immature and three independent preparations for mature oligodendrocytes were carried out (n = 3; N = 9).

|
For quantification of filopodia formation, HEK293H cells were co-transfected with the indicated PRG construct and a membrane-targeting CFP (CFP-MEM) for visualizing the overall cell morphology to allow filopodia counting. All CFP marked membrane protrusions ≥2 μm of one cell were counted. Data were collected from n = 3 independent transfection experiments with at least 10 cells of each experiment and with N = number of total cells counted.

| Immunocytochemistry
After transfection, HEK293H cells were washed twice with Â1 PBS, and fixed with 4% paraformaldehyde and 15% sucrose in Â1 PBS for 20 minutes at room temperature (RT). Cell membranes were permeabilized with 0.1% Triton-X100 and 0.1% sodium citrate in Â1 PBS for 3 minutes at RT, coverslips were then washed three times with Â1 PBS, and nonspecific antibody binding was blocked with 10% normal goat serum (Vector Laboratories, Burlingame, CA, USA) in Â1 PBS for 1 hour at RT. Primary antibody incubation was conducted in 5% normal goat serum in Â1 PBS overnight at 4 C, followed by three washing steps with Â1 PBS, and secondary antibody and 0.5 μg/μL DAPI (Carl Roth, Karlsruhe, Germany) incubation in the same solution as primary antibodies for 2 hours at RT. Coverslips were again washed three times with Â1 PBS and mounted on microscope slides using Mowiol/DABCO (Carl Roth). The following primary and secondary antibodies and concentrations were used: mouse-anti-FLAG (Merck, Darmstadt, Germany, F3165) 1:1500, rabbit-anti-GFP (Abcam, Cambridge, UK, ab6556) 1:2500, goat-antimouse-Alexa568 (Thermo Fisher Scientific, A11004), goat-anti-rabbit-Alexa488 (Thermo Fisher Scientific, A11008).

| Microscopy
Confocal images of transfected HEK293H cells were acquired as z-stacks with a Leica SP8 upright laser microscope equipped with Â40 (oil-immersion, 1.3 NA) and Â63 objectives (oil-immersion, 1.4 NA), using sequential scanning with the 488 nm line of an argon-ion laser and the 552 nm line of a helium laser. Images are presented as maximum projections of z-stacks. Orthogonal views, background correction, and brightness and contrast adjustments were performed using either LasX software (Leica Microsystems, Wetzlar, Germany) or ImageJ (NIH, Bethesda, MD, USA).

| Statistical analysis
Statistical evaluation of the qRT-PCR results of developmental stages in different brain areas ( Figure 1) and primary cultured cells (Figure 2A) was performed using a one-way analysis of variance (ANOVA) followed by a Bonferroni's multiple comparisons test. Developmental stages E14, P0, and P30 were chosen for statistical evaluation of embryonic, birth, and adult expression differences. qRT-PCR results of primary oligodendrocyte maturation were analyzed with GraphPad Prism 7 by an unpaired one-tailed t-test. Data are shown as mean ± SD and were considered significant for P ≤ .05 (*P ≤ .05, ***P ≤ .01). Normal distribution of qRT-PCR results was assessed by the Shapiro-Wilk test for small samples sizes (P > .05). Quantification of filopodia formation ( Figure 4C) is visualized in a boxplot graph including all data points and was statistically analyzed by using a Kruskal-Wallis test followed by a Dunn's multiple comparison test. All statistical analyses and graphs were created using GraphPad Prism 7 (GraphPad Software, San Diego, CA, USA).
Universitäten to Anja U. Bräuer and Mark S. Hipp. Open access funding enabled and organized by Projekt DEAL.