Selective neuronal expression of progranulin is sufficient to provide neuroprotective and anti-inflammatory effects after traumatic brain injury

Abstract

Progranulin (PGRN), which is produced in neurons and microglia, is a neurotrophic and anti-inflammatory glycoprotein. Human loss-of-function mutations cause frontotemporal dementia, and PGRN knockout (KO) mice are a model for dementia. In addition, PGRN KO mice exhibit severe phenotypes in models of traumatic or ischemic central nervous system (CNS) disorders, including traumatic brain injury (TBI). It is unknown whether restoration of progranulin expression in neurons (and not in microglia) might be sufficient to prevent excessive TBI-evoked brain damage. To address this question, we generated mice with Nestin-Cre-driven murine PGRN expression in a PGRN KO line (PGRN-KO^NestinGrn) to rescue PGRN in neurons. PGRN expression analysis in primary CNS cell cultures from naïve mice and in (non-) injured brain tissue from PGRN-KO^NestinGrn revealed expression of PGRN in neurons but not in microglia. After experimental TBI, examination of the structural brain damage at 5 days post-injury (dpi) showed that the TBI-induced loss of brain tissue and hippocampal neurons was exacerbated in PGRN-KO^Grnflfl mice (PGRN knockout with the mGrn fl-STOP-fl allele, Cre-negative), as expected, whereas the tissue damage in PGRN-KO^NestinGrn mice was similar to that in PGRN-WT mice. Analysis of CD68⁺ immunofluorescent microglia and Cd68 mRNA expression showed that excessive microglial activation was rescued in PGRN-KO^NestinGrn mice, and the correlation of brain injury with Cd68 expression suggested that Cd68 was a surrogate marker for excessive brain injury caused by PGRN deficiency. The results show that restoring neuronal PGRN expression was sufficient to rescue the exacerbated neuropathology of TBI caused by PGRN deficiency, even in the absence of microglial PGRN. Hence, endogenous microglial PGRN expression was not essential for the neuroprotective or anti-inflammatory effects of PGRN after TBI in this study.

Keywords: CD68; Microglia; Neuroinflammation; Neuropathology; Neuroprotection; Progranulin; Therapy; Traumatic brain injury.

Fig. 1

Progranulin expression in PGRN-KO^NestinGrn mice. (A, B) Grn mRNA expression in cell cultures from adult brains enriched for non-microglial neural cells (neurons and astrocytes) and microglia. The data are presented as box/scatter plots. The line is the median, the box shows the interquartile range, the whiskers show the minimum to maximum, and the scatters show the individual results of 7–9 mice per genotype, measured in duplicate. (C) Images of the contralesional hippocampal GCL (Bregma − 1.86 mm) showing anti-PGRN immunostaining and nuclear counterstaining by DAPI in PGRN-WT, PGRN-KO^NestinGrn, and PGRN-KO^Grnflfl mice. Genotype-dependent PGRN expression in cells with microglial or neuronal morphology indicates PRGN expression in GCL neurons but not in microglia in PGRN-KO^NestinGrn mice. No specific signal was detected in PGRN-KO^Grnflfl mice. (D) Triple fluorescence staining of the ipsilesional cortex (Bregma − 1.86 mm) at 5 dpi using anti-PGRN/anti-CD68/DAPI revealed PGRN expression in CD68⁺ microglia in PGRN-WT mice but not in PGRN-KO^NestinGrn or PGRN-KO^Grnflfl mice. Arrows point to cells shown at higher magnification

Fig. 2

Exacerbated structural brain damage in PGRN-KO^Grnflfl mice is rescued in PGRN-KO^NestinGrn mice. (A) Representative images of cresyl violet-stained coronal brain sections from PGRN-WT, PGRN-KO^NestinGrn, and PGRN-KO^Grnflfl mice at 5 dpi (Bregma − 1.86 mm). (B) Relative brain tissue loss (% of ipsilesional hemisphere) calculated from 16 consecutive sections (Bregma + 3.14 mm to − 4.36 mm). PGRN-WT and PGRN-KO^NestinGrn mice exhibit attenuated brain tissue loss compared to PGRN-KO^Grnflfl mice. (C) Images of coronal brain sections (Bregma − 1.86 mm) showing anti-NeuN immunostaining in the hippocampal dentate gyrus of the contra- and ipsilesional hemispheres at 5 dpi. Arrows indicate exacerbated loss of GCL neurons in the ipisilesional suprapyramidal blade of PGRN-KO^Grnflfl mice. (D) The number of NeuN⁺ neurons in the GCL was higher in PGRN-WT and PGRN-KO^NestinGrn mice than in PGRN-KO^Grnflfl mice. The data points represent individual mice, PGRN-WT (n = 12), PGRN-KO^NestinGrn (n = 8) and PGRN-KO^Grnflfl (n = 6), and the data are expressed as the mean ± SEM. One-way ANOVA with Holm–Šidák post hoc correction, *p < 0.05**p < 0.01, ***p < 0.001, ns = not significant

Fig. 3

TBI-induced excessive Cd68 gene expression in PGRN-KO^Grnflfl mice is attenuated in PGRN-KO^NestinGrn mice. (A-G) Gene expression analysis of inflammation-associated markers in ipsi- or contralesional brain tissues (Bregma + 0.64 mm to − 2.86 mm) was performed via RT‒qPCR. (A) Grn expression is highest in PGRN-WT mice and is reduced in PGRN-KO^NestinGrn and PGRN-KO^Grnflfl mice. Grn expression in ipsilesional brain tissue was mildly greater in PGRN-KO^NestinGrn mice than in PGRN-KO^Grnflfl mice but was substantially lower than that in WT mice because PGRN upregulation after TBI mainly occurs in microglia. (B-G) Column charts showing the mRNA expression of inflammation-associated markers (Aif1, Gfap, Cd68, Lyz2, Spp1, and Tnf). (D) Augmented ipsilesional Cd68 gene expression in PGRN-KO^Grnflfl mice was partially rescued in PGRN-KO^NestinGrn mice. Two outliers were identified by Rout’s test and excluded (F). The data points represent individual mice, PGRN-WT (n = 12), PGRN-KO^NestinGrn (n = 8) and PGRN-KO^Grnflfl (n = 6), and the data are expressed as the mean ± SEM. One-way ANOVA with Holm–Šidák post hoc correction, *p < 0.05, **p < 0.01, ***p < 0.001, **** p < 0.0001, ns = not significant

Fig. 4

Cd68 gene expression is a surrogate marker of microglial PGRN deficiency and associated brain damage. Linear regression analyses to assess the relationship between Cd68 expression and ipsilesional brain tissue loss in PGRN-WT (n = 12), PGRN-KO^NestinGrn (n = 8), and PGRN-KO^Grnflfl (n = 6) mice. Scatter plots with regression lines, 95% confidence intervals, correlation coefficients (r²), and p values are shown. The data points represent individual mice

Fig. 5

Iba1⁺ microglial infiltration of the injured brain in PGRN-KO^Grnflfl mice is reduced in PGRN-KO^NestinGrn mice. (A) Scheme illustrating the positions of the imaged brain regions. (B) Double immunostaining of the ipsi- and contralesional cortex at 5 dpi (Bregma − 1.86 mm) using anti-Iba1 and anti-GFAP antibodies showing TBI-evoked activation of microglia and astrocytes. (C) Higher magnification images of the boxed regions. (D, E) Column plots showing Iba1⁺ and GFAP⁺ counts in the ipsi- and contralesional cortices. The number of ipsilesional Iba1⁺ microglia was lower in PGRN-KO^NestinGrn mice than in PGRN-KO^Grnflfl mice, whereas the number of contralesional GFAP⁺ astrocytes was greater in PGRN-KO^NestinGrn and PGRN-KO^Grnflfl mice than in PGRN-WT mice. (F, G) Column plots showing the total anti-Iba1⁺ or anti- GFAP⁺ immunostained areas in the ipsi- and contralesional cortices. The area of ipsilesional Iba1⁺ immunostaining was smaller in PGRN-WT mice and PGRN-KO^NestinGrn mice than in PGRN-KO^Grnflfl mice, whereas the immunostaining area of GFAP⁺ astrocytes was not different between genotypes. The data are expressed as the mean ± SEM, and the values are shown for individual mice, PGRN-WT (n = 12), PGRN-KO^NestinGrn (n = 8) and PGRN-KO^Grnflfl (n = 6). One-way ANOVA (D, contra, E, ipsi), Brown-Forsythe ANOVA test (D, ipsi) and Kruskal‒Wallis test (E, contra) and post hoc Holm–Šidák, Dunnett T3 or Dunn’s corrections were used to calculate p values (*p < 0.05, **p < 0.01, ns = not significant)

Fig. 6

Excessive CD68⁺ microglial infiltration of the injured brain in PGRN-KO^Grnflfl mice is reduced in PGRN-KO^NestinGrn mice. (A) Triple-fluorescence staining of ipsilesional cortex at 5 dpi (Bregma − 1.86 mm) with anti-CD68/BODIPY/DAPI showed fewer CD68⁺ microglia in PGRN-WT and PGRN-KO^NestinGrn than in PGRN-KO^Grnflfl mice and partial overlap of the BODIPY signal with that in CD68⁺ microglia. (B) Column plots showing reduced area occupancy by CD68⁺ microglia in PGRN-WT and PGRN-KO^NestinGrn mice compared to PGRN-KO^Grnflfl mice. (C) Column plots showing reduced average size of CD68⁺ microglia in PGRN-WT mice compared to PGRN-KO^Grnflfl mice. Differences between PGRN-KO^NestinGrn mice and PGRN-KO^Grnflfl mice were statistically not significant (p = 0.07). (D) Column plots showing that the percentage of CD68⁺ microglia colabeled with BODIPY had the highest mean percentage in PGRN-KO^Grnflfl mice, a reduced mean percentage in PGRN-KO^NestinGrn and a significant reduction in PGRN-WT mice. The data are expressed as the mean ± SEM, and the values from individual mice are shown, PGRN-WT (n = 12), PGRN-KO^NestinGrn (n = 8) and PGRN-KO^Grnflfl (n = 6). One-way ANOVA and post hoc Holm–Šidák corrections were used to calculate p values (*p < 0.05, **p < 0.01, ns = not significant)