Antibody-assisted selective isolation of Purkinje cell nuclei from mouse cerebellar tissue

Abstract

We developed a method that utilizes fluorescent labeling of nuclear envelopes alongside cytometry sorting for the selective isolation of Purkinje cell (PC) nuclei. Beginning with SUN1 reporter mice, we GFP-tagged envelopes to confirm that PC nuclei could be accurately separated from other cell types. We then developed an antibody-based protocol to make PC nuclear isolation more robust and adaptable to cerebellar tissues of any genotypic background. Immunofluorescent labeling of the nuclear membrane protein RanBP2 enabled the isolation of PC nuclei from C57BL/6 cerebellum. By analyzing the expression of PC markers, nuclear size, and nucleoli number, we confirmed that our method delivers a pure fraction of PC nuclei. To demonstrate its applicability, we isolated PC nuclei from spinocerebellar ataxia type 7 (SCA7) mice and identified transcriptional changes in known and new disease-associated genes. Access to pure PC nuclei offers insights into PC biology and pathology, including the nature of selective neuronal vulnerability.

Keywords: CP: Neuroscience; FANS; Purkinje cells; RanBP2; SCA7; cerebellum; isolation; neurodegeneration; nucleus; phosphodiesterase; spinocerebellar ataxia.

Graphical abstract

Figure 1

SUN1-GFP genetic tagging of nuclear envelopes, followed by flow cytometry analysis, reveals a distinctive cluster of PC nuclei (A) We cryosectioned cerebella from Sun1/sfGFP⁺, Pcp2-Cre⁻ and Sun1/sfGFP⁺, Pcp2-Cre⁺ mice and immunostained them for the Myc tag (green), which is fused to the GFP protein, calbindin (red), and Hoechst (blue). The arrows point to the PC nuclei. Scale bar: 50 μm. (B) We loaded nuclei isolated from Sun1/sfGFP⁺, Pcp2-Cre⁺ cerebella into a cell sorter, and used SSC-A vs. Hoechst-A to identify the population of single nuclei (called P1). (C) We examined P1 nuclei for SSC-A and GFP-A and discovered several subpopulations, one of which was distinguished by strong SSC and GFP intensity (called P2). (D) We performed post sort analysis to ensure P2 sorting purity and found 89.2% of P2 nuclei among all nuclei in sorted samples; n = 3 independent purifications. Hoechst-A, Hoechst channel peak area; P1 (population 1), all singlet nuclei; P2 (population 2), PC nuclei.

Figure 2

Morphological and molecular analysis of the P2 subset from Sun1/sfGFP cerebella confirms its PC identity (A) We used qRT-PCR to compare the expression of the most specific PC marker genes in 4,000 P1 and P2 nuclei sorted from Sun1/sfGFP⁺, Pcp2-Cre⁺ cerebella; n = 3 for each group. (B) qRT-PCR quantification of Gabra6, a marker of cerebellar granule neurons; n = 3 for each group. (C) We imaged P1 and P2 populations by reloading previously sorted P1 and P2 nuclei into the ImageStream cytometer. Representative examples are shown; scale bar: 10 μm. (D) Nuclear size distribution in P1 and P2 populations represented as a Hoechst footprint in the ImageStream analysis; n = 227 and 237 for P1 and P2, respectively. (E) We stained nuclei isolated from Sun1/sfGFP⁺, Pcp2-Cre⁺ cerebella with nucleolar dye NBR, counterstained them with Hoechst, and visualized both P1 and P2 populations with the ImageStream cytometer. Representative examples are shown. Scale bar: 10 μm. Quantification of the number of nucleoli per cell is shown below. Data are represented as mean ± SEM.

Figure 3

Flow cytometry analysis of RanBP2-stained cerebellar nuclei reveals a distinct subpopulation of PC nuclei (A) We analyzed nuclear extracts from mouse cerebellum with a cell sorter, and identified single nuclei (dubbed P1i) using SSC-A vs. Hoechst-A. (B) Examining the P1i singlets for SSC-A and AlexaFluor488-A revealed distinct subpopulations, including one prominent P2i cluster with strong side scattering and Alexa Fluor 488 intensity. (C) We reexamined sorted P2i nuclei to ensure purity. Three independent preparations revealed that 94.2% of sorted nuclei fell back into the P2i gate. AlexaFluor488-A, Alexa Fluor 488 channel peak area; Hoechst-A, Hoechst channel peak area; P1i (population 1, immunostained), all singlet nuclei; P2 (population 2, immunostained), PC nuclei.

Figure 4

Morphological and molecular characterisation of RanBP2-stained P2i subset confirms its PC identity (A) We sorted 4,000 P1i and P2i nuclei and compared the expression of the most specific PC marker genes with qRT-PCR; n = 3 for each group. (B) qRT-PCR quantification of Gabra6, a marker of cerebellar granule neurons; n = 3 for each group. (C) We visualized P1i and P2i previously sorted nuclei by reloading them into the ImageStream cytometer. Representative examples are shown. Scale bar: 10 μm. (D) Nuclear size distribution in P1i and P2i populations represented as a Hoechst footprint in the ImageStream analysis. n = 320 and 197 for P1i and P2i, respectively. (E) We stained isolated nuclei with the nucleolar dye NBR, counterstained them with Hoechst, and visualized both P1i and P2i populations with the ImageStream cytometer. Scale bar: 10 μm. Representative examples are shown. Quantification of the number of nucleoli per cell is shown below. Data are represented as mean ± SEM.

Figure 5

qPCR quantification in SCA7 PC nuclei unveils transcriptional changes in both known and new disease-associated genes (A) qPCR quantification of selected disease-linked genes in PC nuclei isolated from SCA7 and WT mouse cerebella. n = 4 cerebella for each group; 4,000 nuclei were sorted per animal. (B) qPCR quantification of Pde1c, Pde4d, Pde9a, and Pde10a PC nuclei isolated from SCA7 and WT mouse cerebella. n = 7 cerebella for each group; 4,000 nuclei were sorted per animal. Data are represented as mean ± SEM.