Preterm birth disrupts cerebellar development by affecting granule cell proliferation program and Bergmann glia

Abstract

Preterm birth is a leading cause of long-term motor and cognitive deficits. Clinical studies suggest that some of these deficits result from disruption of cerebellar development, but the mechanisms that mediate cerebellar abnormalities in preterm infants are largely unknown. Furthermore, it remains unclear whether preterm birth and precocious exposure to the ex-utero environment directly disrupt cerebellar development or indirectly by increasing the probability of cerebellar injury, including that resulting from clinical interventions and protocols associated with the care of preterm infants. In this study, we analyzed the cerebellum of preterm pigs delivered via c-section at 91% term and raised for 10 days, until term-equivalent age. The pigs did not receive any treatments known or suspected to affect cerebellar development and had no evidence of brain damage. Term pigs sacrificed at birth were used as controls. Immunohistochemical analysis revealed that preterm birth did not affect either size or numbers of Purkinje cells or molecular layer interneurons at term-equivalent age. The number of granule cell precursors and Bergmann glial fibers, however, were reduced in preterm pigs. Preterm pigs had reduced proliferation but not differentiation of granule cells. qRT-PCR analysis of laser capture microdissected external granule cell layer showed that preterm pigs had a reduced expression of Ccnd1 (Cyclin D1), Ccnb1 (Cyclin B1), granule cell master regulatory transcription factor Atoh1, and signaling molecule Jag1. In vitro rescue experiments identified Jag1 as a central granule cell gene affected by preterm birth. Thus, preterm birth and precocious exposure to the ex-utero environment disrupt cerebellum by modulating expression of key cerebellar developmental genes, predominantly affecting development of granule precursors and Bergmann glia.

Keywords: Atoh1; Cerebellum; Development; Granule cell precursors; Jag1; Preterm birth; Proliferation.

Figure 1. Reduced number of granule cell precursors but not Purkinje cells in the cerebellum of preterm pigs

(A) Diagram of midsagittal section of pig cerebellar vermis. Cells were quantified in anterior (lobe III) and posterior (lobe VIII) cerebellum (in the regions depicted by red boxes), in naturally born term controls (Nat. born term), term controls delivered via c-section (C-sect. term) and preterm pigs. (B–G, J–O) Sagittal sections of medial cerebellar vermis stained with indicated antibodies. B–D and J–L show anterior cerebellar vermis (the region corresponding to the left box in diagram “A”). E–G and M–O show posterior cerebellar vermis (the region corresponding to the right box in diagram “A”). (B–I) Quantification of Pax6+ granule cell precursors (arrowheads in B–G) per 100 µm-long segments of the EGL. In preterm pigs, the number of Pax6+ cells in the EGL was significantly reduced in both anterior and posterior cerebellum relative to term controls. **p<0.01, *p<0.05. n=4 naturally born term pigs, n=4 c-section born term pigs and n=6 preterm pigs. (J–R) Quantification of Purkinje cells (arrowheads in J–O) per 500 µm of the Purkinje cell layer (P, Q) and diameter of Purkinje cells (R). In preterm pigs, neither density nor diameter of Purkinje cells was significantly changed relative to term controls (P–R). NS – not significant (p>0.05), n=4 naturally born term pigs. n=4 c-section born term pigs and n=5 preterm pigs. All data are mean ± sd. Scale bars: 100 µm (B–G); 250 µm (J–O).

Figure 2. Reduced number of Bergmann glial fibers but not molecular layer interneurons in the cerebellum of preterm pigs

(B–D, F–H, J–L, N–P) Sagittal sections of medial vermis stained with indicated antibodies. Panels correspond to the box in diagram A and show entire thickness of the molecular layer (ML) (outer cerebellar surface is up). (A) EGL – external granule cell layer, PC –Purkinje cell layer. Cells were quantified in anterior (lobe III) and posterior (lobe VIII) cerebellum (in the regions depicted by red boxes in Fig. 1A), in naturally born term controls (Nat. born term), term controls delivered via c-section (C-sect. term) and preterm pigs. (B–I) Quantification of Pax2+ ML interneurons (arrowheads in B–D, F–H) per 300 µm-long segments of the ML did not reveal a statistically significant difference between term and preterm pigs in either anterior or posterior cerebellum (E, I). NS – not significant (p>0.05). n=4 pigs per group. (J–Q) Decreased number of GFAP+ Bergmann glial fibers (arrowheads in J–L, N–P) in preterm pigs in both anterior (M) and posterior cerebellum (Q) relative to term controls. ***p<0.001, **p<0.01, *p<0.05. n=4 naturally born term pigs, n=4 c-section born term pigs and n=5 preterm pigs. All data are mean ± sd. Scale bars: 100 µm (B–D, F–H), 200 µm (J–L, N–P).

Figure 3. Reduced number of Ki67+ cells but normal differentiation in the EGL of preterm pigs

(A1–C2, F1–H2, K1–M2, P1–R2) Sagittal sections of medial vermis stained with DAPI and indicated antibodies. Cells were quantified in anterior (lobe III) and posterior (lobe VIII) cerebellum (in the regions depicted by red boxes in Fig. 1A), in naturally born term controls (Nat. born term), term controls delivered via c-section (C-sect. term) and preterm pigs. White solid line marks outer surface of the cerebellum, white dashed line shows inner boundary of the EGL. (A1–E) Decreased number of Ki67+ cells (arrowheads in A1–C2) per 100 µm-long segment of the EGL was detected in both anterior and posterior preterm cerebellum relative to term controls. **p<0.01, *p<0.05. (F1–J) Quantification of NeuN+ differentiating cells (arrowheads in F1–H2) per 100 µm-long segment of the EGL did not reveal a statistically significant difference between term and preterm pigs in either anterior (I) or posterior (J) cerebellum. NS – not significant (p>0.05). (K1–T) Thickness of the inner Dcx+ EGL (vertical bars in panels K1–M2) and Tag1+ EGL (vertical bars in panels P1–R2) was not different between term and preterm pigs. NS - not significant (p>0.05). All quantifications were performed in n=4 pigs per group. All data are mean ± sd. Scale bar: 30 µm.

Figure 4. Reduced proliferation in the EGL of preterm pigs

Sagittal sections of medial vermis from naturally born term (A, D) or preterm (B, E) pigs stained with anti-pH3 (phospho-Histone H3) antibody or TUNEL. Pink solid line marks outer surface of the cerebellum, pink dashed line shows inner boundary of the EGL. (A–C) Quantification of mitotic (pH3+) cells (arrows in A and B) per mm of the EGL. The number of pH3+ cells was reduced in the EGL of preterm pigs compared to term controls (**p<0.01), reveling reduced proliferation. (D–F) The number of TUNEL+ cells (D, E, arrowheads) per mm of the EGL was not significantly different between term and preterm pigs, indicating that preterm birth does not affect apoptosis in the EGL. NS – not significant (p>0.05). All quantifications were performed in n=4 pigs per group. All data are mean ± sd. Scale bar: 300 µm.

Figure 5. Reduced number of NeuN+ neurons in the IGL of preterm pigs

Quantification of NeuN+ neurons in the IGL of lobes III and VIII in preterm pigs relative to naturally born term controls. Panels D, E and G, H show NeuN-stained IGL that corresponds to the regions boxed in adjacent diagrams B and C. (B, C) EGL-external granule cell layer, PC-Purkinje cell layer, IGL – internal granule cell layer. Diagrams B and C correspond to the regions depicted by red boxes in panel A. (F, I) There was reduced number of NeuN+ neurons in the IGL in both lobe III and lobe VIII in preterm pigs relative to naturally born term controls. *p<0.05. All quantifications were performed in n=4 pigs per group. All data are mean ± sd. Scale bar: 100 µm.

Figure 6. Reduced expression of genes encoding general regulators of the cell cycle - cyclins D1 and B1 in the EGL of preterm pigs

(A) Diagram of sagittal section of pig cerebellar vermis. From each section, entire EGL (blue area in the diagram) was isolated by LCM and assayed by qRT-PCR. (B–F) Expression of Ccnd1 (encoding Cyclin D1), Ccnb1 (encoding Cyclin B1), Ccna2 (encoding Cyclin A2), Cdkn1b (encoding p27Kip1) and Cdkn2c (encoding p18Ink4c) in the EGL based on qRT-PCR. Ccnd1 and Ccnb1 were significantly downregulated in the EGL of preterm pigs (***p<0.001, **p<0.01, *p<0.05). Expression of Ccna2, Cdkn1b and Cdkn2c was not different between term and preterm EGL (p>0.05, NS). All quantifications were performed in n=4 naturally born term pigs, n=4 c-section born term pigs and n=6 preterm pigs. All data are mean ± sd.

Figure 7. Reduced expression of tissue specific regulators of granule cell development Atoh1 and Jag1 in the EGL of preterm pigs

qRT-PCR analysis of Atoh1, Pax6, Zic1, Zic4, Jag1, Patch1, Gli1 and Smo expression in laser capture microdissected EGL (blue area in panel A). (B–F) Atoh1 and Jag1 were significantly downregulated in the EGL of preterm pigs (***p<0.001, **p<0.01, *p<0.05). Expression of Pax6, Zic1, Zic4, Gli1, Smo and Patch1 was not different between term and preterm EGL (p>0.05, NS). All quantifications were performed in n=4 naturally born term pigs, n=4 c-section born term pigs and n=6 preterm pigs. All data are mean ± sd.

Figure 8. Exogenous Jag1 rescues proliferation defects of preterm granule cell precursors in cerebellar slices

Cerebellar slices from naturally born term (A, D) or preterm (B, C) pigs cultured for 24 hrs with (C, D) or without (A, B) Jag1 and immunostained against pH3 to assess proliferation. Upper pink line marks outer surface of the cerebellum, lower dashed pink line shows inner boundary of the EGL. Arrows point to pH3+ cells in the EGL. (E) Quantification of pH3+ cells per mm of the EGL. When cultured without Jag1, preterm slices had fewer pH3+ cells in the EGL than those from term cerebella (***p<0.001). Preterm slices cultured with Jag1 had more pH3+ cells in the EGL than preterm slices cultured without Jag1 (**p<0.01). Preterm slices cultured with Jag1 had a similar number of pH3+ cells as term slices cultured without Jag1 (NS, p>0.05), showing rescue of granule cell proliferation in preterm EGL by Jag1. All quantifications were performed in n=4 slices per experimental condition. Each slice was taken from a different cerebellum. All data are mean ± sd. Scale bar: 300 µm.