Single cell morphology distinguishes genotype and drug effect in Hereditary Spastic Paraplegia

Abstract

A central need for neurodegenerative diseases is to find curative drugs for the many clinical subtypes, the causative gene for most cases being unknown. This requires the classification of disease cases at the genetic and cellular level, an understanding of disease aetiology in the subtypes and the development of phenotypic assays for high throughput screening of large compound libraries. Herein we describe a method that facilitates these requirements based on cell morphology that is being increasingly used as a readout defining cell state. In patient-derived fibroblasts we quantified 124 morphological features in 100,000 cells from 15 people with two genotypes (SPAST and SPG7) of Hereditary Spastic Paraplegia (HSP) and matched controls. Using machine learning analysis, we distinguished between each genotype and separated them from controls. Cell morphologies changed with treatment with noscapine, a tubulin-binding drug, in a genotype-dependent manner, revealing a novel effect on one of the genotypes (SPG7). These findings demonstrate a method for morphological profiling in fibroblasts, an accessible non-neural cell, to classify and distinguish between clinical subtypes of neurodegenerative diseases, for drug discovery, and potentially for biomarkers of disease severity and progression.

Figure 1

Overview of the workflow of the cellular morphology phenotypic biomarker development approach.

Figure 2

Conventional vs advanced image analysis approaches. We compared the two approaches using images of cells treated with mitochondria respirator chain inhibitors Oligomycin and AntimycinA. Images of cells labelled to identify cell nucleus (A, B), mitochondria (E, F), acetylated α-tubulin (I, J), label-free phase contrast cell images (M, N) and the combination of all markers (Q, R) were analysed. Conventional analysis identified differences in mitochondria morphology between the two cell groups (1.25 fold difference) (G) and did not identify any differences with the other markers (C, K, O). In contrast, the advanced image analysis approach identified an amplified mitochondria morphology difference of 9.80-fold between the two groups (H) and identified significant morphological differences in cell components: nucleus (D), acetylated α-tubulin (L), label-free phase contrast images (P) and the combination of all markers (S). Mean values were compared using students t-test.

Figure 3

Logistic regression analysis of SPAST patient vs control samples. Histogram of single cell logistic regression probability scores is presented for 71,942 cells from 10 SPAST patient and 9 control individuals for multiple cell components: acetylated α-tubulin (A), mitochondria (C), nucleus (E), cell phase contrast (G) and combined markers (I). Dotted lines indicate mean probability score values. The mean logistic regression probability score showing all the individual patient and control cell lines is presented for acetylated α-tubulin (B), mitochondria (D), nucleus (F), the cell phase contrast image (H) and combined markers (J). Mean values were compared using students t-test.

Figure 4

Logistic regression analysis of SPG7 patient vs control samples. Histogram of single cell logistic regression probability scores is presented for 32,268 cells from 5 SPG7 patient and 5 control individuals for multiple cell components: acetylated α-tubulin (A), mitochondria (C), nucleus (E), cell phase contrast (G) and combined markers (I). Dotted lines indicate mean probability score values. The mean logistic regression probability score showing all the individual patient and control cell lines is presented for acetylated α-tubulin (B), mitochondria (D), nucleus (F), the cell phase contrast image (H) and combined markers (J). Mean values were compared using students t-test.

Figure 5

Logistic regression analysis of SPAST patient vs SPG7 patient samples. Histogram of single cell logistic regression probability scores is presented for 26,525 cells from 5 SPG7 patient and 5 control individuals for multiple cell components: acetylated α-tubulin (A), mitochondria (C), nucleus (E), cell phase contrast (G) and combined markers (I). Dotted lines indicate mean probability score values. The mean logistic regression probability score showing all the individual patient and control cell lines is presented for acetylated α-tubulin (B), mitochondria (D), nucleus (F), the cell (phase contrast image) (H) and combined markers (J). Mean values were compared using students t-test.

Figure 6

ROC curve analysis. ROC curve analysis of (A) SPAST vs control samples (B) SPG7 vs control samples (C) SPAST vs SPG7 samples. The AUC for all markers for SPG7 vs control samples is 1.00 and hence all lines overlap and cannot be distinguished. AUC is mentioned for all markers. AaT Acetylated α-tubulin, Mito Mitochondria, Nucleus, Cell phase contrast and Combined: All markers combined.

Figure 7

Logistic regression analysis of noscapine-treated SPAST patient cells. We compared noscapine treated SPAST patient cells (33,764 cells from 10 individuals) to untreated control and SPAST patient cells presented above in the SPAST vs controls section (Fig. 3) for comparison. Noscapine-treated SPAST patient cells had logistic regression probability values comparable to controls for all markers: acetylated α-tubulin (A), mitochondria (B), nucleus (C), phase contrast (D) and all markers combined (E). Mean values were compared using one way ANOVA.

Figure 8

Logistic regression analysis of noscapine-treated SPG7 patient cells. We compared noscapine treated SPG7 patient cells (16,379 cells from 5 individuals) to untreated control and SPG7 patient cells presented above in the SPG7 vs controls section (Fig. 4) for comparison. Noscapine-treated patient cells had logistic regression probability values comparable to controls for acetylated α-tubulin (A). The markers mitochondria (B), nucleus (C), phase contrast (D) and all markers combined (E) were comparable to untreated SPG7 samples. Mean values were compared using one way ANOVA.