Temporal changes in cytoskeletal organisation within isolated chondrocytes quantified using a novel image analysis technique

Computational pathology for musculoskeletal conditions using machine learning: advances, trends, and challenges

Histopathology is widely used to analyze clinical biopsy specimens and tissues from pre-clinical models of a variety of musculoskeletal conditions. Histological assessment relies on scoring systems that require expertise, time, and resources, which can lead to an analysis bottleneck. Recent advancements in digital imaging and image processing provide an opportunity to automate histological analyses by implementing advanced statistical models such as machine learning and deep learning, which would greatly benefit the musculoskeletal field. This review provides a high-level overview of machine learning applications, a general pipeline of tissue collection to model selection, and highlights the development of image analysis methods, including some machine learning applications, to solve musculoskeletal problems. We discuss the optimization steps for tissue processing, sectioning, staining, and imaging that are critical for the successful generalizability of an automated image analysis model. We also commenting on the considerations that should be taken into account during model selection and the considerable advances in the field of computer vision outside of histopathology, which can be leveraged for image analysis. Finally, we provide a historic perspective of the previously used histopathological image analysis applications for musculoskeletal diseases, and we contrast it with the advantages of implementing state-of-the-art computational pathology approaches. While some deep learning approaches have been used, there is a significant opportunity to expand the use of such approaches to solve musculoskeletal problems.

Homogentisic acid induces cytoskeleton and extracellular matrix alteration in alkaptonuric cartilage

Alkaptonuria (AKU) is an ultra-rare disease caused by the deficient activity of homogentisate 1,2-dioxygenase enzyme, leading the accumulation of homogentisic acid (HGA) in connective tissues implicating the formation of a black pigmentation called "ochronosis." Although AKU is a multisystemic disease, the most affected tissue is the articular cartilage, which during the pathology appears to be highly damaged. In this study, a model of alkaptonuric chondrocytes and cartilage was realized to investigate the role of HGA in the alteration of the extracellular matrix (ECM). The AKU tissues lost its architecture composed of collagen, proteoglycans, and all the proteins that characterize the ECM. The cause of this alteration in AKU cartilage is attributed to a degeneration of the cytoskeletal network in chondrocytes caused by the accumulation of HGA. The three cytoskeletal proteins, actin, vimentin, and tubulin, were analyzed and a modification in their amount and disposition in AKU chondrocytes model was identified. Cytoskeleton is involved in many fundamental cellular processes; therefore, the aberration in this complex network is involved in the manifestation of AKU disease.

The rate of hypo-osmotic challenge influences regulatory volume decrease (RVD) and mechanical properties of articular chondrocytes

OBJECTIVES

Osteoarthritis (OA) is associated with a gradual reduction in the interstitial osmotic pressure within articular cartilage. The aim of this study was to compare the effects of sudden and gradual hypo-osmotic challenge on chondrocyte morphology and biomechanics.

METHODS

Bovine articular chondrocytes were exposed to a reduction in extracellular osmolality from 327 to 153 mOsmol/kg applied either suddenly (<5 s) or gradually (over 180 min). Temporal changes in cell diameter and the existence of regulatory volume decrease (RVD) were quantified along with changes in cortical actin and chromatin condensation. The cellular viscoelastic mechanical properties were determined by micropipette aspiration.

RESULTS

In response to a sudden hypo-osmotic stress, 66% of chondrocytes exhibited an increase in diameter followed by RVD, whilst 25% showed no RVD. By contrast, cells exposed to gradual hypo-osmotic stress exhibited reduced cell swelling without subsequent RVD. There was an increase in the equilibrium modulus for cells exposed to sudden hypo-osmotic stress. However, gradual hypo-osmotic challenge had no effect on cell mechanical properties. This cell stiffening response to sudden hypo-osmotic challenge was abolished when actin organization was disrupted with cytochalasin D or RVD inhibited with REV5901. Both sudden and gradual hypo-osmotic challenge reduced cortical F-actin distribution and caused chromatin decondensation.

CONCLUSIONS

Sudden hypo-osmotic challenge increases chondrocyte mechanics by activation of RVD and interaction with the actin cytoskeleton. Moreover, the rate of hypo-osmotic challenge is shown to have a profound effect on chondrocyte morphology and biomechanics. This important phenomenon needs to be considered when studying the response of chondrocytes to pathological hypo-osmotic stress.

Influence of spreading and contractility on cell detachment

Cell adhesion is a key phenomenon that affects fundamental cellular processes such as morphology, migration, and differentiation. In the current study, an active modelling framework incorporating actin cytoskeleton remodelling and contractility, combined with a cohesive zone model to simulate debonding at the cell-substrate interface, is implemented to investigate the increased resistance to detachment of highly spread chondrocytes from a substrate, as observed experimentally by Huang et al. (J. Orthop. Res. 21: 88-95, 2003). 3D finite element meshes of the round and spread cell geometries with the same material properties are created. It is demonstrated that spread cells with a flattened morphology and a larger adhesion area have a more highly developed actin cytoskeleton than rounded cells. Rounded cells provide less support for tension generated by the actin cytoskeleton; hence, a high level of dissociation is predicted. It is revealed that the more highly developed active contractile actin cytoskeleton of the spread cell increases the resistance to shear deformation, and subsequently increases the shear detachment force. These findings provide new insight into the link between cell geometry, cell contractility, and cell-substrate detachment.

Osmotic challenge drives rapid and reversible chromatin condensation in chondrocytes

Changes in extracellular osmolality have been shown to alter gene expression patterns and metabolic activity of various cell types, including chondrocytes. However, mechanisms by which physiological or pathological changes in osmolality impact chondrocyte function remain unclear. Here we use quantitative image analysis, electron microscopy, and a DNase I assay to show that hyperosmotic conditions (>400 mOsm/kg) induce chromatin condensation, while hypoosmotic conditions (100 mOsm/kg) cause decondensation. Large density changes (p < 0.001) occur over a very narrow range of physiological osmolalities, which suggests that chondrocytes likely experience chromatin condensation and decondensation during a daily loading cycle. The effect of changes in osmolality on nuclear morphology (p < 0.01) and chromatin condensation (p < 0.001) also differed between chondrocytes in monolayer culture and three-dimensional agarose, suggesting a role for cell adhesion. The relationship between condensation and osmolality was accurately modeled by a polymer gel model which, along with the rapid nature of the chromatin condensation (<20 s), reveals the basic physicochemical nature of the process. Alterations in chromatin structure are expected to influence gene expression and thereby regulate chondrocyte activity in response to osmotic changes.

The phenotypic characterization of A13/BACii, a novel bovine chondrocytic cell line with differentiation potential

In cartilage research bovine articular cartilage is used as an alternative to human tissue. However, animal material is subject to availability and primary cultures undergo senescence, limiting their use. Here we report the immortalization of primary bovine chondrocytes, which could be used as a surrogate for freshly isolated chondrocytes. Chondrocytes were isolated from cartilage explants and immortalized using 1.0 µg/ml benzo[alpha]pyrene. For 3-dimensional culture, chondrocytes were resuspended in 0.5% low-melt agarose at high density (HD) and cultured for 24 h prior to determining changes in expression profile and morphology. A13/BACii chondrocytes acquired a 'flat' irregular morphology and a foetal-like cell volume (1,509.59 ± 182.04 µm(3)). The human cell line C-20/A4 showed a statistically similar volume and length to A13/BACii. Two-dimensional-cultured A13/BACii expressed elevated levels of type I collagen (col1), reduced levels of type II collagen (col2) compared to freshly isolated chondrocytes and an overall col2 to col1 expression ratio (col2:col1) of 0.11 ± 0.01. Upon 3-dimensional encapsulation, there was a significant rise in col2 expression in both A13/BACii and C-20/A4, suggesting a capacity for redifferentiation in both cell lines with a return of col2:col1 values of A13/BACii to values previously observed in primary chondrocytes. A13/BACii chondrocytes expressed aggrecan, matrix metalloproteinase (MMP)-3, MMP-9 and MMP-13, further supporting indications of the differentiated phenotype. Here we report the creation of a novel chondrocytic cell line and demonstrate its strong potential for redifferentiation upon HD 3-dimensional encapsulation, providing an alternative to conventional dedifferentiated cell lines and primary culture.

Involvement of the cytoskeletal elements in articular cartilage homeostasis and pathology

The cytoskeleton of all cells is a three-dimensional network comprising actin microfilaments, tubulin microtubules and intermediate filaments. Studies in many cell types have indicated roles for these cytoskeletal proteins in many diverse cellular processes including alteration of cell shape, movement of organelles, migration, endocytosis, secretion, cell division and extracellular matrix assembly. The cytoskeletal networks are highly organized in structure enabling them to fulfil their biological functions. This review will primarily focus on the organization and function of the three major cytoskeletal networks in articular cartilage chondrocytes. Articular cartilage is a major load-bearing tissue of the synovial joint; it is well known that the cytoskeleton acts as a physical interface between the chondrocytes and the extracellular matrix in 'sensing' mechanical stimuli. The effect of mechanical load on cytoskeletal element expression and organization will also be reviewed. Abnormal mechanical load is widely believed to be a risk factor for the development of osteoarthritis. Several studies have intimated that the major cytoskeletal networks are disorganized or often absent in osteoarthritic cartilage chondrocytes. The implications and possible reasoning for this are more widely discussed and placed into context with their potential relevance to disease and therapeutic strategies.

Chondrocyte response to cyclic hydrostatic pressure in alginate versus pellet culture

Cells are often cultured at high density (e.g., confluent monolayer and as pellets) to promote chondrogenic differentiation and to maintain the chondrocyte phenotype. They are also frequently suspended in hydrogels such as agarose or alginate for the same purposes. These culture techniques differ markedly with respect to frequency of direct contact between cells and overall intercellular spacing. Because these factors may significantly affect mechanotransduction, the purpose of this study was to determine if the response of articular chondrocytes to cyclic hydrostatic pressure would depend on the culture condition. Primary articular chondrocytes from young and mature pigs were cultured either as pellets or suspended in alginate beads. Both groups were exposed to dynamic hydrostatic pressure (4 MPa, 1 Hz, 5400 cycles per day) for 7 days. Cell proliferation was unaffected by pressure, but pressurized chondrocytes in pellet culture had significantly greater sGAG content and incorporated [3H]proline at a higher rate than nonpressurized controls. Electron microscopy revealed a fibrous extracellular matrix (ECM) surrounding pellets, but not cells in alginate. In addition, expression of Connexin 43 (Cx43) mRNA was slightly lower in alginate than in pellet cultures and was not significantly altered by loading. Thus, metabolic response of chondrocytes to dynamic hydrostatic pressure was affected by culture technique; chondrocytes cultured as pellets exhibited the classical anabolic response to dynamic hydrostatic pressure, but those in alginate did not. Although cell-ECM interaction could be important, the differential response is not likely attributable to differential expression of Cx43 mRNA.

Classifying structural alterations of the cytoskeleton by spectrum enhancement and descriptor fusion

A classifier capable of ranking structural alterations of the cytoskeleton is developed. Images of cytoskeletal microtubules obtained from the epifluorescence microscopy of primary culture rat hepatocytes are analyzed. Morphological descriptors are extracted by contour and mass fractal analysis, direct methods, and spectrum enhancement. All methods are designed and tuned to make the extracted morphological descriptors insensitive to absolute fluorescence intensities. Spectrum enhancement is a nonlinear filter that involves spatial differentiation of the gray-scale image followed by conversion of power spectral density to the logarithmic scale and averaging over arcs in the reciprocal domain. Enhanced spectra exhibit local maxima that correspond to the structured microtubule bundles of a normal cytoskeleton. Descriptor fusion for classification is achieved by means of multivariate analysis. The classifier is trained by image sets representing normal ("negative control") microtubules and those altered by exposure to a fungicide at the highest dose of the experiment design. Some sensitivity and validation tests, including discriminant functions analysis, are applied to the classifier. The latter is applied to recognize images of microtubules not used in the training stage and comes from treatments at lower concentrations and shorter times. As a result, structural alterations are ranked and structural recovery after treatment is quantified. The method has potential use in quantitative, morphology-based tests on the cytoskeleton treated either by anticancer drugs or by cytotoxic agents.

Quantitative kinetics of damage and recovery of cytoskeletal structure by means of image analysis

The cytoskeleton is a network of proteins which structurally and dynamically organise the cytoplasm of living cells. Microtubules are among its constituents. Morphological alterations of microtubules are related to functional impairment. Therefore cytoskeletal morphology is a valuable indicator of cell injury and functionality. This paper focuses on the comparison between normal and altered cytoskeletal microtubules by means of image analysis and classification with the aim of replacing visual assessment. Morphology has been quantified by the extraction of some descriptors yielded by spatial differentiation, fractal analysis and Fourier analysis followed by non-linear filtering. The principal component analysis of these descriptors has led to image recognition and has been applied to hepatocytes and fibroblasts exposed to some xenobiotics. In the case of hepatocytes, images have been ranked according to the severity of cytoskeletal damage, a dose-response relation has been derived from the regression of the first principal component and the percentage of structural recovery after exposure has been estimated.

Mechanical compression and hydrostatic pressure induce reversible changes in actin cytoskeletal organisation in chondrocytes in agarose

In numerous cell types, the cytoskeleton has been widely implicated in mechanotransduction pathways involving stretch-activated ion channels, integrins and deformation of intracellular organelles. Studies have also demonstrated that the cytoskeleton can undergo remodelling in response to mechanical stimuli such as tensile strain or fluid flow. In articular chondrocytes, the mechanotransduction pathways are complex, inter-related and as yet, poorly understood. Furthermore, little is known of how the chondrocyte cytoskeleton responds to physiological mechanical loading. This study utilises the well-characterised chondrocyte-agarose model and an established confocal image-analysis technique to demonstrate that both static and cyclic, compressive strain and hydrostatic pressure all induce remodelling of actin microfilaments. This remodelling was characterised by a change from a uniform to a more punctate distribution of cortical actin around the cell periphery. For some loading regimes, this remodelling was reversed over a subsequent 1h unloaded period. This reversible remodelling of actin cytoskeleton may therefore represent a mechanism through which the chondrocyte alters its mechanical properties and mechanosensitivity in response to physiological mechanical loading.

Quantification of stretch-induced cytoskeletal remodeling in vascular endothelial cells by image processing

BACKGROUND

Reorientation of the cell axis induced by cyclic stretching is an early response to mechanical forces in vitro. However, quantitative assay for this phenomenon has been difficult due to lack of robust methods. We hypothesized that cell orientation may be redefined by the orientation of actin fibers. We developed image processing methods to quantitate the orientation and density of actin fibers.

METHODS

A convolution filter using Sobel kernels was adapted to determine the orientation and density of actin fibers in human endothelial cells. Unidirectional stretching (10%, 0.5 Hz) was applied to induce cytoskeletal remodeling by varying the duration of stimulation (control, 0.5, 1, 2, 5, 10, and 20 h). Actin fibers were visualized by fluorescent phalloidin. The image processing method was compared with the manual method for reproducibility. Both confluent and subconfluent cells were tested to assess the efficacy of the methods.

RESULTS

Cyclic stretch-induced dense and uninterrupted actin cabling formed across the cell body and, later, the actin fibers became aligned perpendicular to the stretch direction. The variance of actin fiber orientation became smaller after 2 h of stretch (F < 0.01). The actin fiber density index, a derived parameter related to the density of actin fibers, increased as early as 30 min of stretching (P < 0.05) and decreased after 10 h of stretching. Reproducibility of our method was extremely good. Applicability of the method was not compromised by cell density.

CONCLUSIONS

Our method is reliable for quantifying cytoskeletal remodeling induced by mechanical force.

Increased presence of cells with multiple elongated processes in osteoarthritic femoral head cartilage

OBJECTIVE

This study examined the morphology of chondrocytes in articular cartilage from osteoarthritic (OA) and non-OA human femoral heads and in particular the appearance of a sub-population of cells with multiple elongated processes radiating up to 30 microm into the extracellular matrix.

METHODS

Cartilage explants were removed from 8 anatomical sites over the surface of OA (n=6) and non-OA (n=5) femoral heads. Cells were labeled for vimentin intermediate filaments and visualized using epi-fluorescence and confocal microscopy. The percentage of cells with elongated processes was correlated with macroscopic and histological indicators of osteoarthritis.

RESULTS

Cells with processes accounted for less than 10% of the total cell population in non-OA cartilage. By contrast, in the peripheral regions of the OA femoral head these cells accounted for 20-45% of the total cell population, the differences being statistically significant. These peripheral areas are habitually non-load bearing and were also the most likely to show gross fibrillation and pannus formation. A statistically significant correlation was demonstrated between the percentage of cells with processes and the histological extent of the OA degradation, quantified in terms of the Mankin score.

CONCLUSIONS

The extension of cell processes, which may be associated with localized breakdown of the pericellular matrix, will undoubtedly alter numerous aspects of cell function including phenotypic expression and mechanotransduction. Hence these significant changes in chondrocyte morphology are likely to have important implications for the aetiology of osteoarthritis and the development of potential treatment strategies.