MechanoMaST – a multimodal pipeline for spatially registering mechanical and transcriptomic tissue data

Linda Decker, Dmitrii Olisov, Nikolai Schleußner, Hendrik Wiethoff, Thomas Schmidt, Henrik Nienhüser, Thomas Moritz Pausch, Jan Oliver Korbel, Alba Diz-Muñoz

Posted on: 28 September 2026

Preprint posted on 31 August 2026

and

Paired-surface spatial mechanomics links tissue stiffness maps to spatial transcriptomics

Huan Ting Ong, Yuting Lou, Jake Turley, Ranmadusha M. Hengst, Md Faris H. Ramli, Xingyu Shen, Jennifer Marlena, Jin Zhu, Rong Li, Chii Jou Chan, Jennifer L. Young

Posted on:

Preprint posted on 20 September 2026

Separated by only a few hours! these two novel preprints detail new and complementary ways to match mechanical readouts from tissues to OMICs

Selected by Felipe Del Valle Batalla

Categories: biophysics, cell biology

Background to the preprints

Single-cell and spatial transcriptomics have revolutionised the understanding of tissue heterogeneity. However, transcript abundance alone doesn’t directly reflect physical properties like tissue stiffness. Tissue stiffness is a critical physical property that is pathologically altered in conditions like fibrosis or cancer and changes progressively with age. Understanding these mechanical changes is essential, yet structural extracellular matrix (ECM) gene expression, particularly fibrillar collagens, has frequently been used as an indirect proxy for mechanical rigidity. However, physical stiffness is influenced by far more than transcript abundance. Protein abundance, post-translational modifications, collagen crosslinking, matrix fibre alignment and architecture, enzymatic degradation, and interstitial fluid pressures all contribute to a tissue’s mechanical state. Consequently, transcript levels offer an incomplete and potentially misleading picture of tissue mechanics, motivating the development of workflows that directly integrate absolute mechanical measurements with spatial transcriptomics.

Combining mechanical measurements with spatial transcriptomics has been technically challenging due to incompatible sample preparation requirements, making performing both measurements on a single tissue section a major hurdle. Nanoindentation requires fresh, unfixed, and hydrated tissue maintained under physiological conditions to measure mechanical properties, while spatial transcriptomics generally requires rapid fixation and permeabilisation to preserve RNA integrity. To overcome this barrier, both approaches utilize adjacent tissue surfaces for each respective measurement and correlate the resulting physical and molecular datasets.

Decker et al.’s MechanoMaST pipeline and Ong et al.’s Paired-Surface Spatial Mechanomics, tackle this bottleneck using distinct but complementary strategies. By mapping absolute mechanical measurements onto spatial transcriptomic coordinates, these studies establish a framework for directly linking tissue biomechanics with molecular states.

Shared Key Findings

Despite using different biological systems and registration procedures, the two studies converge on important conclusions.

Structural collagen transcripts are not reliable proxies for stiffness

The most notable shared finding is the breakdown of the assumption that fibrillar collagen transcription should directly reflect tissue stiffness. In both human colorectal cancer liver metastases and aging mouse ovaries, classical structural ECM transcripts failed to consistently track localized Young’s modulus. In the CRC liver metastasis dataset, COL1A1, COL3A1, and COL5A1 did not show positive correlations with stiffness, while COL1A2 and FN1 exhibited negative correlations. Similarly, in aged ovaries, collagen transcript abundance did not explain the increased physical stiffness detected by nanoindentation. The physical state of the ECM reflects accumulated protein deposition, crosslinking, degradation, and fibre organization, whereas mRNA abundance represents a much more transient molecular state.

Physical and molecular maps can be registered across tissue surfaces

Both studies demonstrate that mechanical measurements can be spatially associated with transcriptomic measurements despite their incompatible experimental requirements. MechanoMaST from Decker et al. achieves this using a custom pipeline of affine transformations and template matching in adjacent serial cryosections. They rigorously quantified the level of spatial alignment and performed error propagation to obtain a refined, high-confidence dataset. Ong et al. generate two opposing surfaces from a single vibratome cut and register them by matching αSMA-positive vascular features visible on both faces, placing the two datasets within a common coordinate system in SpatialX. The mapping in both cases is based on landmarks, with mapping errors carefully estimated and discussed in the manuscripts. These approaches demonstrate that mechanical information measured on one tissue surface can be mapped onto spatial transcriptomic information acquired from another with sufficient spatial correspondence for biological interpretation at the resolved tissue scales.

Tissue mechanics involve non-structural and context-dependent molecular programmes

The studies also shift attention away from structural ECM transcription alone toward secreted factors, inflammatory programmes, and tissue-specific regulatory mechanisms. In tumour tissue, the strongest stiffness-associated molecular signature consisted of secreted factors rather than classical structural collagens. In the aging ovary, matrix-associated genes were enriched among negative stiffness correlations, classical collagen genes did not correlate with stiffness, and immune-associated relationships became more prominent with age. Thus, the molecular determinants of tissue mechanics appear to be considerably more complex than simply ‘how much collagen is being transcribed.

Individual Key Findings

Decker et al. – MechanoMaST

Decker et al. developed MechanoMaST, applying the approach to human colorectal cancer liver metastasis (CRC LM) samples.

Figure 1 from the Decker et al. preprint v2. illustrating the workflow behind MechanoMaST

Mechanical stability across serial cryosections and the gene stiffness signature

A critical methodological question is whether mechanical measurements from one cryosection can meaningfully represent the adjacent section used for spatial transcriptomics. Across consecutive 20-µm cryosections, the authors found that Young’s modulus measurements remained statistically stable, supporting the premise that adjacent sections can be paired for mechanical and molecular analysis. Rather than focusing exclusively on structural ECM genes, the authors combined differential expression across stiffness tiers with Random Forest machine learning. This analysis identified a stiffness signature with TGFBI as the strongest candidate. The broader analysis therefore points toward a mechanoresponsive transcriptional programme associated with tissue remodelling rather than simply increased collagen production.

Functional connection to tumour biology

The association of TGFBI with stiff tumour stroma is particularly interesting. Previous studies have already established that a stiff microenvironment correlates with altered collagen microarchitecture and reduced CD8⁺ T-cell infiltration. Describing TGFBI as the top candidate in MechanoMaST underscores how mechanoadaptive programmes contribute to matrix organization and the immune landscape of the tumour. Furthermore, the reported association of stiffness-linked candidates TGFBI, TFF3, and PRAP1 with metabolic dysfunction-associated steatohepatitis (MASLD) suggests that stiffness-associated transcriptional programmes intersect with broader pathological states involving tissue remodelling and metabolic dysfunction.

Ong et al. – Paired-surface spatial mechanomics

Ong et al. developed a complementary strategy to investigate the relationship between tissue mechanics and transcription during reproductive aging in mouse ovaries. The authors combine large area map nanoindentation on fresh tissues with 10x Genomics Visium HD, providing substantially greater transcriptomic spatial resolution than conventional Visium.

Figure 1 from the Ong et al. preprint showing the steps behind the paired surface mechanomics workflow

Age-dependent reconfiguration of mechanical-transcriptional relationships

The most striking finding is that gene expression tracked local stiffness at both ages, but the associated genes differed. Similar numbers of genes were associated with stiffness in young and aged ovaries, arguing against a general loss of stiffness-associated transcription with age. This suggests that the relationship between the physical and transcriptional states of the tissue is reconfigured rather than progressively disrupted with age. This interpretation is consistent with the broader observation that the relationship between stiffness and transcription is highly compartment-specific and changes with reproductive age. High-stiffness regions were associated with cell cycle programs, while softer regions were associated with cholesterol biosynthesis and matrix-related expression. This demonstrates that physical stiffening does not necessarily coincide with increased transcription of classical structural matrix components. Instead, stiff regions may represent a state characterized by inflammatory signalling, altered matrix maintenance and metabolism, and a loss of normal tissue homeostasis.

Paired-surface registration methodology shows distinct compartment responses

Rather than relying on adjacent serial cryosections, the method uses a single vibratome cut to generate two physically complementary tissue surfaces. One surface is subjected to nanoindentation, while the opposing surface is processed for spatial transcriptomics after cryopreservation. Each 50 µm nanoindentation grid point was associated with a median of 89 4-µm transcriptomic bins, enabling mechanical properties to be examined within discrete anatomical microenvironments. Registration uses immunofluorescence landmarks, specifically αSMA-positive vascular features at the cut plane visible on both sides. Matching these features in SpatialX sets the orientation and scale independently for each tissue, allowing nanoindentation regions to be placed into transcriptomic coordinates. Compartment-resolved analyses showed that stiffness-expression relationships differed across ovarian microenvironments. The corpus luteum, the temporary endocrine structure that forms after ovulation, had the clearest signal. Within this compartment, cholesterol biosynthetic enzymes and the lipoprotein receptor Scarb1 decreased with stiffness, while the steroidogenic regulator Nr5a1 and collagen receptor Ddr2 increased with stiffness, illustrating how local mechanical-molecular relationships can be revealed.

What I Like About These Preprints

A new way of understanding mechanosensing in spatial biology

Spatial mechanomics shifts the focus from simply identifying gene expression locations to exploring how expression correlates with a directly measured physical property at the same anatomical site. A significant finding is the failure of structural collagen mRNA to reliably predict physical stiffness, highlighting a fundamental distinction between gene expression and material properties. Interpreting collagen expression as a direct indicator of tissue stiffness could lead to crucial biological misinterpretations.

Complementary engineering solutions

The two papers provide elegant solutions to the same experimental problem. MechanoMaST prioritizes quantitative uncertainty estimation, spatial alignment quantification, error propagation, and robust registration between continuous cryosections. Ong et al. instead exploit fresh tissues and geometry, using paired surfaces to minimize the separation between mechanical and transcriptomic measurements while taking advantage of the high spatial resolution of Visium HD.

Possible mechanistic and therapeutic applications derived from new methods

The identification of specific molecular signatures associated with mechanical states raises the possibility of testing how these relationships contribute to tissue function in different contexts. For example, the association between TGFBI, tumour stiffness, matrix architecture, and reduced CD8⁺ T-cell infiltration implies that mechanobiological pathways could eventually become therapeutic targets. In the ovary, where interventions targeting fibrosis are already being explored to extend reproductive lifespan, identifying which compartments and programs track local stiffness could help determine where such treatments act and monitor their local mechanical and molecular effects.

Future Directions and Questions for the Authors

1) Both pipelines demonstrate that multiple spatial modalities can be registered across tissue surfaces. Q: Given these registration frameworks, how readily could a third or fourth modality be incorporated—for example, spatial proteomics or other spatial OMICS?

2) The discrepancy between transcript abundance and stiffness raises a fundamental temporal question. Collagen and other ECM components can persist and undergo extensive post-translational modification long after the transcriptional programme that produced them has changed. Q: How much of the transcript–stiffness mismatch reflects temporal lag between transient mRNA expression and the much longer lifetime of deposited matrix?

3) Both studies are fundamentally spatial and correlational. A gene associated with a stiff region may either contribute to tissue stiffening or simply represent a cellular response to an already stiff environment. Q: Could perturbation experiments determine whether these factors actively drive matrix remodelling and tissue stiffening, rather than simply responding to mechanical stress?

1 Star (No Ratings Yet)

Author's response (for both preprints)

The author teams shared about MechanoMaST – a multimodal pipeline for spatially registering mechanical and transcriptomic tissue data

Linda Decker & Alba Diz-Muñoz: Since MechanoMaST combines data from adjacent tissue sections, additional sections could be readily used for a straightforward integration of other spatial OMICS methods. The prerequisites for a successful integration are that the observed modality is conserved across adjacent tissue sections (similar to what we have shown for tissue stiffness) and that the data can be registered to an image of the tissue, ideally a hematoxylin and eosin (H&E) staining. For spatial proteomics specifically, there is already published data available on how well the proteome of human breast cancer and the mouse kidney is conserved across adjacent cryosections (see [DOI: 10.3390/cancers13174419], [DOI: 10.1038/s41591-025-04060-4]; and related methodology [DOI: 10.1002/jms.1926]). Furthermore, both multiplexed immunostaining-based approaches and mass spectrometry-based methods have already been performed together with H&E stainings in the same section. Therefore, spatial proteomics and other approaches fulfilling the prerequisites can be readily incorporated as a third modality into the MechanoMaST workflow. For the addition of a fourth and fifth modality, we recommend reducing the section thickness from 20 µm to 10 µm.

Huan Ting Ong & Jennifer Young: Our paired-surface spatial mechanomics approach is compatible with additional modalities because the cryopreserved block remains available for serial sectioning. While we take the first two 10-µm sections for spatial transcriptomics, we envision combining spatial proteomics, metabolomics, or structural ECM imaging on further serial 10-µm sections and registering using shared landmarks . On the nanoindented fresh tissue, cyclic or high-plex immunofluorescence, as well as label-free imaging (such as second-harmonic generation), could also be applied to the tissue after mechanical mapping to infer ECM structure or protein localization. The main limitation is not the number or type of modalities, but their increasing distance from the measured surface. In the ovary, smaller follicles, vessels, or luteal structures can change substantially over micron distances. Thus, incorporating additional modalities would require robust registration quality control. Spatial proteomics would be particularly informative because it could distinguish transcript-level from protein-level regulation, where we would expect to see a tighter correlation between ECM abundance and mechanics, as has been previously demonstrated ([DOI: 10.1038/s41467-023-39085-1]).

Linda Decker & Alba Diz-Muñoz: We suspect that timing is the key to this mismatch. Colorectal cancer liver metastases grow silently for years before the primary tumour is even detected, meaning that by the time of surgery, a dense, stiff ECM has already been long established, likely well past the transcriptional activity that built it. Untangling whether this discrepancy stems from the long lifetime of ECM proteins, ECM architecture, shifting transcription over time, or cell migration into the tissue will require capturing multiple time points along the way. Future work, perhaps using mouse models where disease progression can be tracked from the very start, will let us catch these dynamics in the act.

Huan Ting Ong & Jennifer Young: We expect the temporal lag to be an important contributor to this discrepancy. Transcriptomic analysis captures a relatively transient molecular state, while collagen, elastin, and other ECM proteins can persist long after the transcriptional programs that produced or modified them . Our observation that collagen transcripts were enriched in younger tissues and matrix-related transcripts in softer regions, despite reports of matrix accumulation and ovarian stiffening with age ([DOI: 10.1111/acel.13259]), is consistent with matrix deposited earlier and being retained as turnover slows with age. However, protein lifetime is unlikely to be the only explanation because tissue stiffness also depends on ECM crosslinking, fiber architecture, and poroelastic fluid movement. Resolving these contributions with time will require staged time course experiments that combine spatial transcriptomics with measurements of protein abundance, matrix organization, ECM turnover, and mechanics, ideally across reproductive age and estrous cycles for ovaries.

Linda Decker & Alba Diz-Muñoz: Yes. For one of our candidates, TGFBI, there is already promising data on the corresponding protein . Mouse tumour models treated with an antibody against TGFBI showed reduced tumour stiffness and collagen fibre thickness ([DOI: 10.1136/gutjnl-2018-317570]), suggesting that TGFBI actively promotes tissue stiffening rather than simply responding to it. As the other candidates are novel, it remains to be seen whether they play the same active role, or whether their association with stiffness is more of a downstream consequence. Perturbation experiments in cultured cancer-associated fibroblasts or mouse tumour models would be an exciting next step to find out.

Huan Ting Ong & Jennifer Young: Yes, this would require perturbations in both directions. For the first scenario, perturbing identified pathways in cell culture, organoid, or mouse models followed by mechanical measurements could test whether candidate molecular programs drive matrix remodeling and tissue stiffening. For the second, perturbing tissue mechanics itself (e.g., via compression or direct matrix modification) followed by temporal transcriptional profiling alongside mechanical measurements could identify which of these factors respond to mechanical cues rather than drive them. There is also the possibility of a feedback relationship, in which both directions work concurrently. This is a key question and validation experiments will be important follow-ups to this work.

Have your say Cancel reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Sign up to customise the site to your preferences and to receive alerts

Also in the biophysics category:

Reciprocal mechanochemical feedback couples neural crest migration and neurulation

Kai Weißenbruch, Lucas Alvizi, Jemima J. Burden, et al.

Selected by 04 September 2026

Béryl Laplace-Builhé

Organoids reveal niche-specific mechanotransduction-guided human cortical patterning and cell fate acquisition.

Alex J. Kingston, Annabelle A. M. Wurmser, Aditi S. Kulkarni, et al.

Selected by 19 August 2026

Kanishka Parashar

Matrix Rigidity Governs Switch-Like Oscillatory State Transitions of the Segmentation Clock in Isolated Presomitic Mesoderm Cells

Chun-Yen Sung, Yuqian Yang, Usha Kadiyala, et al.

Selected by 18 August 2026

Felipe Del Valle Batalla

Also in the cell biology category: