How do stem/progenitor cells within an organ know how big the organ is, vs. how big it should be?
The control of organ size during development and repair is one of the last frontiers of biology, as we lack a fundamental understanding of how organ size is encoded, sensed and regulated.
We use limb development as a model to study this type of questions, with a focus on the long bones. Sophisticated mouse genetics and avian embryo manipulation are used to challenge the system and study its recovery afterwards. We use techniques such as multi-colour lineage tracing, single-cell barcoding, spatio-temporal characterisation of proliferative and differentiation potential, and bulk and single-cell multi-omics to reveal the local mechanisms that restore tissue size and integrity, and the systemic ones that mantain body proportions.
Growth of the long bones
Bone elongation is driven by a transient cartilage template, the growth plate (GP). In this structure, cells known as chondrocytes transition from resting (R) to proliferative (P) and then hypertrophic (H), which then die or transdifferentiate into bone-forming cells, such that the cartilage is replaced by bone (primary ossification centre).
Long-lived cartilage progenitors are key in the control of bone growth, but their fetal origin and their regulation by intrinsic and extrinsic factors are poorly understood.
The Figure has been adapted from our review. R/P/H Z: Resting, Proliferative, Hypertrophic zones.
The two challenges we study in the lab
Catch-up growth (funded by NHMRC APP2002084 and HFSP CDA 00021/2019-C)
To uncover molecular and cellular mechanisms controlling organ size, one powerful approach is analysing the recovery of a normal growth trajectory after a developmental insult, which is known as catch-up growth. The big questions are:
1) how is the initial insult/defect detected?
2) how is overall bone size controlled?
3) do the injury-activated responses also play a role in achieving precision during normal growth?
Our approach: refined spatiotemporal perturbation of long-bone growth
Our DRAGON approach (Figure modified from our Development paper) requires both Cre and tTA/rtTA to drive the expression of a Gene of Interest (GoI). By using a left limb-biased Cre and cartilage-driven rtTA, we achieve exquisite control of expression: the GoI in the left cartilage, and tdTom in the contra-lateral one, in an inducible and reversible manner (due to tTA/rtTA being controlled by Doxycycline).
We also use a recombined version of the Dragon allele that only requires tTA/rtTA. This allows us to use Tamoxifen-inducible Cre lines to follow or alter cell lineages of interest (red cells in the Figure below) in the presence/absence of the growth perturbation caused by the Dragon allele (mustard in the Figure).
1) Determine the cell populations and signalling pathways that underlie catch-up growth and the regulation of cartilage architecture
During postnatal growth, the population of cartilage progenitors switches from having short-lived potential, forming short chondrocyte columns that end up being consumed, to being able to self-renew, forming long chondrocyte columns. However, it is unclear how this happens at the single-cell level:
there is a true switch from one type to another (due to intrinsic and/or extrinsic factors)
both types co-exist early on, but the long-lived ones are quiescent until exposed to a new environment
there is only one type, but their physical location determines their short- vs. long-lived behaviour.
Shortly after cell death (P0)
Through mosaic (salt&pepper) perturbations, our published and pre-printed work has uncovered that cell-nonautonomous mechanisms play a key role in the regulation of chondro-progenitor behaviours. In a nutshell, we previously found that mosaic cell-cycle arrest in the cartilage triggers hyper-proliferation of some of the spared cells, leading to almost perfect compensation (Rosello-Diez et al. 2018, PLoS Biol). We are now studying which specific cell population(s) respond to the insult, and what signalling pathways are involved. Moreover, we recently found that the mTORC1 pathway participates in the compensatory response after transient mosaic cell death in the cartilage. Interestingly, mTORC1 is important not only for the recovery of cartilage architecture, but also for establishing the stereotypical cytoarchitecture during normal growth. See summary on 𝕏 (a 𝕏ummary?). We are determining how mTORC1 is activated, and how it coordinates chondrocyte proliferation/ differentiation across the different cartilage layers.
Lastly, to determine the foetal origin of the long-lived progenitors and their regulation, we:
a) follow the fate of foetal candidate populations over long periods, to identify those contributing to long chondrocyte columns (Figure bellow). Interestingly, we have found that some of these candidates expand in response to growth challenges, and we are exploring the molecular pathways involved in this recruitment.
b) use single-cell transcriptomics and barcoding technologies to determine the lineages and fate decisions leading to cartilage progenitors in the presence or absence of injury.
2) Do chondroprogenitors know when a certain size for age has been achieved? How?
When one of the two growth plates in a bone is damaged, the other one can partially compensate for its absence (Hall-Craags 1968), suggesting that there is a target size for age, and that a feedback mechanism informs cartilage cells of bone size, so that a response is generated in the spared growth plate.
We will explore whether negative feedback could explain this behaviour, AND the attainment of a species-specific bone size. This model is based on 2 assumptions:
bone length is translated into a biochemical or biomechanical factor that feeds back to the cartilage
the sensitivity to that factor changes across species and different bones
The model that results from both hypotheses would work like a rheostat for size (a sizostat), allowing for correcting mechanisms to operate when the actual size does not match the set-point.
3) Are catch-up growth mechanisms repurposed to achieve developmental stability?
Paired organs exhibit remarkable left-right size similarity. This is especially true for the limbs. Lewis Wolpert once stated that the average difference between our arms is no bigger than 0.2%, and his lecture as a Waddington medal awardee on the 13th of April 2015 included an inspirational discussion about this topic.
This remarkable symmetry is one of the main manifestations of developmental stability, the ability to buffer any sources of noise during development. We hypothesise that the same mechanisms that operate during catch-up growth act and interact during normal growth, such that developmental stability arises as an emergent property. As part of our functional validation of catch-up growth mechanisms, we are also studying uninjured mice, which allows us to determine whether developmental stability is altered by the intervention.
The late Lewis Wolpert demonstrates the remarkable symmetry of his limbs. Minute 28 of the BSDB video https://www.youtube.com/watch?v=3pAvvGo3np8.
4) Do local injuries trigger a response in other organs? How? What role does this play?
An interesting follow-up that has emerged from the catch-up growth projects is the communication of local injuries to distant organs, which is becoming a trending topic in the regeneration field. It is posited that this systemic activation upon local injury leads to an alert state that facilitates regeneration, likely via activation of cell proliferation and metabolism. We are currently exploring how the long-range communication works in our models, and the role they play in the whole catch-up process.
Besides circulating signals, we will determine whether the nervous system is also involved in this long-range response, as suggested recently.
In summary, our work is located within a theoretical framework in which the coordination of growth and regeneration happens at 3 levels: cell-cell communication within tissues, inter-tissue communication within organs, and inter-organ communication within the organism.
A classic approach to study limb size determination is to cross-graft limb primordia (a.k.a. buds) between close species that differ in size. Initial experiments suggested that the size of the donor limb was unaffected or only slightly modulated by the host (Twitty 1931, see Figure below).
The two main problems with these grafting experiments
1- We and others showed that when the whole limb bud is taken, its own epithelium exerts a strong instructive influence, so that limb bud cells can only be reprogrammed by extrinsic host signals when they are isolated from the influence of the epithelium (Rosello-Diez et al. 2011, Science).
2- Fate plasticity declines with developmental age, so that by the time a limb bud is big enough to be grafted, it might already be impervious to reprogramming by extrinsic signals.
Therefore, the question remains as to whether the outcome of the inter-species transplantation would be different if donor limb cells developed from the beginning within the environment of the host.
In this project, we will generate for the first time chimeric animals in which rat stem cells are incorporated early on in the host mouse embryo, giving rise to the limb tissues in the context of the host’s signals (See Figure below). Gene expression and chromatin accessibility will be compared for the donor limb cells in their endogenous vs. the chimeric context, to uncover the key molecular events underlying this modulation.
We are also capitalising on the unique Australian fauna we have access to, and our international collaborators in order to obtain pluripotent stem cells from a variety of animal models characterised by different developmental tempos, limb proportions and body sizes.
One way to generate chimeras is to inject or aggregate pluripotent stem cells (PSCs) into early embryos. If done into an embryo genetically incapable of forming an organ of interest, the injected cells will mostly contribute to form such organ. This process, when done in the rodent blastocyst, is called blastocyst complementation, and we will use it to generate mice in which the limb mesenchyme is derived from PSCs of other species (see left Figure). To that goal, we have generated a unique mouse model in which the limbs cannot form, and have established several international collaborations to complement this model with PSCs from multiple species.