# Tumor morphological evolution: directed migration and gain and loss of the self-metastatic phenotype

- Heiko Enderling
^{1}, - Lynn Hlatky
^{1}and - Philip Hahnfeldt
^{1}Email author

**5**:23

**DOI: **10.1186/1745-6150-5-23

© Enderling et al; licensee BioMed Central Ltd. 2010

**Received: **20 January 2010

**Accepted: **20 April 2010

**Published: **20 April 2010

## Abstract

### Background

Aside from the stepwise genetic alterations known to underlie cancer cell creation, the microenvironment is known to profoundly influence subsequent tumor development, morphology and metastasis. Invasive cluster formation has been assumed to be dependent on directed migration and a heterogeneous environment - a conclusion derived from complex models of tumor-environment interaction. At the same time, these models have not included the prospect, now supported by a preponderance of evidence, that only a minority of cancer cells may have stem cell capacity. This proves to weigh heavily on the microenvironmental requirements for the display of characteristic tumor growth phenotypes. We show using agent-based modeling that some defining features of tumor growth ascribed to directed migration might also be realized under random migration, and discuss broader implications for cause-and-effect determination in general.

### Results

Considering only the properties of random migration in tumors composed of stem cells and committed cells, we are able to recapitulate a characteristic clustering feature of invasive tumor growth, a property we attribute to "self-metastatic" growth. When the additional influence of directed migrations under chemotactic environments are considered, we find that tumor growth and invasive morphology are supported while the tumor is distant from the source, but are progressively discouraged as the tumor converges about that source.

### Conclusions

We show that invasive clustering can derive from basic kinetic assumptions often neglected in more complex models. While higher-order mechanisms, e.g. directed migration upon chemotactic stimuli, may result in clustering growth morphologies, exclusive attributions of this phenotype to this or other structured microenvironments would be inappropriate, in light of our finding these features are observable in a homogeneous environment. Furthermore, directed migration will result in loss of the invasive phenotype as the tumor approaches the attractor source. Reviewers: This article was reviewed by Mark Little and Glen Webb.

## Background

Cancer development and tumor growth are complex dynamic phenomena whose spatio-temporal evolution depends on intrinsic properties of the cancer cells as well as on environmental factors. Among the environmental dependencies, the angiogenic switch is probably the most well established bottleneck a tumor has to overcome before a tumor can progress to clinical disease [1, 2].

Here we show that the overall effect of the microenvironment on tumor growth may be largely anticipated by its influence on migration-enabled seeding of new clones at the tumor periphery. In particular we show a biphasic impact of directed migration up gradients in the extracellular matrix. Directed migration yields liberation of cancer stem cells in tumor clusters as progeny continuously migrate away and thus release space constraints. However, as a tumor approaches the source of the gradient, clustering away from the tumor is discouraged and tumor growth slows.

## Results

### Directed migration towards an attractor source

*W*and height

*H*(bottom left corner, (

*0,0*)) with an exponentially decreasing gradient as either x or y is increased.

and the initial cancer stem cell being located in the center of the top-right quadrant of the domain at P(*W**3/4, *H**3/4). We simulate tumor growth for t = 360 days in 25 independent simulations for different taxis response probabilities *G* = 1 - exp(-*ξz*)), which apply above and beyond random migration (see Methods for details). Here, *z* is the product of |
| and another factor related to the available sites *S* and the angular direction of the gradient relative to those sites. The parameter *ξ* is the strength of the relationship between the value of *z* and the probability of directed cell movement - the chemotactic responsiveness of a cell.

Comparable to simulations of random motility alone, initially a small tumor cluster forms around the seeded cancer stem cell. When directed migration is strong (*ξ* = 1), the offspring continuously migrate towards the attractor, which has the effect of further loosing space constraints in the primary tumor cluster (Figure 2B). As a result, the cancer stem cells proliferate frequently. Here and throughout, it is assumed that when stem cells divide, they reproduce themselves by "symmetric division" at a (low) frequency p_{s} (estimated to be = 1% of stem cell divisions). The rest of the time, their divisions are assumed to be asymmetric, producing a stem and a non-stem cell. Non-stem cells divide until their generational lifetimes are exhausted, here taken to be *ρ* = 10 divisions. For *ξ* = 1, the tumors after t = 360 days consist on average of 21,765 ± 1,505 cells (110 ± 9 cancer stem cells), a significant increase (*p* < 10^{-6}) compared to tumors developing with random motility alone (7,023 ± 484 (43 ± 4) cells, Figure 2D). If the contribution of directed migration to tumor growth and morphology evolution is lowered to *ξ* = 0.5, yields approaching the tumor growth rates and morphologies observed without directed migration are obtained (12,374 (66) and 9,808 (56) cells, respectively, Figure 2E).

### Directed migration at an attractor source

*W*/2,

*H*/2) with a radially exponentially decreasing gradient , where:

and the initial cancer stem cell is located right at the attractor source. We simulate tumor growth for t = 360 days in 25 independent simulations for different taxis response strengths *ξ*.

Comparable to simulations of random motility alone, initially a small tumor cluster forms around the seeded cancer stem cell. A strong directed gradient with *ξ* = 1 towards the attractor source and the tumor core results in a continuous inbound migration of cells on the periphery. This results in a persistent spatial inhibition of cancer stem cells in the core of the cluster (Figure 2C). The cancer stem cells divide only infrequently, and the tumor population does not grow. After t = 360 days the tumors consist on average of 1,151 ± 102 cells (10 ± 1 cancer stem cells), a significant decrease (*p* < 10^{-13}) compared to tumors developing without directed migration solely dependent on random motility (Figure 2D). If the contribution of directed migration to tumor morphology evolution is lowered to *ξ* = 0.5 and *ξ* = 0.25, yields approaching the tumor growth rates and morphologies observed without directed migration (1,723 (14) and 2,822 (20)) cells, respectively (Figure 2F) are again observed.

### Biphasic influence of directed migration

## Discussion

Tumors that consist of stem cells and non-stem progenitors and develop with random motility alone, i.e. without environmental gradients, exhibit a self-metastatic phenotype. Stem cells are spatially inhibited by their mortal progeny and thus can only form clusters of a limited size. For tumor progression, space has to be freed around the stem cell to allow for symmetric division and migration to seed a new cluster nearby.

Here we presented a study of the impact of cell migration directed by environmental gradients on a growing tumor population. The continuous migration of tumor cells away from the tumor boundary and towards an attractor source relaxes spatial constraints within tumor clusters, which in turn facilitates increased stem cell division and self-metastatic seeding. If the tumor cluster is sufficiently far away from the attractor, directed migration significantly promotes tumor growth. On the other hand, if the tumor cluster is located around the attractor source, preferred inbound migration introduces spatial constraints on the tumor cluster and the stem cells within. This will result in loss of stem cell division, loss of the self-metastatic phenotype and significantly reduced tumor growth. Depending on the position of a developing tumor cluster to an attractor, directed migration can therefore have a biphasic impact on tumor growth.

We would not expect second-order effects relating to gradients to change these findings. Random perturbations to either gradient strength or cell responsiveness, such as may occur from enzymatic degradations or cell mutations, respectively, would be expected to increase random cell drift away from the tumor mass (if the source is at a distance from the tumor), as there are fewer random ways for cells to compact together than there are for them to drift apart. For this reason it would be expected that undirected second-order effects modulating chemotactic gradients would produce more migration overall, and thus, as here shown, to accelerated tumor growth.

## Conclusions

Directed migration of a heterogeneous population of tumor stem cells and progenitor cells in response to environmental gradients alters tumor growth rate as well as intrinsic tumor morphology obtained by random cell motility alone. Dependent on the relative position of tumors to the attractor source tumor growth is promoted or inhibited, and the intrinsic self-metastatic morphology can be enhanced or completely lost. When analyzing the dynamic behavior of a tumor population, attention should be paid to what morphologies and growth rates are due to the intrinsic properties of tumor cells, and which reflect the environment where the tumor develops. By using minimally-parameterized models that better capture fundamental tumor compositions and kinetics, one may, as here, uncover properties intrinsic to tumor cells that have already been attributed by other models to more complex environmental factors. With the improved understanding of the role of intrinsic factors gained from these models, we might more reliably account for how environmental perturbations really alter the overall tumor growth dynamic.

## Methods

We use an agent-based computer model [12, 13] to simulate single cell dynamics, and derive complex population dynamics from cells interacting with each other and their environment. Model details have been previously published [9, 14]. Briefly, in line with the cancer stem cell hypothesis we assume all cancer cells being able to proliferate a certain number of times, *ρ*_{max}, before dying and thus ultimately being removed from the simulation. For cancer stem cells, we assume *ρ*_{max} = ∞. Cells need to mature through the cell cycle before division can occur, which takes cell type dependent time, in our simulations set to be *τ* = 24 hours. Dependent on available space cells can migrate with rate *μ*. Without available space, i.e. none of the adjacent eight lattice points on a two-dimensional grid is vacant, we assume the cell to be inhibited by their neighboring cells and forced to rest in a quiescent state until again exposed to space. We initiate a single cell in the center of our 3,500 *μ*m × 3,500 *μ*m computational domain, composed of 350^{2} equal-sized blocks of 100 *μ*m^{2} each that can hold at the most one cell at any time. By simulating cell proliferation, migration and cell death at discrete time intervals Δ*t*, complex population behaviors emerge. The dynamics of these populations can be observed by tracking the number of quiescent and proliferating cells over time, as well as their spatio-temporal morphological evolution.

### Random motility

*x*

_{0},

*y*

_{0}) where to move to is based upon available lattice points in the immediate neighborhood

*N*

_{(x0, y0)}. The probabilities are calculated according to the following rule:

With probability 1/(1+number of unoccupied neighboring lattice points) the cell remains temporarily stationary. If a cell is completely surrounded by other cells the probability of not moving is the only choice.

### Directed migration

Let *F* be a function whose gradient
describes a chemotactic stimulus due to the distribution of an attractant in the extracellular matrix. Let
be the vector connecting the center of the lattice point (*x*_{0}, *y*_{0}) where a cell resides at time *t*, to the center of one of the eight adjacent lattice points (*x*_{
i
}, *y*_{
i
}). We now consider the values of
in each surrounding lattice point that are *positive* and correspond to an unoccupied site. We call these the Available Sites (*S*). Note that migrations in negative directions do not occur as *directed* movements. Of the *S* nontrivial movements, we weight each by the factor cos(angle *θ* between the gradient
and the movement direction
for each possible movement), for the sake of calculating the angle-weighted relative opportunities for movement in each available direction. The cos(*θ*) are summed over the *S* available sites to get *AS* (*A* is the average cosine value). Treating the effective "cos(*θ*)" of non-movement to be 1, the total sum of cosines is *AS*+1, where *AS* corresponds to actual movement. The total angle-weighted relative opportunity for movement (to one of the *S* sites) due to directed migration is thus *AS*/(*AS*+1) where *A* is the average of the cos(*θ*) over these sites. This value, times the gradient strength |
|, is now mapped to a probability of mobilization *G* through an arbitrary chemotactic/haptotactic responsiveness function *H*(*z*) for the cells, defined such that *H*(0) = 0 and *H* → 1 monotonically as *z* → ∞. The result is *G* = *H*{ |
|*AS*/(*AS*+1)}. *G* is the probability the gradient succeeds in mobilizing the cell, and takes into account the proportion of movement possible and the strength of the gradient. Once mobilized, where the cell then goes is weighted by the respective cosines for movement, so that cos(*θ*)*G*/(*AS*) is the probability of movement to the available site at angle *θ*. We chose *H*(*z*) = 1 - exp(-*ξz*), where *ξ* is a parameter that scales the contribution of [(gradient strength) × (movement opportunity)] to the actual probability of cell movement.

Algorithmically, directed migrations are considered at the same frequency *μ* as random migrations, where *μ* = 0.00635 mm h^{-1} or about 15 lattice moves (cell widths) per day [15]. For weak local gradients a cell is less likely to move at all due to the gradient than for strong local gradients, so random movements will be the only reason for movement and will thus dominate. By contrast, for strong gradients, the probability of directed movement approaches 1 when favorable lattice points are available, so it will compete strongly with the random process.

## Reviewers' Comments

### Reviewer 1: Mark Little, Imperial College Faculty of Medicine, London, UK

#### General comments

This is an interesting and generally clearly-written paper, although short of methodological detail. Arguably, the model presented is merely schematic, useful nevertheless as illustrating the idea that random+directed motility could aid tumour growth, at least in early stages of tumour development. The detailed mathematical assumptions made (e.g., in relation to the expressions on pp.10-11 describing the chemo-taxis process in the Methods) I suspect have no biological basis, even more so the particular parameter values assumed. What is missing is any biological justification of the gradient-following mechanism. Do tumour cells follow-gradients?

There are a few missing, or poorly explained, things in the model, such as the relation between cancer stem cells and other tumour cells. [Perhaps I missed this somewhere.] I assume that most of the progeny of cancer stem cells are non-stem cells, but clearly there must be some proliferation of stem cells for the model to give the results it does.

#### Specific comments (page/line)

p.4 l.-6 Should this be "source of the gradient, clustering"?

p.10-11 I guess this chemotaxis process is equivalent to the standard Keller-Segel model [16, 17] but this should be referenced.

p.11 l.6 What is the point of the *S*/(*S*+1) term in this expression? It is similar to that in the expression at end of 2nd para on p.10? However, the point in the expression on p.10 was clear (to make the probabilities sum to 1). These *G* do not sum to 1, so I assume they are renormalized to make sure that this is the case. Also, where has the
vector gone?

### Authors' Response

The conclusions are surprisingly robust over substantial variations in parameters, highlighting the power of the arguments. Our point is essentially what the reviewer has already concluded - random plus directed motility is a universal promoter of tumor development for reasons that can be traced to simple cell kinetics. One does not need to invoke higher-order processes e.g. angiogenesis to see these results. That said, the actual values we assumed for the frequency of symmetric division of stem cells (1%), the migration rate of tumor cells (15 cell widths per day), and the generational lifetimes of non-stem cells (*ρ* = 10) are reasonably in line with values determined elsewhere.

That tumor cells follow gradients has been known for some time. The notion has its origin as far back as Paget, who 120 years ago put forward the concept of "seed" and "soil" - that cancer cells metastasize nonrandomly to sites of growth. Since then it has been suspected that tumor cells "home" to specific sites. We now know that "chemokine receptors displayed on tumor cells allow the tumor cell to follow a gradient of chemokine to its target organ milieu that expresses the ligand for this chemokine receptor" [18]. It has even been shown that glioma cells follow chemorepellent gradients produced by the cells themselves, arguably to facilitate the freeing of space for better proliferation [19].

We agree that most progeny of cancer stem cells are non-stem cells. We stated that stem cells undergo "symmetric division" (i.e. reproduce themselves) at frequency p_{s}. The rest of the time, their divisions would be asymmetric, i.e., produce a stem and a non-stem cell. We clarified this further in the manuscript by emphasizing that p_{s} is low, and defining what symmetric and asymmetric division mean. We also define *ρ* as the generational lifetime of non-stem cells, and clarify that the estimate *ρ* = 10 is used. Thank you.

Our model, originally conceived, is quite different. The chemotactic stimuli in our model are not produced by the cells themselves but pre-exist in the milieu (in the spirit of Richmond et al.'s experimental observations [18]). Additionally, the basic premise of the Keller-Segel model is that the cell responds to fluctuations in estimates of the concentration of the critical substrate, rather than to the average concentration. By contrast, we assume the cell reads the local concentration gradient deterministically, but responds stochastically.

There are *S*+1 possible sites into which a cell may move in directed fashion (including the 'null movement'). Of the *S* nontrivial movements, we weight each by the factor *cos(angle* *θ* *between the gradient*
*and the movement direction*
*for each possible movement)*, for the sake of calculating the angle-weighted relative opportunity for movement in each available direction. The *cos(θ)* are summed over the *S* available sites to get *AS* (*A* is the average cosine value). Treating the effective "*cos(θ)*" of non-movement to be 1, the total sum of cosines is *AS*+1, where *AS* corresponds to actual movement. The total angle-weighted relative opportunity for movement (to one of the *S* sites) due to directed migration is thus *AS*/(*AS*+1) (we correct a misprint) where *A* is the average of the *cos(θ)* over these sites. This value, times the gradient strength |
|, is now mapped to a probability of mobilization *G* through an arbitrary chemotactic/haptotactic responsiveness function H(z) for the cells, defined such that *H*(0) = 0 and *H* → 1 monotonically as *z* → ∞. The result is *G* = *H*{ |
|*AS*/(*AS*+1)}. *G* is the probability the gradient succeeds in mobilizing the cell, and takes into account the proportion of movement possible and the strength of the gradient. Once mobilized, where the cell then goes is weighted by the respective cosines for movement, so that *cos(θ)G/(AS)* is the probability of movement to the site at angle *θ*. The probabilities of actual movement sum to *G*, and *1-G* is the probability there is no directed movement.

These points have been added to the paper to clarify the origin of *G*. Thank you.

### Reviewer 2: Glenn Webb, Department of Mathematics, Vanderbilt University, Nashville, Tennessee, USA

This paper makes a very strong case for the importance of differential cancer stem cell proliferation vs. committed cell proliferation in tumor invasive growth, independent of directed migration effects and independent of spatial heterogeneity. The results are established using agent based cellular automata models. Can similar results be obtained with continuum partial differential equations models?

### Authors' Response

We certainly ascribe to this general view, as we have before elaborated in a recent article [9]. We extend the concept here by showing that differential cancer stem cell proliferation vs. committed cell proliferation is merely advanced further by directed migration (except in the singular case where the point of attraction is in the tumor itself) in the same general manner as was seen with undirected migration without imposition of spatial heterogeneity.

In a follow-up collaboration we have indeed recapitulated these key results using continuum partial differential equations and hope to submit this work shortly.

## Declarations

### Acknowledgements

The authors gratefully acknowledge financial support from the AACR Centennial Postdoctoral Fellowship in Cancer Research 08-40-02-ENDE (HE) and National Aeronautics and Space Administration Specialized Center of Research grant NNJ06HA28G.

As a one time exception to the publishing policy of Biology Direct the articles in this series are being published with two reviewers.

## Authors’ Affiliations

## References

- Folkman J: Tumor angiogenesis: therapeutic implications. N Engl J Med. 1971, 285 (21): 1182-6.PubMedView ArticleGoogle Scholar
- Almog N, Ma L, Raychowdhury R, Schwager C, Erber R, Short S, Hlatky L, Vajkoczy P, Huber PE, Folkman J, Abdollahi A: Transcriptional Switch of Dormant Tumors to Fast-Growing Angiogenic Phenotype. Cancer Res. 2009, 69 (3): 836-44. 10.1158/0008-5472.CAN-08-2590.PubMedView ArticleGoogle Scholar
- Hanahan D, Weinberg R: The hallmarks of cancer. Cell. 2000, 100 (1): 57-70. 10.1016/S0092-8674(00)81683-9.PubMedView ArticleGoogle Scholar
- Al-Hajj M, Wicha MS, Benito-Hernandez A, Morrison SJ, Clarke MF: Prospective identification of tumorigenic breast cancer cells. Proc Natl Acad Sci USA. 2003, 100 (7): 3983-8. 10.1073/pnas.0530291100.PubMedPubMed CentralView ArticleGoogle Scholar
- Fioriti D, Mischitelli M, Di Monaco F, Di Silverio F, Petrangeli E, Russo G, Giordano A, Pietropaolo V: Cancer stem cells in prostate adenocarcinoma: a target for new anticancer strategies. J Cell Physiol. 2008, 216 (3): 571-5. 10.1002/jcp.21493.PubMedView ArticleGoogle Scholar
- Anderson ARA, Weaver AM, Cummings PT, Quaranta V: Tumor morphology and phenotypic evolution driven by selective pressure from the microenvironment. Cell. 2006, 127 (5): 905-915. 10.1016/j.cell.2006.09.042.PubMedView ArticleGoogle Scholar
- Macklin P, Lowengrub JS: Nonlinear simulation of the effect of microenvironment on tumor growth. J Theor Biol. 2007, 245 (4): 677-704. 10.1016/j.jtbi.2006.12.004.PubMedView ArticleGoogle Scholar
- Roose T, Chapman SJ, Maini PK: Mathematical models of a vascular tumor growth. Siam Review. 2007, 49 (2): 179-208. 10.1137/S0036144504446291.View ArticleGoogle Scholar
- Enderling H, Hlatky L, Hahnfeldt P: Migration rules - tumours are conglomerates of self-metastases. Br J Cancer. 2009, 100 (12): 1917-1925. 10.1038/sj.bjc.6605071.PubMedPubMed CentralView ArticleGoogle Scholar
- Prehn RT: The inhibition of tumor growth by tumor mass. Cancer Res. 1991, 51 (1): 2-4.PubMedGoogle Scholar
- Enderling H, Anderson AR, Chaplain MA, Beheshti A, Hlatky L, Hahnfeldt P: Paradoxical dependencies of tumor dormancy and progression on basic cell kinetics. Cancer Res. 2009, 69 (22): 8814-8821. 10.1158/0008-5472.CAN-09-2115.PubMedView ArticleGoogle Scholar
- Anderson ARA, Chaplain MAJ, Rejniak KA: Single-cell-based models in biology and medicine. Birkhauser, Basel. 2007View ArticleGoogle Scholar
- Galle J, Hoffmann M, Aust G: From single cells to tissue architecture--a bottom-up approach to modelling the spatio-temporal organization of complex multi-cellular systems. J Math Biol. 2009, 58: 261-283. 10.1007/s00285-008-0172-4.PubMedView ArticleGoogle Scholar
- Enderling H, Park D, Hlatky L, Hahnfeldt P: The importance of spatial distribution of stemness and proliferation state in determining tumor radioresponse. Math Model Nat Phenom. 2009, 4 (3): 117-133. 10.1051/mmnp/20094305.View ArticleGoogle Scholar
- Maini PK, McElwain DLS, Leavesley DI: Traveling wave model to interpret a wound-healing cell migration assay for human peritoneal mesothelial cells. Tissue Eng. 2004, 10 (3-4): 475-482. 10.1089/107632704323061834.PubMedView ArticleGoogle Scholar
- Keller EF, Segel LA: Model for chemotaxis. J Theor Biol. 1971, 30 (2): 225-234. 10.1016/0022-5193(71)90050-6.PubMedView ArticleGoogle Scholar
- Keller EF, Segel LA: Travelling bands of chemotactic bacteria: a theoretical analysis. J Theor Biol. 1971, 30 (2): 235-248. 10.1016/0022-5193(71)90051-8.PubMedView ArticleGoogle Scholar
- Richmond A: CCR9 homes metastatic melanoma cells to the small bowel. Clin Cancer Res. 2008, 14 (3): 621-623. 10.1158/1078-0432.CCR-07-2235.PubMedView ArticleGoogle Scholar
- Werbowetski T, Bjerkvig R, Del Maestro RF: Evidence for a secreted chemorepellent that that directs glioma invasion. J Neurobiol. 2004, 60 (1): 71-88. 10.1002/neu.10335.PubMedView ArticleGoogle Scholar

## Copyright

This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.