A chronology of multicellularity evolution in cyanobacteria 3 4 5 6 7

Abstract The transition from unicellular to multicellular organisms is one of the most significant events in the history of life. Key to this process is the emergence of Darwinian individuality at the higher level: groups must become single entities capable of reproduction for selection to shape their evolution1–4. Evolutionary transitions in individuality are characterized by cooperation between the lower level entities5–7 and by division of labour8,9. Theory suggests that division of labour may drive the transition to multicellularity by eliminating the trade-off between two incompatible processes that cannot be performed simultaneously in one cell1,9,10. Here we examine the evolution of the most ancient multicellular transition known today, that of cyanobacteria11,12. We developed a novel approach for the precedence polarization of phenotypic traits that employs gene phylogenies and does not require a species tree. Applying our procedure to cyanobacterial genomes we reconstruct the chronology of ecological and phenotypic trait evolution in cyanobacteria. Our results show that the prime driver of multicellularity in cyanobacteria was the expansion in metabolic capacity offered by nitrogen fixation, which was accompanied by the emergence of the filamentous morphology and a reproductive life cycle. This was followed by a range of niche expansions and interactions with other species, and the progression of multicellularity into higher complexity in the form of differentiated cells and patterned multicellularity.


Introduction
combined forms of nitrogen are scarce in most environments (e.g., open oceans or 118 terrestrial habitats 21 ). Combined nitrogen, which is critical for the biosynthesis of amino 119 and nucleic acids, was likely a limiting resource in the early Earth environment 22 . Hence, 120 the capability of N FIXATION was key for cyanobacterial radiation into new habitats and 121 subsequent diversification (phases ii and iii). 122 The realization of the full metabolic potential of N2 fixation, however, faced the 123 challenge of the incompatibility of nitrogenase with intracellular oxygen 23  When compared to the more transient associations in spatially structured communities, 139 such as in extracellular polymeric substance (EPS) imbedded biofilms, the development 140 of filaments opens possibilities for a more direct exchange of molecules with high 141 specificity. Metabolic exchange could have evolved as described for the evolution of metabolic cross-feeding 26 , as the exchange of carbon and nitrogen against other 143 products is generally common in photosynthetic or nitrogen-fixing organisms 27   The data underlying this study consists of the genomic sequences and phenotypic traits 211 of 199 cyanobacterial species. These were selected from the available genomes so that 212 the number of represented taxa will be as large as possible and genus-level redundancy 213 will be reduced (See Supplementary Table 1 for the complete list of species).

Phenotypic traits 215
Phenotypic traits were chosen for their potential relevance to the evolution of 216 multicellularity in cyanobacteria, such as environmental factors that might facilitate 217 multicellularity and markers that are indicative for the transition to multicellularity 218 (Table 1) The pairwise polarity inference is applied to each pair of traits and summarized 295 in a trait-pair polarity matrix. We again apply the FDR correction, this time over all trait-296 pairs and polarity contrasts. In the present study, with 25 traits, we apply FDR over 297 3,480 tests. To derive an ordering of the traits, we apply 'Topological sort' 60 to the 298 significant polarities of type 'A precedes B', or vice versa. In the present study the 299 significant polarities form a partial order, i.e., there are no self-contradicting precedence 300 cycles, and the topological sorting order is used to order the polarities matrix and to 301 reduce it to a feed-forward network (Fig. 2).

HORMOGONIA*
Motile reproductive cells that result from repeated rounds of fission without intermittent growth phases. They break of the mother filament, ensuring the reproduction and dispersal of benthic species. Necridia* Dead cells resulting from PCD for hormogonia release.

HETEROCYSTS*
Thick-walled cells that are specialized in fixing N2.

BAEOCYTES*
Reproductive cells that result from repeated rounds of fission without intermittent growth phases.

Morphological and physiological traits UNICELLULAR
Single-celled morphology. After cell division cells separate.

FILAMENTOUS*
Multi-celled morphology. Cells remain attached after cell division.

NITROGEN FIXATION
Fixation of N2 into ammonium.

SHEATH
Part of the cell envelope, located outside the cell wall.

MUCILAGE
Part of the envelope, located outside the cell wall, comprised of EPS, without a defined structure. GAS VESICLES* Intracellular gas-filled chambers for regulating buoyancy in the water column.

MOTILITY
Movement across surfaces or through a liquid medium.

PLANES
Cell division in two or three perpendicular planes.

TRUE BRANCHING*
Fission in multiple planes leads to branches that remain attached to the main filament.

Habitat and life style FRESHWATER
Aquatic environments with salinity between 0-0.5ppt.

PLANKTONIC
Organism that lives in the plankton (not attached).

SESSILE/ BENTHIC
Attachment to a substrate.

MATS
Growth inside thick, laminated, microbial structures.

FREE-LIVING
Organism that lives autonomously, in contrast to:

NOT FREE-LIVING
Organism that lives in a symbiotic relationship. EPILITHIC/ ENDOLITHIC Growth on or inside rocky substrates.

EPIPHYTIC
Growth on plants.

PERIPHYTIC
Attachment to underwater substrates.
* Multicellularity markers: traits that are adaptations on the level of the filament.

457
SMALL CAPS indicate the traits that have been used in the analysis.

N2 fixation -No N2 fixation
Whereas some studies claimed the last cyanobacterial common ancestor to fix N2 45 , there are others that concluded that it could not fix N2 and that cyanobacteria must have acquired this trait several times independently 42,46,47 .
Trait-pair polarity tests show that N FIXATION is a derived trait and that the ancestor of cyanobacteria was lacking the ability to fix N2 (FDR adjusted U-test pvalue 2.6×10 -125 ).

Freshwater -Marine
Some studies suggest that early cyanobacteria lived in freshwater and subsequently diverged into marine environments 39,42,44 , whereas others provide evidence in support of a marine origin 48 .
The cyanobacteria ancestor most likely inhabited an aquatic environment and colonized both environments early.
Our results show that there is no evidence for either MARINE or FRESHWATER environments as ancestral or derived habitat (Simultaneous polarity with FDR adjusted U-test p-value 5.4×10 -148 ).

Akinetes -Heterocysts
There is a common agreement that these cell types appeared late in cyanobacterial evolution 49 , but there is a controversy about whether they shared a common ancestor and appeared simultaneously 50 or successively 42,49 .
Our results show that AKINETES and HETEROCYSTS emerged simultaneously (FDR adjusted U-test p-value 1.8×10 -124 ). polarities. Thus, the traits form a partial order that is used to determine the trait order in 480 the matrix (a) and is visualized as a feed-forward network (b). Shades of green mark the 481