Coordination Problems in Evolution: The Rise of Eukaryotes


This is a se­ries of posts about co­or­di­na­tion prob­lems, as they ap­pear in the course of biolog­i­cal evolu­tion. It is based on the book “The Ma­jor Tran­si­tions in Evolu­tion” by John May­nard Smith and Eörs Sza­th­máry. Pre­vi­ous part, dis­cussing Ei­gen’s para­dox as well as the ori­gin of chro­mo­somes, can be found here.

In this part we are go­ing to look at the ori­gin of eu­kary­otic cell, speci­fi­cally at its ac­qui­si­tion of en­dosym­biotic or­ganel­les, and at the ori­gin of mul­ti­cel­lu­lar­ity.

Prokary­ota vs. eukryota

While all sin­gle-cel­led or­ganisms may look like similar wig­gly lit­tle crea­tures to us, there is a huge differ­ence be­tween prokary­ota (like bac­te­ria) and eu­kary­ota (like pro­to­zoa or, for that mat­ter, our own cells). The cell wall is differ­ent. The in­te­rior of the cell is differ­ent. One has rigid cell wall, in the other it’s the cy­toskele­ton that holds the cell to­gether. One has a sin­gle-ori­gin DNA strand at­tached to the cell wall, the other has nu­cleus con­tain­ing chro­mo­somes. One has mi­to­chon­dria and chloro­plasts, the other does not. Even the mechanism of cell di­vi­sion is differ­ent. If we haven’t known that we share part of the genome, it would be easy to make a mis­take and be­lieve that the life on Earth had origi­nated at two sep­a­rate oc­ca­sions.

The tran­si­tion from prokay­otes to eu­kary­otes is likely the most com­plex tran­si­tion in the en­tire course of evolu­tion. It took two billion years to hap­pen. More than the emer­gence of life it­self.

All that be­ing said, we are go­ing to look only at a sin­gle part of the evolu­tion of eu­kary­otes, namely at their ac­qui­si­tion of mi­to­chon­dria. Mi­to­chon­dria were free-liv­ing cells once. But then they’ve be­came an in­sep­a­rable part of eu­kary­otic cell. Hence, back to the co­or­di­na­tion prob­lems!

How did it come to be that some cells started liv­ing within other cells? Well, as­sum­ing that flex­ible cell wall and phago­cy­to­sis evolved be­fore the do­mes­ti­ca­tion of mi­to­chon­dria, get­ting them in­side wouldn’t be a big prob­lem. It hap­pens each time one cell eats an­other.

What’s more in­ter­est­ing is how did the guest cell sur­vive and how did the co­op­er­a­tive be­hav­ior be­tween the host and the guest evolve.


Let’s make a di­gres­sion and think about sym­bio­sis for a sec­ond. If we as­sume that there are only two strate­gies for the host (cul­ti­vate the sym­biont or try to kill it) and two strate­gies for the sym­biont (co­op­er­ate with the host or par­a­sitize) then the prob­lem gets re­duced to var­i­ants of the pris­oner’s dilemma game.

Con­sider this kind of ar­range­ment of pay­offs. The num­bers spec­ify the fit­ness of the host (left) and the sym­biont (right):

It can be eas­ily seen that there is only one equil­ibrium: What­ever the host does it’s bet­ter for the sym­biont to par­a­sitize. And if the sym­biont is a par­a­site it’s always bet­ter for the host to kill it.

Un­der what con­di­tions do we see this kind of game? The au­thors point out that this hap­pens when each in­di­vi­d­ual host ac­quires some ge­net­i­cally differ­ent sym­bionts from the en­vi­ron­ment. The rea­son is that it doesn’t pay for the sym­biont to in­vest in the co­op­er­a­tion with the host if the host is go­ing to be kil­led by a differ­ent sym­biont any­way.

How about a differ­ent sce­nario?

The ideal strat­egy for the sym­biont is not clear in this case. If the host is cul­ti­va­tor it may pay to the sym­biont to co­op­er­ate. If, on the other hand, the host tries to kill it, the best thing for the sym­biont to do would be to mul­ti­ply as fast as pos­si­ble, re­gard­less to any dam­age to the host.

This kind of setup is ex­pected if each host ac­quires only a sin­gle sym­biont from the en­vi­ron­ment:

How­ever, with hosts in­fected by a sin­gle sym­biont, co­op­er­a­tive mu­tu­al­ism is likely to be sta­ble once it evolves. The evolu­tion from par­a­sitism to mu­tu­al­ism will be fa­vored if the hosts kil­ling re­sponse is in­effec­tive, and if the fur­ther spread of the sym­biont is greater if the host does sur­vive. It will not oc­cur if the host can rapidly rid it­self of the par­a­site, or if the par­a­site spreads only by kil­ling the host.

Fi­nally, let’s have a look at the the fol­low­ing sce­nario:

Again, there’s only one equil­ibrium. It’s always bet­ter for the sym­biont to co­op­er­ate and once it’s co­op­er­at­ing, it’s bet­ter for the host to cul­ti­vate it.

This hap­pens when the host ac­quires one or a few sym­bionts from one of its par­ents.

It makes sense: If the only place you can dis­perse to are your host’s chil­dren you re­ally don’t want to kill it.

So there’s a rule of thumb emerg­ing here: If the trans­mis­sion of the sym­biont hap­pens be­tween un­re­lated in­di­vi­d­u­als (hori­zon­tal trans­mis­sion) the sym­bio­sis will evolve to­wards par­a­sitism. If the sym­biont is passed only from one par­ent to its chil­dren, then the re­la­tion­ship will evolve to­wards mu­tu­al­ism.

In fact, both ex­per­i­ments and ob­ser­va­tions in the wild show that ver­ti­cal trans­mis­sion of the sym­biont leads to mu­tu­al­ism and hori­zon­tal trans­mis­sion leads to par­a­sitsm. There are some ex­cep­tions though. For ex­am­ple, the trans­mis­sion of lu­mi­nous bac­te­ria in deep-see fish is hori­zon­tal, yet the sym­bionts are es­sen­tial to the sur­vival of their hosts.

Par­a­sites or live­stock?

Now, let’s get back to the ori­gin of eu­kary­otic cell. What was the re­la­tion­ship be­tween the early host cells and early mi­to­chon­dria?

It may have been that the mi­to­chon­dria were par­a­sites. Maybe they some­times es­caped the host cell and in­fected differ­ent cells. How­ever, the au­thors hint at an in­ter­est­ing al­ter­na­tive: The host cells may have farmed the mi­to­chon­dria for the later con­sump­tion, just like we do with the cat­tle.

One im­por­tant point to un­der­stand here is that, how­ever we feel about slaugh­ter­ing cows, from the pop­u­la­tion gener­ics point of view it’s a mu­tu­ally benefi­cial ar­range­ment. Homo Sapi­ens gets steaks. Bos Tau­rus be­comes one of the most com­mon ter­res­trial ver­te­brates around.

So, the host cells may have first con­sumed mi­to­chon­dria, but then learned to keep them around (or rather in­side) so that they can be con­sumed later.

And we do see some ev­i­dence that the host cell adopts ac­tive mea­sures to keep the re­la­tion­ship mu­tu­al­is­tic. In sex­u­ally re­pro­duc­ing species the trans­mis­sion of mi­to­chon­dria hap­pens from one par­ent only. When hu­man egg merges with hu­man sperm, all the mi­to­chon­dria from the sperm are dis­carded and only those from the egg make their way into the em­bryo. That, ac­cord­ing to the model de­scribed above, pre­vents com­pe­ti­tion be­tween the differ­ent strains of mi­to­chon­dria at the ex­pense of the host cell.

Later on in the evolu­tion, straight­for­ward con­sump­tion of the sym­bionts must have been re­placed by pro­tein “taps” that we see in­stalled into the cell wall of mi­to­chon­dria to­day. The taps al­low the nu­tri­ents pro­duced by the mi­to­chon­dria to flow into the host cell. Think of Maa­sai punc­tur­ing the flesh of a cow and drink­ing the blood with­out kil­ling the an­i­mal. The fact that the tap pro­tein is always en­coded in the DNA of the host cell rather than in mi­to­chon­drial DNA is a hint that the idea of the host cells “farm­ing” mi­to­chon­dria may not be that im­plau­si­ble.

Gene trans­fer to the nucleus

Once the mi­to­chon­dria were liv­ing in­side the host cell a cu­ri­ous pro­cess be­gan. The genes from the mi­to­chon­dria started “jump­ing” into the host cell’s nu­cleus.

By los­ing their genes, mi­to­chon­dria lost the chance to break free from the eu­kary­otic cell for good. So why did it hap­pen? And who benefited?

I re­ally like this pro­cess be­cause it shows how com­plex the in­ter­play be­tween differ­ent lev­els of se­lec­tion can be. In par­tic­u­lar, we have to do with three dis­tinct lev­els of se­lec­tion here: Selec­tion on the level of the host cell, se­lec­tion on the level of mi­to­chon­dria and se­lec­tion on the level of a sin­gle mi­to­chon­drial gene.

First, we can imag­ine a mi­to­chon­drial gene get­ting at­tached to a nu­clear chro­mo­some. It would be clearly ad­van­ta­geous for the gene: One more copy! Hooray! What’s not to like?

But why didn’t the gene got dis­carded from the nu­clear DNA given that it performed no use­ful func­tion? Well, it turns out that nu­clear DNA can con­tain hu­mungous amounts of dead code and yet the code doesn’t get dis­carded by nat­u­ral se­lec­tion. Con­trast that with prokary­otes which tend to keep their ge­netic code short, sweet and stream­lined.

But wait. The gene would still be trans­lated into pro­tein. That would be a use­less ex­pen­di­ture of en­ergy and thus it would be se­lected against. To make it ad­van­ta­geous for the cell there must have been a mechanism to trans­port the pro­tein back into mi­to­chon­dria. Luck­ily, all that is needed for that is to add to the pro­tein a “tran­sit pep­tide”, a han­dle which would be rec­og­nized by a re­cep­tor in the mi­to­chon­drial mem­brane and used to carry the pro­tein in­side. Creat­ing such a han­dle is easy. Baker & Schatz pasted ran­domly cho­sen pieces of E. coli and mouse DNA in front of pro­tein genes, and found that 2.7 per cent of the bac­te­rial in­serts and 5 per cent of the mam­malian ones were suc­cess­ful tran­sit pep­tides.

Another hint that the tran­si­tion may be easy is that there is no dis­tinct pat­tern to the tran­sit pep­tides. In other words, the tran­sit pep­tides prob­a­bly evolved 700 times in­de­pen­dently — once for each mi­to­chon­drial gene that was trans­ferred into the nu­cleus.

When the gene is in nu­cleus and the pep­tide han­dle in its place, the mi­to­chon­dria can gain ad­van­tage by dis­card­ing the genes that they don’t need any­more (they are im­port­ing those pro­teins from the out­side any­way). And, as already men­tioned, prokary­otes are very good at strip­ping their genome of any un­nec­es­sary bag­gage. Nick Lane does some back-of-the-en­volope calcu­la­tions and con­cludes that the en­ergy sav­ings are truly huge.

All of that be­ing the case, the ques­tion is rather why all the genes haven’t been trans­ferred to the nu­cleus.

Why the gene trans­fer stopped

For mi­to­chon­dria, the pro­cess was stopped by the change in mi­to­chon­drial ge­netic code (see only kind-of-re­lated but fun-to-read column by Dou­glas Hofs­tadter). As soon as one of the mi­to­chon­drial codons be­gan cod­ing for a differ­ent amino acid, the genes could no longer jump to the nu­cleus. When they did they were turned into defec­tive pro­teins by the old, un­mod­ified nu­clear trans­la­tion ma­chin­ery.

But that can’t be the en­tire story. The chloro­plast genes are en­coded in the plain old generic code. Yet the chloro­plasts still keep some of the genes for them­selves. This may (or may not) in­di­cate that there’s still some level of sep­a­rate iden­tity to the or­ganel­les, that they may have goals of their own not fully al­igned with the goals of the en­clos­ing eu­kary­otic cell. An ex­am­ple of that would be, for ex­am­ple, mi­to­chon­dria try­ing to dis­tort the sex ra­tio of the host species.

(As a side note, there are or­ganel­les called per­ox­i­somes that were once thought to be en­dosym­bionts, very much like mi­to­chon­dria. Ex­cept that they had no genes at all. It has been sug­gested that they may be en­dosym­bionts that have trans­ferred all of their genes to the nu­cleus. How­ever, that idea has been re­cently challenged.)

Mul­tic­ul­lu­lar life and Orgel’s sec­ond rule

Mul­ti­cel­lu­lar life sure looks like it has a co­or­di­na­tion prob­lem. All those billions and trillions of cells have to co­op­er­ate some­how. Most of them have to give up in­di­vi­d­ual re­pro­duc­tion and rather work for the benefit of all. Hell, there’s even pro­grammed cell death where the cell is ex­pected to will­ingly die when there’s no use for it any more.

But when you take a step back the ar­gu­ment doesn’t make sense. All those cells are ge­net­i­cally iden­ti­cal. There aren’t mul­ti­ple en­tities en­gag­ing in a co­or­di­na­tion prob­lem. There’s just one en­tity: The multi-cel­lu­lar or­ganism it­self.

Or is there?

It may be in­struc­tive to pause here for a while and do an ex­er­cise in evolu­tion­ary think­ing.

Con­sider what hap­pens if a so­matic cell mu­tates.

The mu­ta­tion may cause the cell to di­vide in un­reg­u­lated man­ner.

But there’s even more in­trigu­ing pos­si­bil­ity: Imag­ine that a the cell mu­tates in such a way that it’s more likely to give rise to a germ cell. For ex­am­ple, a plant cell that would oth­er­wise pro­duce a leaf would give rise to a flower in­stead. By do­ing so it would lower the over­all fit­ness of the or­ganism: The plant would now have less leaves than what’s op­ti­mal. How­ever, at the same time the rene­gade cell would in­crease its own fit­ness be­cause any pol­len or seed pro­duced by that flower would carry the mu­tated gene in­stead of the origi­nal one.

So what do you think? Does the above make sense or does it not? Is it re­ally an in­tra-or­ganism con­flict? Think about that. I’ll wait.

Smith and Sza­th­máry ap­proach the prob­lem by split­ting it into two parts.

First, they dis­cuss whether mu­ta­tion that in­creases a chance of giv­ing rise to a ga­mete cre­ates an in­ter­nal con­flict and, con­se­quently, se­lec­tion pres­sure for other cells to evolve a mechanism to pre­vent such mu­ta­tions. They con­clude that it doesn’t. After all, this is not much differ­ent from when the mu­ta­tion oc­curs in a germ cell. The child will be slightly ge­net­i­cally differ­ent from the par­ent, but that’s just how evolu­tion works. If the child hap­pens to be more fit than the par­ent, it will even­tu­ally pre­vail in the com­pe­ti­tion on the or­ganism level. If not so, the new strain will be elimi­nated by the nat­u­ral se­lec­tion.

The sec­ond part of the ques­tion is what hap­pens if the mu­tant is ma­lig­nant, i.e. if it causes un­reg­u­lated cell pro­lifer­a­tion. We call that can­cer. In that case, the au­thors con­clude, there will be an ac­tual se­lec­tive force to pre­vent, or de­lay, the ma­lig­nancy.

Have you got that right? If not so, don’t be dis­ap­pointed and think about Orgel’s sec­ond rule. The rule says: “Evolu­tion is clev­erer than you are.”

If you want to re­mem­ber just one thing about evolu­tion, Orgel’s sec­ond rule may be a good choice.

In fact, Smith and Sza­th­máry, as good evolu­tion­ary biol­o­gists, have the rule in­grained and con­clude the sec­tion by hedg­ing their bets:

There is, there­fore, no rea­son to think that [spe­cific mechanisms dis­cussed in the book] evolved to sup­press cell-cell com­pe­ti­tion. But the ques­tion is im­por­tant, and we do not re­gard our ar­gu­ments as de­ci­sive.

To be continued

In the fol­low­ing parts I will cover the ori­gin of sex and of so­cial species. In the end I am go­ing to spec­u­late about pos­si­ble par­allels be­tween co­or­di­na­tion prob­lems in evolu­tion and co­or­di­na­tion prob­lems in hu­man so­ciety.