K+ Channel Pore Structure: An open & shut case

The Nobel Prize in Chemistry 2003

The Paradox

• K channels selective for K+

over Na+ by >10,000 ו High energy barrier• K channels work at 108

ions/second; nearly diffusion limit

• Low energy barrier

1.33 Å

0.95 Å

How can one molecule satisfy both requirements?

Figure 1: Alignment of K channel core sequences

Why bacterial?

Does this alignment

convince you that KcsA is a good substitute

for Shaker?

The alignment is not enough: evidence that KcsA≈Shaker

Flaw?

Toxins bind specifically to the outer pore. If

KcsA≈Shaker, maybe a toxin will bind both

Induced fit: Maybe the toxin shapes the channel

A real toxin: Agitoxin from Scorpion

KcsA vs Shaker

Problem: AgTx does not bind KcsAHypothesis?

1. KcsA is nothing like Shaker2. KcsA lacks key amino acids

for AgTx binding

Solution: Change KcsA

Changing just 3 amino acids NOT in the most

conserved region allowed toxin binding

Test with a column

Scorpion venom

Mutant KcsA attached to

column Toxin binds

Most flows through

Does everyone know what a column is?

Make a column: Run what comes off on an HPLC

Missing controls?

• What about WT channel?• This is what eluted; what

flowed through?

Test Affinity

Mutant toxin does not bind Shaker, so it follows that it should not bind to mutant KcsA.

Test with a competition assay done in with radio labeled WT toxin.

Shaker

Mutant KcsA

TxmTx

mTx

mTx

Unless in the process of mutating KcsA, a whole new interaction surface was created.

How to crystallize a protein

Reservoirsolution

Slide

Pure protein in dilute

reservoir soln

Seal

Brew of salts , buffers, polymers, & detergents that is empirically determined

H2OSolution is

concentrated by diffusion

Requirement: Lots of protein

Why had this not been done

before?

Great, now to the Doyle paper: Fig 2

1. What does this depict?2. What are the red balls?3. Why are they useful4. In the crystal, what diffracts X-rays?5. How does one convert this electron

cloud into structure?

See Branden & Tooze, introduction to protein structure for a decent introduction to crystallographic analysis

Figure 3

1. Where is S5 & S6 (M1/M2)?2. Is the pore helix part of S5 or S6?3. What is the significance of the

aromatic amino acids?4. How is this shape described?

A Tee Pee

Powassan

Road trip!

Extra Credit: 5 points

286 mi

Have your picture taken in front of the tee pee

Figure 4

White OrangeRedYellow GreenPurple

Toxin binding TEA (i)Selectivity FilterTEA(o)Sulfhydryl modification: closed or openSulfhydryl modification: open

Why show all this?

Figure 5

• What type of model• What is the red & why

is that important?• The yellow?

• Space filling model• Red: anion (neg)

amino acids; attract cations (a bit).

• Hydrophobic amino acids—don’t want K stuck in the middle.

Figure 5

10 Å

1 Å

What is the implication of these widths?

1 Å is the width of a K ion. There is room for single file in the filter, but several below the filter

Figure 6

1. What is this?2. What is a & c?3. Why can they see one

& not the other?4. Interpretation?

Rb+ K+

We’ll come back to this issue

Figure 7

Lower electrostatic barriers to K crossing a membrane

“The Lake” and open area for ions

to sit

What is this (not the green ball)?

1. View?2. Where is the filter?3. What is the function of

the Y?4. Why did the other

amino acids have to be G?

5. Why are the G’s so conserved?

Figure 8

4. Big point #1: It is not the “R” groups that are required, it is the hydroxyls

5. Glycine is the only amino acid that will allow this arrangement

How does the channel discriminate large K from small

Na?1.33 Å

0.95 Å +hydration shellNa+ doesn’t fit the rigid

selectivity filter, so the energy barrier to striping off H20 is

much higher

In the pore

MacKinnon left us with a Model

His structures were of insufficient resolution to give a detailed picture of K in

the poreAlthough that sure isn’t the

impression you get from the paper

Group 1, you’re up

He increased his resolution to 2 Å with a trick:

Protein flexibility decreases resolution; using a rigid antibody & co-crystallizing the Ab-protein complex is a possible way to hold it still

Why co-crystallize with Fab complex

Fab is a fragment of IgG

Any problems with this method?

And what do you get for all this?

When crystallized in 300 mM K, the structure contains:

534 amino acids469 waters2 partial lipidsAnd how many K+ ions?

7 K+ ions!

Extracellular & intracellular K conc in vivo?

How do we come up with 7 K?

The one at the bottom is ready to enter the selectivity filter, its hydration shell is held in place by weak H-bonding to surrounding amino acids, which holds the K still

• What is the purpose of the one at the bottom?

• What are those red things?

How about the top 2?

1. At the top of the pore, those two ions are actually one, as they are too close together to co-exist

2. Half the time it’s the upper position, half the lower

2 main points:

Now it gets cool

1. So when the ion is at the top, there is room for K #1, and the next at #3

2. When the lower “top” K is present, the next K down must be at #2, then #4

How are these views different?

Now, back to Rb+: what was really going on?

Rb+K+

If you can answer this question, I will nominate you as the

“Smartest Student in UB History”

Rb+, not normally found in biology (that’s a hint), was found in only two places in the pore, as was

K+ at low concentrations. However, Rb+ is present in two places at high concentrations in the original

structure. So why the different result?

HINTS:

• At low ion concentration, they look similar

• At higher ion concentrations, they differ

• Anyone?• OK, one more:

K

Rb

This is the last hint

K+ is 1.33 Å Rb+ is 1.48 ÅThey are not quite the same

OK, so my nomination as the “Smartest Student in UB History” won’t exactly be

your ticket to fame & fortune

There is more of an energy barrier to the movement of Rb+, much less to K+.

Therefore, Rb+ will “pause” or be localized to certain regions of the selectivity filter, which is

caught by the crystal

You’d have to share credit with this person anyway

Ananya Mitra, Ph.D. (2005)

(You’d be in good company)

One more thing; remember figure 5?

300 mM KCl

2 mM KCl

• This kink or collapse of the pore is due to loss of K+ in the pore

• This narrowing is proposed as a possible gating mechanism

• Which it is not• Does anything strike

you as not making sense?

For those of you who are still dealing with the crushing blow of not being nominated as the

“Smartest Student in UB History”

Here is one more chance

We (Ok, I) said that the rigid structure was necessary for

selectivity

But if it can collapse in the absence of K+, why can’t it accommodate the lowly Na+

atom?

NaO

O O

OK

O

O O

O

There is no certain answer: the best comes from molecular dynamics

simulations

MD is a computational technique where each atom in a protein is simulated in a model with all of its physical

properties

If you want a better explanation than that, you’re

talking to the wrong guy

These would be the right guys

Electrostatic interactions are the key property

• In retrospect, this seems obvious

• No protein structure known can keep from flexing on scales of <0.1 Å as proposed for KscA

• All those negatively charged carbonyls pointing toward one another can’t be happy

• It turns out that the electrostatic interaction of the carbonyls with K+ is much more favorable than for Na+

• This is a much more important factor than geometry

O

O O

O

This explains why Na+

channel pores are not just narrower versions of K+

channel pores

Even if the geometry worked, the electrostatics

would not

The next paper is based on this paper:

1. Why bother with another one?

2. Was the KcsA channel open or closed?

Group #2, your turn

The next step: Another channel: MthK

• Ligand-gated K channel. • Gating means something applies force to a close

channel to open it. – For a voltage gated channel, it is voltage that moves the

sensor that is translated to work– For a ligand-gated channel, it is the energy of ligand

binding that is translated to work.

Some EP

What kind of channel?

Inward rectifier

What are these?

How is it gated?

Conservation with other K channels: Blocked by CTX

CTX=charybdotoxin, which blocks Shaker & Ca-gated K channels. 1 µM is a very high concentration.

What is CTX?

Simple view of structure

Pore

Cytoplasmic Ca-binding

domainsRCK

What are these?What are these?

Monomer

Disordered, so does not clearly show in the structure

What about these?

How do Ca2+ & the RCK domains control the gating?

+Ca2+

No Ca2+

• The RCK domains have big conformational shifts upon Ca binding

• This could provide the energy for gating

• Based that observation, this model was proposed.

• But there’s a big problem…• They do not know the

structure of the region between RCK & the inner pore helix!

Personally, I find this more interesting

+Ca2+No Ca2+

Closed Open

Last Paper

Again, MacKinnon gets a second full length paper in Nature to explain data he presented in the first paper

Time for Group 3 to step up

We better make sure we’re comparing closed to open

This is KcsA. Closed or Open?

How do we know?

1. Opening at the bottom too small

2. Nobody can get the C-terminal deleted form of the channel to open

3. In general, it is pretty hard to get KcsA to open under any circumstances

Figure 2

How does this rigid

helix move?

What is this?

A little easier to see open v closed

Rotates into the board

KcsA model

Figure 3

Glycine allows rotationAla does not block the pore

Which are the key amino acids & why?

Another key amino acid?

• I’ve previously mentioned that in addition to opening & closing, there is INACTIVATION

• What is inactivaition?• See any amino acids that might be involved?

Figure 4

Open channel has room for a drug or the N-type inactivation peptide

A fancy calculation belabors the obvious

Using dielectric constant values for water & lipid,

there is much less of an electrostatic barrier for the ion

to cross in an open channel

Bottom line

• The selectivity filter structure is understood in about as much detail as any dynamic structure in any protein

• We have a good idea of the difference between open & closed.

• We have a picture of how the gate might open & close, but this is far from a sure thing, even in that channel

• And it is gating that is the next frountier of these detail mongers

Since everyone has been so good:

Here’s a version of the notes with the answers