Nanoscale analysis of

structural synaptic plasticity

Kristen M. Harris

Department of Neuroscience

University of Texas at Austin

synapseweb.clm.utexas.edu

NSE 2016

Hippocampal CA1 Pyramidal Cell

Neuron

Cell Body

Dendrites

DendritesDendritic

Spine

Axonal

Bouton

Adult Rat

Rapid Golgi preparation (Harris, 1980)

Axons

Dendritic Spines in Hippocampal CA1 Neuropil

1 micron

PSD

Glia

Axon

Spine

Dendrite

Glia

Adult Rat

Dendritic Spines in Hippocampal CA1 Neuropil

1 micron

PSD

Glia

Axon

Dendrite

Glia

Adult Rat

Dendritic Spines in Hippocampal Neuropil

SER

Spine

Apparatus

dcv

0.5 m

Do synapse size and composition

reflect function?

John Fiala

http://synapseweb.clm.utexas.edu > Tools > Software

Reconstruct

Introductory Cruisethrough

Hippocampal CA1

neuropil

Credits

University of Texas at Austin

Josef Spacek, Kristen Harris, Larry Lindsey, Patrick Parker

Chandra Bajaj, Jarred Bowden

The Salk Institute

Justin Kinney*, Tom Bartol, Terry Sejnowski

Dan Keller, Varum Chaturvedi

3D – Blender

Music by: Camille Saint Saëns, Carnival of the Animals, Aquarium

Josef Spacek

Charles University

Hradec Kralovè

Czech Republic

CA1 densely reconstructed volume.

Perfused 1984(~180 µm3)

Harris et al. (2015) (Nature) Sci. Data

~500 (498) synapses in volume = 1 RBC!!

(~180 µm3) Perfused 1984

Harris et al. (2015) (Nature) Sci. Data

Learning from 3DEM and realistic

MCell models.

Tom Bartol, Cailey Bromer,

Terry Sejnowski, Salk Institute

Mary Kennedy, CalTech

Bartol et al. (2015) Nanoconnectomic upper bound... eLife: 10.7554.

Bartol et al. (2015) Computational reconstitution of spine calcium... Front. Syn. Neu.:10.3389

Nature’s experiment provides coupled

spines with “same” activation histories.

Bartol et al. 2015b, eLIFE

Axon-coupled spines have nearly

equal head and synapse sizes.

Enhanced Cell Blender Axon-coupled spines

Fig. 3: Coupled spines

have nearly equal head

sizes. Variation across

population gives ~4.7

storage bits per spine.

Bartol et al. 2015, eLIFE

CV (0.083)

Uncoupled spines show no correlations.

Non-axon-coupled

different dendritesNot different from random

Bartol et al. 2015, eLIFE

26 distinct intervals in synapse size =

4.7 information bits available to CA1 synapses.

• CV (0.083) per distribution matches axon-coupled same dendrite spines.

• Intervals spaced to achieve 31% overlap with adjacent intervals.

• Gives 69% between peaks (discrimination threshold of information theory).

60 – fold range in size

Bartol et al. 2015b, eLIFE

MCell reconstitution models reveal fast Ca+2

transients missed by Ca+2 indicator dyes.

Small Spine 6x greater

[Ca2+] thanlarge spine

Bartol et al. 2015, Front. Synaptic Neuroscience

Conclusion 1

• Synapse size is a reasonable predictor of strength.

• Synapses are not simple on/off switches, but graded responders.

www.openconnectomeproject.org/synapseweb

Harris et al. (2015)

Scientific Data doi:

10.1038/sdata.2015.46

Control and Test pulses

at 1 pulse per 2 min

From looking and modeling to experiments.

Theta-Burst Stimulation

(TBS) – 8 trains

S1

S2

0

50

100

150

200

250

300

350

-40 -20 0 20 40 60 80 100 120

% B

as

elin

e

Time (min)

Time series 3DEM of LTP

ControlLTP

Bell et al. (2014)

Bourne and Harris (2011)

Jensen and Harris (1989)

Microwave-enhanced fix at key times.

5 min 30 min 2 hr

Chair, Dept. Neurology

Perelman Sch. Med. Univ. Penn.

Frances Jensen

Jen Bourne

Univ. Colo.

Denver

TBS

Beth Bell

EA, Austin

~120 µm from LTP or Control Stimulating Electrodes

Investigate regions with

differential activation.500 µm

70 µm

CODED:

HCSBRUnbiased

length

Dendrites with TBS-LTP had fewer

spines with larger PSDs than controls.

Control TBS-LTP

50th

percentile

dendrites

At 2 hours,

but not at

5 or 30 min.

Conclusion 2

Many small

synapses

Fewer large

synapses

Adult dendrites have limited total synaptic input.

Control TBS-LTP

Equal total PSD area across

time and conditions.

50

100

150

200

250

-40 -20 0 20 40 60 80 100 120

% B

as

eli

ne

Time (min)

What structural changes underlie

‘silent’ growth of the PSD?

Control

LTP

TBS

Synapse strength

does not increase

beyond the first 5

minutes after TBS.

Typical view of a spine synapse with

thickened postsynaptic density (PSD).

50-60 nm

Section

Thickness

250 nm

AZ = PSD across from the active zone with vesicles.

Serial sections through that PSD.

AZ = PSD across from the active zone with vesicles.

250 nm

Nascent Zone (NZ) revealed in serial

sections through the PSD.

250 nm

NZ = PSD across from no presynaptic vesicles.

Nascent Zone (NZ) revealed in serial

sections through the PSD.

250 nm

NZ = PSD across from no presynaptic vesicles.

Nascent Zone (NZ) revealed in serial

sections through the PSD.

250 nm

NZ = PSD across from no presynaptic vesicles.

Role of Nascent zones in

“silent” PSD growth.

Bell, et al., (2014)

Spacek and Harris, (1998)

Beth Bell

Josef Spacek

Arial ViewLateral View

NZ AZ

Silent synapse growth by addition of

nascent zones?

Stochastic M-Cell modeling: Probability of AMPA-type channel opening falls

off exponentially from 0.4 at the center of release to 0.1 at 200 nm away.

(Franks et al., 2002, 03)

Small dense core vesicles (DCV) carry

presynaptic active zone proteins.

DCVs transport AZ proteins

(Piccolo, Bassoon)DCV spicules

Sorra, Mishra, Kirov, Harris (2006)

Karin Sorra

Arroscience

DCV in Transport Packet

1 µm

(C. Garner, N. Ziv and Others)

Surprisingly, DCVs occur throughout

mature CA1 axons.

Sorra, Mishra, Kirov, Harris (2006)

1 µm3

DCVs are recruited to presynpatic

boutons by 5 minutes after TBS.

Return to control levels

by 30 min.

Sometimes, DCVs ‘inserting’

near edge of the PSD.

Bell et al., (2014)

1 µm

Addition of presynaptic vesicles converts

NZs to AZs, with no change in PSD area.

NZ: LTP < ControlPSD area:

LTP = Control

Control TBS-LTP

Growth of nascent zones is primary source

of postsynaptic growth in PSD area.

PSD: LTP > ControlNZ: LTP > Control

50

100

150

200

250

-40 -20 0 20 40 60 80 100 120

% B

as

eli

ne

Time (min)

Purpose of the silent

synapse growth following LTP?

Control

LTP

TBS

Synapse strength

does not increase

beyond the first 5

minutes after TBS.

LTP Saturation lasts >1 HR (LE Hoods)

No Augmentation - 30 Minutes No Augmentation - 1 hr

Cao and Harris, J. Neurophys. 2014

Adult Adult

Reliable augmentation of LTP

by 4 hours.

Third TBS at 4 hours

produces augmentation.

Augmentation begins at

90 minutes

Cao and Harris, 2014

Conclusion 3: Structural mechanism for

advantage of spaced vs massed learning.

Augment LTPSaturates LTP8 TBS

by 5 min by 30 min by 2 hours

NZ conversion to AZ underlies LTP.

Silent NZ growth prepares synapse for

enhanced LTP augmentation and learning.

Improving Electron Microscopy:

Scanning Electron microscopy in

the Transmission mode (tSEM)

Kuwajima , Mendenhall , Lindsey Harris (2013) PLOS 1

John MendenhallMasa Kuwajima

Neuropil in tSEM at

nanometer scale

Kuwajima M, Mendenhall JM, Lindsey LF, Harris KM (2013)Randy Chitwood, UT-Austin

Standard

TEM field

tSEM

field

Kuwajima , Mendenhall , Lindsey Harris (2013)

Rationale for expanding tSEM.

• Large fields at high resolution with little or no

montaging.

• Low dosage kV and small beam size produce

little or no section shrinkage during imaging.

• Automated image collection across serial

sections at <2 nm x-y resolution.

• Transmission retains all information

throughout the section thickness.

Bottlenecks and improvement.

• Fragile thin sections.

• Disambiguation of obliquely sectioned objects

having borders within thin sections.

• Alignment of large tSEM fields.

Thicker sections are less fragile but

have more ambiguous membranes.

150 nm

ves?

ecs?

sa?

mt?

Tomography reveals buried structures in

15 nm virtual sections.

Thicker sections plus tomography solution.

sasa

Automated large field alignments

being solved!

Wetzel, Bartol, Sejnowski

Large field alignment, yellow box

– prior size limit.

Conclusion 4

• The future is bright for experimental 3DEM!

0.5 m

Current Lab:

Lyndsey Kirk

Masa Kuwajima

John Mendenhall

Patrick Parker

Clayton Smith

Seth Weisberg

Corey Haines

Paola Gonzolez

Marshall Drake

Dakota Hankama

Sindy Ventura

Dusten Hubbard

Kate Dempsey

Alyssa Norbert

Thanks!

Sergei Kirov

Yelena Kulik

Anusha Mishra

Linnaea Ostroff

Lara Petrak

Chris Risher

David Selig

Gordon Shepherd

Bitao Shi

Heather Smith

Karin Sorra

Beatrice Tsao

Rachel Ventura

Mikayla Waters

Deborah Watson

Mark Witcher

NS21184, MH095980, NS074644,

Kavli and Brain Research Foundations

Texas Emerging Technologies Fund

Collaborators:

Josef Spacek

Cliff Abraham

Terry Sejnowski

Tom Bartol

Mary Kennedy

Cailey Bromer

Chandrajit Bajaj

Randy Chitwood

Dan Johnston

Boris Zemelman

Joshua Vogelstein

Randal Burns

Alex Baden

Former:

Jennifer Bourne

Beth Bell

Jared Bowden

Albert Cardona

Guan Cao

Marina Chicurel

Michael Chirillo

Mitya Chklovskii

James Cooney

Mike Ehlers

John Fiala

Alex Goddard

Jamie Hurlburt

Frances Jensen

Paul Jackson

Larry Lindsey

Olga Ostrovsky (3/17)

Max Snodderly

Guan’s Goodbye Party