four calcium event clusters and for all events
(Fig. 1, I and J).
We quantified the anatomical-functional
correlations using three measures: First, we
performed principal component analysis (PCA)
to embed the activity of each ROI across time
and trials into a two-dimensional space of the
two leading components that explain 84.42 ±
9.67% of the variance. Second, we used a
Mantel test ( 38 – 40 ), which quantifies the degree
of correlation between the functional activ-
ity and structural distance matrices. Third, we
SCIENCEscience.org 15 APRIL 2022•VOL 376 ISSUE 6590 269
Fig. 2. Compartmentalized
activity in tuft dendrites of
type 1 thick-tufted layer 5
PTNs during running on
treadmill and hand reach.
(A) Calcium events histogram
from one type 1 PTN during
treadmill session. Arrows
indicate values separating the
four different event clusters.
(B) Examples of calcium
events (DF/Fheatmap) from
the four event clusters in type
1 PTN. ROIs are arranged by
the tree structure as indi-
cated by the dendrogram,
shown on the left (R, green; L,
orange). (C) Two-dimensional
tree diagram (left) and the
corresponding structural dis-
tance matrix and dendrogram
(right) of a type 1 PTN. Dots
represent recorded ROIs.
(D) Two-dimensional PCA
embedding of all ROIs activ-
ity; each dot represents a
single ROI. (E) (Top) Matrices
showing pairwise Pearson
correlation coefficients com-
puted from the calcium
signals arranged by the tree
structure shown in (C).
(Bottom) Pearson correlation
values as a function of
shortest path distance fitted
with a linear regression
model. [(A) to (E)] Same
neuron and session. ROIs
compared within left hemi-
tree (orange); within right
hemi-tree (green) and
between R/L hemi-trees
(red). Black line represents
linear regression model fit.
(FtoH) As in (C) to (E) for
a different type 1 PTN during
a hand reach session. (Ito
M) Box plots of the following
parameters: Mantel statistics
comparing Pearson correla-
tion and structural distance
matrices (I),R^2 of linear
regression model that pre-
dicted Pearson correlations
by distance (J), slope of linear regression model that predicted Pearson correlations by distance (K),R^2 of linear regression model that predicted Pearson correlations
by distance calculated for the hemi-trees separately (L), and Z-score of the experimental Pearson correlations of within compared to between hemi-trees (mean
Pearsonwithin−mean Pearsonbetween) calculated in relation to the shuffled distribution (M). p< 0.05, p< 0.01, p< 0.001; blue asterisk, mean value. One-way
analysis of variance (ANOVA) with Tukey post-hoc test (12 neurons, 8 animals, 27 sessions).
Calcium events histogram
0
10
20
30
40
50
60
0 0.5 1 1.5 2 2.5 3
F/F
Events coun
t
A B
Type 1 PTNs
Treadmill
#
Linear regression model
0
0.2
0.4
0.6
0.8
1
R²
1234
Event cluster
*** **
***
**
J
Linear regression model
Event cluster #
-1.2
-1
-0.8
-0.6
-0.4
-0.2
0
1234
Slope
**** ***
***
I K
2
4
6
8
#
***
Activity matrix vs. distance matrix
0
0.
0.
0.
0.
1
Mantel
1234
Event cluster
**
***
***
Linear regression model
within hemi-tree
0
0.2
0.4
0.6
0.8
1
R²
1234
Event cluster #
*
L
Left
Right
Hand reach
0 0.2 0.3 0.4
-0.3
-0.1
0.1
0.3
0.5
0.1
PC1
PC2
Distance (μm)
F
G
Structural distance matrix H Pearson correlation
R² =0.314, Slope=-0.378
200 600 1000
-0.2
0.2
0.6
1
Correlation
Event cluster 1
ROIs
(^111) ROIs 2
21
R²=0.404, Slope=-0.415
-0.2 200 600 1000
0.2
0.6
1
Event cluster 2
(^1112)
21
R²=0.639, Slope=-0.4
-0.2 200 600 1000
0.2
0.6
1
Event cluster 3
112
1
21
R²=0.13, Slope=-0.104
200 600 1000
0.2
0.6
1
-0.2
0
Event cluster 4 1
112
1
(^021)
900
ROIs (μm)
ROIs
1121
21
Mantel =0.575, p< 0.001 Mantel =0.638, p< 0.001 Mantel =0.8, p< 0.001 Mantel =0.36, p< 0.001
-0.4 0 0.10.20.30.40.50.6
-0.2
0
0.2
0.4
PC1
PC2
Distance (μm)
C
D
Structural distance matrix E Pearson correlation
R² =0.146, Slope=-0.208
200 600 10001400
-0.6
-0.2
0.2
0.6
1
-1
Correlation
Event cluster 1
ROIs
(^119) ROIs 1
19
R²=0.779, Slope=-0.836
200 600 10001400
-0.2
0.2
0.6
1
Event cluster 2
(^1191)
19
R²=0.839, Slope=-0.942
200 600 10001400
-0.2
0.2
0.6
1
Event cluster 3
(^1191)
19
R²=0.136, Slope=-0.054
200 600 10001400
0.2
0.6
1
-0.2
0
Event cluster 4 1
(^1191)
(^019)
1200
ROIs (μm)
ROIs
1911
19
Mantel =0.34, p< 0.001 Mantel =0.875, p< 0.001 Mantel =0.926, p< 0.001 Mantel =0.368, p< 0.001
Examples of events by cluster
Cluster 2
1 234
Trial #
Cluster 3
123 4
Trial #
Cluster 4
1234
Trial #
-0.3
F/F 4
Cluster 1
123 4
Trial #
M
Within-between permutation
test
**
Event cluster #
-2
2
4
6
10
12
Z-score
1234
8
14
0
RESEARCH | RESEARCH ARTICLE