Table 1. Summary of the nanomechanical analysis of NPs.

Protein			n	NM (%)	M (%)	hM (%)
PolyQ	Q19		111	100	0	0
	Q35		100	95.0	5.0 (4)	1
	Q62	−QBP1	107	92.5	7.5 (5)	2.8
		+QBP1	124	96.7	3.3 (3)	0
VAMP2	I27-VAMP2		188	100	0	0
	Ubi-VAMP2		234	100	0	0
Aβ	Aβ42	−SV111	116	67.2	32.8 (9)	0
		+SV111	128	69.8	30.2 (8)	0.8
	Arc Aβ42	−QBP1	102	37.9	62.1 (10)	1
		+QBP1	108	42.6	57.4 (9)	0
	F19S/L34P Aβ42		396	100	0	0
α-syn	wt		98	55.1	44.9 (10)	2±1
	A30P		92	41.0	59.0 (10)	3.3
	A53T	−QBP1	96	39.6	60.4 (10)	5.2±1.0
		+QBP1	259	86.5	13.5 (4)	0.4
Sup35NM		−QBP1	100	36.0	64.0 (10)	8
		+QBP1	100	73.0	27.0 (9)	1

SMFS experiments were performed on pFS-2 polyprotein constructs, although the data reported only refer to the guest NPs. The calculation of the associated experimental errors is described in Text S1 and they are only indicated when they are different to zero. Due to the small number of events, the frequency of hM conformers in the NPs (a subset of the M set, operationally defined with a high F cut off) is not statistically significant. However it correlates very well with the %M. Furthermore, statistical analysis shows that, with the exception of polyQ tracts (for which the number of M, events is too low), the differences in the %M events for NP guests are statistically significant for the following pairs: NP and non-NP, NP and familial NP, NP (except for Arc Aβ42) and NP+QBP1. These differences are not statistically significant for the following pairs: Aβ42 and Aβ42+SV111, Arc Aβ42 and Arc Aβ42+QBP1, and non NPs. The numbers in parenthesis are half of the 95% confidence interval for the %M (or NM, modeled as a Bernoulli distribution) while the numbers after ± are the SMFS experimental errors. Note that the % numbers here are the estimation for the population while the % numbers in Figures 2– 5 correspond to the raw samples.

n, sample size. NM, no force peaks detected: F≤20 pN; M, at least one force peak: F>20 pN; hM, subset of M conformers with at least one force peak with F≥400 pN.