Thrombocytopenia Microcephaly Syndrome - a novel phenotype associated with ACTB mutations

Sharissa L. Latham#,*,1, Nadja Ehmke#,2,3, Patrick Y.A. Reinke1,4, Manuel H. Taft1, Michael J. Lyons5,

Michael J Friez5, Jennifer A. Lee5, Ramona Hecker6, Michael C. Fruehwald7, Kerstin Becker8, Teresa

M. Neuhann8, Denise Horn2, Evelin Schrock9, Katharina Sarnow9, Konrad Grützmann10, Luzie

Gawehn9, Barbara Klink9, Andreas Rump9, Christine Chaponnier11, Ralf J. Knöfler12, Dietmar J.

Manstein*,1,4 and Natalia Di Donato*,9

Supplementary notes.

Variant ACTB mRNA and β-CYA are present in patient primary fibroblasts

All ACTB-TMS patients display heterozygous ACTB mutations. We confirm by RNA-Seq that these

ACTB-TMS mutations do not result in nonsense-mediated mRNA decay and show that the variant

mRNA accounts for 50% of the ACTB mRNA present in P3 and P5 primary fibroblasts

(Supplementary Fig. 7a). With mass spectrometry, we can identify the altered C-terminal frame shift peptide in P3 only (Supplementary Fig. 7b). The 4 amino acid deletion of P5 could not be detected by this method. However, this peptide region was also not detected in the control.

Whole-transcriptome analysis (RNA-Seq)

To further understand the mechanisms of action, we studied the effect of ACTB-TMS mutations on the global expression profile of fibroblasts obtained from P3, P5 and a healthy control. RNA was isolated from three independent fibroblast cell culture flasks for each individual and subsequent whole transcriptome sequencing was performed. Principal component analysis showed clear grouping of replicates for one individual and clear separation between samples from different individuals. Sample distance analysis and unsupervised hierarchical clustering showed a clear separation between patient samples and control in two main clusters, with a further separation between the two patients within the ACTB-mutation cluster (Supplementary Figures 8). Gene expression analysis revealed 2912 significantly differentially expressed (DEG, Benjamini-Hochberg adjusted-value ≤ 0.01, FC >1)

1

when comparing both patients with mutation vs control, illustrating the far-reaching impact of ACTB-

TMS mutations also on the transcriptome. Interestingly, P5 had many more significantly deregulated genes compared to normal control than P3 (10750 versus 8381 genes, respectively; DEG, Benjamini-

Hochberg adjusted-value ≤ 0.01, FC >1), which is in line with the more severe phenotype.

Pathway analysis using all genes significantly differentially expressed between both patients’ fibroblasts compared to control showed, as expected, overexpression of coding genes associated with the KEGG “regulation of ” pathway in both ACTB-TMS cell lines

(Supplementary Fig. 9). In addition, we also observed significant upregulation in the “cell cycle” pathway.

86 protein coding genes were massively overexpressed in both ACTB-TMS cell lines with the log2 FC

> 4 (Supplementary table 4). 18 of 86 were part of the KEGG cell cycle, DNA replication, RNA

transport and DNA repair pathways. 19 further genes encode for kinetochore complex components,

centrosomal and proteins involved in spindle formation and checkpoint activity. We also

found significant overexpression of 9 different family members that are -

dependent motors, also required for spindle formation. We hypothesize that this strong overexpression

of multiple genes encoding nuclear and cytoplasmic proteins involved in DNA replication and cell

division is due to the impaired function of nuclear β-actin1. Specific characterization of nuclear β- actin was beyond the scope of the study and will be assessed in the follow up projects.

In Silico analysis of mutation-induced structural changes in β-CYAP3 and β-CYAP5

Structural changes in β-CYAP3 and β-CYAP5occur in the region of actin SD1 that is formed by

residues 338-374. This region is highly conserved in all actin isoforms (Supplementary Fig. 1). The

P3 mutation changes 26 amino acids and removes 19 residues (Fig. 5a, left). A deletion of four amino

acids in P5 results in a shift of residues G342-S358 by ~5Å, while residues K359-F377 are predicted to be structurally unaffected (Fig. 5a, right).

In both P3 and P5, the binding interfaces for interaction with the CM loop (T417 – T432) and supporting loop (E542 – L550) of are affected. The CM loop is essential for strong binding of myosin’s upper 50kDa domain to actin via hydrophobic (NM-2C: V422, V427, A430; β-CYA: V30,

2

P333, Y337) and electrostatic interactions (NM-2C: K429; β-CYA: D25, E334). The absence of C-

terminal residues P333, E334 and Y337 in β-CYAP3 is predicted to perturb CM loop binding. A recent

cryo-EM structure of the human cytoplasmic actomyosin complex shows that this binding interface is

stabilized by internal interactions of W3402. This residue interacts with V9, D11 and K18 of the central SD1 β-sheet and forms ring stacking interactions with P27. Our model predicts that absence of

W340 in both β-CYAP3 and β-CYAP5 perturbs the CM loop binding surface (Fig. 5b, top). This view receives further support from the results of MD-simulations that compare the folding stability and behavior of β-CYAWT and β-CYAP5.

Mutations within the supporting loop alter the actin-activated ATPase activity, actin affinity and in

vitro sliding motility of a Dictyostelium discoideum myosin 2 motor domain construct3. Based on

biochemical results, the supporting loop was reported to interact with actin N-terminal residues4.

However, this interpretation is not compatible with the structure of a human rigor actomyosin complex that was obtained using electron cryomicroscopy2. The structure shows two actin C-terminal glutamines (Q353, Q354) to be within the interaction range of this NM-2 loop. According to our modelling results, these residues are despite the frameshift spatially conserved in β-CYAP3, but out of

the interaction range in the case of β-CYAP5 (Fig. 5b, bottom).

For the actin-α- complex, a low resolution cryo-EM density is available (PDB: 3LUE, described in 5), which shows that the binding sites for α-actinin and myosin overlap. The actin binding

domain of α-actinin is commonly referred to as calponin homology domain, as it shares its actin

binding domain with a large superfamily of proteins including calponin, , , ,

, and . The interface includes the interaction of β-CYA SD1 I345 with F101 of α-

actinin. For both β-CYAP3 and β-CYAP5 the hydrophobic interactions of I345 are conserved in the form of leucine residues (Fig. 5c). However, in β-CYAP3, D349 and D351 are within the interaction

range of α-actinin H102 and D95, introducing additional electrostatic interactions (Fig. 5c, left).

Regarding α-SMA, despite conservation of C-terminal residues between α-SMA and β-CYA, N- terminal residues are highly divergent (α-SMA: MCEEEDSTAL, β-CYA: MDDDIAAL). These residues are also part of the interaction interfaces of actin with NM-2 (NM2C: R661-R664 within

3 loop2/W-helix) and α-actinin (R93, K96) and may also contribute to the phenotype observed in sub- nuclear bundles.

4

List of Supplementary Figures and Tables

Supplementary Figure 1. Schematic representation of ACTB-TMS mutations

Supplementary Figure 2. Peripheral blood and bone marrow smears of affected individuals with

ACTB-TMS

Supplementary Figure 3. Craniofacial appearance of affected individuals with ACTB-TMS

Supplementary Figure 4. Characterization of ACTB-TMS patient derived primary fibroblasts

Supplementary Figure 5. Localization of actin isoforms in ACTB-TMS patient primary fibroblasts

Supplementary Figure 6. Assessment of ABP in ACTB-TMS primary fibroblasts and peripheral thrombocytes

Supplementary Figure 7. ACTB-TMS variant mRNA and protein expression

Supplementary Figure 8. Whole-transcriptome analysis of control, P3 and P5 primary dermal fibroblasts

Supplementary Figure 9. KEGG pathways analysis of RNA-Seq data

Supplementary Table 1. Detailed clinical information for ACTB-TMS patients

Supplementary Table 2. Protein coding genes significantly deregulated in ACTB-TMS P3 vs. healthy control primary fibroblasts

Supplementary Table 3. Protein coding genes significantly deregulated in ACTB-TMS P5 vs. healthy control primary fibroblasts

Supplementary Table 4. Protein coding genes upregulated (log2 FC >4) in both ACTB-TMS patient fibroblasts

Supplementary Table 5. List of ABP assessed in ACTB-TMS patient fibroblasts

Supplementary Table 6. Assessment of 124 ABP encoding genes in P3 and P5 vs. healthy control fibroblasts

Supplementary Table 7. Summary of ACTB-TMS P3 and P5 clinical, cellular and structural data

Supplementary Table 8. List of antibodies used in this study

5

α-SMAWT 311 377 γ-C YAWT 309 375 β-C YAWT 309 375 319 329 338 348 358 368 β-C YAP1/P2 309 375 β-C YAP3/P4 309 357 β-C YAP5 309 371 β-C YAP6 309 379 β-C YAPN 309 374

Unique to α-SMA Patient mutations Conserved glutamines

Supplementary Figure 1. Schematic representation of ACTB-TMS mutations Peptide sequence

alignment of wildtype (WT) α-SMA, γ-CYA, β-CYA and ACTB-TMS patient β-CYA isoforms starting at β-CYAWT residue 309.

6

Supplementary Figure 2. Peripheral blood and bone marrow smears of affected individuals

with ACTB-TMS. (a-c) May-Gruenwald-Giemsa-stained stained peripheral blood smears of (a) P3,

(b) P4 and (c) P5, with anisocytosis demonstrating normal sized thrombocytes (arrowheads) and macrothrombocytes (arrows); (d) May-Gruenwald-Giemsa-stained bone marrow smear of P5 demonstrating numerous megakaryocytes (arrows).

7

Supplementary Figure 3. Craniofacial appearance of affected individuals with ACTB-TMS. (a, d) Photographs of P3 at 31 years show flared and straight eyebrows, mild telecanthus, up-slanted palpebral fissures, bulbous nose, broad mouth, thin lips and prominent chin; (b, e) photographs of P4 at 5 years with the overlapping features (flared eyebrows, telecanthus, epicanthic folds, up-slanted palpebral fissures, broad nasal tip, and thin upper vermillion border); (c, f) photographs of P5 at 4 years 10 months with overlapping but more noticeable dysmorphism, such as straight eyebrows, prominent telecanthus with the sharp descent of the upper medial canthal fold, epicanthic folds, broad nasal tip, and thin upper vermillion border.

8

Supplementary Figure 4. Characterization of ACTB-TMS patient derived primary fibroblasts.

(a) Representative micrographs of control, P3 and P5 primary fibroblasts at 60 min intervals during

8h migration. Colored markers indicate contacts made between neighboring cells, showing that control cells can migrate past one another, P5 cells aggregate following cell contact and P3 cells are intermediate in their behavior. Scale bars represent 100 µm; (b) Growth curve of control (blue) vs patient (red) primary fibroblasts collected 24 h, 48 h and 96 h after seeding 30,000 cells per well, indicates no significant differences in ACTB-TMS proliferation rate.

9

Supplementary Figure 5. Localization of actin isoforms in ACTB-TMS patient primary

fibroblasts. (a) β-CYA (green) and γ-CYA (magenta) distribution in P3 and P5 fibroblasts shows

discrete lateral and axial isoform segregation consistent with that observed in control (see Fig. 2a).

Maximum intensity projections for indicated z-stack slices are given; (b) Representative Western Blot

of α-SMA in control, P3 and P5 fibroblasts shows that the protein is not detected in control cells, yet

is enriched in both ACTB-TMS cultures. Ponceau staining shows equal sample loading. (c) P3 cells co-stained for α-SMA (magenta) with β-CYA (top, green) or γ-CYA (bottom, green), showing greater overlap of α-SMA with β-CYA within the cells’ basal plane, consistent with the findings in P5 cells

(see Fig. 3e). All scale bars represent 20 µm.

10

Supplementary Figure 6. Assessment of ABP in ACTB-TMS primary fibroblasts and peripheral

thrombocytes. (a) Tpm4.2, identified as a candidate gene from RNA-Seq and found to be

upregulated by Western Blot analysis, does not localize to basal sub-nuclear filaments, as shown in

the maximum intensity projections of z-slices 1-2 from healthy control, P3, P5 and BWCFF control

primary fibroblasts. Scale bars represent 10 µm; (b) Immunofluorescence staining of P1 peripheral

blood smears labelled for β-CYA (green), γ-CYA (magenta) and NM-2A (cyan) is consistent with that observed in P5 (see Fig. 3d). Scale bars represent 2 µm.

11

Supplementary Figure 7. ACTB-TMS variant mRNA and protein expression. (a) Screenshot of

the ‘Integrative Genomics Viewer’ software interface showing equal expression of both mutant and

native ACTB mRNA in RNA-Seq data, indicates in red boxed regions; (b) LC-MS analysis of control,

P3 and P5 primary fibroblast protein samples was performed in the proteomics core facility of

Hannover Medical School, as described by Erdmann et al.6. β-CYA peptides detected by mass

spectrometry are shown in color. The peptide unique to the P3 variant is identified only in the P3

sample (red). The peptide region including residues ‘SVWI’ was not detected in control fibroblasts and accordingly, the ‘SVWI’ deletion of P5 could not be identified by mass spectrometry.

12

Supplementary Figure 8. Whole-transcriptome analysis of control, P3 and P5 primary dermal fibroblasts. (a) Heat map from hierarchical clustering of Euclidian sample distances based on regularized logarithmic (rlog) transformed expression values of all genes of all samples. Each row and column denote a sample. The dendrogram on the x-axis shows the grouping of samples according to replicates (triplicates) and patients/control. The bottom color bar represents the patients/controls; (b)

Plot of principal component analysis (PCA) shows an individual separating pattern with grouping of

13 replicas; (c) Heat map showing unsupervised clustering using the 500 genes with the highest variance in gene expression over all samples. For all graphs, the control fibroblasts (culture ID 31660) are shown in green, P3 fibroblasts (culture ID 34012) are shown in blue and P5 fibroblasts (culture ID

30563) are shown in red.

14

Supplementary Figure 9. KEGG pathways analysis of RNA-Seq data. A significant upregulation of genes implicated in both (a) actin cytoskeleton and (b) cell cycle regulation pathways is observed in ACTB-TMS patient fibroblasts. Normalized log2 FC (P3 and P5 vs control) are color coded (red = upregulated, green = downregulated).

15

Supplementary References

1. Viita, T. & Vartiainen, M.K. From Cytoskeleton to Gene Expression: Actin in the Nucleus.

Handb Exp Pharmacol 235, 311-329 (2017).

2. von der Ecken, J., Heissler, S.M., Pathan-Chhatbar, S., Manstein, D.J. & Raunser, S. Cryo-

EM structure of a human cytoplasmic actomyosin complex at near-atomic resolution. Nature

534, 724-8 (2016).

3. Furch, M., Remmel, B., Geeves, M.A. & Manstein, D.J. Stabilization of the actomyosin

complex by negative charges on myosin. Biochemistry 39, 11602-8 (2000).

4. Varkuti, B.H. et al. A novel actin binding site of myosin required for effective muscle

contraction. Nat Struct Mol Biol 19, 299-306 (2012).

5. Galkin, V.E., Orlova, A., Salmazo, A., Djinovic-Carugo, K. & Egelman, E.H. Opening of

tandem calponin homology domains regulates their affinity for F-actin. Nat Struct Mol Biol

17, 614-6 (2010).

6. Erdmann, J. et al. Glucosyltransferase-dependent and -independent effects of TcdB on the

proteome of HEp-2 cells. Proteomics 17(2017).

16