Conditional deletion of SMN in cell culture identifies functional SMN alleles

Abstract

Spinal muscular atrophy (SMA) is caused by mutation or deletion of survival motor neuron 1 (SMN1) and retention of SMN2 leading to SMN protein deficiency. We developed an immortalized mouse embryonic fibroblast (iMEF) line in which full-length wild-type Smn (flwt-Smn) can be conditionally deleted using Cre recombinase. iMEFs lacking flwt-Smn are not viable. We tested the SMA patient SMN1 missense mutation alleles A2G, D44V, A111G, E134K and T274I in these cells to determine which human SMN (huSMN) mutant alleles can function in the absence of flwt-Smn. All missense mutant alleles failed to rescue survival in the conditionally deleted iMEFs. Thus, the function lost by these mutations is essential to cell survival. However, co-expression of two different huSMN missense mutants can rescue iMEF survival and small nuclear ribonucleoprotein (snRNP) assembly, demonstrating intragenic complementation of SMN alleles. In addition, we show that a Smn protein lacking exon 2B can rescue iMEF survival and snRNP assembly in the absence of flwt-Smn, indicating exon 2B is not required for the essential function of Smn. For the first time, using this novel cell line, we can assay the function of SMN alleles in the complete absence of flwt-Smn.

Introduction

Spinal muscular atrophy (SMA) is an autosomal recessive disorder resulting in motor neuron loss. SMA is caused by a deletion or mutation in the survival motor neuron 1 (SMN1) gene and retention of the SMN2 gene (1,2). Both genes make SMN protein; however, SMN2 contains a mutated splice modulator that results in the exclusion of exon 7 from 90% of the transcripts (3–7). This truncated SMN protein (SMN∆7) inefficiently oligomerizes and is rapidly degraded (8–10). Approximately 10% of SMN protein made by 2 copies of SMN2 is full-length, which is sufficient for survival, but not for motor neuron function (11,12). Hence, SMA is a disease of SMN protein deficiency (11,12). The severity of phenotypic presentation correlates with SMN2 copy number, such that more copies of SMN2, which results in more SMN protein, are correlated with a milder phenotype (13–16).

SMN is an essential protein; its loss results in lethality in all species (17–19). How SMN deficiency leads to the selective loss of motor neurons in SMA is unknown. SMN forms an oligomeric complex that functions in the assembly of the spliceosomal and U7 small nuclear ribonucleoprotein (snRNP) complexes (20–24). snRNP assembly is SMN dependent and one can directly measure this activity in both tissue and cell samples (20–23,25). Other functions for SMN have been proposed in the regulation of translation, axonal transport and cytoskeleton dynamics, but what specific role SMN has in these functions is unclear (24,26–29).

In most SMA cases SMN1 is nonfunctional, therefore patients are fully reliant on the small amount of full-length wild-type SMN (flwt-SMN) produced from SMN2 (1,30). Humans are the only species that have an SMN2 gene. Approximately 1% of SMA cases are caused by a missense mutation in SMN1 and are categorized as mild or severe (2,31). Severe mutations are found in patients who clinically present according to their SMN2 copy number, whereas mild mutations are found in patients who present with a milder phenotype than predicted by their SMN2 copy number (2,32). SMN1 missense alleles are always found in the presence of SMN2 in SMA patients, even though 10–15% of the healthy population lacks SMN2 (15,30). This implies that the missense mutations are not functional on their own. Mice have one Smn gene, which is equivalent to SMN1 and produces flwt Smn protein (33). Knockout of the mouse Smn gene results in embryonic lethality (17). Transgenic expression of SMN1 missense alleles does not rescue lethality in Smn^KO/KO mice indicating these alleles are non-functional in the absence of flwt-Smn (34–36). Yet, transgenic expression of mild SMN1 missense alleles in Smn^KO/KO mice in the presence of flwt-SMN from SMN2 can rescue survival, weight, snRNP assembly and electrophysiology (34–36). This suggests that the SMN1 missense alleles are functional when flwt-SMN is present due to intragenic complementation. Intragenic complementation occurs when heteromeric protein complexes, composed of different mutant alleles of a gene, exhibit function greater than that found in homomeric complexes of each mutant alone (37). Furthermore, co-expression of N-terminal and C-terminal SMN missense alleles, like A111G with T274I, in Smn^KO/KO mice shows complete rescue in survival, weight, electrophysiology and snRNP assembly (V.L. McGovern et al., manuscript in preparation) (31). In fact, the ability of SMN1 missense mutants to complement flwt-SMN and other SMN missense alleles, thereby restoring full activity of the SMN complex, indicates that the SMN oligomer is the functional unit of SMN in the cell.

Currently, there are no cell systems in which functional SMN can be completely removed on a conditional basis. Previous experiments used cell lines with a doxycycline repressible promoter or constructs expressing shRNA/siRNAs to modulate flwt-Smn or flwt-SMN expression (38–40). Both of these methods can result in residual levels of flwt-Smn/-SMN expression due to incomplete knockdown or leaky promotors. Given that SMN mutant alleles can undergo intragenic complementation with flwt-SMN, it is impossible to assess SMN mutant functionality and to identify specific suppressors of SMN loss of function, in the cells described to date. Here, we describe a cell line derived from immortalized mouse embryonic fibroblasts (iMEFs), which, upon the expression of Cre-recombinase, excises mouse Smn exon 7, resulting in the elimination of functional Smn in the cell. Using these iMEFs, we show that both mild and severe SMN missense mutants are non-functional and cannot rescue the loss of flwt-Smn on their own. Additionally, co-expression of specific combinations of missense alleles can complement and rescue survival in the conditional cell line. Lastly, an Smn lacking exon 2B, identified from a reported Smn knockout NSC-34 line, rescues iMEF survival and snRNP assembly indicating that exon 2B is not essential for Smn function (41).

Results

Generation of a cell line that can be conditionally deleted for Smn exon 7

Smn^WT/F7;tdTomato mice were crossed with Smn^WT/KO;tdTomato mice to yield embryos with the Smn^KO/F7;tdTomato genotype. The Smn^KO is a widely used Smn null allele (17). The Smn^F7 allele consists of Smn exon 7 flanked by loxP sites (42). In the presence of Cre recombinase, exon 7 is removed and the resultant Smn^D7 allele expresses the non-functional Smn∆7 protein (42,43). Additionally, the tdTomato reporter allele was included to monitor Cre activity, which removes a floxed stop codon cassette 5′ of tdTomato allowing its expression, marking expressing cells with red fluorescence (44). MEFs were prepared from Smn^KO/F7;tdTomato (KO/F7) embryos at E10–12 and genotyped for the presence of the WT, KO, F7, SMN2 and tdTomato alleles (Fig. 1,Supplementary Material, Fig. S1). MEFs were then immortalized by transfection with the human papilloma virus E6 and E7 proteins (iMEFs) (45–47). iMEFs were also generated from Smn^WT/F7;tdTomato (WT/F7) and Smn^KO/F7;SMN2^+/− (KO/F7;SMN2) mice as controls. The SMN2 allele is the human SMN2 transgene under expression of the human SMN promoter (48). In these cells, the introduction of an improved Cre (iCre) deletes Smn exon 7 from the Smn^F7 allele (D7), but the WT/D7 and KO/D7;SMN2 iMEFs retain production of flwt-Smn or flwt-SMN from Smn^WT (WT) or SMN2, respectively. Conversely, the transduction of iCre in KO/F7 iMEFs via lentivirus results in the deletion of Smn exon 7 and no flwt-Smn is produced (Fig. 1).

Figure 1

Mouse alleles utilized to generate cell lines for the conditional deletion of full-length wild-type Smn (flwt-Smn). The Smn^KO (KO) allele was created by targeted insertion of lacZ-Neo sequence into exon 2A (17). This is a null allele, resulting in knockout of Smn. The Smn^F7 (F7) allele is a conditional Smn knockout allele (43). This was created by targeted insertion of a construct containing intron 6 through intron 7 with loxP sites flanking exon 7. In the presence of Cre-recombinase, Smn exon 7 is removed, converting the Smn^F7 into Smn^D7 (D7), resulting in production of the nonfunctional Smn∆7 protein. The tdTomato reporter transgene contains loxP sites flanking an early termination signal upstream from the tdTomato sequence and prevents protein expression of tdTomato. When Cre is present, the early termination signal is removed and expression of tdTomato occurs. iMEFs were harvested from embryos which contain the tdTomato allele as well as the WT/F7, KO/F7 or KO/F7;SMN2 genotypes. These cells were immortalized by infecting with retrovirus expressing the human papilloma virus E6 and E7 proteins. Cells were then infected with a lentivirus expressing GFP and iCre to delete Smn exon 7, making a conditional Smn knockout cell line.

iMEFs lacking Smn are not viable

We first asked whether iMEFs can survive without a source of flwt-Smn. KO/F7 iMEFs were treated with a lentivirus delivering iCre and green fluorescent protein (GFP), sorted by fluorescence-activated cell sorting (FACS) and seeded at low density in 96-well plates (Fig. 1). WT/F7 and KO/F7;SMN2 iMEFs were treated as flwt-Smn and flwt-SMN expressing controls. The proliferation of these iMEFs was monitored and imaged over 3 weeks (Fig. 2A). As was expected, the WT/D7 and KO/D7;SMN2 iMEFs proliferate and the KO/D7 iMEFs do not survive (Fig. 2A). Therefore, protein produced by Smn lacking exon 7 is incapable of rescuing cell survival. Proliferating colonies from the WT/D7 and KO/D7;SMN2 iMEFs were clonally isolated and genotyped for the deletion of Smn exon 7 by droplet digital PCR (ddPCR) (49,50). Surviving colonies were analyzed for the presence of exon 7 in mouse Smn mRNA using ddPCR (Fig. 2B and C). There was a 61% reduction in Smn transcripts containing exon 7 in the WT/D7 iMEFs after iCre delivery (Fig. 2B). Importantly, none of the KO/D7;SMN2 clones make a Smn mRNA that includes exon 7 (Fig. 2C) indicating these cells are surviving solely on SMN produced from SMN2. Representative western blots confirm the expression of mouse and human SMN (huSMN) protein in these iMEF lines (Fig. 2D). The quantification of total SMN protein expression was determined by western blot in the WT/F7 and WT/D7 cell lines (Fig. 2E). The WT/D7 clones express 60% of the SMN protein detected in the parental WT/F7 cell line. We found huSMN protein expression increased 1.3 fold in KO/D7;SMN2 clones as compared with the parental KO/F7;SMN2 iMEF line (Fig. 2F).

Figure 2

Smn^D7 acts like a null allele in iMEFs (A) Deletion of Smn exon 7 is lethal in iMEFs without an alternative source of flwt-Smn. Representative images of cells seeded at low density in 96-well plate over the course of 3 weeks. Clonal expansion from single cells is evident in lines that have an alternate source of flwt-Smn from mouse (WT/D7) or flwt-SMN from human (KO/D7;SMN2) alleles. (B) The quantification of Smn exon 7 inclusion in mouse Smn mRNA using ddPCR in WT/F7 and WT/D7 samples. WT/D7 iMEFs make less Smn transcripts that include exon 7 than the WT/F7 parent line. WT/F7: 1.00 ± 0.19 RFU, n = 3; WT/D7: 0.39 ± 0.06 RFU, n = 5 (^**P = 0.0097). (C) The quantification of Smn exon 7 inclusion in mouse Smn mRNA using ddPCR in KO/F7;SMN2 and KO/D7;SMN2 samples. No Smn mRNA expressed in KO/D7;SMN2 clones include exon 7. These lines are entirely reliant on flwt-SMN protein produced from SMN2 for survival. KO/F7;SMN2: 1.00 ± 0.03 RFU, n = 5; KO/D7;SMN2: 0.00 ± 0.00 RFU, n = 3 (^****P < 0.0001). (D) Representative western blots probing for human or total SMN and α-Tubulin. huSMN is only detected in the SMN2 lines. (E) Western blot protein quantification of total SMN normalized to α-Tubulin in WT/F7 and WT/D7 cells. WT/D7 iMEFs make less Smn protein than WT/F7 cells. WT/F7: 1.00 ± 0.06 RFU, n = 3; WT/D7: 0.60 ± 0.06 RFU, n = 5 (^**P < 0.0023). (F) Western blot protein quantification of huSMN normalized to α-Tubulin in KO/F7;SMN2 and KO/D7;SMN2 cells. huSMN protein expression is ~ 1.3 fold higher in KO/D7;SMN2 iMEFs when compared with the KO/F7;SMN2 iMEF line. KO/F7;SMN2: 1.00 ± 0.02 RFU, n = 4; KO/D7;SMN2: 1.32 ± 0.09 RFU, n = 3 (^**P = 0.0083). For Figure 2B, C, E and F, the expression is given as mean ± standard error from the mean (SEM), n and the P-value calculated from two-tailed student t-test (^nsP > 0.05, ^**P ≤ 0.01). For F7 lines, n are biological replicates. For D7 lines, n is the number of individual clones.

Expression of a single huSMN missense allele fails to rescue cell survival in the absence of flwt-Smn

Next, we wanted to test patient identified SMN1 missense alleles in this system to determine if they can rescue cell survival in the absence of flwt-Smn. The missense alleles we chose are located across known SMN domains, including the Gemin2-binding domain, Tudor domain and YG-box/oligomerization domain (Fig. 3A). KO/F7 iMEFs were transfected with plasmids expressing huSMN cDNAs with the missense mutations A2G, D44V, A111G, E134K, or T274I. Alleles A2G, D44V, A111G, and E134K are driven by the ubiquitous expressing cytomegalovirus promotor (CMV). The T274I allele is driven by the SMN promoter and D44V has an N-terminal HA-tag (HA-D44V) and is driven by the CMV promoter (Supplementary Material, Fig. S1) (51). Stable cell lines were selected and genotyped for the SMN missense allele using ddPCR and locked nucleic acid (LNA) probes (Supplementary Material, Fig. S2). We selected iMEFs that express high and low levels of the SMN missense mutants. iMEFs expressing high levels of the SMN missense mutants are shown in Figure 3. huSMN mRNA expression was tested by ddPCR (Fig. 3B), and huSMN protein expression was confirmed and quantified by western blot (Fig. 3C and D,Supplementary Material, Fig. S3). Mutants expressing low levels of the SMN missense mutants are given in Supplementary Material, Figure S4. The A2G allele produces a faster migrating protein band. HA-D44V is not detected by the antibody used to measure total SMN (BD, 610647), which is likely due to the location of the mutation within the epitope region (14-174 amino acids). However, the HA-D44V protein is detected by the human-specific SMN antibody (Millipore, MABE230).

Figure 3

mRNA and protein expression in KO/F7 iMEFs expressing a single huSMN missense mutant allele. (A) All mild and severe huSMN missense mutations used in this study are found in SMA patients (1,32,82). Mutations assayed are in known SMN domains including Gemin2-binding, Tudor and the YG-box/oligomerization domains. Numbers designate the amino acids that end each exon or are before the translational termination signal. (B) Droplet-ddPCR cDNA quantification of huSMN mRNA expression in KO/F7 iMEFs selected to stably express a single huSMN missense mutant. huSMN expression is normalized to Smn mRNA. (C) Western blot quantification of huSMN protein normalized to α-Tubulin in KO/F7 iMEFs selected to stably express a single huSMN missense mutant. The mean protein expression ± SEM is included in the top half of Table 2. For both B and C, mean expression ± SEM, n and P-values are given in Supplementary Material Figure S3. P-values are calculated from two-tailed student t-test comparing mean huSMN expression of the missense mutant iMEF to the non-expressing iMEF each line is derived from (^*P ≤ 0.05, ^**P ≤ 0.01, ^***P ≤ 0.001, ^****P ≤ 0.0001). (D) Representative western blots probing for human and total SMN and α-Tubulin. D44V has an N-terminal HA-tag, adding an additional 25 amino acids therefore it migrates slower (Supplementary Material Fig. S1). The HA-D44V is not detected by the antibody used for total SMN. A2G migrates faster and A111G always includes an additional faster migrating band.

To determine whether the transfected SMN allele was functional, the absence of Smn exon 7 must be confirmed in both genomic DNA and mRNA in all surviving KO/D7 iMEFs. iMEFs expressing both high and low protein levels for each of the missense mutants were transfected with the lentivirus delivering iCre and GFP, sorted by FACS and monitored for the growth of surviving colonies for 3 weeks. No viable colonies were obtained that express a single huSMN missense mutation upon deletion of Smn exon 7 despite high levels of huSMN protein expression. To summarize, all SMN1 missense mutants tested lack the essential function of SMN required to rescue KO/D7 iMEF survival.

Co-expression of two huSMN missense alleles rescues cell survival in the absence of flwt-Smn

Intragenic complementation occurs when two mutant proteins of a gene form a heteromeric complex with increased function as compared with a complex composed of either mutant alone (37). Consequently, we asked whether co-expression of combinations of huSMN missense mutants can complement and restore SMN function. KO/F7 iMEFs expressing low protein levels of SMNA2G, A111G, or T274I were transfected with plasmids expressing one other huSMN missense allele. These iMEFs were utilized to avoid over-expression of huSMN when adding a second missense mutant. Quantifications of huSMN mRNA and protein expression for these low expressing iMEFs are in Supplementary Material, Figure S4. None of the low expressing iMEF lines survive upon deletion of Smn exon 7. Stable cell lines were genotyped using LNA probes specific for the missense mutation single nucleotide polymorphism (SNP) with ddPCR to confirm the presence of both missense alleles. iMEFs co-expressing two huSMN missense mutants were infected with lentivirus expressing iCre and GFP, sorted by FACS and seeded to determine whether KO/D7 iMEF survival could be rescued. All allele combinations tested resulted in the rescue of iMEF survival except for the T274I + A2G combination.

The contribution of both huSMN missense mutants to the total huSMN mRNA expression was confirmed by RT-ddPCR in all surviving KO/D7 iMEFs (Table 1,Supplementary Material, Fig. S5). The huSMN protein expression for individual clones for each mutant allele combination indicates the amount of SMN necessary to rescue survival in iMEFs is broad (Fig. 4A). Importantly, in a number of the combinations, the huSMN protein expressed by several clones is significantly lower than that which is detected in the iMEFs that express a single huSMN missense mutant (Fig. 4B,Table 2). For instance, the huSMN protein expression from the lowest A2G + HA-D44V KO/D7 iMEF is 0.02 ± 0.00 relative fluorescence units (RFU), which is significantly lower than the expression measured in the high singly expressing A2G iMEF at 0.69 ± 0.03 RFU (P < 0.0001) and HA-D44V iMEF at 0.17 ± 0.04 RFU (P = 0.0147) (Table 2). The same is true for the A2G + E134K combination with huSMN expression at 0.09 ± 0.01 RFU compared with the singly expressing A2G iMEF at 0.69 ± 0.03 RFU (P = 0.0409) and E134K iMEF at 1.57 ± 0.24 RFU (P = 0.0006) (Table 2). The huSMN expression levels in A111G + E134K, T274I + HA-D44V, and T274I + E134K are significantly lower than one of their respective singly expressing iMEFs, further showing rescued cell survival is not dependent on the level of huSMN protein, but due to the presence of the second missense allele (Fig. 4B,Table 2). In only two combinations—A111G + HA-D44V and T274I + A111G—are the huSMN protein levels expressed significantly higher than their respective singly expressing iMEFs, but for each of the present missense alleles there is a co-expressing KO/D7 iMEF that is significantly lower in expression (Fig. 4B,Table 2). Representative western blots for huSMN and total SMN protein expression in KO/D7 iMEFs co-expressing two SMN missense mutants, as well as the KO/F7 dual-expressing and KO/F7 single SMN mutant expressing iMEFs they are derived from are given in Figure 5.

Table 1

Contribution of each SMN missense mutant to huSMN mRNA expression in KO/D7 iMEFs co-expressing two SMN missense mutant alleles

	Mean % ± SEM (RFU)
Sample (Allele 1 + Allele 2)	Allele 1	Allele 2
A2G + HA-D44V	11 ± 5	87 ± 5
A2G + E134K	78 ± 4	24 ± 5
A111G + A2G	20 ± 11	84 ± 8
A111G + HA-D44V	93 ± 2	7 ± 3
A111G + E134K	91 ± 2	5 ± 2
T274I + HA-D44V	44 ± 7	57 ± 6
T274I + A111G	65 ± 5	35 ± 7
T274I + E134K	60 ± 4	36 ± 4

All means are calculated from n ≥ 3 individual clones.

n/a no clones were identified from KO/D7;T274I + A2G, thus allele contribution could not be measured.

Figure 4

Range of huSMN protein expression in KO/D7 iMEFs co-expressing two huSMN missense mutants indicates very little SMN is required for cell survival. (A) Range of huSMN protein expression in KO/D7 iMEFs co-expressing two huSMN missense mutants represented in box plots. huSMN expression is normalized to α-Tubulin and the resultant expression levels are normalized to KO/F7;SMN2. huSMN expression in KO/F7 and KO/F7;SMN2 is calculated from 5 and 4 biological replicates, respectively. Each co-expressing KO/D7 iMEF clone is represented by a circle, the expression of which is the mean of 3 biological replicates. The filled circle is the clone used in the western blot shown in Figure 4C. Horizontal dashed lines designate the huSMN protein expression levels in the highest expressing single huSMN missense mutant iMEFs tested (Fig. 3,Table 2). The means plotted for the dashed lines are A2G: 0.69 RFU; D44V: 0.17 RFU; A111G: 0.55 RFU; E134K: 1.57 RFU; T274I: 0.24 RFU. The lowest huSMN protein expressing clones for each missense mutant combination are graphed in Figure 4B and the values are given in Table 2. (B) Rescue of cell survival is not due to expression level of huSMN, but due to intragenic complementation occurring between the two huSMN missense mutants expressed in KO/D7 iMEFs. Mean huSMN protein expression ± SEM is plotted for the lowest expressing KO/D7 iMEF co-expressing clones for each surviving combination. Horizontal dashed lines are the same as Figure 4A. Refer to Table 2 for mean ± SEM, n and P-values for comparing expression levels of the single mutant expressing iMEFs to the lowest protein co-expressing KO/D7 iMEFs. All means are calculated from at least 3 biological replicates. (C) Representative western blots probing for huSMN, total SMN and α-Tubulin in KO/F7 and surviving KO/D7 iMEFs stably expressing two huSMN missense mutants and the single huSMN missense mutant KO/F7 iMEFs they are derived from. KO/D7 iMEFs co-expressing two huSMN missense mutants used for representative blots are identified by closed circles in Figure 4A.

Table 2

Several KO/D7 iMEF clones co-expressing two SMN missense mutants make significantly less huSMN protein than the iMEFs which express a single missense mutant and do not survive post-deletion of flwt-Smn

KO/F7 Single Missense iMEF	Mean huSMN expression ± SEM (RFU)
A2G	0.69 ± 0.03
HA-D44V	0.17 ± 0.04
A111G	0.55 ± 0.01
E134K	1.57 ± 0.24
T274I	0.24 ± 0.01
KO/D7 Co-expressing iMEFs (Allele 1 + Allele 2)	Mean huSMN expression ± SEM (RFU)	P-value for comparison to huSMN expression in single missense iMEF
		Allele 1	Allele 2
A2G + HA-D44V	0.02 ± 0.00	<0.0001****	0.0147*
A2G + E134K	0.09 ± 0.01	0.0409*	0.0006***
A111G + A2G	0.76 ± 0.11	0.1252ns	0.7149ns
A111G + HA-D44V	1.84 ± 0.28	0.0022**	0.0003***
A111G + E134K	0.88 ± 0.05	0.2632ns	0.0252*
T274I + HA-D44V	0.07 ± 0.01	0.0063**	0.0545ns
T274I + A111G	0.93 ± 0.04	<0.0001****	<0.0001****
T274I + E134K	0.05 ± 0.02	0.5648ns	0.0005***

Means are calculated from ≥3 biological replicates. Significance calculated from one-way ANOVA with Dunnet corrections for multiple testing comparing the means of iMEFs expressing the single SMN missense alleles to the surviving KO/D7 iMEFs expressing both of the corresponding mutant alleles, i.e. huSMN protein expression from A2G and HA-D44V singly expressing iMEFs were compared with the iMEF co-expressing A2G + HA-D44V. (^nsP > 0.05, ^*P ≤ 0.05, ^**P ≤ 0.01, ^***P ≤ 0.001, ^****P ≤ 0.0001)

Figure 5

KO/D7 iMEFs co-expressing two huSMN missense mutants perform snRNP assembly. (A) U4 snRNP assembly activity in KO/D7 iMEFs co-expressing two huSMN missense mutants is similar or greater than the assembly activity produced by the flwt-SMN made from one copy of SMN2. U4 snRNP assembly is normalized to total SMN protein expression to give snRNP assembly activity in terms of SMN abundance. All activities are then normalized to the activity measured in KO/D7;SMN2 iMEF clones. Refer to Table 3 for means, SEM, n and P-values. For KO/D7;SMN2, the mean is calculated from 4 biological replicates of 2 individual clones. For KO/D7 iMEFs co-expressing two huSMN missense mutants, the mean is calculated from at least 3 individual clones. One-way ANOVA with Dunnett’s correction comparing each co-expressing group to KO/D7;SMN2 indicates U4 snRNP assembly activity is equivalent or greater than flwt-SMN for the huSMN missense hetero-oligomers. (B–D) Total U4 snRNP assembly is linearly correlated with SMN protein abundance in KO/D7 iMEFs co-expressing two huSMN missense mutants. (B) Table providing the huSMN missense allele combinations, correlation coefficients (R²) and slopes of linear regressions fitted to the total snRNP assembly versus SMN protein plots. (C) Plots for KO/D7 iMEFs co-expressing two huSMN missense mutants, excluding those combinations involving E134K. (D) Plots for KO/D7 iMEFs co-expressing two huSMN missense mutants only involving E134K. As indicated in (B-D), all allele combinations which include E134K have snRNP assemblies, which do not correlate linearly with SMN abundance. Additionally, those combinations in which the huSMN mRNA contribution is <10% for one huSMN missense allele (Table 1) also do not linearly correlate with SMN abundance.

All combinations of huSMN missense alleles that rescue iMEF survival in the absence of flwt-Smn also assemble U4 snRNPs (Fig. 5A,Table 3). All snRNP assemblies were normalized to the amount of SMN in the cytoplasmic extract. Previously, Wan et al. (25) showed snRNP assembly has a positive linear correlation with SMN protein levels in a cell line induced to have variable SMN expression by a doxycycline repressible promoter. We asked whether this was also the case within our surviving KO/D7 iMEF clones that co-express two huSMN missense mutants. The A2G + HA-D44V, T274I + HA-D44V, A111G + A2G, and T274I + A111G combinations all have total U4 snRNP assembly, which linearly correlates with the abundance of SMN, indicating our huSMN missense hetero-oligomers conform to the Wan et al. observation (Fig. 5B and C) (25). Interestingly, the slopes of the linear regressions fitted to the activity plots for each allele combination are different from one another, suggesting each hetero-oligomer composition has a unique snRNP assembly activity (Fig. 5B and C). For instance, the A111G + A2G KO/D7 iMEFs have a snRNP activity slope of 2.44 (R² 0.998) and the T274I + A111G KO/D7 iMEFs have a slope of 0.654 (R² 0.999) (Fig. 5B and C). This suggests, for each increase in SMN protein abundance, one would expect the A111G + A2G combination to assemble ~ 3.7 times the number of snRNPs as the T274I + A111G combination. Although we obtain rescue in KO/D7 iMEFs co-expressing A111G + HA-D44V and A111G + E134K, the snRNP assembly does not correlate with the SMN abundance (Fig. 5B–D). Interestingly, the huSMN mRNA expressed in both of these combinations is composed of <10% of one of the missense mutants—7% HA-D44V in A111G + HA-D44V and 5% E134K in A111G + E134K—which may account for the lack of correlation between snRNP assembly and SMN abundance (Table 1). Conversely, the A2G + E134K and T274I + E134K combinations have >10% composition of each missense mutant in the huSMN mRNA expression—24% E134K for A2G + E134K and 36% E134K forT274I + E134K—but also display a snRNP assembly uncoupled from SMN abundance (Table 1,Fig. 5B and D).

Table 3

snRNP assembly activity in KO/D7 iMEFs surviving on two huSMN missense mutant alleles

Sample	Mean ± SEM (RFU)	n	P-value
KO/D7;SMN2	1.00 ± 0.42	4
A2G + HA-D44V	5.56 ± 2.13	3	0.0027**
T274I + HA-D44V	5.06 ± 1.37	3	0.0077**
A111G + A2G	2.21 ± 0.24	4	0.7061ns
A111G + HA-D44V	0.99 ± 0.19	3	0.9998ns
A111G + T274I	1.71 ± 0.41	3	0.9672ns
A2G + E134K	0.86 ± 0.35	3	>0.9999ns
A111G + E134K	1.13 ± 0.25	3	>0.9999ns
T274I + E134K	1.09 ± 0.25	5	0.9996ns

For KO/D7;SMN2 n = 4 biological replicates of two individual clones. For all others, n = number of individual clones.

Significance calculated from One-way ANOVA with Dunnet corrections for multiple testing comparing to KO/D7;SMN2. (^nsP > 0.05, ^*P ≤ 0.05, ^**P ≤ 0.01)

Exon 2B is not essential for SMN function in cell survival or snRNP assembly

Bernabo et al. (41) previously reported the generation of an NSC-34 line deleted for SMN using CRISPR and guide RNA targeted to Smn exon 2B. In their publication the authors refer to this NSC-34 line as SMN 0%; we designate it here as NSC-34-Smn∆2B. The CRISPR-generated mutations were uncharacterized and western blot showed the absence of a Smn band at the expected 38 kDa size. We confirmed absence of the 38 kDa Smn band by western blot, yet identified a faster migrating band suggesting the presence of a truncated Smn protein. Amplification of the genomic DNA from this cell line from intron 2A to intron 2B resulted in two bands suggesting a truncation in at least one Smn allele (Fig. 6A). This was confirmed to be a 131 base-pair deletion from the gRNA target area in exon 2B into intron 2B, removing the U1 snRNP recognition site (Supplementary Material, Fig. S6). The other mutations identified are frameshift mutations, which predict early termination of SMN protein translation (Supplementary Material, Fig. S6).

Figure 6

Detection of exon 2B deletion and expression in NSC-34 and iMEFs. (A) Genomic amplicon spanning mouse Smn intron 2A to Smn intron 2B showing a 131 bp deletion in the NSC-34-Smn∆2B (SMN 0%) cell line published by Bernabo et al. 2017 (41). This deletion removes the U1 snRNP recognition site, making an mRNA skipping exon 2B in frame. (B) Stable iMEFs were generated which express Smn cDNA lacking exon 2B (KO/F7;Smn∆2B) and were deleted for flwt-Smn (KO/D7;Smn∆2B). The quantification of expression was performed by ddPCR using a locked-nucleic-acid probe for the Smn∆2B product. Expression is given as percentage of Smn transcripts. KO/F7: 0.57 ± 0.57% RFU, n = 7; KO/F7;Smn∆2B: 37.8 ± 9.3% RFU, n = 5 (^*P < 0.0148 to KO/D7); KO/D7;Smn∆2B: 77.4 ± 9.2% [23–104%] RFU, n = 9 (^****P < 0.0001 to KO/F7, ^**P = 0.0064 to KO/F7;Smn∆2B). For F7 lines the mean ± SEM is given, n = biological replicates. For KO/D7;Smn∆2B, the mean ± SEM [min–max] is given, n = number of individual clones. P-values were calculated by one-way ANOVA with Dunnett’s correction comparing each group to one another. (C) KO/D7;Smn∆2B iMEFs do not make FL Smn mRNA. cDNA amplification from the exon 1- exon 2 junction through the 3’UTR of Smn mRNA in Bernabo et al. NSC-34 cells and KO/D7;Smn∆2B iMEFs. Three products are detectable: flwt-Smn 887 bp, Smn∆7839 bp and Smn∆2B 767 bp. (D) KO/D7;Smn∆2B iMEFs make a truncated Smn protein that migrates at the same size as the Bernabo et al. NSC-34-Smn∆2B cell line. Representative western blots probing for total SMN and α-Tubulin.

Amplification of cDNA from this cell line revealed another truncated band corresponding to a Smn transcript skipping the entirety of exon 2B in-frame (Supplementary Material, Fig. S7). This cDNA lacking exon 2B (Smn∆2B) was cloned into an expression vector and transfected into KO/F7 iMEFs. Stable, expressing lines were obtained, denoted as KO/F7;Smn∆2B, (Fig. 6B) treated with lentivirus expressing iCre and GFP, sorted by FACS and assayed for proliferating colonies deleted for flwt-Smn. Several KO/D7;Smn∆2B colonies were obtained and expressed the mouse Smn∆2B (Fig. 6B). The removal of Smn exon 7 was confirmed in cDNAs generated from these colonies by PCR amplification from Smn exon 2A through the 3′ untranslated region (UTR) (Fig. 6C). KO/D7;Smn∆2B iMEFs do not make a flwt-Smn transcript but do make transcripts corresponding to either skipping of exon 2B or removal of exon 7 (Fig. 6C). Western blots were performed to confirm Smn∆2B migrated similar to the product as described by Bernabo et al. (Fig. 6D). These experiments show the NSC-34 cell line published by Bernabo et al. does make a Smn product—Smn∆2B—and this product is capable of rescuing cell survival in the absence of flwt-Smn. Furthermore, Smn∆2B is a competent snRNP assembler, showing comparable assembly activity to flwt-Smn in KO/D7;Smn∆2B iMEFs (Fig. 7A) as well as in NSC-34 cells (Fig. 7B and C). Thus, exon 2B is not essential for Smn function in cell survival, nor for snRNP assembly, in NSC-34 cells and iMEFs.

Figure 7

Cells solely expressing Smn∆2B have equivalent U4 snRNP assembly activity as flwt-Smn in iMEFs and NSC-34 cells. (A) U4 snRNP assembly activity in WT/F7 and KO/D7;Smn∆2B iMEFs. U4 snRNP assembly activity for each sample is normalized to the activity of the WT/F7 iMEF. WT/F7: 1.00 ± 0.04 RFU, n = 3; WT/D7: 0.83 ± 0.04 RFU [0.500–1.50], n = 4 (^nsP = 0.9101); KO/D7;Smn∆2B: 1.24 ± 0.26 RFU, [0.98–1.49] n = 4 (^nsP = 0.8516). (B) U4 snRNP assembly activity in Bernabo et al. NSC-34 cells. U4 snRNP assembly is normalized to total SMN protein expression. All activities are then normalized to the activity given by the NSC-34 cell line. NSC-34: 1.00 ± 0.02 RFU, n = 3; SMN 20%: 0.75 ± 0.03 RFU, n = 3 (^*P = 0.0489); NSC-34-Smn∆2B: 1.08 ± 0.10 RFU, n = 3 (^nsP = 0.5843). For each sample the mean ± SEM is given. For F7 lines, n = biological replicates. For D7 lines, n = individual clones, and range in activity from the lowest and highest clone is given in brackets. P-values are calculated from a one-way ANOVA with Dunnett’s correction comparing to WT/F7 iMEF or NSC-34 cell line (^nsP > 0.05, ^*P ≤ 0.05). (C) Total U4 snRNP assembly plotted against SMN protein expression for Bernabo et al. NSC-34, SMN 20% and NSC-34-Smn∆2B cell lines. Linear regression was fitted showing strong correlation between snRNP assembly and SMN protein expression in these cell lines indicating Smn∆2B is functionally equivalent to flwt-Smn in regard to snRNP assembly.

Discussion

SMA is caused by mutation or deletion in SMN1 gene and retention of SMN2, which results in SMN protein deficiency (1,2,11,12). SMN2 primarily makes an SMN protein lacking exon 7, which inefficiently oligomerizes and is rapidly degraded (8,10,52). Homozygous deletion of Smn exon 7 in mice is embryonic lethal and is the same phenotype presented by animals homozygous for the Smn null allele Smn^KO (17,42). The embryonic lethality in mice lacking a source of flwt-SMN is consistent with the primary function of SMN in assembly of the spliceosomal snRNPs and the histone mRNA processing U7 snRNP (21–23). Invertebrate and zebrafish Smn knockouts are also lethal, though the large maternal contribution of Smn mRNA in the yolk extends lethality to later stages when the maternal Smn becomes depleted (18,19,53,54).

Consistent with the previous Smn knockout experiments in animals, here we report loss of functional Smn by deletion of exon 7 in iMEFs results in cell death. This is in accordance with other cell models which modulate flwt-SMN expression. In the DT40 chicken pre-B cell line, targeted disruption of the single endogenous chicken Smn gene fails to produce viable clones (38). Therefore, the group first introduced a tet-repressible construct controlling chicken or huSMN production before disrupting the endogenous chicken Smn gene. When SMN is reduced in these cells, the proliferation rate decreases, and when severely reduced, the cells die (38). Another group has shown severe induction of an siRNA against flwt-Smn in mouse NIH-3 T3 cells results in a failure to proliferate (40). Surprisingly, expression of SMN∆7 rescued the DT40 cell line when flwt-SMN is repressed. In fact, virtually all mutations, including loss of SMN exon 7, deletion of the Tudor domain, a complete N-terminal deletion, and Y272C mutation, rescued proliferation of the DT40 cell line (55). This is inconsistent with several studies using these mutant alleles (12,42,43,52,56–58). First, mice exclusively expressing SMN∆7 die early in the embryonic stages, hence this is an embryonic lethal allele (42). In fact, targeted deletion of Smn exon 7 in specific tissues results in destruction of that tissue—including liver, muscle and neurons (43,57). Additionally, SMN∆7 was not detectable in patient lymphoblasts using an antibody raised against the amino acids encoded in exon 7 suggesting SMN∆7 protein is very unstable (52). Furthermore, the Y272C mutation affects SMN oligomerization leading to an unstable SMN mutant, which is degraded (12). Lymphoblasts from patients with this mutation show no increase in SMN protein levels over what is produced by SMN2, suggesting rescue of DT40 cell proliferation cannot be due to the this mutant SMN alone (12). Moreover, the loss of the Tudor domain is critical to the canonical function of SMN in snRNP assembly (56). However, the expression of SMN∆7 in Smn^KO/KO;SMN2^+/+ mice extends survival from 5 to 13 days (58). This is attributed to the over-expression of SMN∆7 and the ability of the SMN∆7 product to interact with WT SMN from SMN2, contributing to oligomer formation inhibiting SMN∆7 degradation (58). In the DT40 cell lines, SMN∆7 could complement with a low level of flwt-SMN leaking from the repressible promoter, contributing to a rescue effect.

On a technical note, assays performed in DT40 cells were concluded after 1 week in culture, whereas we required cells to survive a minimum of 3 weeks. The discordance between our assays may be resolved if the DT40 lines were cultured longer and flwt-SMN expression was fully repressed. However, unlike the DT40 and NIH-3 T3 cell lines, there is no attenuator of wild-type SMN levels in our iMEFs. The removal of flwt-Smn in our iMEFs is binary—either iCre removes Smn exon 7 or it does not. We have created assays that reliably detect the presence of Smn exon 7 at both genomic and transcript levels. Thus, we selected surviving clones that completely lack the capability to produce flwt-Smn. This distinction separates our studies from those previously published, as our system avoids any chance of oligomerization and complementation with flwt-SMN producing a functional complex. Furthermore, our iMEFs do not require constant induction of SMN repression or knockdown, allowing for the selection process to yield stable clones.

Even though SMN2 is absent in 10–15% of the normal human population, SMN1 missense mutations are never found in the absence of SMN2 (15,30). This indicates that SMN1 missense mutations are only functional when they can complement WT SMN produced from SMN2. In fact, the expression of A2G, D44V, A111G, T274I ,or Q282A rescues Smn^KO/KO;SMN2^+/+ mice, but fails to rescue Smn^KO/KO mice (34,35,36). Similar to the results in mice, we find that expression of a single missense allele—A2G, HA-D44V, A111G, E134K, and T274I—fails to rescue iMEF survival when Smn exon 7 is deleted. Yet, certain paired combinations of SMN mutant alleles can rescue Smn^KO/KO mice due to intragenic complementation. Intragenic complementation occurs in oligomeric protein complexes, such as arginosuccinate-lyase (37,59,60) and Calpain 3 (61), in which heteromers exhibit function not replicated by homomers of each respective mutant singly. Presence of the A111G mutant with either T274I or Q282A can completely rescue survival of Smn^KO/KO mice resulting in normal electrophysiological properties of the motor unit (V.L. McGovern et al., manuscript in preparation). Our iMEFs were used to extend these results showing multiple pairs of huSMN missense alleles can rescue cell survival and snRNP assembly when co-expressed. The huSMN protein levels in several of the huSMN missense mutant co-expressing KO/D7 iMEFs is not greater than iMEFs expressing a single missense mutant (Fig. 4B,Table 2), indicating rescued survival is due to the presence of a second allele and not merely due to expression levels of SMN.

All mutant allele pairs that do show complementation and survive are also capable of snRNP assembly, which suggests that snRNP assembly is the essential function of SMN in the cell. It is well established that the SMN complex is necessary for snRNP assembly. Structures have been determined for the SMN Tudor domain using nuclear magnetic resonance and the 6S complex using cryo-electron microscopy, though details of what specific functions are performed by different SMN domains in the assembly reaction remain vague (62–65). Our study provides insight into how the SMN oligomers interact within the full SMN complex. SMN has been shown to have at least 3 distinct functional domains: Gemin2-binding, Tudor, and a self-oligomerizing domain (56,66,67). Our study shows there may be additional domains. For instance, rescue when co-expressing the A2G mutation with HA-D44V indicates an additional domain on the SMN N-terminus distinct from the Gemin2-binding domain. However, the A2G mutation does not complement T274I, indicating the N- and C-termini between different SMN monomers may interact to perform some concerted function. Though T274I occurs in the oligomerization domain, the isoleucine substitution only weakly affects oligomerization potential (8,68,69). Co-expressing the T274I mutation with HA-D44V as well as the Tudor domain mutations A111G and E134K results in survival of the cell, further highlighting the SMNT274I oligomerization potential.

Interestingly, E134K complemented the A2G, A111G, and T274I alleles. E134K is a severe mutation that does not disrupt the structure of SMN and occurs in patients whose clinical presentation tightly associates with their SMN2 copy number (70,71). The E134K mutation affects the charge density on the binding surface of the Tudor domain and has been shown to be a weak snRNP assembler in in vitro snRNP assembly assays (72,73). Consistent with E134K disrupting snRNP assembly, it does not produce viable cells when expressed on its own. Conversely, SMN missense hetero-oligomers, which include E134K, exhibit snRNP assembly activity but this activity does not have a positive linear relation with total SMN abundance (Fig. 5B and D). Currently, we do not know the exact reason for this lack of linearity. It is interesting that this was also observed in two of the combinations—A111G + D44V and A111G + E134K—in which one of the expressed mutant alleles is <10% of the total huSMN mRNA composition. This suggests the composition of the hetero-oligomer may have an effect on snRNP assembly and that a disproportionate amount of E134K within the hetero-oligomer may have an inhibitory effect. However, given the complexity of this question, we are currently unable to determine whether this is, in fact, the case and may be better addressed in in vitro systems (73,74).

We found that the NSC-34 cell line reported to produce no Smn makes an Smn lacking exon 2B (41). Expressing Smn∆2B in iMEFs rescues cell survival and is equivalent to flwt-Smn at assembling the U4 snRNP. Interestingly, the region encoded by exon 2B contains di-lysine motifs responsible for the binding to α-COP and subsequent transport of SMN and mRNAs to neuronal axon terminals (39,75–77). The mutation of these di-lysine motifs to alanine results in the ablation of α-COP binding and when expressed in SMA mice showed no benefit to survival (77). We found Smn∆2B assembles snRNPs, but we did not determine if the mutation of just the di-lysine residues results in an allele capable of snRNP assembly. We would predict that the expression of Smn∆2B in SMA mice would rescue snRNP assembly uncoupled from α-COP binding function within motor neuron axons, determining the contribution of each pathway to SMA pathogenesis. Exon 2B is also one of the least conserved exons in Smn, with 45% of residues conserved between mouse and zebrafish, decreasing to 18% in Drosophila and 0% in Caenorhabditis. elegans (78,79). Additionally, no non-truncating SMA causing mutations have been identified, further suggesting this exon is not essential for SMN function (80).

In summary, our KO/F7 iMEF cell lines can determine which SMN alleles are capable of essential function in the absence of flwt-Smn. We showed that SMA patient SMN1 missense mutations lack the ability to rescue KO/D7 iMEF survival on their own, but essential function can be restored by co-expression with other mutant SMN alleles. Rescue by co-expression of two mutant huSMN alleles occurs through intragenic complementation and is not due to increased expression of the mutant huSMN protein. Alleles, which rescue survival in our cells, like Smn∆2B, can also be assayed in SMA mice to define functions that are critical to SMA pathogenesis. Furthermore, iMEFs expressing a single huSMN missense mutant can be screened for suppressors, identifying mutations in associated or substrate proteins, which can restore essential function to the nonfunctional huSMN missense mutant. These can then be tested in SMA mouse models to determine how the rescue of that function contributes to SMA pathogenesis.

Materials and Methods

MEF isolation and immortalization

Smn^WT/F7;tdTomato mice were crossed with Smn^WT/KO;tdTomato mice to yield embryos with the Smn^KO/F7;tdTomato or Smn^WT/F7;tdTomato genotype. Smn^WT/F7;tdTomato mice were crossed with Smn^WT/KO;SMN2^+/+ mice to yield embryos with the Smn^KO/F7;tdTomato;SMN2^+/− genotype. Embryos were harvested 10–12 days post-conception. The head and visceral organs were removed and remaining tissue was minced and dissociated with 0.5% trypsin–ehtylenediaminetetraacetic acid (EDTA). Trypsin activity was quenched with 12% fetal bovine serum in Dulbecco’s Modified Essential Medium (FBS DMEM) and cells were pelleted by centrifugation at 200 x g for 4 min. Supernatant was decanted, and the pellet was resuspended in 12% FBS DMEM supplemented with L-glutamine, Penicillin and streptomycin. Cells were seeded then incubated at 37°C, 5% CO₂ until confluent. MEFs were immortalized by transduction with retrovirus delivering human papilloma virus genes E6 and E7 at passage 12–15. As neomycin resistance was conferred by the Smn^KO insertion allele, we could not use neomycin resistance to select infected immortalized cells. Rapidly proliferating lines were kept post-infection yielding iMEFs. iMEFs were cultured in 12% FBS DMEM supplemented with L-glutamine, Penicillin and streptomycin (12% FBS Atlanta S11550, 10% PenStrep/L-Glutamine Gibco 10 378-016, DMEM Gibco 11 960 077) under 37°C, 5% CO₂. Cells were routinely passaged every 2–3 days.

Genotyping and expression assays

DNA isolation and purification were performed by incubating cells in sodium dodecyl-sulfate (SDS)-lysis solution supplemented with proteinase K at 55°C overnight, followed by phenol:chloroform extraction and ethanol precipitation (81). Primers used to genotype SMN alleles are found in Supplementary Material, Figure S8. Assays for SNP detection and ddPCR are found in Supplementary Material, Figure S2. RNA samples were isolated from cells with TRIzol reagent and ethanol precipitation followed by DNaseI-treatment as described by the manufacturer (Invitrogen, 15 596 026 and Ambion, AM2222). cDNA synthesis was performed with AMV-RT (NEB, M0277L) according to the manufacturer’s instructions. Primers and assays used for cDNA detection are found in Supplementary Material, Figure S5. Droplet-digital PCR (ddPCR) using BioRad QX-200 was performed according to the manufacturer’s instructions. Additionally, a 0.25 μM non-extending oligonucleotide was annealed at 50°C before the initial denaturing step to increase the specificity of the LNA probe (82,83). A concentration of 5 ng–300 ng per reaction of genomic DNA template was sufficient. All expression assays used 1 μL–4 μL cDNA.

Lentivirus production and infection

pCDH-CB-copGFP-T2A-iCre, psPAX2 and pMD2.1 plasmids were purchased from Addgene (cat#72256, 12 260, 12 295, respectively). Low passage 293HEK cells were cultured and expanded, using 10% FBS DMEM media supplemented with L-glutamine, Penicillin and streptomycin, passaging with 0.5% Trypsin–EDTA and incubators at 37°C, 5% CO₂. 293HEKs were expanded to 30–36 15 cm dishes for lentivirus production. Plasmids were added together in the following amounts: 60 μg pCDH-CB-copGFP-T2A-iCre, 45 μg psPAX2 and 18 μg pMD2.1 per 15 cm culture. Cultures were transfected with overnight incubation of plasmids and cells in 300 mM CaCl₂ and 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES)-buffered saline, pH 7.05. Culture media was replaced with 2% FBS DMEM media, and virus was collected every 24 h over the next 4 days. Media was filtered through low-protein bind polyethersulfone (PES) 0.40 μm filter flask followed by ultracentrifugation at 75 000 x g for 2 h at 4°C. Viral pellets were resuspended and pooled in 2 ml 2% FBS media or in 10% glycerol-phosphate-buffered saline for final resuspension. Virus was titered by treating 20 000 cells from the WT/F7 iMEFs with 1–20 μL aliquots. Cells were cultured for 72 h after which they were analyzed by FACS (BD Aria III) to determine the percentage of GFP and tdTomato double-positive cells. The amount of virus used in colony assays was determined from the lowest volume that gave ~ 70% GFP and tdTomato double-positive population.

Colony assay

Transduced iMEFs were seeded at 20 000 cells per well in a 24-well dish with 300 μL culture volume. The titrated amount of Lv-CB-copGFP-T2A-iCre was added to each well (see lentivirus production method). Cells were incubated at 37°C, 5% CO₂ in conditions for 72 h. Remaining cells were sorted by FACS to enrich for GFP and tdTomato double-positive population using BD Aria III. A total of 5000 to 20 000 cells were seeded into 10 cm dish and allowed to grow for 3 weeks. Surviving red and green colonies were picked and transferred into 96-well plates to keep clonal populations separate. Colonies were expanded into larger well sizes for analysis of exon 7 deletion, RNA expression, genotyping and protein quantifications.

Cloning of SMN missense mutations

SMN2 cDNA including the 3′ UTR was cloned between the BamHI and EcoRI sites within pcDNA3/Neo vector. Missense mutations A2G, D44V, A111G, E134K and T274I were made by site-directed mutagenesis (34–36,51). All are cloned 3′ of the standard CMV promoter except for T274I, which is driven by the 4.1 kb SMN promoter (84). The HA-D44V cDNA is the only mutant used in which the amino HA-tag was not removed by restriction digest, resulting in SMN protein including an extra 25 amino acids (Supplementary Material, Fig. S1). Smn∆2B cDNA was cloned into pcDNA3 using NEB HiFi DNA assembly reaction (NEB, E2621S) in between BamHI and EcoRI restriction sites. Bleomycin and hygromycin resistance genes were cloned in place of the neomycin resistance gene between the AvrII and BstBI restriction sites for each missense allele by conventional restriction enzyme approaches as well as through the NEB HiFi DNA assembly reaction.

Generating stable cell lines

Parent KO/F7 or single huSMN missense mutant KO/D7 iMEFs were seeded in 24-wells and grown to 70% confluency. Cells were transfected with FspI, MluI or ScaI linearized plasmids containing conferring A2G, HA-D44V, A111G, E134K, T274I or Smn∆2B mutants using Invitrogen Lipofectamine 3000 (500 ng plasmid, 1 μL P3000 reagent and 1.5 μL Lipofectamine 3000). Cells were incubated in OptiMEM overnight at 37°C, 5% CO₂ (ThermoFisher, 31 985 062). Cells were then passaged into 10 cm dishes supplemented with 150–200 μg/ml hygromycin (ThermoFisher, 10 687 010) or 500 μg/ml Zeocin (ThermoFisher, R25005). Hygromycin was supplemented every other day, zeocin was supplemented every 2 days. A total of 30–50 colonies were picked for each condition and seeded into 96-well plates.

Western analysis

Whole cell lysates were prepared from iMEF pellets in 6% SDS-blending buffer (81). Samples were sonicated for 30 s, boiled 5 min and centrifuged at 10 000 x g to remove debris. Protein concentration was quantified by bicinchoninic acid assay (BCA) (ThermoFisher, 23 250) and the assay was read at 562 nm on the Synergy HT spectrophotometer (BioTek) using Gen5 software. A total of 50 μg total protein was mixed with 5x loading buffer and loaded into wells of a 6% stacking gel and 12% separating denaturing gel. Electrophoresis was performed at 20 mV through the stacking gel and 60 mV through the separating gel. Transfer was performed in standard glycine-methanol buffer onto polyvinylidine fluoride (PVDF)-FL or PVDF-P membrane at 24 V for 1.5 h (Millipore, IPFL00010, IPVH00010). Membranes were immediately incubated with 5% dry milk, 0.2% Tween-20 PBS blocking solution for 1 h. Antibody treatments were all performed in 1/5th blocking solution 0.2% Tween-20 PBS. Primary antibodies: mouse anti-SMN at 1:1000 (BD, 610647), anti-SMN2 clone SMN-KH at 1:2000 (Millipore, MABE230) (85), mouse anti-α-Tubulin at 1:15 000 (Sigma, T8203). Primary antibody solutions were incubated overnight at 4°C. Secondary antibody incubations lasted 1 h at room temperature, using donkey anti-mouse at 1:15 000 IRDye800CW (ODYSSEY) or donkey anti-mouse-HRP at 1:15 000 (Jackson Immunology, 20 404). Washes were performed with 0.5% Tween-20 PBS. SuperSignal West or Fempto ELISA Chemiluminescent Substrate Kit was used for secondary antibody detection (Pierce, 34 080, 37 075, respectively). Images were captured using Sapphire Molecular Imager (Azure Biosystems) and bands were quantified in AzureSpot 2.0 using rolling-ball method of background subtraction. Results are recorded as volume under peak for the ~ 38 kDa SMN band normalized to the α-Tubulin 52 kDa band unless otherwise stated. For representative blots in Figure 2D, membranes were probed with the huSMN detecting KH antibody and α-Tubulin first, then stripped and re-probed with the BD antibody and α-Tubulin. All other representative blots were performed on samples blotted in parallel.

snRNP assembly assay

snRNP assembly reactions were carried out as described by Wan et al. 2005 (25). The Y12 antibody used is from LS Bio (B8621) at 4 μL per reaction. Luminescence was measured with the Tecan Infinite F200 using I-Control software, no attenuation, 1000 ms integration, 0 ms settle. Luminescence for U4 snRNP assembly was divided by the SMN protein expressed in 50 μL of cytoplasmic extract, then normalized to the activity of the KO/D7;SMN2 clones, WT/F7 iMEF or NSC-34 cell line.

Statistical analysis

All statistics were performed in GraphPad Prism version 8.4.3. Unpaired parametric student t-test was used to determine if two means were statistically different from one another (Figs 2B, C, E and F,3B and C, and Supplementary Material, Figs S3 and S4). For comparing the means of multiple groups to a single sample, one-way analysis of variance (ANOVA) was performed with Dunnett’s correction for multiple comparisons (Figs 4B,5A and 7A and B,Table 2,Table 3). For comparing the means of multiple groups to each other, one-way ANOVA was performed with Dunnett’s correction for multiple comparisons (Fig. 6B). (^nsP > 0.05, ^*P ≤ 0.05, ^**P ≤ 0.01, ^***P ≤ 0.001, ^****P ≤ 0.0001).

Acknowledgements

We would like to thank Eliot Androphy for gifting cells that produce the immortalizing retrovirus. Furthermore, we thank Aurelie Massoni-Laporte, Eileen Workman and Narasimhan Madabusi for assistance in cloning. We thank Alex Cornwell and Bryan McElwain of The Ohio State University Wexner Medical Center Flow Cytometry Core for their expertise. Additionally, Eileen Workman assisted with the snRNP assembly assays. Lastly, kind regards to Carlos Miranda, Kathrin Meyer, Shibi Likhite for guidance in producing lentivirus and use of the GE InCell imager.

Conflict of Interest statement. None declared.

Funding

National Institutes of Child Health and Human Development (NICHD) (R01HD060586 to A.H.M.B.); CureSMA (BUR1617 to A.H.M.B.); the Marshall Heritage Foundation.