Characterization of RNA binding properties of RBM5 triple domains

Next, I wanted to study the RNA binding characteristics of RRM1-Zf1-RRM2 C191G protein. I used the prior knowledge about the RNA sequence specificities for recognition via different domains. I know that the Zf1 domain requires a GG motif for RNA binding while RRM1 can bind to a C/U rich RNA sequence. On the other hand, Song et al. (Song, Wu et al.

2012) showed that RRM2 can bind to a C/U or A/G rich RNA sequence with similar binding affinity. They had also used the intronic region upstream of ln100 element in Caspase-2 pre-mRNA to derive the RNA sequences (5’-CUCUUC-3’ and 5’-GAGAAG-3’).

The RNA sequence used for characterization of RNA binding properties of RRM1-Zf1 is GGCU_12 (5’-UGGCUCCUUCUCU-3’), consisting of 12 bases which should be long enough for binding to all three domain as well. I designed another 13 bases long RNA sequence by extending the 5’ site and shortening the 3’ end- ne_GGCU_13 (5’-GAACUUGGCUCUU-3’). The idea behind testing these two different RNA sequences for binding to the triple domain construct was to achieve domain reorientation in case of linear recognition of the RNA. This simply means that due to the specificity in RNA recognition provided by Zf1, the other two domains have to re-organize with respect to each other depending on the availability of RNA bases as illustrated in Figure 62.

Figure 62 Hypothetical model of RNA recognition

A hypothetical model of RNA recognition by the triple domain construct whereby base-pairing specificity is provided by the Zf1 domain which recognizes the GG motif is for GGCU_12 RNA (A) and ne_GGCU_13 (B). RRM1, Zf1 and RRM2 are color coded in blue, green and pink while GGCU_12 and ne_GGCU_13 RNAs are denoted with red and light green colors, respectively.

To learn if the two different RNA sequences are recognized differently by the triple domain (RRM1-Zf1-RRM2 C119G) protein, titrations were made into the protein and ¹H,¹⁵

N-143

HSQC spectra were subsequently recorded (Figure 63). The complex formation between the protein and RNA occurs in fast to intermediate regime on the NMR chemical exchange timescale whereby some signals shift in position with the titration point while some others disappeared during the titration. There are also a handful residues which show two amide signals with different relative intensities. The overall peak intensities for the residues becomes weak owing to exchange broadening processes.

A comparison of the CSP plots of the RRM1-Zf1-RRM2 C191G protein bound to GGCU_12 RNA and ne_GGCU_13 RNA indicates that apart from a few local differences in residues Met 132, Val 138 and Phe 144 of RRM1, both the RNA sequences show similar overall chemical shift perturbation plots. It is noteworthy that if the different RNA sequences would have caused a domain re-organization as I expected, it would be clearly visible in the linker between Zf1 and RRM2 (residues 211-230).

It is therefore unlikely that the RNA is recognized in a linear fashion by the protein.

Additionally, we also learn that the presence of purines instead of pyrimidines at the 5’-end of the RNA sequence is also tolerated and the bases are recognized in a similar manner. From the CSP plots, it also becomes clear that there are chemical shift changes observed in the linker between Zf1 and RRM2 domains, indicating either a direct interaction of the linker with RNA or due to allosteric changes.

144

Figure 63 Overlay of free and RNA bound ¹H,¹⁵N-HSQC spectra of RRM1-Zf1-RRM2 C191G

(B) An overlay of ¹H,¹⁵N-HSQC spectra of the triple domain construct in free form, in complex with GGCU_12 and ne_GGCU_13 RNAs is shown in black, red and green, respectively. (B) The chemical shift perturbation plots between the free and different RNA bound forms of triple domain construct are shown, color coded according to the ¹H,¹⁵N-HSQC spectra in red and green , respectively.

Next, I investigated if the C/U rich RNA recognition is conserved in the single domains (RRM1 and RRM2), in the context of the triple domain RRM1-Zf1-RRM2 C191G. To this end, an overlay of CSP plots of RRM1/RRM2 bound to C/U rich RNA and RRM2-Zf1-RRM2 C191G bound to an RNA containing C/U rich and GG motifs (GGCU_12) was made (Figure 64A). The overall pattern of CSPs is conserved between the single versus triple domain constructs.

145

Figure 64 CSP plots comparing free and RNA bound forms of RBM5 protein constructs

(A) The chemical shift perturbation plots of free and RNA bound forms of RRM1 (bound to C/U rich RNA), RRM2 (bound to C/U rich RNA) and RRM1-Zf1-RRM2 C191G (bound to GGCU_12 RNA) are shown in black, cyan and red, respectively. (B) The chemical shift perturbation plots of free/RNA bound forms of RRM1-Zf1 C191G and RRM1-Zf1-RRM2 C191G bound to GGCU_12 RNA are shown in purple and red, respectively.

A similar comparison of the tandem domains RRM1-Zf1 C191G free and RNA bound versus triple domains RRM1-Zf1-RRM2 free and RNA bound proteins is made in the presence of the same RNA (GGCU_12) in Figure 64B. It is apparent that the chemical shifts occurring upon RNA binding are different in the tandem and triple domain constructs, with greater changes in the Zf1 than in RRM1. Since we have already seen that there are large changes between the tandem and triple domain constructs already in the free form, it could well be true that these differences are also translated into the RNA bound forms. As stated previously, these changes could either be a result of additional inter-domain (or between with the linker between Zf1 and RRM2) contacts in the triple domain construct which are absent in the tandem domain and/or due to disturbance of RRM1-Zf1 interface in the triple domain construct. In section 5.6, the possibility of partial destabilization of the RRM1-Zf1 interface in the RNA bound form of the tandem domains was discussed. In light of the differences between NMR titration data of the triple domains versus tandem domains with RNA, it becomes tempting to speculate that the RRM1-Zf1 interface exists in the free form of tandem domain RRM1-Zf1 while being at least

146

partly disturbed in the presence of RNA or in the context of the triple domain construct (in both free and RNA bound forms). Validation of this hypothesis by obtaining either high resolution structure of the tandem domains-RNA complex or structural models of the triple domains (free/RNA bound) using PREs, RDCs and SAXS would be essential for conclusion of this study.

Figure 65 ITC binding isotherms of RRM1-Zf1 and RRM1-Zf1-RRM2 C191G mutants

ITC binding isotherms of tandem domains RRM1-Zf1 and triple domains RRM1-Zf1-RRM2 C191G mutants for binding to GGCU_12 RNA are shown in (A) and (B), respectively. The binding dissociation constants are indicated.

Furthermore, ITC binding isotherms were acquired to obtain the binding affinities of tandem and triple domain C191G mutants for GGCU_12 RNA (Figure 65). A 2-fold gain in affinity for GGCU_12 RNA is observed in the presence of the additional RRM2 domain in triple domain construct compared to the tandem domain construct. The gain in affinity due to RRM2 indicates that all three domains are able to bind the RNA, consistent with NMR titration data. In theory, it is expected that the addition of individual domains should have a multiplicative effect on the gain in affinity, which is clearly not the case here. This could be possibly attributed to either the use of a sub-optimal RNA for binding studies or the partial occlusion of one of the domains in the triple domain construct which could hinder RNA binding.

147