**3. Peptide fusion inhibitors target the gp41 NHR core**

Peptides, especially C-peptides (sequence see Fig. 5), can efficiently block the gp41 NHR-CHR interaction to inhibit HIV-cell membrane fusion and infection. They act in a dominant-negative manner by binding to the transiently exposed coiled-coil N-peptide region in the PHI (Eckert & Kim 2001). The wild-type C-peptide sequences have been shown to have low nanomolar IC50 values for HIV-1 ENV mediated membrane fusion and viral infection. Peptide engineering has been employed on wild-type C-peptide sequences to obtain structure activity relationship (SAR) data for the peptide fusion inhibitors, resulting in peptides with an improved anti-HIV profile and a better understanding of the mechanism of gp41 mediated virus-cell membrane fusion (Otaka et al. 2002; Dwyer et al. 2007). The insight gained from these works was finally tested by the artificial design of peptide fusion inhibitors with few sequence homologies to natural peptides or protein sequences (Qi et al. 2008; Shi et al. 2008).

#### **3.1 Peptides from the wild-type gp41 sequence**

44 Biochemistry

generally applicable to different interacting peptide pairs, as long as the two peptide sequences are matched. This has been confirmed by the development of a longer gp41 Npeptide/C-peptide pair, Env5.0 and CP5; the peptide pair showed nanomolar binding affinity and can be used to screen more potent fusion inhibitors. Env5.0 contains the whole groove, and it has been used to identify ligands that interact with the range of the groove outside of the deep pocket by designing suitable probes (Cai et al. 2009). In addition, Env2.0 has been successfully used as a target for a screening assay to identify small molecule fusion

In summary, peptides have been used to construct the HIV-1 gp41 fusion core, which is a 6- HB. Crystal structure analysis showed that the gp41 NHR trimer forms an interior core, which contains three hydrophobic grooves as the binding site for C-peptide. A deep pocket in the groove is a hot spot for the NHR-CHR interaction, and can be a target for small molecule fusion inhibitors. The NHR target can be constructed by adding physicochemical constraints in the N-peptides to promote the formation of a soluble and discrete NHR trimer, which can be

Peptides, especially C-peptides (sequence see Fig. 5), can efficiently block the gp41 NHR-CHR interaction to inhibit HIV-cell membrane fusion and infection. They act in a dominant-negative manner by binding to the transiently exposed coiled-coil N-peptide region in the PHI (Eckert & Kim 2001). The wild-type C-peptide sequences have been shown to have low nanomolar IC50 values for HIV-1 ENV mediated membrane fusion and viral infection. Peptide engineering has been employed on wild-type C-peptide sequences to obtain structure activity relationship (SAR) data for the peptide fusion inhibitors, resulting in peptides with an improved anti-HIV profile and a better understanding of the mechanism of gp41 mediated virus-cell membrane fusion (Otaka et al. 2002; Dwyer et al. 2007). The insight gained from these works was finally tested by the artificial design of peptide fusion inhibitors with few sequence homologies to natural peptides or protein sequences (Qi et al. 2008; Shi et al. 2008).

used for screening HIV-1 fusion inhibitors targeting gp41 NHR or the deep pocket.

**3. Peptide fusion inhibitors target the gp41 NHR core** 

inhibitors (Cai & Gochin 2007; Zhou et al. 2010).

Fig. 4. Designed soluble and discrete NHR target.

The first highly potent HIV-1 fusion inhibitors were independently discovered by two groups in the early 1990s, including SJ-2176 (gp41630-659) (Jiang et al. 1993; Jiang et al. 1993) and DP178 (gp41638-673, later named T20) (Wild et al. 1994), which were both derived from the gp41 CHR wild-type sequence. Due to their stronger anti-HIV activity compared with N-peptides, most of the exploited fusion inhibitors were C-peptides, among them, T20 (**7**) and C34 (gp41628-661, **8**) were extensively studied. C-peptide fusion inhibitors are usually unstructured in solution by themselves, and form α-helical structures in a 6-HB after interaction with NHR.

T20, originally named DP178, was developed into the first HIV-1 fusion inhibitor with the brand name Enfuvirtide (Lazzarin et al. 2003; Walmsley et al. 2003; Su et al. 2004). It has low nanomolar antiretroviral activity. Under physiological conditions, it is unstructured and cannot form a stable 6-HB with N-peptide; however, it is highly soluble, making it a good drug candidate. Its mechanism of action has been controversial until now, since it cannot form a 6-HB with N-peptide, which is an established interaction model of HIV-1 peptide fusion inhibitors that has been supported by x-ray crystallography (Chan et al. 1997). T20 does not contain the WWI motif necessary to bind with the primary NHR deep pocket. This may account for its relatively weak binding with NHR and the resulting loss of activity against emerging drug resistant HIV-1 isolates. The 8-residue C-terminus of T20 contains three Trp residues and is highly hydrophobic, which enables T20 to bind with the lipid membrane; thus, this 8-residue motif is called the lipid binding domain (LBD). The hydrophobic residues in the LBD are critical for T20 to maintain high anti-HIV activity, although the LBD elicits no anti-HIV activity by itself. It seems that T20 may interact with both the gp41 NHR groove and the lipid membrane to interfere with 6-HB formation, thus inhibiting HIV-1 infection (Liu et al. 2005; Liu et al. 2007).

C34 forms a stable 6-HB with NHR, thus preventing productive 6-HB formation, a mechanism well supported by x-ray crystallography (Chan et al. 1997). It also displays stronger antiretroviral activity than T20, while its poor solubility under physiological conditions hinders it as a promising drug candidate (Otaka et al. 2002). Like all other wild-type gp41 Cpeptides, C34 is unstructured under physiological conditions, while it adopts a nearly full αhelical structure when interacting with N-peptide to form a 6-HB. C34 contains the WWI motif, so it can interact with the primary binding pocket in NHR to form a stable complex with N-peptides. N-PAGE has shown that C34 can form a stable 6-HB in the presence of N36 or N46; and thermal denaturation has shown that the N36/C34 complex displays typical twostate denaturation behavior with a *Tm* value of ~61 °C (Pan et al. 2009). In a 6-HB, C34 uses residues *a* and *d* to interact with NHR. The *a* and *d* residues in the N-terminal half of C34 are uniformly hydrophobic and elicit a predominantly hydrophobic interaction with residues *e* and *g* in NHR and bury these residues in the 6-HB; and the *a* and *d* residues in the C-terminal half of C34 form a hydrophilic layer spanning four α-helical turns, which is assumed to match the similar hydrophilic layer in the related NHR sequence. Thus, C34 is widely used as a tool to study the mechanism of HIV fusion inhibitors, as well as the lead or template for next generation fusion inhibitor design, and will be discussed in Section 3.2.

CP32 (gp41621-652, **14**) is another identified highly potent wild-type gp41 C-peptide fusion inhibitor that targets NHR sequences other than T20 and C34 (He et al. 2008). It contains a 7 residue motif upstream of the C34 sequence. Interestingly, the CP32 sequence matches the T21 sequence, the first identified peptide HIV-1 fusion inhibitor from gp41 under the name DP107 (Wild et al. 1992); the match is expected based on the N36/C34 complex and the antiparallel interactions between gp41 NHR and CHR. CP32 contains the WWI motif, so it can interact with the NHR deep pocket to form a stable 6-HB. The CP32/T21 complex is ~100% α-helical with a *Tm* of 82 °C, which is more stable than the N36/C34 complex. The discrete 6- HB conformation of the CP32/T21 complex was supported by N-PAGE, size exclusion chromatography, as well as analytical ultracentrifugation. CP32 showed an IC50 value of 4.2 nM against HIV-1 ENV mediated cell-cell fusion and an IC50 value of 4.6 nm against HIV-1NL4-3wt infection of MT-2 cells. Although it has a similar potency as C34 against wild-type HIV-1 isolates, CP32 is ~20-fold and >500-fold more potent than C34 and T20, respectively, against the drug-resistant HIV-1NL4-3-V38SE/N42S isolate, possibly due to the fact that it targets a different sequence in gp41 NHR.

There are also longer C-peptides fusion inhibitors, such as C43 and C52, which include both partial or complete sequences of C34 and T20; however, none of these longer peptides have improved anti-HIV potency compared with T20 and C34 (Deng et al. 2007). In the PHI, a long groove may expand throughout the gp41 NHR and beyond, and it may be targeted by its CHR counterpart. The C-terminal half of the gp41 ectodomain may make contact with the Nterminal half at a certain time during fusion processes, so the C-peptide sequence may expand to the whole C-terminal half of the gp41 ectodomain and interact with the PHI to inhibit gp41 mediated virus-cell membrane fusion. The energy along the gp41 NHR-CHR interface is not evenly distributed; the WWI motif and the LBD serve as hot spots in the gp41 NHR-CHR interaction. A highly potent C-peptide fusion inhibitor must contain at least the WWI motif or the LBD; in addition, a suitable length of total peptide sequence is required to provide additional interactions in the NHR groove to stabilize the C-peptide-NHR interaction. Though they form stable α-helical structures in the 6-HB, C-peptides and N-peptides from the wildtype gp41 sequence are usually unstructured in solution; thus, they are prone to proteolysis. The viral strains resistant to T20 also required the development of a highly potent fusion inhibitor to overcome drug resistance. The use of protein/peptide engineering to improve the physicochemical properties of the wild-type C-peptide sequence and to increase the stability of the C-peptide-NHR complex is discussed below.

#### **3.2 Engineered peptides**

New generations of peptide fusion inhibitors have been developed by engineering C34 related sequences in order to increase the *in vivo* stability and NHR binding affinity, and to overcome T20 resistance. It is well accepted that increasing the helicity of the peptide fusion inhibitor will increase its antiretroviral potency by increasing its binding affinity with NHR and the *in vivo* stability (Otaka et al. 2002). In a 6-HB, C-peptides interact with NHR with their *a* and *d* residues, which are considered to be critical for molecular recognition between CHR and NHR; while amino acid residues at the *b*, *c*, *f*, and *g* positions are exposed to solution and are not considered to be critical for the gp41 NHR and CHR interaction (Chan et al. 1997). However, the solvent exposed residues have a global effect on the solubility, stability, and other physicochemical properties of the C-peptides, so they affect the *in vivo* activity and the druggability of peptide fusion inhibitors. Salt bridges and helical enhancers have been engineered by replacing the solvent exposed residues with the desired residues in order to get more potent HIV-1 fusion inhibitors.

T21 sequence, the first identified peptide HIV-1 fusion inhibitor from gp41 under the name DP107 (Wild et al. 1992); the match is expected based on the N36/C34 complex and the antiparallel interactions between gp41 NHR and CHR. CP32 contains the WWI motif, so it can interact with the NHR deep pocket to form a stable 6-HB. The CP32/T21 complex is ~100% α-helical with a *Tm* of 82 °C, which is more stable than the N36/C34 complex. The discrete 6- HB conformation of the CP32/T21 complex was supported by N-PAGE, size exclusion chromatography, as well as analytical ultracentrifugation. CP32 showed an IC50 value of 4.2 nM against HIV-1 ENV mediated cell-cell fusion and an IC50 value of 4.6 nm against HIV-1NL4-3wt infection of MT-2 cells. Although it has a similar potency as C34 against wild-type HIV-1 isolates, CP32 is ~20-fold and >500-fold more potent than C34 and T20, respectively, against the drug-resistant HIV-1NL4-3-V38SE/N42S isolate, possibly due to the fact that it targets a

There are also longer C-peptides fusion inhibitors, such as C43 and C52, which include both partial or complete sequences of C34 and T20; however, none of these longer peptides have improved anti-HIV potency compared with T20 and C34 (Deng et al. 2007). In the PHI, a long groove may expand throughout the gp41 NHR and beyond, and it may be targeted by its CHR counterpart. The C-terminal half of the gp41 ectodomain may make contact with the Nterminal half at a certain time during fusion processes, so the C-peptide sequence may expand to the whole C-terminal half of the gp41 ectodomain and interact with the PHI to inhibit gp41 mediated virus-cell membrane fusion. The energy along the gp41 NHR-CHR interface is not evenly distributed; the WWI motif and the LBD serve as hot spots in the gp41 NHR-CHR interaction. A highly potent C-peptide fusion inhibitor must contain at least the WWI motif or the LBD; in addition, a suitable length of total peptide sequence is required to provide additional interactions in the NHR groove to stabilize the C-peptide-NHR interaction. Though they form stable α-helical structures in the 6-HB, C-peptides and N-peptides from the wildtype gp41 sequence are usually unstructured in solution; thus, they are prone to proteolysis. The viral strains resistant to T20 also required the development of a highly potent fusion inhibitor to overcome drug resistance. The use of protein/peptide engineering to improve the physicochemical properties of the wild-type C-peptide sequence and to increase the stability of

New generations of peptide fusion inhibitors have been developed by engineering C34 related sequences in order to increase the *in vivo* stability and NHR binding affinity, and to overcome T20 resistance. It is well accepted that increasing the helicity of the peptide fusion inhibitor will increase its antiretroviral potency by increasing its binding affinity with NHR and the *in vivo* stability (Otaka et al. 2002). In a 6-HB, C-peptides interact with NHR with their *a* and *d* residues, which are considered to be critical for molecular recognition between CHR and NHR; while amino acid residues at the *b*, *c*, *f*, and *g* positions are exposed to solution and are not considered to be critical for the gp41 NHR and CHR interaction (Chan et al. 1997). However, the solvent exposed residues have a global effect on the solubility, stability, and other physicochemical properties of the C-peptides, so they affect the *in vivo* activity and the druggability of peptide fusion inhibitors. Salt bridges and helical enhancers have been engineered by replacing the solvent exposed residues with the desired residues in

different sequence in gp41 NHR.

the C-peptide-NHR complex is discussed below.

order to get more potent HIV-1 fusion inhibitors.

**3.2 Engineered peptides** 

T1249 (**13**) was developed by Trimeris as a second generation peptide HIV-1 fusion inhibitor after T20 (Miralles et al. 2003; Eggink et al. 2008; Pan et al. 2009). It was designed to include both hot spots, the WWI motif of C34 and the LBD of T20. To keep the peptides a suitable length, the seven residues following the WWI motif were considered to be not critical for the NHR interaction and were deleted; thus, the WQEWEQKI motif remained. It also contained the conserved amino acid residues from SIV and HIV-2 that are essential for fighting contains the T20 resistant virus. In addition, alanine substitutions and salt bridges were added to increase the α-helicity, resulting in a 39-mer highly mutated peptide based on the wild-type HIV-1 gp41 sequence. T1249 showed enhanced antiretroviral activity against the T20 resistant virus, and ~50% α-helicity compared to the unstructured character of the wild-type gp41 peptide. It entered into phase II clinical trials, but it was terminated due to side effects (Miralles et al. 2003).

T1144 (**9**) and T2635 (**31**) are third generation peptide fusion inhibitors developed by Trimeris (Pan et al. 2011). They fully exploited the strategy used in the development of T1249, however, they are based on the gp41626-663 sequence (Dwyer et al. 2007). T1144 and T2635 showed strong activity against both native and highly T20 resistant HIV-1 strains. They form stable α-helices in solution with a helical content of 97% and 75% for T1144 and T2635, respectively. In addition, they both form a very stable 6-HB with NHR under physiological conditions. Analytical ultracentrifugation showed that these highly helical peptides form trimers in solution, which may make them more resistant to proteolysis and increase their *in vivo* stability. Ultra stable C-peptides from the same CHR sequence were also obtained, which showed ~100% helical content in solution and formed ultra stable 6- HBs with N-peptide with *Tm* values >100 °C, even in 8 M urea solutions. However, these peptides showed very weak antiretroviral activity. This indicated that it required a suitable degree of stability and α-helical content for the C-peptide to efficiently inhibit 6-HB formation to stop the HIV-1-cell fusion process.

Sifuvirtide (SFT, **12**) was developed by FusoGen and was based on the C34-related sequence gp41627-662 (He et al. 2008; Liu et al. 2011). It was derived from the HIV-1 subtype E sequence and was engineered to mutate the exposed residues to salt bridges to increase the helical content and solubility. Like C34, Sifuvirtide is featureless under physiological conditions, while it forms a nearly full α-helical 6-HB with N36, with a *Tm* of 72 °C, 10 °C higher than that of N36/C34. As expected from its sequence origin, Sifuvirtide does not interact with the lipid membrane. Sifuvirtide showed low nanomolar inhibitory activity against HIV-1 ENV mediated cell-cell fusion and HIV-1 infection, including T20-resistant HIV-1 isolates. It showed an *in vivo* half-life of 20 h in a single dose administration in 12 healthy volunteers, which is much more stable than T20 and suitable for a once daily administration. It has finished phase IIb clinical trials in China and has shown promising antiretroviral profiles against both T20 resistant and T20 sensitive HIV-1 strains (Wang et al. 2009). The same strategy was applied to CP32 and resulted in CP32M with an improved anti-HIV profile (He et al. 2008).

SC35EK (**10**), also based on C34, was developed by Fujii's group. Most of the *b*, *c*, *f*, and *g* residues were substituted with glutamic acid and lysine residues in order to form EE-KK double salt bridges to fortify the α-helical structure (Otaka et al. 2002). SC35EK showed a little bit more potency than C34 in a multinuclear activation of galactosidase indicator (MAGI) assay (IC50 from 0.68 to 0.39 nM), while the salt bridge greatly enhanced its solubility and made it a suitable drug candidate. Its structure is still largely random in solution, while the *Tm* of its 6-HB formed with N36 increased from 57 °C to 77 °C, which is 20 °C higher than C34. SC35EK was further shortened to SC29EK (**11**), with similar potency (Naito et al. 2009). The same strategy was applied to T20, the resulting T20EK (**16**) showed eight times more potency than T20 and can efficiently inhibit T20 resistant HIV-1 strains (Oishi et al. 2008).

In summary, the C-peptide fusion inhibitor could be engineered to improve the anti-HIV profile. Exposed residues in the 6-HB were substituted to build salt-bridges to significantly stabilize the C-peptide-NHR complex. This type of substitution can improve the solubility of the peptide fusion inhibitor to improve its druggability. The substitution also improved the pharmacokinetic profile, resulting in a longer *in vivo* half life. The helicity of isolated Cpeptides were greatly increased by replacing the *a* and *d* residues in the hydrophilic layer, resulting in thermally stable C-peptide fusion inhibitors with high α-helical content; they formed an extremely stable complex with NHR. Some of these structured C-peptides showed high anti-HIV potency, especially against highly drug-resistant HIV-1 isolates; while too thermally stable C-peptides of this type caused abolishment of their inhibitory activities.


Fig. 5. C-peptide fusion inhibitors

#### **3.3 Artificially designed peptides**

Artificial design was employed to design unknown peptide sequences with few homologies to natural peptide sequences (Qi et al. 2008; Shi et al. 2008). Based on the crystal structures of the HIV-1 gp41 fusion core, C-peptide uses it hydrophobic *a* and *d* residues to interact with the NHR. An EEYTKKI heptad unit (HR) was designed, with the heptad repeat 'bcdefga', as the building block. The *d* and *a* positions in the HR were hydrophobic Tyr and Ile residues, respectively, which were expected to form a hydrophobic face to interact with the hydrophobic NHR grooves. The residues at the *b* and *c* positions in the HR were negatively charged Glu, which were expected to form an intrahelical salt bridge with positively charged Lys at the *f* and *g* positions to stabilize the helical structure; these highly polar residues also form a highly hydrophilic face that increases the solubility of the peptides.

solubility and made it a suitable drug candidate. Its structure is still largely random in solution, while the *Tm* of its 6-HB formed with N36 increased from 57 °C to 77 °C, which is 20 °C higher than C34. SC35EK was further shortened to SC29EK (**11**), with similar potency (Naito et al. 2009). The same strategy was applied to T20, the resulting T20EK (**16**) showed eight times more potency than T20 and can efficiently inhibit T20 resistant HIV-1 strains

In summary, the C-peptide fusion inhibitor could be engineered to improve the anti-HIV profile. Exposed residues in the 6-HB were substituted to build salt-bridges to significantly stabilize the C-peptide-NHR complex. This type of substitution can improve the solubility of the peptide fusion inhibitor to improve its druggability. The substitution also improved the pharmacokinetic profile, resulting in a longer *in vivo* half life. The helicity of isolated Cpeptides were greatly increased by replacing the *a* and *d* residues in the hydrophilic layer, resulting in thermally stable C-peptide fusion inhibitors with high α-helical content; they formed an extremely stable complex with NHR. Some of these structured C-peptides showed high anti-HIV potency, especially against highly drug-resistant HIV-1 isolates; while too thermally stable C-peptides of this type caused abolishment of their inhibitory activities.

Artificial design was employed to design unknown peptide sequences with few homologies to natural peptide sequences (Qi et al. 2008; Shi et al. 2008). Based on the crystal structures of the HIV-1 gp41 fusion core, C-peptide uses it hydrophobic *a* and *d* residues to interact with the NHR. An EEYTKKI heptad unit (HR) was designed, with the heptad repeat 'bcdefga', as the building block. The *d* and *a* positions in the HR were hydrophobic Tyr and Ile residues, respectively, which were expected to form a hydrophobic face to interact with the hydrophobic NHR grooves. The residues at the *b* and *c* positions in the HR were negatively charged Glu, which were expected to form an intrahelical salt bridge with positively charged Lys at the *f* and *g* positions to stabilize the helical structure; these highly polar residues also form a highly hydrophilic face that increases the solubility of the peptides.

(Oishi et al. 2008).

Fig. 5. C-peptide fusion inhibitors

**3.3 Artificially designed peptides** 

A 35-mer 5HR (**17**) (Fig. 6), which contains five copies of the HR, based on the length of most highly potent HIV-1 fusion inhibitors, was used as a template to build peptides to disrupt the HIV-1 gp41 NHR-CHR interaction. The interaction between 5HR and N46 (gp41536-581) was modeled by using a PyMOL program based on the crystal structure of the N36/C34 6-HB, and compared with that of C34. The binding between the residues of 5HR and N46 was less complementary than that of the residues between C34 and N46. 5HR showed weak anti-HIV-1 activity (IC50 = 156 ± 8 μg/mL), as measured by a dye transfer HIV-1-mediated cell-cell fusion assay. The WWI motif and LBD were used to replace the HR unit at the N- or C-terminus of 5HR, respectively, or both, based on the SAR of the Cpeptide fusion inhibitors, resulting in PBD-4HR (**18**), 4HR-LBD (**19**), and PBD-3HR-LBD (**20**). Inserting a LBD or WWI motif in the 5HR sequences resulted in 2-fold and 6-fold increased potency, with an IC50 value of 74 ± 4 and 26 ± 0.4 μg/mL for 4HR-LBD and PBD-4HR, respectively. The increasing potency was synergistic and PBD-3HR-LBD had an IC50 value of 4.8 ± 0.3 μg/mL, a striking 33-fold increase over 5HR. As expected from the design, peptides containing PBD, e.g. PBD-4HR and PBD-3HR-LBD, could form a stable 6-HB with the N-peptide N46 and effectively blocked gp41 core formation, as measured by CD spectroscopy and N-PAGE; peptides containing the LBD, including 4HRLBD and PBD-3HR-LBD, were bound tightly to lipid vehicles, with an association constant of 6.80 × 104 and 1.27 × 105 M-1, respectively, as determined by isothermal titration calorimetry (ITC). These results suggest that the HR sequence can be efficiently docked into the NHR groove and act as a structural domain; and the interaction can be greatly increased by including the WWI motif and LBD in the sequence. Thus, 4HR-LBD, PBD-4HR, and PBD-3HR-LBD are artificial fusion inhibitors that mimic T20, C34, and T1249 – the three typical highly potent HIV-1 fusion inhibitors target different sites of gp41 NHR, respectively.

The anti-HIV-1 activities of 4HR-LBD and PBD-4HR are lower than those of T20 and C34, which may be due to less sequence complementarity between the artificially designed HR and HIV-1 gp41 NHR. The resulting less tight binding suggests that a specific interaction should be uncovered and be addressed for the design of highly potent fusion inhibitors targeting specific viruses.

Fig. 6. Artificially designed peptide fusion inhibitors

In summary, C-peptide fusion inhibitors interact with gp41 NHR to prevent fusogenic 6-HB formation, and thus terminal HIV-1 ENV mediated virus-cell membrane fusion. A WWI motif in C-peptide that interacts with the NHR deep pocket is critical to the C-peptide-NHR interaction, and an extended interaction between C-peptide and the rest of the groove in the NHR trimer provides additional energy to stabilize the 6-HB. Artificial peptide design, based on the knowledge learned from SAR studies of the C-peptides, provides an alternative for peptide fusion inhibitor design; it also provides a stringent test for the knowledge gained and sets a new starting point for fully understanding the fundamentals of virus-cell membrane fusion in order to guide future fusion inhibitor design against HIV and other viruses with class I fusion proteins.

### **4. Peptidomimetics as probes and inhibitors to study the gp41 NHR-CHR interaction**

Several SAR studies of highly potent peptide fusion inhibitors have provided an efficient way to disrupt the HIV-1 gp41 NHR-CHR interaction for anti-HIV therapy; they have also deepened our understanding of the gp41 NHR-CHR interaction. Peptide drugs have their intrinsic weaknesses, however, such as high-cost, and unsuitability for oral administration due to *in vivo* proteolysis. Peptidomimetics that use unnatural building blocks may overcome the *in vivo* instability of peptide drugs, leading to orally bioavailable drugs. Peptidomimetics are more like small molecules than peptide drugs, so highly potent peptidomimetic fusion inhibitor studies can be useful for guiding small molecule fusion inhibitor design. Peptidomimetic fusion inhibitors that target gp41 NHR, including Dpeptides, foldamers, and covalently linked restrained α-helical peptides (sequences or structures see Fig. 7), are discussed in this section.

#### **4.1 D-peptides**

As enantiomers of natural L-peptides, D-peptides are not degraded by proteases and have the potential for oral bioavailability. D-Peptides that target a specific protein or peptide target can be discovered by mirror-image phage display (Eckert et al. 1999). The target is synthesized chemically with D-amino acids, resulting in a product that is the mirror image of the natural L-amino acid form, which is used to screen phage that expresses a peptide library of phage coat proteins, to select phage clones with L-peptide sequences that specifically bind to the D-target. The mirror images of the phage-expressed L-peptide sequences are chemically synthesized with D-amino acids. By symmetry, these D-peptides should bind to the natural L-amino acid target.

Cyclic D-peptide HIV-1 fusion inhibitors targeting IQN17 have been identified by mirrorimage phage display (Eckert et al. 1999). The phage-expressed peptide library contained ten random amino acid residues flanked by either a cysteine or a serine on both sides. Of the 12 identified IQN17-specific phage clones, nine were pocket specific binders, and eight contained the consensus sequence CXXXXXEWXWLC. The corresponding D-peptides were synthesized and were oxidized to form disulfide bonds. Lysines were added to improve the solubility. An intramolecular disulfide bond was critical for pocket binding and viral inhibition by these D-peptides, since cysteines were selected from an initial phage library containing either Cys or Ser at these positions. Replacing the Cys with Ala in the most potent derivative D10-p5-2K (**22**, IC50 of 3.6 μM) caused complete loss of inhibitory activity in a gp41 mediated cell/cell fusion assay.

A IQN17/D10-p1 (**21**) co-crystal was obtained and resolved to 1.5 Å resolution by x-ray crystallography. Structural superposition showed that the overall architecture of the gp41 NHR deep pocket in the IQN17/D10-p1 complex is almost identical to that in the wild-type N36/C34 structure (Chan et al. 1997), with a Cα rmsd of 0.65 Å. D10-p1 forms a circular structure and binds only to the gp41 region of IQN17. Ala-2 to Ala-5 and Ala-11 to Ala-16 form short left-handed α-helices, and the middle region is unstructured. The overall

of virus-cell membrane fusion in order to guide future fusion inhibitor design against HIV

Several SAR studies of highly potent peptide fusion inhibitors have provided an efficient way to disrupt the HIV-1 gp41 NHR-CHR interaction for anti-HIV therapy; they have also deepened our understanding of the gp41 NHR-CHR interaction. Peptide drugs have their intrinsic weaknesses, however, such as high-cost, and unsuitability for oral administration due to *in vivo* proteolysis. Peptidomimetics that use unnatural building blocks may overcome the *in vivo* instability of peptide drugs, leading to orally bioavailable drugs. Peptidomimetics are more like small molecules than peptide drugs, so highly potent peptidomimetic fusion inhibitor studies can be useful for guiding small molecule fusion inhibitor design. Peptidomimetic fusion inhibitors that target gp41 NHR, including Dpeptides, foldamers, and covalently linked restrained α-helical peptides (sequences or

As enantiomers of natural L-peptides, D-peptides are not degraded by proteases and have the potential for oral bioavailability. D-Peptides that target a specific protein or peptide target can be discovered by mirror-image phage display (Eckert et al. 1999). The target is synthesized chemically with D-amino acids, resulting in a product that is the mirror image of the natural L-amino acid form, which is used to screen phage that expresses a peptide library of phage coat proteins, to select phage clones with L-peptide sequences that specifically bind to the D-target. The mirror images of the phage-expressed L-peptide sequences are chemically synthesized with D-amino acids. By symmetry, these D-peptides

Cyclic D-peptide HIV-1 fusion inhibitors targeting IQN17 have been identified by mirrorimage phage display (Eckert et al. 1999). The phage-expressed peptide library contained ten random amino acid residues flanked by either a cysteine or a serine on both sides. Of the 12 identified IQN17-specific phage clones, nine were pocket specific binders, and eight contained the consensus sequence CXXXXXEWXWLC. The corresponding D-peptides were synthesized and were oxidized to form disulfide bonds. Lysines were added to improve the solubility. An intramolecular disulfide bond was critical for pocket binding and viral inhibition by these D-peptides, since cysteines were selected from an initial phage library containing either Cys or Ser at these positions. Replacing the Cys with Ala in the most potent derivative D10-p5-2K (**22**, IC50 of 3.6 μM) caused complete loss of inhibitory activity

A IQN17/D10-p1 (**21**) co-crystal was obtained and resolved to 1.5 Å resolution by x-ray crystallography. Structural superposition showed that the overall architecture of the gp41 NHR deep pocket in the IQN17/D10-p1 complex is almost identical to that in the wild-type N36/C34 structure (Chan et al. 1997), with a Cα rmsd of 0.65 Å. D10-p1 forms a circular structure and binds only to the gp41 region of IQN17. Ala-2 to Ala-5 and Ala-11 to Ala-16 form short left-handed α-helices, and the middle region is unstructured. The overall

**4. Peptidomimetics as probes and inhibitors to study the gp41 NHR-CHR** 

and other viruses with class I fusion proteins.

structures see Fig. 7), are discussed in this section.

should bind to the natural L-amino acid target.

in a gp41 mediated cell/cell fusion assay.

**interaction** 

**4.1 D-peptides** 

positions of the D10-p1 and C34 helices closely overlap, but most of the side chains are significantly different, corresponding to the opposite handedness of the inhibitors. Of the 16 residues in D10-p1, only six interact directly with the gp41 pocket of IQN17, including Trp-10, Trp-12, and Leu-13 in the conserved EWXWL sequence, and Gly-1, Ala-2, and Ala-16 in the invariant original flanking phage sequence. The side chains of Trp-10, Trp-12, Leu-13, and Ala-16 are deeply buried in the hydrophobic pocket of IQN17. A hydrogen bond is formed between a pocket residue Gln-577 and Trp-12 in D10-p1. The packing difference between the Trp-12 and Leu-13 side chains in D10-p1 and Trp-631 and Ile-635 in C34 results in slight changes in the shape of the pocket. Overall, however, the hydrophobic pocket maintains its integrity between the N36/C34 and IQN17/D10-p1 structures. NHR chemical shift differences showed that, for all of the identified D-peptides, Trp-10, Trp-12, and Leu-13 are buried in the IQN17 pocket, validating the pocket as a target for drug development.

In follow-up work, the consensus residues in the sequence (CX5EWXWLC) reported above were fixed so that a constrained library was constructed in which the other six positions were randomized (Welch et al. 2007). The mirror-image phase display using IQN17 as the target identified, incidentally, the potent 8-mer D-peptide 2K-PIE1 (**23**) in the 10-mer template phage library. The x-ray crystal structure showed that 2K-PIE1 interacts in a similar manner as D10-p1 to IQN17, and 2K-PIE1 forms a more compact structure with IQN17. So, a comprehensive 1.5 × 108 member 8-mer phage library of the form CX4WXWLC (3.4 × 107 possible sequences) was generated, and was screened using IZN17 as the target (Eckert & Kim 2001). The resulting PIE7 (**24**) was the most potent inhibitor (IC50 = 620 nM) and is 15-fold more potent than the best first-generation D-peptide (D10-p5). Comparison of the crystal structures of 2K-PIE1 and PIE7 complexed with IQN17 reveals several interesting differences. First, an intramolecular polar contact between the hydroxyl of D-Ser7 and the carbonyl of D-Gly3 in 2K-PIE1 is lost in PIE7 but is replaced with a new interaction between the side chain carboxylate of D-Asp6 and the amide of D-Gly3. Second, new hydrophobic interactions are created in PIE7 between the ring carbons of D-Tyr7 and the pocket residue Trp-571. Third, the carbonyl of D-Lys2 of PIE7, although somewhat flexible in orientation, forms a direct hydrogen bond with the ε nitrogen of Trp-571 in some of the structures. Fourth, in some of the structures the hydroxyl of D-Tyr7 in PIE7 forms a new watermediated hydrogen bond with the pocket residue Gln-575, and this interaction cannot be formed in the 2K-PIE1 structure. Dimerized or trimerized PIE7 was constructed via PEG cross-linkers. The resulting (PIE7)2 and (PIE7)3 have IC50 values of 1.9 nM and 250 pM against HXB2, respectively. In contrast, PIE7 inhibits both JRFL, a primary R5-tropic strain, (IC50 = 24 μM) and BaL (IC50 = 2.2 μM) entry, although ~40- and 4-fold less potently than HXB2 entry, respectively; the PIE7 trimer is a moderately potent inhibitor of this strain (IC50 = 220 nM) and an extremely potent inhibitor against BaL (IC50 = 650 pM).

Structure-guided phage display was used to optimize the flanking residues for further improvement of PIE7 (Welch et al. 2010). The crystal structure shows significant contacts between the presumed inert flanking residues (Gly-Ala on the N-terminus and Ala-Ala on the C-terminus) and the NHR deep pocket. A new phage library was designed using XXCDYPEWQWLCXX as the template. PIE12 (**25**) was identified as the most potent (40-fold more potent than PIE7 against the JRFL strain). The x-ray crystal structure showed similarity between PIE12/IZN17 and PIE7/IZN17 structures with a RMSD of 0.6 to 1.2 Å on all C<sup>α</sup> atoms. In PIE12/IZN17, new N-terminal flank residues (His1 and Pro2) form favorable ring stacking interactions with the pocket (IQN17-Trp571), the substitution of Leu for Ala in the C-terminal flank sequence causes it to be buried an additional 50 Å into the hydrophobic surface area of the pocket, and the new interactions with the flanking sequence do not perturb the pocket-binding structure of the core PIE7 residues. These differences may account for the improved activity of PIE12 over PIE7. CD thermal denaturation showed that the PIE12-trimer forms the more stable complex with IZN17 with a *Tm* of 81 °C in 2 M Guanidine chloride (Gua.HCl), 8 °C higher than that of the PIE7-trimer complex. The anti-HIV-1 breadths of the PIE7-trimer, PIE12-trimer, and PIE12 were tested by a pseudovirion assay against a panel of 23 pseudotyped viruses representing clades A to D, several CRFs, and enfuvirtide-resistant strains. Both PIE7 and PIE12-trimers potently inhibited all strains tested, though PIE12-trimer was generally a superior inhibitor.

Fig. 7. Peptidomimetics used to disrupt the HIV-1 gp41 NHR-CHR interaction. (A) Dpeptides; (B) β-foldamers; (C) α/β-foldamers; (D)/(E) linked peptides; (F) stapled peptides.

Viral passage studies were conducted to select for resistant strains. A strain bearing E560K/V570I mutations, which conferred a 400-fold resistance to PIE7-dimer, was selected with 20 weeks of propagation. These mutations dramatically weaken the binding of Dpeptides to the gp41 pocket but not the C-peptide inhibitor C37. Despite this loss of affinity, the escape mutations had a minimal effect on the potencies of PIE12-dimer and PIE12 trimer. PIE12-dimer and PIE12-trimer resistant virus were identified after 40 and 65 weeks of propagation, respectively, using a much slower escalation strategy; only a Q577R single substitution was identified. Interestingly, this substitution is present in nearly all group O isolates but is rare among group M isolates. Examination of the PIE12 crystal structure shows that Q577 makes hydrogen bonds with Glu7 and Trp10 in PIE12, which may explain the disruptive effects of this mutation.
