In my honest opinion it is risky to solely trust on a TLS Connection to achieve full transport layer security. Imagine an Attacker gathers access to the Servers Domain / DNS-Entries, Server Hardware (the content signing Private Keys SHOULD NEVER BE placed directly on the Server to lower a complete Security Breach) or successfully executes a similar Attack. Exchanged Document SHOULD BE signed by the author and verified by the requesting Party. The Signature (using the servers own Private Key) SHOULD BE added to the Plaintext, otherwise (adding it outside the encryption) it may be manipulated when the connection is compromised.

Allow me to suggest the following Procedural Improvements:

Server
1. Content is signed with the Authors Private Key and available on the server, ready to be further processed
2. A random one-time encryption key is ... 
... generated ...
... signed using the servers Private Key ...
... and encrypted with the User Agents Public Key
2. Content and Signature are encrypted using the previously generated one-time encryption key resulting as the Ciphertext
3. Signed & encrypted one-time encryption Key and Ciphertext are sent to the User Agent via TLS

User Agent
1. Receives Signed & encrypted one-time encryption Key and Ciphertext
2. User Agent decrypts one-time encryption Key and verifies Signature (Servers Public Key)
3. Try to decrypt the Ciphertext with the previously decrypted key, when the signature verification (Authors Public Key) was successful
4. Verify signature found in the Ciphertext - Continue ONLY, when verification was successful
5. Display Document